Ved Stranden 18
DK-1061 Copenhagen K
TemaNord 2016:503
Nordic Council of Ministers
Nordic Economic Policy Review
TemaNord 2016:503
ISBN 978-92-893-4450-0 (PRINT)
ISBN 978-92-893-4452-4 (PDF)
ISBN 978-92-893-4451-7 (EPUB)
ISSN 0908-6692
NUMBER 2 / 2015
The review appears once a year. It is published
electronically on the website of the Nordic Council of
Ministers: On that website, you can
also order paper copies of the Review (enter the name
of the Review in the search field, and you will find all the
information you need).
The Nordic Economic Policy Review is published by the
Nordic Council of Ministers. This year’s issue is part of the
Danish presidency programme for the Nordic Council of
Ministers in 2015. The review addresses policy issues in a
way that is useful for in-formed non-specialists as well as
for professional economists. All articles are commissioned
from leading professional economists and are subject to
peer review prior to publication.
Torben M. Andersen and Jesper Roine
The Nordic welfare model in an open European labor market
Bernt Bratsberg and Knut Roed
Future Pathways for Labour Market Policy: Including the Excluded
Michael Svarer and Michael Rosholm
Economics of Innovation Policy
Tuomas Takalo and Otto Toivanen
Taxing mobile capital and profits: The Nordic Welfare States
Guttorm Schjelderup
Nordic family policy and maternal employment
Julian V. Johnsen and Katrine V. Løken
Education and equality of opportunity: What have we learned from educational reforms?
Helena Holmlund
Retirement and Health in the Nordic Welfare State
Nabanita Datta Gupta and Bent Jesper Christensen
The future of Welfare services: How worried should we be about Wagner, Baumol and Ageing?
Andreas Bergh
Ethnic fractionalization and the demand for redistribution – Potential implications for the Nordic model
Johanna Mollerstrom
The social upper classes under Social Democracy
Kalle Moene
NUMBER 2 / 2015
TN2016503 omslag.indd 1
09-02-2016 07:49:39
TN2016503 omslag.indd 2
09-02-2016 07:49:39
Managing Editor
Professor Torben M. Andersen, Department of Economics,
University of Aarhus, Denmark.
Special Editors for this volume
Professor Torben M. Andersen, Department of Economics, Aarhus University and
Professor Jesper Roine, Stockholm Institute of Transition Economics, Stockholm
School of Economics.
Papers published in this volume were presented at the conference “Whither the
Nordic Welfare Model” hosted by the Ministry of Finance, Finland, October 2015.
This project was financed as part of the Danish presidency flagship program
“The Nordic ‘supermodel’?”
Nordic Economic Policy Review
Whither the Nordic Welfare Model?
Torben M. Andersen, Jesper Roine, Bernt Bratsberg, Knut Roed,
Michael Svarer, Michael Rosholm, Tuomas Takalo,
Otto Toivanen, Guttorm Schjelderup, Julian V. Johnsen,
Katrine V. Løken, Helena Holmlund, Nabanita Datta Gupta,
Bent Jesper Christensen, Andreas Bergh, Johanna Mollerstrom
and Kalle Moene
TemaNord 2016:503
Nordic Economic Policy Review
Whither the Nordic Welfare Model?
Torben M. Andersen, Jesper Roine, Bernt Bratsberg, Knut Roed, Michael Svarer,
Michael Rosholm, Tuomas Takalo, Otto Toivanen, Guttorm Schjelderup, Julian V. Johnsen, Katrine V. Løken, Helena Holmlund, Nabanita Datta Gupta, Bent Jesper Christensen,
Andreas Bergh, Johanna Mollerstrom and Kalle Moene
ISBN 978-92-893-4450-0 (PRINT)
ISBN 978-92-893-4452-4 (PDF)
ISBN 978-92-893-4451-7 (EPUB)
TemaNord 2016:503
ISSN 0908-6692
© Nordic Council of Ministers 2016
Layout: Hanne Lebech
Cover photo: PUB, UNIT/NCM
Print: Rosendahls-Schultz Grafisk
Printed in Denmark
This publication has been published with financial support by the Nordic Council of Ministers.
However, the contents of this publication do not necessarily reflect the views, policies or recommendations of the Nordic Council of Ministers.
Nordic co-operation
Nordic co-operation is one of the world’s most extensive forms of regional collaboration, involving Denmark, Finland, Iceland, Norway, Sweden, and the Faroe Islands, Greenland, and
Nordic co-operation has firm traditions in politics, the economy, and culture. It plays an important role in European and international collaboration, and aims at creating a strong Nordic community in a strong Europe.
Nordic co-operation seeks to safeguard Nordic and regional interests and principles in the
global community. Common Nordic values help the region solidify its position as one of the
world’s most innovative and competitive.
Nordic Council of Ministers
Ved Stranden 18
DK-1061 Copenhagen K
Phone (+45) 3396 0200
1. Introduction .......................................................................................................................................... 7
Torben M. Andersen and Jesper Roine
2. The Nordic welfare model in an open European labor market.................................... 19
Bernt Bratsberg and Knut Røed
3. Future Pathways for Labour Market Policy: Including the Excluded ........................ 43
Michael Rosholm and Michael Svarer
4. Economics of Innovation Policy ................................................................................................ 65
Tuomas Takalo and Otto Toivanen
5. Taxing mobile capital and profits: The Nordic Welfare States ..................................... 91
Guttorm Schjelderup
6. Nordic family policy and maternal employment ..............................................................115
Julian V. Johnsen and Katrine V. Løken
7. Education and equality of opportunity: What have we learned from
educational reforms? ...................................................................................................................133
Helena Holmlund
8. Retirement and Health in the Nordic Welfare State .......................................................171
Nabanita Datta Gupta and Bent Jesper Christensen
9. The future of Welfare services: How worried should we be about
Wagner, Baumol and Ageing? ...................................................................................................197
Andreas Bergh
10. Ethnic fractionalization and the demand for redistribution – Potential
implications for the Nordic model..........................................................................................219
Johanna Mollerstrom
11. The Social Upper Class under Social Democracy .............................................................245
Kalle Moene
1. Introduction
Whither the Nordic Welfare Model?
Torben M. Andersen, Department of Economics, Aarhus University and
Jesper Roine, Stockholm Institute of Transition Economics, Stockholm
School of Economics
The Nordic Welfare Model frequently attracts international attention
and is by many seen as a social model to be inspired by or even to be
copied. In recent years, the “Nordic Way” has been a topic for discussion
at the World Economic Forum and it has even appeared on the cover of
The Economist under the heading “The next supermodel”. 1 Somewhat
paradoxically, however, the debate in the Nordic countries often features skepticism on the future of the model. Does this reflect a timely
concern voiced by those best placed to see what is going on, or are the
doubts on the contrary a result of model-hypochondria?
A first caveat – or perhaps part of the answer – has to do with the
meaning of the very concept “Nordic welfare model”. Is it really meaningful to talk about the existence of such a thing? Is the term welldefined given how large the differences are between the Nordic countries and given the major policy shifts in the past decades? The answer
clearly depends on what one includes in the meaning of the model concept. If one thinks that it is associated with a certain set of specific policies or certain levels of tax rates or benefits, then clearly the concept is
questionable. These things have indeed changed over time and are also
different across the Nordic countries. For example, unemployment in1 The Economist, 2 February 2013,
Nordic Economic Policy Review
surance is voluntary in Denmark, Finland and Sweden, but mandatory
in Norway. Pension systems are fundamentally different spanning from
a large role to funded occupational pensions in Denmark to a notionally
defined contribution scheme in Sweden. While tax burdens are high in
the Nordic countries (except for Iceland), the tax structure differs with
Denmark having the larger share of tax revenue accruing from direct
income taxes and value added tax, while Sweden raises much more tax
revenue from social contributions.
However, in a longer perspective such a focus on certain policies
would seem odd. If one were to look at reforms and levels of tax rates
and benefits over the whole history over which the concept of a “Nordic
model” has been identified and discussed, these have varied a lot. Indeed, continuous change has been a distinguishing feature of the model,
and the changes over the past decades are not in any obvious way larger
in magnitude than those in the preceding decades.
If one instead identifies the Nordic model as being concerned with a
number of broader principles and goals in terms of outcomes, the concept becomes more well-defined. What matters then are the overall objectives and the overall design of the package. Here the complementarity between policies and institutions is crucial. It is not the ingredients,
but the overall packaging, which makes a difference in terms of final
outcomes. With this kind of perspective it also becomes clear that the
naïve “copy and paste” perspective often taken in comparative policy
discussions focusing on a single or few policy instruments is misleading
since it overlooks the complementarities between the different policy
elements. From this point of view the Nordic model should not be defined or assessed in terms of specific policy instruments, what matters
is the overarching objectives. Goals – such as equal opportunities in life
regardless of family background, the eradication of poverty, gender
equality, the lowering of income inequality, etc. – as well as some principles – such as individually based universal rights to things such as
health care and education, well-organized labour markets, etc. – have
Nordic Economic Policy Review
remained largely stable, while the specific policies and instruments to
reach them differ across time and countries. 2
In terms of economic performance the Nordic countries, like most
others, have seen good and bad periods, but the Welfare Model – defined in terms of its goals rather than a fixed set of policies – has proved
resilient. The Nordic countries stand out today as they did decades ago
as being countries with comparatively high living standard and a relatively equal distribution of income. In the jargon of economics, the Nordics seem to have found a way to balance concerns for efficiency and
equity. The public sector is large, hence the tax burden is high, and yet
the Nordics rank in the top for various indicators of economic performance and competitiveness. 3 Figure 1a–c depicts a few select indicators
often used to compare countries along the efficiency and equity dimension. The Nordic countries are high income countries, and have high
employment rates, especially for women. Income inequality and poverty is low in international comparison.
2 Of course these things are (and have been) debated (see e.g. Andersen, Roine and Sundén (2014), Chapter 2, for an overview of different views of the Nordic welfare state). The main point here is to emphasize
that the model should be understood in terms of broad goals rather than in terms of specific policies.
3 In the most recent version of the Global Competitiveness Index 2014–2015 ranking Finland placed 4, Sweden 10, Norway 11, Denmark 13 and Iceland 30, out of 144 countries.
Nordic Economic Policy Review
Figure 1a: Performance indicators for Nordic countries – Per capita income
Per capita income
GDP per capita
GDP per capita,
mainland Norway
Note: Income per capita is measured in USD PPP corrected. Norway-mainland is GDP corrected for
the importance of off-shore oil and gas extraction, 2010.
Source: Data from
Figure 1b: Performance indicators for Nordic countries – Employment rates
Employment rates
Women - OECD
Men - OECD
Note: Employment rates for the age group 15–64, 2011.
Source: Data from
Nordic Economic Policy Review
Figure 1c: Performance indicators for Nordic countries – Inequality
Note: Gini-coefficient defined over equivalised disposable income 2010.
Source: data from
Figure 1d: Performance indicators for Nordic countries – Poverty
Poverty rate
Note: Poverty measured as the share of individuals with equivalised income below 50% of median income.
Source: Data from
Nordic Economic Policy Review
Taxes are, no doubt, high in the Nordic countries, and taxes distort economic incentives. However, the effects of taxes cannot be seen independently of what taxes are financing The two broad expenditure types
relate to the social safety net and provision of welfare services, cf. Figure 2. The social safety net plays an important distributional role but it
is also provides insurance. The latter may have a direct welfare effect
but also be conducive to flexibility and ensures that the costs of changes
at the level of society are not fully carried by specific individuals. Welfare services include education, health and care. They are provided universally and at contemporary standards and meeting the requirements
of most people. Welfare services are important from a distributional
point of view, and in terms of ensuring equal opportunity. Clearly, these
activities are also important for labour supply along both the quantitative and qualitative dimension. As examples, day care – which is also
associated with other values in relation to family policy and social integration – promote labour supply, especially for women. Education is
obviously associated with productivity but is also associated with e.g.
later retirement. The complicated interrelation between the effects of
taxes and welfare spending underlines the need to continuously recalibrate policies to find the right balance between concerns for efficiency and equity alongside various changes in society.
Nordic Economic Policy Review
Figure 2: Public sector activities: social expenditures and public consumption
Net social expenditure
Net social
% of GDP
Public consumption
% of GDP
OECD collective
OECD collective
Note: Net social expenditures correct gross social expenditures for taxes on transfers to make data
comparable between countries, where in some transfers are taxable income and in others they are
not taxable income, see Adema et al. (2011). Public consumption is split between traditional collective expenditures, and expenditures on activities which can be attributed to specific individuals
(welfare services). Data applies to 2011.
Discussions on the future of the welfare model often has “competitiveness” in the broad sense of the term as a common denominator. Can the
Nordic countries remain competitive and thereby sustain a high living
standards? Often this debate has the implicit premise that to remain
Nordic Economic Policy Review
competitive we cannot deviate too much from other countries, and
therefore the Nordic model is particularly vulnerable.
The concern for competitiveness is not new to the small and open
economics of the Nordics. It has always been an overriding concern. But
“being competitive” is not tantamount to “being alike” and implying that
all social models have to converge. This view has no support in, for instance, trade theory stressing the importance of differentiation and
comparative advantages. There is also an increasing understanding that
different social structures and institutions can be a source of comparative advantages. A recent literature levy a critique on traditional analyses for having a too one-sided focus on identifying the optimal institutional setting, see e.g. Nunn and Trefler (2013). There is no specific institutional setting which is optimal. The reason is that various institutional arrangements have pros and cons, which may be a source of
comparative advantage. Countries with flexible employment protection
legislation and generous unemployment insurance may have a comparative advantage in industries with substantial short-term variation in
demand and thus production, while countries with more strict employment protection legislation and less generous unemployment insurance
may have a comparative advantage in production of commodities with
less variability. As an example of this Cuñat and Melitz (2012) find in a
cross-country study empirical support that countries with more flexible
labour markets have a higher degree of specialization in sectors more
frequently exposed to sector-specific shocks. This may be interpreted in
the sense that the nature of shocks or needs for adjustment to some extent is endogenous, meaning that countries (or rather its companies in
the private sector) specialize in the activities for which their particular
institutional setting has a comparative advantage. This type of research
is still in its infancy, but it is highly suggestive of why different institutional settings (welfare regimes) survive. The important lesson – repeating basic insights from trade theory – is that competitiveness is a
question about comparative advantages.
Past performance is important, but the pertinent question is whether the Nordic Welfare Model is robust and resilient to various challenges including changing demographics, globalization, new technologies
and environmental changes?
Nordic Economic Policy Review
Society is undergoing various changes, some small, some larger,
some come gradually and others in clusters. In the debate they are often
labelled challenges, signalling that policy initiatives are required.
Whether they are a challenge or a threat to the Nordic model as such is
a different issue. To take an example, the issue of ageing is undoubtedly
a very important policy challenge. The age composition of populations
are changing, not least because longevity goes up (and is in itself associated with huge welfare gains). Clearly, the social contract has to be
adopted to such a change. Although this may be politically difficult, it is
not difficult in a technical sense, and a solution is clearly feasible without changing any fundamental properties of the model. One solution
may be to increase retirement ages alongside increases in longevity and
thereby ensure that the fraction of life spent in the labor market remains unchanged. It is not possible in any meaningful way to interpret
this as a change in the basic principle of the model.
Another much discussed area is that of automation and digitalization and the impact this will have on all aspects of the economy, in particular on the future of work. This is a vast debate with many dimensions but it is interesting to note that some aspects of the challenge and
some suggested solutions turn out to be much aligned with basic principles of the Nordic model. Predictions such as the need for continuous
education throughout life and the need for individuals to be able to hold
several jobs over a life-time, in fact, place demands on policy similar to
those of a small open economy in an increasingly globalized world. Ideas such as “protect individuals, not firms” and “make sure the workforce
has continuous possibilities to educate and re-educate themselves to
meet new challenges” are not less familiar to the Nordic model than to
other countries, rather the opposite. For sure, policy will have to change
to adapt to new realities, but again, it is not obvious that these challenges make the model obsolete. 4
Some challenges may be related to properties of the model. Immigration of unskilled or low-skilled individuals may be a particular probJust as an example, in a recent issue of Foreign Affairs Colin and Palier (2015) outline some challanges in
the “digital age” and arrive at the conclusion that aspects of the Nordic model are, in fact, better suited to
“fostering a more fluid and entrepreneurial economy” than many of the alternatives discussed.
Nordic Economic Policy Review
lem in societies with high minimum wages (no working poor), high
qualification requirements for jobs, high employment rates for both
genders etc.
In other cases solutions may also be model specific, as e.g. in relation to provision of welfare services like education, health and care,
which are essential elements of the welfare state. Such welfare services
may be subject to both Baumol’s cost disease (relative costs increases
over time since productivity increases are typically lower than for manufactured products) or Wagner effect (increasing demand for service
alongside improvements in material living standard). While the drivers
are universal, the solution is model specific, since the Nordic countries
have opted for a larger public role in the provision of services
Society is undergoing large changes – as it also has in the past – not
least those arising in the intersection between globalization and technological change which changes modes of production which on impact
creates both winners and losers in the labour market. The derived effects also include new forms of employment, less stable employment
relations etc. Left on its own this may be a source of increasing inequality. This raises questions for traditional distribution policies running via
taxes and the social safety net, but also for the possibilities to actively
counteract these changes via education, labour market and social policies. Maintaining a high employment level is both a value in itself related to social inclusion and equality, but the financial viability of the welfare model also depends on maintaining a high employment level.
These changes also have wider effects on the interface between markets, civil society and the family. Changed employment relations and demands in the labour market may affect the possibilities the individual has
in balancing work-life and family-life. Families may also undergo changes
(divorces) and there is an increasing trend in single-families.
Perhaps the most important aspect of the Nordic model should not
be found in economic details but rather in the political economy sphere.
It is sometimes denoted the “consensus” tradition which permeates industrial relations and politics. The political capital is large and this is
reflected in an ability to undertake reforms. Rather wide ranking reforms of pension and retirement schemes have been implemented
Nordic Economic Policy Review
smoothly in e.g. Denmark and Sweden, and they are among the few
countries who have addressed the ageing problem.
In discussing changes or challenges, it is also worth pointing out
that many of these are common to most countries or global in their nature. The need for changes and adjustments should therefore be seen in
the perspective of the changes needed in other countries. It is not clear
that the challenges are posing a larger problem for the Nordic countries.
To list just a few, the US is facing a problem of steeply increasing inequality and segregation. Southern European countries experience an
outflow of well-educated young and strong protests and retrenchment
of reforms to address the ageing problem.
Globalization and technological changes are associated with collective gains but an unequal distribution of gains and losses. Welfare arrangement may contribute to compensate the losers and (re) distribute
the gains, which in turn may be conducive to reforms. Clearly there is a
hen-and-egg issue in the interdependence between welfare arrangements and ability to reform – the welfare state may create an environment in which it is easier to undertake reforms, but the reforms are also
crucial for the development and viability of the model. How this relationship has been established and developed historically is in itself and
interesting question, but beyond this volume to consider.
Rather than looking backward at past performance it is important in
due time to consider changes in society and discuss how to address
them. This volume addresses some topical issues on the future of the
Nordic welfare model. It is, of course, by no means exhaustive, but instead covers a number of broad issues outlining what recent research
has to say on them. Each paper is relatively short, given the width of
each topic, and the titles of each contribution explains what the covered
topic is, so rather than trying to summaries the contributions we hope
that they all, together or individually will contribute to a better and
more informed discussion about the future challenges, reform needs
but also possibilities of the Nordic model.
Nordic Economic Policy Review
Adema, W., Fron, P., & Ladaique, M. (2011). Is the European Welfare State Really
More Expensive?: Indicators on Social Spending, 1980–2012; and a Manual to the
OECD Social Expenditure Database (SOCX), OECD Social, Employment and Migration Working Papers, No. 124, OECD Publishing.
Andersen, T., Roine, J., & Sundén, A. (2014). Hur får vi råd med välfärdsstaten? SNS
Konjunkturrådsrapport 2014, SNS förlag, Stockholm.
Colin, N. & Palier, B. (2015). The next safety net: Social Policy for a digital age.
Foreign Affairs 94(4): 29–33.
Cuñat, A. & Melitz, M. (2012). Volatility, Labor Market Flexibility, and the Pattern of
Comparative Advantage. Journal of the European Economic Association, 10(2):
Nunn, N. & Trefler. D. (2014). Domestic Institutions as a Source of Comparative Advantage, In G. Gopinath, E. Helpman & K. Rogoff (Ed.) Handbook of International
Economics,. Volume 4, chapter 5: 263–315.
Nordic Economic Policy Review
2. The Nordic welfare model in
an open European labor
Bernt Bratsberg, The Ragnar Frisch Centre of Economic Research,
[email protected] and Knut Røed, The Ragnar Frisch Centre
of Economic Research, [email protected]
Is it possible to sustain an ambitious and redistributive Nordic welfare
state in a Europe with open borders? Drawing on longitudinal administrative records spanning four decades, we first present discouraging historical
evidence showing that labor migrants from low-income source countries
tend to have unstable employment careers with marked overrepresentation in welfare programs. This pattern extends to post-accession labor migrants from Eastern Europe, who quickly experience high rates of unemployment. The article discusses possible avenues for making the welfare
state “migration robust”. We argue that there are alternatives to reclosing
borders and/or cutting down welfare state ambitions, and recommend policies based on strengthening of activity requirements in social insurance
programs, raising minimum job standards, and substitution of workoriented services for cash-based family allowances.
1 We are grateful to Gregory Clayes, participants at the 2015 NEPR conference in Helsinki, and the editors
for valuable comments. The article draws on research funded by the Norwegian Research Council (projects
“Work Life Challenges – workforce management and worker involvement solutions” and “European
Strains”) and is part of the research activities of the Centre of Equality, Social Organization, and Performance, University of Oslo. Data made available by Statistics Norway have been essential for this research.
The recent enlargements of the EU Single Market represent new opportunities for growth and prosperity. The liberalized migration regime that
followed allows workers to flow more easily toward their best potential
use (Kahanec et al., 2014; see also Clemens, 2011 and Kennan, 2012). Improved labor mobility removes bottlenecks in production processes and
facilitates dynamic cushioning of regional cyclical fluctuations. Left unhindered, the open-border policy is a powerful tool for raising and equalizing living standards across Europe, and thus for promoting economic
and social cohesion. However, at least in the short and intermediate
terms, the integrated labor market also presents some political and economic challenges. Large cross-country differences in labor productivity,
wages, and social insurance standards may trigger migration flows that
place pressure on present welfare state institutions (Sinn, 2002; Kvist,
2004). In particular, the fact that social insurance benefits in the Nordic
countries by far exceed typical wages in most accession countries may
distort migration flows and weaken labor migrants’ incentives to remain
in productive employment over the long haul. As we show below, prior
empirical evidence points to a considerable risk that labor immigrants
from low-income countries fail to establish lifecycle employment and
earnings careers on par with natives, but instead exhibit high rates of
premature labor market exit and welfare uptake. 2 Beyond their mere fiscal implications, such processes may well lead to economic marginalization of minority groups and, as discussed by Mollerstrom (this volume),
reinforce any decline in popular support for redistribution linked to
growing demographic heterogeneity as many natives will perceive it as
less likely that they will be on the receiving end. This brings to the fore
questions of how labor market institutions and social insurance systems
should adapt in order to reap the full benefits – and avoid the perils – of
an integrated European labor market.
2 A large literature studies welfare uptake among immigrants and natives across Europe; see Nannestad
(2004) for a review and Boeri (2010) and Barrett and Maitre (2013) for recent cross-country comparisons.
Nordic Economic Policy Review
Based on Norwegian administrative register data, this paper first
reexamines past experiences with labor immigration. Labor immigration
benefits the welfare state in the short run through its immediate expansion of labor supply. But, the impacts on the fiscal sustainability of the
welfare state also depend on the migrants’ long-term integration in the
labor market and their rate of return migration. Our brief review of prior
studies, paired with new evidence on labor market outcomes of recent
European labor migrants, indicates grounds for concern: Labor immigrants from countries with low economic development have more unstable employment patterns, and face a much higher probability of becoming
reliant on social insurance transfers, than natives. We move on to discuss
mechanisms that can explain these patterns, such as vulnerability to cyclical fluctuations; lack of language skills needed to adapt to new
jobs/occupations in response to structural change; high effective replacement ratios in the social insurance system; and employer incentives
to recruit low-skilled immigrant workers to jobs with low wages and
poor working conditions. Finally, we discuss some policy options. We argue that a reintroduction of migration barriers is not the way to move
forward. Instead we recommend policies aimed at making the Nordic
welfare model more “migration robust”: First, by establishing (or raising)
minimum standards/wages in the labor market in order to prevent social
dumping at the tax payers’ expense, and, second, to make the social insurance system more participation oriented – essentially by substituting
job offers and/or various forms of activation for pure cash transfers.
Experiences prior to the enlargements of the
European labor market
Between 1975 and 2004, work-based immigration to Norway from outside the Nordic region (the EEA area since 1994) was subject to strict
regulations. Hence, in order to examine the long-term labor market performance of labor immigrants from low-income countries, we have to
go back to the waves that arrived just prior to the 1975 legislation. Although this obviously raises questions about comparability with today’s
migrants and today’s labor market institutions, the exercise has the adNordic Economic Policy Review
vantage that it facilitates assessments of immigrant performance over
four decades.
Bratsberg et al. (2010; 2014) have examined the lifecycle employment and earnings patterns of these early cohorts of labor migrants to
Norway. Their analyses distinguish between immigrants from countries
with similar earnings levels and living standards to those of Norway (i.e.,
Western Europe) and immigrants from countries with considerably lower earnings and living standards (e.g., Pakistan and Turkey). A key finding
of these studies is that whereas labor immigrants from Western Europe
had employment and earnings patterns similar to those of natives, the
labor immigrants from low-income countries had a disproportional tendency to drop out of labor market after 10–15 years of employment.
The left-hand panel of Figure 1 summarizes and updates some of the
main findings of the prior studies by showing the annual employment
rates of male immigrants who entered Norway during the early 1970s,
and then remained in the country until 2013. Similarly, the right-hand
panel shows employment rates for female immigrants who entered during the late 1970s (with the delayed entry period explained by the fact
that very few women from low-income countries arrived during the
early 1970s; the majority of those from the late 1970s being spouses of
the male cohorts of the early 1970s). Here, we distinguish between
three different regions of origin that dominated labor-related migration
to Norway during this period: i) Pakistan and Turkey, ii) the Nordic
countries, and iii) the rest of Western Europe (denoted EEA in the figure). In addition, we show employment rates for a comparison group of
natives, stratified to have the exact same age composition as the three
immigrant groups put together. Since the age compositions of these
groups were roughly the same (with the typical immigrant being 24–25
years of age at the time of arrival), we can compare the dynamic employment patterns directly. It is clear that lifecycle employment was
much lower for labor immigrants from Pakistan and Turkey than for
immigrants from Western Europe and natives. Employment levels
tended to be high during the first years in Norway, but after around 10–
15 years of residency, they started to drop precipitously. Immigrants
from the Nordic countries and the rest of Europe, on the other hand,
had employment patterns very similar to natives.
Nordic Economic Policy Review
Figure 1: Annual employment rates 1972–2013, conditional on continued residency in Norway –
Immigrants from the 1970s and native comparison groups
Employment rate (%)
Note: Employment is defined as having annual labor or business earnings exceeding the base
amount of the social insurance system (currently NOK 90,068). The figure shows annual averages
for those aged 25–64. The “EEA” group consists of immigrants from the following countries (with
sample share in parenthesis); The United Kingdom (47), Germany (14), France (12), the Netherlands
(10), Spain (4), Switzerland (4), Italy (3), Austria (2), Belgium (2), Ireland (1), Portugal (1), Greece (1),
and Luxembourg (0).
Figure 2 illustrates how the low employment rates among immigrants
from Pakistan and Turkey translate into correspondingly high participation rates in disability insurance programs. These data are available from
1992 only, but we see that already at this point more than 20% of the Pakistani and Turkish immigrants who arrived 15–20 years earlier had become recipients of disability insurance benefits. By the end of our observation period, more than 60% of the labor migrants still in Norway, as
well as their spouses, had become disability insurance claimants, compared to around 20% of the immigrants from high-income countries.
Nordic Economic Policy Review
The reason why we focus on disability insurance benefits here, and
not, say, unemployment benefits or social assistance, is that the disability program had already become the major social insurance program for
the 1970s cohort of labor immigrants when the data series for social
insurance transfers start in 1992. Likewise, the unemployment insurance register data are first available in 1989 and we are unable to study
the dynamic transitions between employment, unemployment insurance, and disability program enrollment during the critical downturn of
the late 1980s. Unemployment insurance was presumably an important
income source for many of the immigrants during the slump, but since
these benefits are subject to time limits (currently a maximum of two
years) the welfare state had to find other ways to ensure lasting solutions for persons who persistently failed to find new employment. Existing empirical evidence (Rege et al., 2009; Bratsberg et al., 2013) shows
that there is a large “grey area” between unemployment and disability
insurance programs: Job loss raises the probability of becoming a disability insurance claimant considerably, and Bratsberg et al. (2013) argue that disability insurance is sometimes unemployment insurance in
disguise. For the cohorts under study, the underlying data indeed show
a strong correlation between unemployment and subsequent disability
program enrollment: For the men who first entered disability in 1993,
unemployment insurance rates in 1989 were twice those of men who
did not enter disability insurance.
Nordic Economic Policy Review
Figure 2: Disability program participation 1992–2013, conditional on continued residency in
Norway. Immigrants from the 1970s and native comparison groups
Disability program participation (%)
Note: Figure entries are restricted to those aged 25–64.
Given that we examine a wave of immigrants who arrived during the
1970s, it is now possible to assess their patterns of employment and
earnings over their whole potential working lives. On average, the male
labor immigrants from Pakistan and Turkey were employed 61.9% of
all years between the ages of 25 and 66. To put this number into perspective, we have computed the corresponding number for native men
of the same birth cohorts, who were employed 85.5% of their potential
working lives. Examining annual earnings instead, we find that the labor
immigrants from Pakistan and Turkey on average earned 177,791 NOK
per year (not conditional on employment, and inflated to 2012 currency), whereas the comparable group of native men earned 328,464 NOK.
Hence, the labor immigrants’ lifetime earnings were on average only
54% of those of native men of the same birth cohorts.
Nordic Economic Policy Review
Moving on to the spouses of these labor immigrants, we find that the
women were employed in only 21.9% of all years between the ages of
25 and 66, with average annual earnings as low as 43,737 NOK. By
comparison, native females of the same birth cohorts were employed
68.0% of their potential working-life years, with average earnings equal
to 176,772 NOK. Hence, the lifetime earnings of the cohort of immigrant
women were only 25% of those of native-born women.
Lower lifetime earnings than natives do not necessarily indicate
that immigrants represent a fiscal burden for the welfare state, however. In order to provide a broader assessment of the fiscal consequences
of migration, one has to include contributions and expenditures over
the complete lifecycle, taking into account that tax payers do not have to
pay the costs of child care and education before immigrants arrive and
that some immigrants will spend the last – and the most cost intensive
in terms of health care – years of their life in their country of origin.
Hence, the break-even point of direct taxes paid versus benefits received likely involves lower lifetime labor earnings for immigrants than
for natives (Preston, 2014).
Nevertheless, it is of paramount importance to understand why the
immigrants from Pakistan and Turkey performed so poorly in the Norwegian labor market over the long term. Unfortunately, we are not able
to provide complete evidence-based answers to this question. We know
that business cycle fluctuations played an important role, as many of the
immigrants lost their foothold in the labor market around the major
cyclical downturns in the early and late 1980s; see Bratsberg et al.
(2010). A large fraction was originally recruited to declining (and, to
some extent, dying) industries and they did not possess the human capital and language skills typically required by the new and growing industries. Thus, dependency on temporary social insurance became prevalent. Since social insurance benefits are more generous for persons with
children and dependent spouses, many of the immigrants experienced
that social insurance gave as high, and in some cases even higher, family
income than fulltime work (Bratsberg et al., 2010). This situation might
have undermined incentives to provide the effort required for regaining
a foothold in the labor market, thereby transforming temporary insurances into permanent ones.
Nordic Economic Policy Review
The presence of a relatively comprehensive social security net in
Norway, combined with large differences in living standards between
Norway and the source country, further weakened the incentives for
return migration among the labor migrants from low-income countries,
even in cases were new employment could not be found in Norway. This
illustrates an important asymmetry in labor-motivated migration patterns between countries with very different levels of development:
Whereas high labor demand during economic booms in the wealthy
country will attract workers from poor countries, there is no reason to
believe that a subsequent economic bust will set the migration flows in
reverse. At this point, there is a significant difference between labor migration flows across countries with similar and countries with very different living standards. The discrepancy also shows up in our data:
While as many as three in four of the 1970s immigrants from the Nordic
and other Western European countries had left Norway by 2013, this
was the case for only one in four of the Pakistani and Turkish immigrants – despite the latter group’s much poorer performance in the
Norwegian labor market.
For those who did stay in Norway, the long-term labor market performance of immigrants from the Nordic and other Western European countries is actually a completely different story: As Figures 1 and 2 showed,
their lifecycle employment patterns are hardly distinguishable from those
of natives, and for females, participation in disability insurance programs is
even considerably lower than for similarly aged native women.
Experiences since the expansions of the
European labor market
So, for the issue of how the new European flows of labor migrants will
affect the long-term fiscal sustainability of the Nordic welfare states, a
key question is whether the eastwards extensions of the European labor market will lead to migration flows and long-term employment
patterns that resemble the 1970s experiences with migration from
poor or from rich source countries. Geographical and cultural closeness suggest that past immigration from Europe is the most relevant
Nordic Economic Policy Review
reference. However, since the new European labor market covers
countries with large differences in economic development, and with
very different labor market and social insurance institutions, the answer to the question is not obvious.
Given that the first eastwards extension of the European labor market took place only 11 years ago (in 2004), it is obviously too early to
paint a complete picture of the lifecycle employment and earnings patterns of the new immigrants. What we can do, however, is to examine
economic outcomes over a 10-year period after entry. To do this, we
look at three groups of recent European immigrants to Norway: i) those
from the nearby Nordic countries, ii) those from other countries in
Western Europe (for simplicity denoted EEA), and iii) those from the
2004 accession countries in Eastern Europe (denoted EU8). Specifically,
we examine immigrants from these regions who arrived in Norway between 2005 and 2008 and were 17 to 46 years of age at the time of entry. Figure 3 first shows how long these migrants stayed in Norway.
While fewer than 40% of the Nordic immigrants remained in Norway
10 years after arrival, this was the case for 70–80% of the migrants
from EU8, and around 50–60% of the migrants from other countries in
Western Europe, depending on gender. In other words, the return migration patterns of the new immigrants from the accession countries in
Eastern Europe are closer to those we saw among 1970s immigrants
from low-income countries than from countries in Western Europe.
Nordic Economic Policy Review
Figure 3: Share of immigrants residing in Norway, by years since entry and region of origin
B. Women
Share in Norway (%)
A. Men
Years since immigration
Note: Immigrant populations consist of 2005–2008 arrivals age 17–46 at entry. The EU8 group consists of immigrants from the following countries (with sample share in parenthesis); Poland (75),
Lithuania (14), Slovakia (3), Latvia (3), Estonia (2), Hungary (1), Czech Republic (1), and Slovenia (0);
and the “EEA” group of Germany (50), the United Kingdom (14), the Netherlands (10), France (9),
Spain (4), Italy (4), Portugal (2), Austria (2), Switzerland (1), Belgium (1), Greece (1), Ireland (1),
Liechtenstein (0), Luxembourg (0), Andorra (0), and San Marino (0).
Figure 4 displays, separately by gender, annual employment rates for
each of the three immigrant groups. The employment figures are computed for persons aged 25 or more, conditional on continued residency
in Norway (at the end of each calendar year), and also conditional on
not being enrolled in education during the year. Again, we add a native
comparison group with the same age composition as the various immigrant groups combined. (There is some variation in age across the three
immigrant groups. To illustrate, among those in Norway at the end of
2010 the average age was 34.5 for the EU8 group, 35.7 for the EEA
group, and 31.6 for those from the Nordic countries. When we reweight
Nordic Economic Policy Review
the native data to reflect the age distribution of each immigrant group,
we uncover however only minor differences from the overall native
employment numbers displayed in Figure 4.) The figure shows that employment rates for Nordic immigrants again tend to be high and indistinguishable from those of natives. Male immigrants from Eastern Europe also had very high employment rates in 2007 and 2008, but their
employment rates apparently took a serious blow during the financial
crisis in 2009–2010. In 2013, however, their employment was almost
back to native levels. Male immigrants from other Western European
countries have consistently had employment rates somewhat below
native levels. For female immigrants from both Western and Eastern
Europe, we see patterns of relatively low employment rates the first
years after immigration, followed by a gradual convergence toward native levels. The underlying data reveal that many of the female nonNordic immigrants were admitted as family immigrants (44% of the
EU8 women and 39% of the EEA women, compared to only 1% and 7%
of the male groups). The data also reveal that the convergence of female
employment rates over time foremost is attributable to strong employment growth for those who entered on a family visa. A second point to
note here is that, while Figure 1 showed similar employment profiles
for natives and the 1970s immigrants from both the Nordic countries
and other countries in Western Europe, the recent data indicate some
differences between these groups. In particular, the recent Western European immigrants from outside the Nordic countries exhibit lower return migration rates and lag somewhat behind the Nordic group in the
labor market. Plausible explanations for these patterns relate to the
stronger links between admission and job offers for labor immigrants
from outside the Nordic countries during the 1970s, with a greater emphasis on skills considered to be needed in the Norwegian labor market,
along with the much stronger economic growth in Norway compared to
the United Kingdom and continental Europe between 1970 and 2010.
Hence, it is probable that the 1970s’ cohorts of immigrants from Western Europe were particularly favorably selected in terms of their job
opportunities in Norway.
Nordic Economic Policy Review
Figure 4: Employment rates 2006–2013, immigrants arriving in 2005–2008 and native
comparison groups
B. Women
Employment (%)
A. Men
Note: Employment is defined as having annual labor or business earnings exceeding the base amount
of the social insurance system (in 2013 NOK 84,204). The data include persons at least 25 years of age,
not enrolled in education, and with residency in Norway at the end of the calendar year.
Figure 5 shows average annual earnings for those who were employed
each year. It is clear that labor earnings tend to be much lower for Eastern European immigrants than for all the other groups. Moreover, the
figure gives no indication of the assimilation effects typically found elsewhere in the literature, whereby immigrant earnings grow more rapidly
than those of natives during the first years in the host country. Instead,
the earnings gap between natives and EU8 immigrants remained constant over the eight-year period considered, with the 2013 earnings of
male EU8 immigrants 34% below those of native men and the earnings of
female EU8 immigrants 24% below those of native women.
Nordic Economic Policy Review
Figure 5: Annual labor and business earnings 2007–2013, conditional on employment.
Immigrants arriving in 2005–2008 and native comparison groups
B. Women
Earnings (1000 2013 NOK)
A. Men
One possible explanation for the relatively low earnings of EU8 immigrants is that they were recruited into occupations with particularly low
wages. If we focus on immigrants and natives employed in the major
immigrant occupations, the earnings gaps are reduced considerably; see
Figure 6. The earnings differential between EU8 immigrants and natives
remains significant, however, and again there is no indication of assimilation effects on immigrant earnings. When we compare natives and
immigrants from accession countries who in 2008 worked in the five
most common immigrant occupations, the earnings gap in 2013 stood
at 22% for males and 13% for women.
Nordic Economic Policy Review
B. Women
A. Men
Earnings 5 main occupations (1000 2013 NOK)
Figure 6: Annual labor and business earnings 2007–2013, conditional on employment in one of
the five main 2008 immigrant occupations
Note: The five main occupations and their employment share in the immigrant data are, for men,
carpenter (11), clerical (10), construction laborer (6), cabinet maker (6), and brick layer (4), and, for
women, cleaner (18), child care (6), sales (6), clerical (6), and waiter (5).
Even though most of the new Eastern European immigrants managed
to maintain a foothold in the Norwegian labor market through the financial crisis, a relatively large fraction also claimed unemployment
insurance (UI) benefits. Figure 7 shows the uptake of UI benefits
month by month for immigrants still residing in Norway. For men,
there was a huge spike in benefit claims around the financial crisis,
starting late 2008, particularly for men from Eastern Europe. The EU8
claimant rates came back down around 2012, but have remained at
much higher levels than those of other European immigrants and natives. And since 2012 they have again displayed an increasing pattern.
For Eastern European women, we have seen a more monotonous increase in UI claimant rates after the financial crisis, albeit with slower
Nordic Economic Policy Review
growth since 2012. Of particular concern is that the UI claimant rates
of the EU8 group seem to have stabilized at very high levels when
compared to natives. In fact, in December 2014 (the last entry in Figure 7) the UI claimant rate among men from accession countries who
had immigrated to Norway in 2005–2008 was five times that of native
men of the same age, and the claimant rate among women four times
that of similarly aged native women. 3 Even immigrants from the Nordic and other Western European countries have had higher UI claimant rates than natives in the aftermath of the financial crisis, but at
much lower levels than those of the EU8 immigrants.
3 As with earnings (see Figures 5 and 6), immigrant-native differences in UI uptake are smaller when we
consider workers in the same occupation. To illustrate, when we restrict the sample to those in the major
immigrant occupations used in Figure 6, the January 2011 uptake rate among male EU8 immigrants was
2.8 times that of native men (14.8 vs. 5.2%) and the uptake rate of EU8 women twice that of native women
(7.7 vs. 3.8%). Bratsberg et al. (2014) study overall UI uptake during the financial crisis and find that differences in age, tenure, industry and occupation account for 40% of the observed difference in uptake between
immigrant men from Eastern Europe and native men and 60% of the observed difference in UI uptake
among women.
Nordic Economic Policy Review
Figure 7: Monthly unemployment insurance claimant rates 2007–2014, immigrants arriving in
2005–2008 and native comparison groups
B. Women
Share with UI benefits (%)
A. Men
Note: The data include persons at least 25 years of age who are not enrolled in education and with
residency in Norway at the end of each calendar year.
Structural challenges
While it is too early to draw any firm conclusions regarding lifecycle
employment in Norway, we do see some discouraging medium and
long-term labor market performance patterns among the postenlargement immigrants from lower-income countries in Europe. Why
do we apparently fail to achieve full labor market integration on par
with natives? We will focus here on three interrelated explanations.
The first is that immigrants often are recruited to jobs with low
general skills requirements, and, in particular, to jobs where Norwegian
language skills is not a key ingredient. These jobs are often temporary
in nature and disproportionally found in cyclically sensitive industries
Nordic Economic Policy Review
such as construction. And since most firms practice a last-in-first-out
principle in case of downsizing, immigrant workers generally have a
high risk of becoming unemployed. Once unemployed, the lack of general qualifications and language skills obviously become a serious handicap in attempts to find other types of work.
Second, those who become unemployed are in most cases entitled to
unemployment insurance (UI). In principle, UI entitlements are fully
transferable within Europe. Among labor immigrant in Norway, entitlement will in any case follow if their labor earnings during the prior
calendar year exceeded 1.5 times the base amount of the social security
system (in 2013, earnings above NOK 126,000 – or one third of the average earnings of male EU8 immigrants depicted in Figure 5). For
workers from, say, Poland or Lithuania, Norwegian UI benefits will typically exceed earnings in the home country by an order of magnitude. In
Table 1, we report average monthly UI benefits and wages in the home
country, collected from the OECD Social and Welfare Statistics database
(see columns I and II). These data show that typical Norwegian UI benefits are 7 to 15 times average UI benefits – and 4 to 5 times average
wages – in Eastern Europe. Because the preceding section showed that
Eastern European migrants earn less than natives and other migrants,
we have also computed monthly UI benefits for those who actually
claimed benefits in Norway in 2010 (see column III) as well as monthly
pay among wage earners (column IV) and average monthly labor earnings for those employed during the year (column V). Although both
benefits and wages of Eastern European migrants fall below those of the
other groups considered, they remain much higher than UI benefits and
wages at home. Hence, incentives for returning home to look for employment there are weak. A probable consequence is that many immigrants from accession countries remain registered as unemployed in
Norway, despite being poorly qualified for new employment. For some
of the unemployed, it will also be tempting to bring the insurance money back to the home country, where costs of living may be less than half
of those in Norway, and thereby obtain a higher standard of living than
what even a fulltime job could deliver in Norway. The rules of the UI
program allow for such stays within the European labor market for a
period of up to three months, but the absence of border controls be36
Nordic Economic Policy Review
tween European countries obviously implies that it can be done to a
much larger extent in practice. 4 Such opportunities may undermine incentives for active job search in Norway, and raise the reservation wage
of the unemployed.
Table 1: Unemployment benefits and average earnings at home and in Norway, immigrants from
the Baltic States, Poland, and the Nordic countries, 2010
At home
In Norway
Monthly UI
Monthly wage
income if
Monthly UI
benefits among
pay, wage
Note: Benefits and wages are converted to Euros using average exchange rate for 2010. Source of
entries in columns (I) and (II) is OECD iLibrary, OECD Social and Welfare Statistics. Entries in columns
(III)–(V) are authors’ calculations based on the register data for the immigrant and native samples
used in Figures 4 and 5.
Third, since European legislation implies that welfare state entitlements
are transferred to the country of employment, a job in, say, Norway, not
only grants membership in the Norwegian social insurance system, but
also entails eligibility to various family allowances. For families with
children, this entails that a job in Norway may be attractive even if the
4 Although the higher costs of living in Norway will mitigate some of the cross-country differences uncovered in Table 1, the possibility of exporting benefits justifies comparisons without adjustments for purchasing power parity.
Nordic Economic Policy Review
offered wage is extremely low. For example, the Norwegian cash-forcare subsidy for a one-year old child now amounts to NOK 6,000 per
month, which adjusted to the 2010 wage levels and exchange rates used
in Table 1 corresponds to EUR 629, or around 80% of average earnings
in Poland. Such features give employers and prospective immigrant
employees incentives to agree on very low wages and poor working
conditions. While this can be a win-win situation for the employer and
the immigrant worker – at least in the short run – it may stimulate the
creation of poor jobs with high subsequent unemployment or disability
risk and substantial (expected) costs for the welfare state.
A more robust welfare model
There is now an ongoing policy debate in several European countries
about reestablishing elements of the previously existing migration barriers; either by making eligibility of economic transfers from a particular country conditional on past social security contributions to that
same country (i.e., limit the transferability of eligibility), or by adjusting
benefits with a country specific cost-of-living index when they are exported to another country. The latter would mean, for example, that
Norwegian UI benefits paid out in Poland would be cut by more than
one half compared to the current level.
Requiring a country-specific contribution period before benefits can be
claimed may reduce some of the incentive distortions discussed in the
previous section, particularly those related to creation and acceptance
of very poor and short-lived jobs. However, this potential advantage
must be balanced against the disadvantage of also reducing welfareenhancing labor mobility within Europe. Further, introducing cost-ofliving adjustments to social insurance payments appears to us to be a
“dead end”. Given that people can travel freely across Europe without
notifying authorities of their whereabouts at each point in time, the
scope for circumventing downwards cost adjustments appear almost
limitless (unless draconian control measures are implemented).
In any case, we will argue that a strategy designed to strengthen the
sustainability of the Nordic welfare model primarily must consist of pol38
Nordic Economic Policy Review
icies aimed at making the welfare system robust to the existence of an
open European labor market, and not of policies aimed at reversing it.
How can this be achieved in practice? One obvious policy option is to
scale down on social insurance programs and thus make the welfare
state less generous for everyone. Such a policy might trigger a “race to
the bottom”, as it seeks to ensure that a country’s own welfare state
provisions are not too generous relative to those in other countries.
Thriftier social insurance would improve work incentives for natives as
well as immigrants, and reduce migration distortions generated by
cross-country differences in social insurance systems. However, the relatively generous social insurance programs in the Nordic countries are
in place for a reason. They reflect voter preferences for a low-risk society with sound insurance arrangements in case of sickness, disability, or
involuntary unemployment. They also reflect preferences for a relatively egalitarian society with little poverty. Viewed as a whole, the “Nordic
model” has been successful in delivering economic growth and high
standards of living for the vast majority of its citizens.
Is there an alternative way? We will argue that there is, but obviously not without its pros and cons. We will sketch a policy based on
three pillars:
A first pillar consists of ensuring minimum standards in the labor
market, including a minimum wage and possibly limitations on the use of
temporary contracts. In the absence of such minimum standards, employers will have an incentive to recruit foreign workers with a high expected future income flow from the welfare state as such workers are
willing to accept lower wages, ceteris paribus. This may result in an “adverse selection” of foreign workers (from a fiscal point of view), and also
imply particularly high social insurance replacements among those who
do arrive, as the progressive nature of social insurance entails that the
replacement ratio declines with earnings. In a worst case scenario, firms
could repeatedly recruit new immigrant workers to temporary and poorly paid jobs, based on the premise that their “real pay” would come from
the welfare state. Hence, a benefit of minimum standards is that they may
remove externalities arising from the fact that a job contract in, say, Norway, not only commits the employer, but also the Norwegian welfare
Nordic Economic Policy Review
state, to future payments. Another benefit of minimum standards is that
they will reduce average replacement ratios of welfare benefits.
A second pillar consists of transforming the social insurance system
from being based on passivating and easily exportable cash transfers to
being based on active participation. There is now ample empirical evidence showing that the moral hazard problems in social insurance can
be considerably reduced by offering income through active participation
rather than cash benefits only; see Røed (2012) for a recent overview
and discussion of the literature. Participation-based social insurance
further leads to more favorable selection by attracting persons who
wish to work and contribute while deterring persons with low workmorale. In relation to longer term spells of both unemployment and sick
leave/disability, active participation entails that social insurance payments to a larger extent become contingent on participation in job
search, training, communal work, or vocational/medical rehabilitation
programs. In particular, with active participation the design of temporary and permanent disability insurance programs will encourage and
support the use of remaining (partial) work capacity, if necessary
through the establishment of sheltered employment. A job offer is obviously also more place-bound than a cash transfer, and cannot readily be
exported to a home country with lower costs of living.
The third pillar consists of transforming family allowances from being based on cash transfers to being based on the supply of
free/affordable family-friendly and work-oriented services. For example, instead of offering (exportable) cash-for-care subsidies, the welfare
state can offer high-quality child care directly.
Now, all of these policies also have some downsides. Higher minimum standards in the labor market distort the price-adjustment mechanism in the labor market and may raise unemployment among lowskilled workers. A more activity-oriented social insurance system may
come with high administration costs and may require a large number of
sheltered workplaces adapted for persons with reduced work capacity.
Offering publicly provided childcare instead of cash transfers reduces
the families’ freedom of choice. Finding the optimal policy inevitably
involves some tradeoffs. It is about balancing conflicting arguments. The
point we wish to emphasize here is that policy makers actually have a
Nordic Economic Policy Review
range of options. There are viable alternatives to the reintroduction of
migration barriers and to benefit-cutting competition. We will also argue that the extension of the common European labor market to include
countries with lower economic development implies, ceteris paribus,
that the optimal balancing points shift toward higher employment
standards and more place-bound social insurance and family support
programs. Hence, if the policy was close to its optimum prior to the enlargements of the European labor market, it probably needs considerable adjustment now.
The recent enlargements of the European labor market represent a considerable challenge for welfare state economies. Provided that we wish
to preserve the freedom of movement across European national borders, welfare state institutions in rich countries need to adapt. We have
argued that business as usual is not a viable option, and that we either
need to scale down on income insurance and family support programs
(a race to the bottom), or make welfare state institutions more migration robust by i) raising minimum standards in the labor market, ii)
making income insurance programs more participation oriented, and
iii) substitute place-bound services like free childcare for exportable
cash transfers in family support programs.
Barrett, A., & B. Maître (2013). Immigrant welfare receipt across Europe, International Journal of Manpower 34(1), 8–23.
Boeri, T. (2010). Immigration to the Land of Redistribution, Economica 77(308),
Bratsberg, B., Fevang, E., & Røed, K. (2013). Job Loss and Disability Insurance, Labour Economics 24, 137–150.
Nordic Economic Policy Review
Bratsberg, B., Raaum, O., & Røed, K. (2010). When minority Labor Migrants Meet the
Welfare State, Journal of Labor Economics 28(3), 633–676.
Bratsberg, B., Raaum, O., & Røed, K. (2014). Immigrants, Labour Market Performance, and Social Insurance, The Economic Journal 124(580), F644–F683.
Clemens, M.A. (2011). Economics and Emigration: Trillion-Dollar Bills on the Sidewalk?, Journal of Economic Perspectives 25(3), 83–106.
Kahanec, M., Pytlikova, M., & Zimmermann, K.F. (2014). The Free Movement in an
Enlarged European Union: Institutional Underpinnings of Economic Adjustment,
IZA DP No. 8456.
Kennan, J. (2012). Open Borders, Review of Economic Dynamics 16(2), L1–L13.
Kvist, J. (2004), Does EU enlargement start a race to the bottom? Strategic interaction among EU member states in social policy, Journal of European Social Policy 14,
Mollerstrom, J. (2015). Ethnic fractionalization and the demand for redistribution
– Potential implications for the Nordic model, Nordic Economic Policy Review
(this volume).
Nannestad, P. (2004). Immigration as a Challenge to the Danish Welfare State? European Journal of Political Economy 20(3), 755–67.
Preston, I. (2014), The Effect of Immigration on Public Finances, Economic Journal
124 (580), F569–592.
Rege, M., Telle, K., & Votruba, M. (2009). The Effect of Plant Downsizing on Disability
Pension Utilization, Journal of the European Economic Association 7(5), 754–785.
Røed, K. (2012). Active Social Insurance. IZA Journal of Labor Policy 1:8 (2012)
Sinn, H-W. (2002). EU enlargement and the future of the welfare state, Scottish Journal of Political Economy 49, 104–115.
Nordic Economic Policy Review
3. Future Pathways for Labour
Market Policy: Including the
Michael Rosholm and Michael Svarer, Department of Economics and
Business Economics, Aarhus University
It is a challenge for the Nordic countries to increase the employment rate
for people on the edge of the labour market. We discuss the potential for
reducing exclusion from employment with a special focus on Denmark. A
series of labour market reforms in the recent decades have reduced the
number of people on public income transfers and are expected to have a
further positive impact on the employment rate in the years to come
when the reforms are fully implemented. The main focus of the reforms
have been on increasing economic incentives to work for people in or
close to the labour market and on fine tuning active labour market policy
for unemployed in the unemployment insurance system. A remaining
challenge for current labour market policy is to increase the participation
rate for those who are at risk of getting excluded from the labour market.
We present new evidence on active labour policies for people on the edge
of the labour market and discuss the potential for increasing employment
rates for this particular group of unemployed.
In comparison to most other OECD countries, the Nordic labour markets
are characterized by high participation rates. Still, the fraction of people
on the edge of the labour market who are either on sickness benefits,
social assistance benefits or disability pensions remains high. It is,
therefore, a major economic political challenge to support inclusion of
people who currently are more or less excluded from the labour market.
This article will concentrate on the Danish case and will focus on the
role of active labour market policy to stimulate the supply side of the
labour market. Alternative measures like in-work tax credits for vulnerable groups in the labour market or firm-based policies to encompass
individuals with low working capacity into firms are not considered.
Although the focus is on Denmark and narrowed to a particular set of
policy instruments, the main conclusions and empirical findings are
clearly relevant for the other Nordic countries as well since they face
similar challenges and also rely heavily on the use of active labour market policies.
The employment rate in Denmark has been remarkably stable in the
past 50 years. Figure 1 shows that close to 75% of the population aged
18–64 were employed in 1960, and the same is true today. The gender
composition of the employed has changed remarkably, though. From
the mid-1960s and onwards, the female labour force participation rate
increased significantly. Despite this huge inflow of female workers, the
overall employment rate barely changed. The implication is, of course, a
similar decline in the male employment rate.
Nordic Economic Policy Review
Figure 1: Employment rate and fraction on public support, age 18–64
Source: Statistics Denmark, ADAM data bank.
At the same time the number of people receiving some kind of public
transfer exhibited an almost exponential growth pattern in the 1970s
and 1980s. In the early 1990s the number of people on public support
had increased from 200,000 in 1960, corresponding to 6% of the population aged 18–64, to almost 1 million, representing almost 27% of the
population in the age group 18–64. Since then a series of labour market
reforms in combination with more favourable macroeconomic conditions have brought the fraction in the age group 18–64 receiving public
transfers down to around 22%. The drop in the number of people on
public support since 1993 has been mirrored in the employment rate,
which has risen from 70 to close to 75%. An interesting point to note is
that almost the full population in working ages in Denmark is either
working or receiving some type of public transfer. In contrast to the
vast majority of other countries, the fraction of people not working is
entitled to some kind of public transfer.
Nordic Economic Policy Review
Currently, around 800,000 Danes between 18–64 years old receive
some type of public transfer. Approximately 100,000 of these are expected to become employed within the next 5 years or so, when the
business cycle normalizes, and when the impacts of recent reforms in
the unemployment insurance system and early retirement systems are
realized (Danish Economic Council, 2015). If this assessment turns out
to be correct, the employment rate will break the 50 years old glass ceiling and reach a level around 78%.
As shown in Figure 2, the employment rate in Denmark is relatively
high in an international comparison, and is only marginally lower than
the other Nordic countries, with the exception of Iceland, which has a
remarkably high employment rate.
Figure 2: Employment rate in selected countries, 2014
Note: In per cent of population between 15–74 years old.
Source: OECD.
In any event, with the expected increase in the employment rate over
the next five years, it will be a challenge for Danish labour market policy
in the future to increase the employment rate even further. The structural unemployment rate is at an all-time low, and it is difficult to perceive policies to lower it even further.
Nordic Economic Policy Review
There are several reasons why persons of working ages do not
work. A common theme of the recent reforms in Denmark related to
unemployment insurance, social assistance and early retirement has
been to increase incentives to work by reducing the generosity of and in
particular the eligibility for public transfers. The assessment of e.g. the
Danish Economic Council (Danish Economic Council, 2015) is that these
reforms have had a positive impact on employment rates.
Another strategy in order to increase employment rates is to improve the employability and qualifications of those receiving public
support. Active labour market policies constitute an important supporting pillar of the flexible labour markets of Denmark and several other
European countries. In Denmark, these policies have contributed to
lowering the gross unemployment rate to remarkably low levels since
the early 1990s (see e.g. Andersen & Svarer, 2007). The active labour
market policy tools have consisted of traditional measures like classroom training, job search assistance, employment subsidies and job
training and have been used quite intensively to lower unemployment
rates (see e.g. Andersen & Svarer, 2012).
Whereas benefits cuts and active labour market policies have been
used intensively in the last 30 years for those close to the labour market, but not in jobs, less has been done to help individuals who face a
longer route into jobs – the excluded or those at risk of exclusion.
In recent years more focus has been devoted to helping unemployed
with a high risk of being excluded from the labour market into employment. One important reason for this is that this group is fairly large
compared to the structurally unemployed. Another reason is, of course,
that a life in social and labour market exclusion is costly, not only to the
individual itself, but also to society. It is obvious that policies aimed at
individuals on the edge of the labour market have a lower probability of
generating higher aggregate employment in the short run than e.g. increasing the early retirement age from 63 to 64 or reducing unemployment benefit periods from 4 to 2 years, and they should therefore be
assessed accordingly. On the other hand, for each person among the excluded who finds stable employment, the gain to society is relatively
large. This is in particular the case if the policy manages to bring down
the inflow into public transfer for young people.
Nordic Economic Policy Review
In this article we present a number of recent Danish measures that
have been introduced to prevent and combat exclusion for the more vulnerable unemployed, and we provide a brief evaluation and discussion of
the potential of including the excluded in the future labour market.
Active labour market policies for unemployed
on the edge of the labour market
It can be argued that for this group, the best way to prevent a life in social exclusion is to provide sufficient capabilities to undertake a qualifying education, as education seems to offer at least partial insurance
against a life on the social margin.
There is a large literature (see e.g. Elango et al. (2015) for an overview) on how early childhood interventions in nurseries, kindergartens
and schools may prevent social exclusion, but this is not the topic of the
present paper. We look at policies designed to assist adolescents and
adults at risk of exclusion or already excluded.
One important issue is the dropouts from the youth educational system. Dropout rates are fairly low in high schools, but in vocational schools
(which provide qualifying educations), the gross dropout rate is around
50%. However, around 20 percentage points of these eventually complete another qualifying education, leaving a net dropout rate of 30%,
which is remarkably high and has not changed much in the past 15 years.
Many of the young persons who drop out end up receiving social assistance, and various policies have been attempted to help prepare
them for the youth education system, but so far none have been successful. A recent analysis from Rangvid et al. (2015) use propensity
score matching and IV methods to show that amongst all the preparatory measures used to help young people prepare for the youth education
system, only one (production schools) might have positive impacts.
However, even this result is not robust (IV-estimation shows positive
effects, while matching shows negative effects), and most of the other
measures show significant negative impacts. Moreover, all these
measures are quite costly. Hence, it would seem that there is room for
Nordic Economic Policy Review
Recently, two interventions have been introduced and tested with
the explicit aim of helping this group of at-risk youth; a “bridge building” intervention and a mentoring intervention. In the following, we
briefly present some of the main results from the evaluations of these
Case: Building bridges for young unemployed
In 2013 the Danish labour market authority initiated an intervention
aimed at young people without a qualifying education who received social assistance. The focus was especially on those perceived to have
some type of disadvantage (academically or socially). The main aim was
to help these young persons into education and to ensure that they
completed the education they had enrolled in. The intervention was
multi-facetted and took place at an ordinary educational institution,
typically a vocational school, hence the term “bridge-building”. The
young persons would meet there each day, attend classes in reading,
writing and mathematics, learn to be there on time every day (an important part of the intervention), learn to feel comfortable at the school,
visit other educational institutions, and so on. Each person would be
assigned a personal mentor, and there would be one (and only one)
contact person to the municipal system during the intervention period.
A total of 2,600 young persons participated in the programme,
spread over 12 different educational institutions located all over Denmark. The intervention was based on voluntary participation both by
the educational institutions and by the unemployed, so there is no exogenous variation in participation that can be exploited to evaluate the
effect of the intervention. To obtain a measure of how the intervention
has effected educational enrolment, Rosholm & Svarer (2015) use a
matching estimator strategy. Based on rich information from administrative registers, grades from compulsory school, labour market and
educational histories, socioeconomic status, age, gender and ethnicity,
marital/cohabitation status, health status (including information on
mental health), drug and alcohol abuse, and a subjective case worker
Nordic Economic Policy Review
evaluation of “readiness for education”, each participant is matched
with up to 10 control persons. 1
Figure 3 shows the estimated effects of the bridge building intervention on the fraction in education. The left hand panel shows the fraction
receiving educational support in the treatment group and a matched
control group. In the matched control group, approximately 20% are in
education 26 weeks after the start of the intervention, while the treatment group has nearly 40% in education after 26 weeks. After 1½
years, the fraction in education has increased to 30% in the matched
control group, while it is still 40% in the treatment group. The figure on
the right hand side shows the estimated impact of the intervention as
well as the 95% confidence interval.
For more information on this study, see Rosholm & Svarer (2015).
Nordic Economic Policy Review
Figure 3: Effects on education
Fraction receiving educ. grant
10 15 20 25 30 35 40 45 50 55 60 65 70 75 80
Weeks since entry into program
Treatment group
Matched control group
Effect on fraction rec. educ. grant
10 15 20 25 30 35 40 45 50 55 60 65 70 75 80
Weeks since entry into program
95% confidence interval
The figure 3a shows the fraction of the treatment- and control group enrolled in education at a
particular time measured since the moment they entered the bridge building project. The figure 3b
the average treatment effect of the bridge building project from a matching analysis. For more details see Rosholm & Svarer (2015).
Nordic Economic Policy Review
First graph in Figure 3 shows that after 26 weeks, the fraction of young
persons who are in the educational system has increased by
17 percentage points relative to the matched control group – around
80% in relative terms. 1½ years after programme start, the effect has
declined to 10 percentage points, which is still a 25% increase in the
fraction in the educational system. Moreover, this effect is statistically
significant. Additional analyses show that the fraction completing the
basic semester in the vocational schools increases from 5 to slightly
above 10%, again statistically significant, and there is also a subsequent
positive impact of entering the main part of the vocational track.
Although the individuals in the intervention have relatively poor
qualifications, additional analyses show that the intervention seems to
work even better for the weaker among these young persons. Figure 4
shows the effects for those who have a grade in Danish from compulsory school leaving exams at grade 9, and for those who do not. Note that
for those who do not have such an exam (47% of the treatment group),
the effect is 15 percentage points after 1½ years, while it is only 5–10%
for those who have such a grade. Although the difference is not statistically significant, it is a notable pattern, since many interventions are
often less effective when aimed at weaker groups in the labour market.
Similarly, we find larger effects for young persons who have spent more
time on social assistance during the past year.
Nordic Economic Policy Review
Figure 4: Effects by school leaving exam (or not) in Danish
Effect, no compulsory Danish exam
10 15 20 25 30 35 40 45 50 55 60 65 70 75 80
95% confidence interval
Effect, has compulsory Danish exam
10 15 20 25 30 35 40 45 50 55 60 65 70 75 80
95% confidence interval
Note: The figures show the average treatment effect of the bridge building project from a matching analysis for people who do not have an exam in Danish from compulsory school (left) and for
those who have an exam in Danish from compulsory school (right). For more details see Rosholm
& Svarer (2015).
Nordic Economic Policy Review
In terms of cost considerations, the per person cost of the bridge building intervention was DKK 45,000. This is a fairly large upfront investment. However, compared to a lifetime of public income support and
associated costs, it may be worthwhile if the substantial effects we find
can be sustained in the long run.
All in all, the findings from the bridge building project suggest that
such an intervention aimed at helping young vulnerable person into vocational education has increased enrolment into vocational education.
The real test for the programme is whether the increased enrolment in
education also leads to more completed education and increased employment rates for this group. Future analysis will reveal whether this is
the case.
Case: Mentoring of young people on social
In 2012, the Danish labour market authority initiated an intervention
that used mentors for young individuals without a qualifying education
who were considered to be at risk of having difficulties in the educational system.
In each of 13 job centres, 200 young persons were randomized into
either a treatment or a control group. The control group would receive
treatment as usual, which would be regular meetings with case workers
(every 13 weeks), and occasional participation in some of the active
measures aimed at this particular group, which would often be the set
of interventions found to be not effective by Rangvid et al. (2015).
The treatment group would receive the treatment as usual plus an
offer of a mentor for 52 weeks. It was voluntary to accept the mentor,
and 92% of the treatment group were assigned a mentor. The mentor is
employed by the job centre to support the activities agreed upon between the job centre and the young person in the job- and education
plan. The mentor then follows the young person until he or she starts an
education or a job, or until a maximum of 12 months has elapsed. The
mentor may focus on helping the young person finding a job or an educational institution, but he may also help with more personal tasks,
which were not specified in advance. This could be anything from get54
Nordic Economic Policy Review
ting up in the morning and attending meetings and activation programmes to helping reach an agreement with the bank on debt repayments, meeting a general physician or a psychologist, or other personal
problems the young persons might have.
Of the 1,299 in the treatment group, 1,193 individuals, corresponding to 92% of the treatment group, participated in at least part of the
intervention. 2
Figure 5 below shows the average number of minutes per week
each mentee spends with the mentor.
Figure 5: Average weekly time spent with a mentor
Figure 6 shows the impact of the programme on the main outcome of
interest, namely the fraction entering the educational system.
For more details on the mentoring intervention, see Svarer et al. (2014).
Nordic Economic Policy Review
Figure 6: Effect on education (receiving study support)
Fraction in education
.1 .15
Weeks since entry
Treatment group
Control group
The dotted lines are 95% confidence bands on the effect
The first thing to note is that, in the control group, only around 12% are
in education one year after being assigned a mentor, while the fraction
in education in the treatment group is around 16%. The effect of the
programme is depicted by the solid black line, which shows that there is
indeed a 4 percentage points effect of the programme, and this effect is
statistically significant after slightly less than a year, although it does
tend to disappear again towards the end of the observation period. In
Figure 7a–c we show the effect of being assigned a mentor on different
sub groups, namely those who have no compulsory Danish exam, those
who have a grade below 4 (the median in the sample among those who
have a grade), and those who have grade 4 or above. Being assigned a
mentor appears to be effective for youth with low grades, but not for
those with higher grades or those without a compulsory school leaving
exam. This is in contrast to the results for the bridge building intervention, which appeared to be more effective for youth without a compulsory school leaving grade in Danish.
Nordic Economic Policy Review
Figure 7a: The effect of the mentoring intervention, by grade in Danish in compulsory school –
Education, Danish GPA≥4
Fraction in education
.05 .1 .15 .2
Education, Danish GPA>=4
Weeks since entry
Treatment group
Control group
The dotted lines are 95% confidence bands on the effect
Figure 7b: The effect of the mentoring intervention, by grade in Danish in compulsory school –
Education, Danish GPA<4
Fraction in education
.1 .15
Education, Danish GPA<4
Weeks since entry
Treatment group
Control group
The dotted lines are 95% confidence bands on the effect
Nordic Economic Policy Review
Figure 7c: The effect of the mentoring intervention, by grade in Danish in compulsory school
Finally, in Figure 8 the impact of the mentor intervention on the fraction
in employment is shown. Around 5% in the control group are in employment after a year. In the treatment group, this is slightly larger,
growing to 7% after 70 weeks. This implies a small but borderline statistically significant effect on employment rates.
Nordic Economic Policy Review
Figure 8: Effect of the mentoring intervention on employment
Fraction in employment
.1 .15
Weeks since entry
Treatment group
Control group
The dotted lines are 95% confidence bands on the effect
The mentoring intervention suggests that there is a potential to increase educational attainment for a group of vulnerable young unemployed. Future assessments will tell whether the increased enrolment
into education results in more completed education and more stable
employment paths. At this stage it will also make more sense to conduct
a proper cost-benefit analysis of the intervention.
For the mentoring intervention, the price per participant was
around DKK 25,000, which is considerably cheaper than the bridge
building intervention which also included mentoring. Still, the bridge
building intervention yielded impacts that were three times larger than
the mentoring intervention, so it may be that the former is still more
cost effective.
Nordic Economic Policy Review
Case: Measuring Employability
A final important policy issue is that helping the group of adult individuals at risk of social exclusion takes time as it is a group with complex
problems and issues. According to the Expert Group (2015), they have
very little education, very little working experience, 36% of them use
psycho-pharmaca, they have various physical health problems, 17% of
them were placed in foster care as children, they have severe debt issues, and so on. Hence, improving their employability requires that they
progress in several dimensions, and we do not know which of these are
more important, or if progress in some dimensions require progress in
other dimensions.
A recent research project (see Rosholm (2015) for details) tries to
measure the progress of persons at risk of exclusion by use of repeated
surveys. These surveys are carried out as part of the meetings taking
place in job centres during meetings with case workers taking place approximately every 13 weeks. Both the case worker and the client is
asked to score the client on issues such as access to a network for job
search, ability to cooperate, self-confidence, ability to take initiative,
extroversion, ability to cope with everyday issues, health and health
coping, reservation wages, job search strategies, subjective employability assessments, etc.
The aim of the employability indicator project is to assess whether
improvements on these indicators can predict subsequent employment.
The problem is that at the outset of this project in late 2013 and early
2014, half of the participants – not immediately employable individuals
receiving social assistance – had been on public income assistance uninterrupted at least since the beginning of 2008; that is, for six years. 90%
of them had been employed less than 20% of the time in the same period. Hence, their prospects are meagre, and it is very difficult to judge
the quality of active labour market policies on an outcome such as employment since they are very far away from employment. Hence, there
is a perceived need for intermediate measures that can point towards
(predict) employment, but that may be easier to affect in the short run.
Figure 9 shows the fraction in employment among the participants
in the project.
Nordic Economic Policy Review
Figure 9: Fraction employed in the employability indicator project
Fraction in enployment
0.00 0.02 0.04 0.06 0.08 0.10 0.12 0.14 0.16 0.18
Time since entry into project (weeks)
Only around 7% of the participants in this project – who do not receive
anything beyond the “treatment as usual” apart from their responding
to the employability survey at the regular meetings with case workers
in the job centre – were employed two years after entry into the programme. This also demonstrates that they are indeed hard to place. Still,
the figure hides the fact that 14% of the participants had actually held a
job at some point during the two years after entry into the programme.
The project also asks questions about job search behaviour, specifically about the types of job search channels used. There are a total of six
different job search channels that can be specified: 1) responding to
newspaper adds, 2) via the internet, 3) unprompted applications, 4) using ones network, 5) via job training sites, and 6) via temporary help
agencies. If we regress the number of search channels used on a set of
individual specific variables in a linear regression model, we get an Rsquared of 3%. When we include the employability indicators (initial
levels as well as subsequent changes), the R-squared increases to 26%.
Nordic Economic Policy Review
Hence, these indicators are certainly highly predictive of commencing
job search. In particular, questions on health coping strategies, selfconfidence, knowledge about the labour market, purposefulness, everyday coping, cooperation, initiative, and the reservation wage affect the
number of search channels used.
Next, when we estimate a linear probability model of an individual
being employed in a given week on a set of individual background variables, the R-squared is below 3%. Once the information on employability
indicators and job search channels is included, the R-squared increases to
8%. This is not quite as dramatic an increase as we found for job search,
but still, we almost triple the explained variation in the model.
The model further reveals that, not surprisingly, starting to search is
a very important predictor of eventually finding employment. Moreover, informal channels are – for this particular group – the most effective. In particular, search via job training sites, temp agencies, and unprompted applications are the ones that are significantly associated
with finding employment.
Using employability indicators has certain implications for labour
market policy. First of all, it enables us to investigate whether different
active labour market policies are effective in creating employability in
the important dimensions. Moreover, in time, it may enable case workers and clients to be much more specific about the aims of the policies
employed, and to design policies to create progress in certain relevant
dimensions. It may thus become easier to tailor active policies to this
group of very vulnerable – and highly heterogeneous – individuals, for
whom we have yet to “crack the nut”.
Conclusion and discussion
As labour market policies become more and more successful, partly due
to continual use of the existing evidence regarding their effectiveness, it
may become increasingly difficult to reap the fruits of additional refinements in the same policies in terms of reducing e.g. the structural
unemployment rates.
Nordic Economic Policy Review
Hence, there is a case to take a closer look at policies aimed at individuals at the edge of the labour market; the excluded and those at risk
of exclusion. If we can prevent some of the young persons at risk of exclusion from becoming excluded, even a fairly large investment in adequate policies may pay off in the longer term. In addition, if we can become better at helping those who have become excluded back into the
labour market through a better understanding of their problem, then
we can better tailor policies to help them that may also pay off.
We have discussed a couple of interventions aimed at preventing
exclusion with the potential to improve active labour market policies
aimed at youth as well as an employability project aimed at better understanding how to create progress for those at the edge of the labour
market. This type of policies may become more important in the future
if we want to continue to strive to include as many as possible and
hence sustain the welfare state in the longer term.
Andersen, T. & Svarer, M. (2007). Flexicurity – Labour Market Performance in Denmark, CESifo Economic Studies, 53 (3), 389–429.
Andersen, T. & Svarer, M. (2012). Active labour market policies in a recession. IZA
Journal of Labor Policy.
De Økonomiske Råd (2015). Dansk Økonomi. Efterår 2015.
Elango, S., García, J. L., Heckman, J. J., & Hojman, A. P. (2015). Early Childhood Education, forthcoming in Means-Tested Transfer Programs in the United States, Volume
II, edited by Robert A. Moffitt.
Ekspertgruppen (2015). Nye veje til job – for borgere i udkanten af arbejdsmarkedet,
Ekspertgruppen om udredning af den aktive beskæftigelsesindsats. Report to the
Ministry of Employment.
Rangvid, B. S., Jensen, V. M., & Nielsen, S. S. (2015). Forberedende tilbud og overgang
til ungdomsuddannelse. SFI-rapport 15:14.
Rosholm, M. (2015). The employability indicator project – some first results, Draft,
Aarhus University.
Rosholm, M. & Svarer, M. (2015). Kvantitativ evaluering af brobygningsindsatsen,
report to the Danish Labour Market Board (
Nordic Economic Policy Review
Svarer, M., Rosholm, M., Havn, L. & Høeberg, L. (2014). Evaluering af mentorindsats
til unge uden uddannelse og job, report to the Danish Labour Market Board
Nordic Economic Policy Review
4. Economics of Innovation
Tuomas Takalo, Hanken School of Economics and HECER and Otto Toivanen,
KU Leuven, Aalto University and CEPR
We argue that the design of innovation policy in the Nordic countries
should better acknowledge i) the uncertainty related to outcomes of
innovative activities, ii) the benefits of agglomeration, iii) the effects of
being small open economies, and iv) the impact of digitization. All these
call for a predictable institutional environment that allows researchresources to agglomerate through a bottom-up process and to flow to
their best, often unexpected, uses. Indirect innovation policies such as
e.g. basic research, education, competition policy, and financial and labor market regulations may be more important than direct innovation
policies such as intellectual property and government support for private R&D, especially in small open economies where benefits from direct support of private R&D and strong domestic intellectual property
rights are low.
1 Tuomas Takalo, Hanken School of Economics, E-mail: [email protected] Otto Toivanen, KU Leuven and CEPR, E-mail: [email protected] We thank Ari Hyytinen, Markku Stenborg, Matti Pohjola,
Jesper Roine, Petri Rouvinen, and the seminar participants at the NPER conference in Helsinki for useful
discussion and comments. Toivanen thanks the Bank of Finland for hospitality. The usual caveat applies.
The key driver of economic growth is innovation (see, e.g., Aghion and
Howitt, 2009). This consensus on the policy goal – to foster innovation –
has not lead to agreement as to the means to achieve it. Recent books
written by academics for the wider audience illustrate the large variation in policy advice: Lerner (2009a) and Acemoglu and Robinson
(2013) argue that governments should focus on creating the right institutional environment for the private sector to work. Mazzucato (2013)
and Atkinson (2015) make strongly the case that governments should
take an active role in choosing the direction of research, development
and innovation activities. The objective of this article is to discuss what
economic research suggests as to what innovation policy should look
like and what role the government should take.
At the heart of the economic approach to innovation policy is the
concept of market failure which creates a wedge between social and
private returns to innovative activity. 2 The main market failure in the
area of innovation is the imperfect appropriability of the returns to research and development (R&D) investments, as innovative firms and
individuals cannot capture all benefits that their innovations provide,
but share them with consumers and other firms and users (Nelson 1959
and Arrow, 1962). 3 Financial market imperfections in relation to the
funding of R&D investments are often mentioned as another important
market failure (see Hall and Lerner 2010 and Kerr and Nanda 2014 for
surveys). As a result of these market failures, the private sector is likely
to invest too little in R&D activities. 4 Roughly speaking, the private sec2 Some scholars such as Nelson (2009) and Mazzucato (2013) forcefully argue that market failure allows a too
narrow role for the government, and advocate the systems of innovation approach instead. We primarily view
this difference as a semantic one. For example, in the cases that Mazzucato (2013) brings forth to argue for an
active government, the government acts to correct market failures such as missing markets, imperfect competition, imperfect information and other systemic problems that are not solved by market forces.
3 We will use the words “R&D”, “invention”, and “innovation” almost interchangeably albeit they do involve
subtle but important differences. See, e.g., Carlino and Kerr (2015) for a discussion.
4 R&D projects may also generate negative social externalities (e.g., competition at the marketplace may lead
to business stealing and duplication of R&D costs). While in theory these adverse effects of R&D investments
could result in overinvestment in R&D, in practice underinvestment due to imperfect appropriability and
financial market imperfections is a much more likely outcome. For example, Jones and Williams (1998)
Nordic Economic Policy Review
tor should take care of activities where social welfare mainly consists of
private profits, and the government should provide those activities with
high social returns but low or non-existing private profits, and in the
possibly large grey area in-between, the government may design policies that complement innovation in the private sector and steer the private sector to choose actions that are closer to the social optimum.
The starting point of our analysis is the fundamental challenge of
innovation policy: how to encourage the development of new innovations while achieving the potentially conflicting goal of ensuring maximal diffusion of those innovations? We stress four features that shape
innovative activity and the government’s role in it. First, there is considerable uncertainty as to who will succeed in research and in commercializing that research and when. The endemic informational problems in innovation create scope for both positive and negative unintended consequences of government policies. 5
The second feature which we stress is agglomeration. Evidence (e.g.,
Jaffe,1989, Cowan and Zinovyeva, 2013, and Carlino and Kerr, 2015)
suggests that agglomeration of innovative activities leads to higher
productivity, and should thus be encouraged. In our view the best option to foster agglomeration is to invest in high-quality basic research
and to build an institutional environment that channels, in a bottom-up
manner, human and financial capital to those geographic and intellectual areas that show signs of success.
Our third key feature is the universal good nature of knowledge. 6
This is a two-edged sword for the Nordic countries: On the one hand, it
means that the Nordic countries should actively suck in new knowledge
generated by the more than 99% of human population living elsewhere.
estimate that the socially optimal level of R&D in the US would be 2–4 times the actual one, despite all the
policies of promoting innovation that are already in place.
5 Holmstrom and Myerson (1983) provide an important analysis of how incomplete information affects the
social planner’s problem of which policy to choose.
6 Admittedly there is evidence that knowledge spillovers are still to some extent local (which provides a
rationale for favoring agglomeration within countries, as discussed in the previous paragraph). At the same
time, there is plenty of evidence of increasingly strong international knowledge flows (see, e.g., Griffith et al.
2011). As an early example, the first Finnish telephone company was established in Helsinki in 1877, only in
a year after Bell got his patent on the telephone in the US.
Nordic Economic Policy Review
On the other hand, this means that a large part of the wedge between
social and private welfare, i.e., the very basis for an active government
role in supporting innovation, disappears. 7 Almost without exception,
the existing literature on innovation policy takes a “large country” approach. Some policy conclusions, however, may change markedly when
a small open economy approach is adopted.
Fourth, the design of innovation policies should take into account
that we are only in the early phases of digitization that is increasing international knowledge flows and bringing other large but unknown
changes to us. The best way to prepare for the future is to provide a
sound institutional structure that allows the economy to adjust. This
calls for increased flexibility at all levels of the institutional set-up, and
especially in education. 8
This takes us to the main point of this article: the most important
innovation policies are likely to be “non-innovation” policies that determine the institutional environment for innovation but are not directly aimed at promoting innovation. Education, basic research, financial
and labor market regulations, competition and regional policies, and
bankruptcy laws are examples of “indirect innovation” policies that may
affect innovation more than direct innovation policies such as intellectual property, and government support for private R&D.
The remainder of the paper follows the above themes, In Section 1
we discuss direct innovation policies. Section 2 is devoted to indirect
innovation policies. We offer conclusions in Section 3.
7 A large part of the wedge between social and private welfare is consumer surplus. In the case of, say, Astra
pharmaceuticals, most of the consumer surplus generated by Astra’s new drugs resides somewhere else but
in Sweden and should be ignored when designing an innovation policy that maximizes the social welfare in
Sweden. Also, technological spillovers contributing to the welfare wedge partially flow abroad.
8 For research on the impact of digitization, see, e.g., Greenstein et al. (2013) and Goldfarb et al. (2015).
Brynjolfsson and McAfee (2015) and Bessen (2015) popularize this research.
Nordic Economic Policy Review
Direct innovation policies
Intellectual property rights
Intellectual property has many facets that have been extensively analyzed (see, e.g., Menell and Scotchmer, 2007, for a survey). Intellectual
property attempts to solve the fundamental tradeoff of innovation policy by legal means, as it confers an innovator a temporary exclusive right
to her innovation. This right provides a possibility to monetize innovation and thereby enhances the incentives to innovate. After the right
expires, the innovation and protected knowledge becomes freely usable.
The basic disadvantages of intellectual property right are the reduced
consumer surplus and technological spillovers that follow when the
property right is in force. Basic economic theory (see, e.g., Takalo,2001,
for a summary) suggests that as a result of these trade-offs, there should
be an inverse-U shaped relationship between social welfare and the
strength of intellectual property protection.
Somewhat puzzlingly, however, to date there is little evidence that
stronger intellectual property generates more innovation (see. e.g., Boldrin and Levine, 2008 and Lerner, 2009b). As a necessary condition for
a welfare improving intellectual property policy is that it enhances incentives to innovate, this suggests that weaker intellectual property
rights would be optimal.
Over the recent decades economic research of intellectual property
has focused on cumulative innovation, which has produced a more nuanced view of the intellectual property system. On the positive side, the
intellectual property system has created a market for knowledge (for
evidence, see, Branstetter et al., 2006, Serrano, 2010 and Galasso et al.,
2013) that in some circumstances may have facilitated knowledge
transfers and financing of innovations. But the literature has documented another major draw-back of the intellectual property system: the
boundaries of intellectual property rights are inherently imprecise and
are ultimately defined by courts. From an innovator’s point of view this
leads to a threat of intellectual property disputes, which acts like as a
tax on innovation. As a result, the basic theoretical result of the positive
effect of stronger intellectual property on innovation may be overNordic Economic Policy Review
turned when innovation is cumulative and boundaries of intellectual
property imprecise (see, e.g., Bessen and Maskin, 2009), potentially explaining the puzzling empirical results. 9
Even when these more complex effects are acknowledged, stronger
intellectual property rights are hardly welfare improving. If anything,
recent empirical research suggests that social costs related to imprecise
boundaries of intellectual property rights are rising and, at least in the
US, may exceed the social benefits of the intellectual property system
(Jaffe and Lerner, 2004, Boldrin and Levine, 2008, Bessen and Meurer,
2008, and Turner et al., 2013). 10
For a small open economy, an optimal intellectual property system
would probably warrant strong intellectual property rights in the rest
of the world but weak intellectual property rights at home (Scotchmer
2004a). This would allow the country’s own citizens and firms to use
and experiment with innovations developed elsewhere more easily but
exporting firms would nonetheless have incentives thanks to strong intellectual property rights abroad. The drawbacks of the strong intellectual property rights would be borne by citizens and firms abroad.
Government funding of private R&D
Public funding of private R&D through subsidies, soft loans, and tax incentives is a widely used policy tool. OECD countries spent almost
USD 50 billion of taxpayers’ money on supporting private R&D in
2013. 11 Governments have also adopted more tools over time, especial9 Some recent empirical papers attempt to test the predictions of the basic theory of intellectual property,
and the theory of intellectual property with cumulative innovation separately: See Izhak et al. (2015) for the
basic theory, and Williams (2013), Sampat and Williams (2015), and Galasso and Schankerman (2015) for
cumulative innovation. The findings of these studies support those of the earlier ones: A positive causal
effect of stronger intellectual property on innovation is difficult to come by.
10 An important exception is Aghion et al. (2014) who show that countries with stronger intellectual property regimes may benefit more from reforms that enhance competition in the marketplace.
11 We arrive at this figure by multiplying Business Enterprise R&D (BERD) measures in 2010 PPP USD by
the percentage of BERD financed by government, obtained from OECD Main Science and Technology Indicators www-site (accessed 16 September 2015). The same figure for the 5 Nordic countries was a little over
USD 1billion.
Nordic Economic Policy Review
ly introducing R&D tax credits (e.g., Finland introduced tax credits for
2013–2014, and Sweden in 2014).
The basic mechanism of most of these support schemes is similar in
that the government pays some fraction of the marginal cost of R&D. 12
Lowering the marginal cost means that a supported firm invests more,
at least partially closing the gap between the privately and socially optimal levels of R&D. There is also a hope that additional finance by the
government would attract new firms to start R&D, but recent research
shows that existing policies merely lowering marginal costs of R&D are
not effective policy tools to this end (see Czarnitzki et al., 2015 and Lach
et al., 2015). Extrapolating the results from the literature on corporate
taxation (e.g., Devereux and Griffith, 1998), it is likely that average R&D
cost, and not the marginal one, is what matters for the firms’ discrete
decisions on whether to start investing in R&D or not.
These financial support policies have also important differences.
First, subsidies can be tailored for each project for which the government receives an application (for research that makes use of this, see
Takalo et al., 2013a), whereas every eligible firm can make a claim for
tax credits. 13 One thus needs to trade off the propensity of firms to apply and receive support with the government’s ability to tailor the support to the particular project. The application process for subsidies also
means that the government may become a focal point for information
on emerging agglomeration patterns.
Second, tax incentives in their purest form only work for firms that
are profitable and pay taxes. This severely hampers their effectiveness
in encouraging start-up innovation. Many countries like Norway and the
Netherlands have therefore resorted to “subsidy-like” tax incentives
where the R&D-performing firm gets what amounts to a discount on
labor-related social costs and taxes. A further problem with tax credits,
especially if they have a cap, is that a large part of government expendiThis is the case e.g. in all the European countries whose schemes we are familiar with.
In several European countries the probability of applying for an R&D subsidy is usually below 10%, and
below 20% even for R&D performing firms (see Czarnitzki et al. 2014). One should however note that the
uptake of R&D tax credits is not universal either. Busom et al. (2012) report a usage rate of less than 50% in
Spain for R&D – performing (i.e., eligible) firms, and in the Netherlands the usage percentage is round 80%
for firms with > 10 employees and round 40% for smaller firms (Verhoeken et al. 2012).
Nordic Economic Policy Review
ture consists of transfers to firms investing beyond the cap, with no incentive effect. 14 For this reason some countries (e.g., US) give tax credits
on incremental R&D. This in turn distorts firms’ investment decisions
over time.
Given the amounts of tax euros channeled to private sector R&D
through these policy tools, it is no surprise that a vast empirical literature studying their treatment effects exists. 15 Takalo et al. (2013b) emphasize that the extent to which government support increases private
R&D do not directly map into social benefits. The reason is that a firm
equates the private benefits of R&D with the marginal cost of R&D, but
ignores consumer surplus and knowledge spillovers. For example, a
small increase in an investment in an R&D project creating large consumer surplus and spillovers may be socially much more beneficial than
a large increase in R&D in a project with small (but still positive) consumer surplus or spillovers.
In small open economies, one should pay attention to the share of
consumer surplus and spillovers flowing outside the borders where
they do not benefit the local tax payers (Conti 2015 and Czarnitzki et al.,
2015). While existing policies typically impose restrictions on offshoring of government funded projects, the open-economy view could call
for more radical changes in policy-thinking. For example, if the outflows
of consumer surplus and spillovers constitute a large share of the welfare effects of R&D beyond private profits, private R&D without support
may be close to the socially optimal level from a national point of view.
For another example, while the standard theory suggests that R&D projects waiving (strong) intellectual property should be prioritized when
granting R&D subsidies, in a small open economy the argument is
weaker in the case of exporting firms. These open-economy considerations also suggest that the benefits from international coordination of
14 For example, both the Finnish R&D tax credit scheme (that was in place 2013–2014) and the Swedish one
introduced in 2014 have such a cap.
15 For literature surveys on the effects of R&D subsidies, see David et al. (2000), Klette et al. (2000), GarciáQuevedo (2004), Cerulli (2010), and Zúñica-Vicente et al. (2014), and on the effects of R&D tax incentives,
see Hall and van Reenen (2000), Mohnen and Lokshin (2010), and European Commission (2013).
Nordic Economic Policy Review
R&D support policies could be large (see Czarnitzki et al., 2015 for a
welfare comparison of national versus EU-wide support policies).
Other innovation policy tools
Prizes and contests are an old way of supporting innovation (see
Scotchmer, 2004b) but over the past century they have been relatively
little used. Using Maurer and Scotchmer’s (2004) classification of prize
types, targeted prizes are posted ex ante by a sponsor (e.g., a public
agency) who has identified a problem to be solved. The prize is awarded
to the first entity that solves the problem. For example, the Clay Mathematics Institute announced in 2000 a USD 1,000,000 prize for the first
solution for each of seven unsolved mathematical problems.
Blue-sky prizes are awarded ex post for innovations that the sponsor
considers valuable. A blue-sky prize could be granted in an ad hoc manner each time the sponsor observes a particularly valuable innovation,
or the sponsor can commit to grant the prize. The Nobel Prize is the
most well-known example of blue-sky prizes, and the Finnish Millenium
Technology Prize another. The incentive of effects of blue-sky prizes are
probably quite small, and they should be seen more as a marketing tool.
In contrast, targeted prizes could constitute an efficient innovation
policy tool. If the rewarded solution is put in the public domain for free
use, the prizes completely solve the ex post problem of diffusion of innovations. The problem with targeted prizes is that the sponsor should
know ex ante what should be invented.
Setting up contests for targeted prizes helps to aggregate information from innovators, as the sponsor can compare the proposals.
Modern information and communication technologies have enabled
both the public and the private sector to set up innovation prize platforms (such as where not only solutions but also problems are posted. Such crowdsourcing, another manifestation of the
changes brought by digitization, provides a new avenue to identify the
right problems for prizes and set up contests.
Another tricky task with prizes is to make sure that they reflect the
social value of innovations so that they are of proper size. Estimating a
proper size for a prize is difficult since this not only depends on the valNordic Economic Policy Review
ue of an innovation but also the costs of creating it. Kremer (1998) proposes an interesting public patent-buyout solution to the problem of
eliciting information: The patent authority could auction a patent right
and use information revealed by bids so as to give an appropriate reward to the patent applicant. To preserve incentives in the auction, a
patent grant should de facto be granted with a small probability, otherwise the invention could be put in the public domain. Shavell and Van
Ypersle (2001) propose a simpler, but less perfect, mechanism to relate
the size of prize to the value of innovation, reminiscent of the royaltybased licensing fees.
Being monetary rewards, prizes are vulnerable to misuse and ex
post opportunism (e.g., once the problem is solved, why should the
sponsor give the reward). 16 Furthermore, contests inherently involve
duplication of R&D costs when the participants race against each other
to obtain the prize.
Nonetheless, targeted prizes provide an underused tool of innovation
policy. For example, there are numerous diseases that are more prevalent
mostly in the Nordic countries. Posting a correctly designed prize would
be a simple means to complement (the small) market incentives.
Public procurement and production also provide tools for innovation
policy. Governments can provide services to complement private sector
innovation, work in partnerships with private entities, buy innovations
from private contractors, or directly produce innovations themselves.
Such public procurement and production of innovations and complementary services have been widely used thorough the economic history (see,
e.g., Scotchmer, 2004b and Mazzucato, 2013), but still may have some
untapped potential for innovation policy (Edler and Georghiou, 2007).
In theory, some public innovation support services, direct public production and procurement share the benefits and costs with targeted prizes. On the one hand, the ex ante incentives to innovate can be inefficient,
since the decision of what to invent and what information to produce is
made by the government. On the other hand, nothing prevents efficient
diffusion of innovations ex post. However, a part of public procurement
A classic example of these problems is the Longitude prize (see., e.g, Sobel, 1995).
Nordic Economic Policy Review
and production is concentrated on nationally strategic sectors such as
defense with the purpose of minimizing the diffusion of research results.
Promotion of research joint ventures (RJVs) and other forms of R&D
cooperation is a widely used tool of innovation policy in industrialized
countries. RJVs allow participating firms to internalize technological
spillovers and thereby they should enhance R&D efforts. Therefore, RJVs
are, for example, prioritized in subsidy allocation decisions in several
countries, and constitute a block exemption under the EU competition
law. There is some evidence (e.g., Branstetter and Sakakibara, 2002) that
RJVs have the stated beneficial effects in enhancing spillovers and R&D
efforts. There is however also evidence that RJVs are primarily motivated
by cost sharing (Röller et al., 2007) and lead to product market collusion
(e.g., Hellman and Sovinsky, 2010 and Duso et al., 2014).
Indirect innovation policies
There is rather little robust empirical evidence on the relation between
education and innovation. 17 One exception is Toivanen and Väänänen
(2015) who find a positive causal impact of education on invention. This
suggests that indeed, a policy reaction to Jones’s (2005) advice of “having more inventors in order to become richer” as a society is to increase
investments in (engineering) higher education.
A key insight from innovation research is the skewed distribution of
innovative outcomes, with a low median but a high mean value of innovations (e.g. Pakes, 1986 and Lanjouw, 1998). To us, this seems to call for an
education system that generates a wide-skill base and allows different
skills to be combined in possibly unexpected ways, i.e., an education system that encourages individuals to acquire a variety of skills and allows
individuals with specialized skills to easily match with each other.
There is a very large literature on the causal effect of education on individual wages.
Nordic Economic Policy Review
It is well known that innovative activity is concentrated geographically and that high-quality universities play a central role in this agglomeration process (see Audretch and Feldman, 1996, for a seminal
paper, and Carlino and Kerr, 2015, for a survey of the empirical evidence). Top universities contribute to the agglomeration of innovative
activity in many ways. One important channel is the supply of educated
individuals on which innovative activity depends: For example, Moretti
(2004) finds a 0.5 percentage point increase in the plant-level productivity as the consequence of a 1 percentage increase in the share of college graduates in the population of a metropolitan area in the US.
As small open economies, the Nordic countries greatly benefit from
the knowledge and innovations created elsewhere. While innovation
continues to exhibit locational economies of scale also in future, digitization and modern ICT are making knowledge flows less dependent on
geography (Griffith et al., 2011), suggesting a crucial role for education
in enhancing absorptive capacity of the countries.
Basic research
There is plenty of anecdotal evidence of successful private sector innovations that are based on research in government funded laboratories
and universities, often without a direct commercial objective in mind
(see, e.g., Mazzucato, 2013). But just as in the case of education, there is
little in terms of rigorous causal evidence. 18 Basic research done at
high-quality universities is a source of significant local knowledge spillovers to the private sector (e.g., Jaffe, 1989, Breschi et al., 2006, and Carlino and Kerr, 2015). As innovative firms seek to benefit from these
spillovers, they locate close to universities (e.g., Jaffe, 1989, Anselin et
al., 1997, and Abramovsky et al., 2007). This forms another important
channel through which universities contribute to agglomeration of innovation (Carlino and Kerr, 2015). In small open economies in particu-
18 Sveikauskas (2007) offers a survey of the scant literature, and Hausmann (2012) and Akcigit et al. (2014)
recent contributions.
Nordic Economic Policy Review
lar, one should not discount the importance of high-quality basic research as a pull-factor of foreign R&D (e.g. Belderbos et al., 2014).
Any government needs to make decisions on how to allocate the resources devoted to basic research. Despite difficulties created by incomplete information, the government may well be in a position to
make high-level decisions regarding allocation of resources across different fields of basic research (e.g., health vs. environment). But it
should delegate resource allocation decision-making within research
fields to its leading experts and allow, through that same system, reallocation across fields as a function of outcomes. Such a bottom-up approach would hopefully lead to a limited number of large, active research centres within each field that would compete against each other
for top researches and funds. This should not only improve the quality
of basic research but also seed up commercialization of that research
(Goldfarb and Henrekson, 2003).
The principles of good (corporate) taxation (see Mirrlees et al., 2011,
pp. 22) minimizes negative effects on welfare and economic efficiency,
has low administrative costs; is distributionally fair, and transparent. In
cases where production or consumption of goods and services is associated with large externalities, it is theoretically justified to make exceptions to these principles. However, in practice corporate tax incentive
schemes tend to become complex and unpredictable and increase tax
planning and avoidance (see Mirrlees et al., 2011). 19 If tax incentives
are used to as an innovation policy tool they should be simple, and focused on innovation or their financing incentives directly. As also concluded by the European Commission (2013), it is therefore much easier
to justify, say, R&D tax credits rather than, say, IPR boxes from an innovation policy point of view.
19 For example, in Finland corporate taxation changes almost annually (e.g., R&D tax credits were in force in
2013–2014, and business angel tax relief was introduced for years 2013–2015).
Nordic Economic Policy Review
Just as there is evidence of countries competing in terms of the level
of corporate taxation (Devereux et al., 2008), they are also likely to use
various R&D incentives for the same purpose. In particular, competition
for intellectual property revenues is tempting since intangible assets
are relatively easy to reallocate from one location to another based on
tax considerations (see, e.g., European Commission, 2013 and Griffith et
al., 2014). In our view, introduction of IPR boxes at best amounts to a
Prisoner’s Dilemma – game among countries where the detrimental
Nash equilibrium should be avoided by international cooperation.
In contrast, tax competition for innovative corporations and individual inventors might be beneficial for the Nordic countries: Danish
evidence (Kleven et al., 2014) suggests that small open economies with
relatively homogenous populations may benefit substantially from tax
schemes that give temporary preferential treatment to foreign highskilled individuals. Akcigit et al. (2014) find that top-inventors are sensitive to top income tax rates in choosing where to locate. Taxation of
individual inventors should also affect their incentives and individuals’
career choices.
Similarly, the effects of (average) corporate taxation are larger at
the extensive margin than at the intensive margin: the possibility to
make money is one of the key drivers of (high-growth) entrepreneurship (Lerner, 2009a and Isenberg, 2013). The example of earlier successful entrepreneurs and their role as business angels are vital in the
creation of a culture of entrepreneurship and risk-taking. However, a
large gap between corporate and personal tax rates is conducive for tax
planning and avoidance efforts, and successful entrepreneurs and associated capital gains will almost by definition increase income inequality.
Other indirect innovation policies
Besides the policies listed above there is a variety regulatory policies
that have a significant impact on innovative activity. We discuss briefly
here some selected regulatory policies.
Competition policy is an important part of an innovation infrastructure (Shapiro, 2002, Encaoua and Hollander, 2002, and Segal and Whinston, 2007). According to an extensive literature, there appears to be an
Nordic Economic Policy Review
inverse-U relationship between market structure and innovation activity created by two opposing forces: On the one hand, competition is bad
for innovation since it reduces the returns to successful innovation; on
the other hand, competition is conducive for innovations since it forces
the firms to innovate so as to escape competition. 20 This suggests that
liberalization of protected and regulated industries might promote innovation. Intensified competition in an upstream industry may also increase innovation in a downstream industry. For example, liberalization
of financial services sector not only generated frantic innovation in the
industry itself but also in the real sector (Amore et al., 2013 and Chava
et al., 2013). 21
Trade policy matters for innovation for several reasons. In particular, countries that are open to trade will reap a larger part of international knowledge spillovers and the potentially greatest benefit of innovation investments made elsewhere: new goods and services. While this
is uncontroversial, we need to understand much better what shapes international knowledge flows. For example, cultural aspects such as ethnicity may importantly shape international knowledge flows (Kerr,
2007). Trade also increases competition, thereby possibly changing incentives to innovate (Bloom et al., 2015).
In general, the beneficial effects of enhanced competition and trade
openness on innovation appear to be the largest in countries like the
Nordic ones where firms are closer to technological frontier and where
corruption does not distort competition (Dabla-Norris et al., 2013 and
Aghion et al., 2013).
From the innovation policy point of view, well-functioning labor
markets would encourage risk taking and reallocate labor from declining industries and regions to rising ones. Also the efficiency of direct
innovation policy tools may depend on the functioning of labor markets.
For example, R&D subsidies and tax credit may affect only the wages of
20 The classic references are Kamien and Schwartz (1975) and Aghion et al. (2005). Kilponen and Santavirta
(2007) document the existence of the inverse-U relationship in Finland. However, Hashmi (2013) finds a
negative relationship between the intensity of competition and innovation in the US.
21 Some financial innovations clearly generated negative externalities.
Nordic Economic Policy Review
R&D personnel if the supply of R&D personnel is inflexible. (e.g., Goolsbee, 1998 and Wolff and Reinthaler, 2008).
Unfortunately the empirical literature on the relation between labor
market regulations and innovation is rather unsettled. On the one hand,
the Danish-type flexicurity with relatively weak employment protection
but relatively high unemployment benefits might be particularly conducive for start-up formation and radical innovation but, on the other
hand, weak employment protection may deteriorate employees’ incentives to innovate in established corporations (see, e.g., Acharya et al.,
2013, Bozkaya and Kerr, 2013, and Griffith and Macartney, 2014 for different results).
As mentioned in the introduction, financial market imperfections constitute an important rationale for an active innovation policy. R&D activities are inherently opaque, human capital intensive, and involve soft information. As a result, innovative start-ups have difficulties to access to
outside finance due to informational asymmetries and lack of collateralizable assets (Hall and Lerner, 2010 and Kerr and Nanda, 2014).
It is notoriously difficult to identify the existence of such financial
constraints (see Hall and Lerner, 2010 for various empirical strategies):
The fact that some firms suffer from lack of finance may just indicate the
financial markets work as they should, and are denying funding of bad
projects. Furthermore, even in theory it is difficult to identify the right
policy response to these financial market imperfections: Informational
asymmetries may even lead to overfinancing, which would call for a punitive taxation of start-up finance (e.g., de Meza and Webb, 1987, Boadway and Keen, 2005, and Takalo and Toivanen, 2013). 22 Despite
these challenges, two broad conclusions emerge. First, bank lending
remains an important source of outside finance, even for start-ups
(Robb and Robinson, 2014, Kerr and Nanda, 2014). Bank lending and
associated credit constraints are also procyclical (Aghion et al., 2012).
Second, the evidence suggests that private sector equity investing is
conducive for innovation (Hall and Lerner, 2010 and Kaplan and Lerner,
22 As illustrated by the dot-com boom and bust at the turn of the millennium, and ongoing financial and
economic crisis that begun from the US subprime mortgage markets, this kind of over-financing is not just a
theoretical curiosity, and may have severe macroeconomic consequences.
Nordic Economic Policy Review
2010). Equity investors have both incentives and human capital for ex
ante screening, interim monitoring and value-enhancing advice Furthermore, because innovative investments are complex and risky, optimal financing contracts become complex, too: Investors need to have
both a share of upside returns in case of a success and control rights in
case of a failure (Kaplan and Strömberg, 2003). Whether private sector
equity financing markets work efficiently or not appear to matter more
for countries close to technological frontier, such as the Nordic countries (Aghion and Mayer-Foulkes, 2005 and Dabla-Norris et al., 2013).
Based on these conclusions, there seems to be a case for policies
that improve early-stage equity financing in the Nordic countries. But
the right policy is hardly based on public equity investing in commercial
projects. Rather one should create the right environment for private
sector equity investors. More generally, if there is need for public innovation finance beyond R&D subsidies, the public sector should not mimic private innovation finance but invest differently, operating when liquidity in financial markets dries up and focusing on projects where the
ratio of social returns to private returns is high.
Besides the many issues discussed above (e.g., taxation, education,
basic research, and labour markets), the legal environment matters for
private sector investors, For example, Hyytinen et al. (2003) show how
a strengthening of the Finnish investor protection legislation enhanced
the role of equity finance in the Finnish corporate finance environment.
But again, identifying the right policies to improve legislation is not
easy. For example, while a lenient bankruptcy legislation clearly encourages entrepreneurial risk-taking by reducing the cost of failure, it also
discourages financing of entrepreneurship. The evidence on which of the
two opposing effects dominates remains inconclusive (see, e.g., Acharya
and Subramanian 2009, and Cerqueiro et al. 2014, for conflicting results)
and likely depend on the institutional context. There is little research on
the Nordic countries regarding the matter, but Koskinen et al. (2007) find
that weakening of strong creditor rights in corporate bankruptcy in Finland boosted corporate investments and firm valuations.
Nordic Economic Policy Review
The wide consensus that innovation is important for economic growth
and thus human well-being is based on a solid theoretical and empirical
basis. The theoretical basis for innovation policy is also solid: Because of
consumer surplus and technological spillovers which are not captured
by innovating firms and individuals, there is too little innovative activity
in the private sector. Unfortunately the empirical knowledge of the efficacy of different innovation related policies remains controversial.
The central feature of innovation is uncertainty, and sound innovation
policies acknowledge the limited ability of even the best-informed agents to
make good choices in tomorrow’s increasingly digitized, interconnected
world. To us, this clearly suggests an emphasis for a bottom-up approach,
rather than vice versa, where resources flow to those sectors and regions
that show signs of success. It may well be that the best governmental innovation policies are the least headline-grabbing ones, focusing on building
the right infrastructure for better informed agents with stronger incentives,
be they academic researchers or corporate inventors.
But in building the better innovation infrastructure the governments
should be bold. The Nordic countries stand to continue to do well but only if they maintain and improve a good basic education system, highquality universities, and an open, competitive and sufficiently flexible environment that enables experimentation and growth of those who succeed in innovation. Sparse public resources can be used more efficiently if
the local economies of scale in innovation are better recognized. Since a
large part of consumer surplus and knowledge spillovers generated by
innovations coming from a Nordic country almost by definition reside
elsewhere, the role for more targeted national innovation policies may be
more limited than has been thought previously: as examples, domestic
intellectual property protection might optimally be weak instead of
strong, and direct support to private R&D should be targeted to those
projects generating high domestic spillovers, including consumer surplus,
rather than those projects aiming at conquering the world.
Nordic Economic Policy Review
Abramovsky, L., Harrison, R. & Simpson, H. (2007). University Research and the Location of Business R&D, Economic Journal, 117, C114-C141.
Acemoglu, D. & Robinson, J. (2012). Why Nations Fail? Deckle Edge.
Acharya, V., a& Subramarian, R. K. (2013). Bankruptcy Codes and Innovation.
Review of Financial Studies, 22, 4949–4988.
Acharya, V., Baghai-Wadji, R. & Subramarian, K. (2013). Labor Laws and Innovation.
Journal of Law and Economics, 53, 997–1037.
Aghion, P., Akcigit, U., & Howitt, P. (2013). What Do We Learn From Schumpeterian
Growth Theory? NBER Working Paper No. 18824.
Aghion, P., Askenazy, P., Berman, N., Cette, G., & Eymard, L. (2012). Credit Constraints and the Cyclicality of R&D Investment: Evidence from France. Journal of
European Economic Association, 10, 1001–1024.
Aghion, P., Bloom, N. Blundell, R., Griffith, R. & Howitt, P. (2005). Competition and Innovation: An Inverted-U relationship. Quarterly Journal of Economics 120: 701–728.
Aghion, P. & Howitt, P. (2009). The Economics of Growth. Cambridge, MA: MIT Press.
Aghion, P., Howitt, P., & Prantl, S. (2014). Patent Rights, Product Market Reforms, and
Innovation. mimeo.
Aghion, P. & Mayer-Foulkes, D. (2005). The Effect of Financial Development on Convergence: Theory and Evidence. Quarterly Journal of Economics, 120, 173–222.
Akcigit, U., Baslandze, S. & Stantcheva, S. (2014). Taxation and International Mobility
of Inventors. NBER wp. 21024.
Akcigit, U., Hanley, D., & Serrano-Velarde, N. (2014). Back to Basics: Basic Research
Spillovers, Innovation Policy and Growth. NBER wp. 19473.
Amore, M., Schneider, C. & Zaldokas, A. (2013). Credit Supply and Corporate Innovations. Journal of Financial Economics 109, 835–855.
Anselin, L., Varga, A., & Zoltan, A. (1997). Local Geographic Spillovers between University Research and High Technology Education. Journal of Urban Economics, 42,
Arrow, K. .J. (1962). Economic Welfare and the Allocation of Resources for Invention. In Nelson, R. R. (ed.), The Rate and Direction of Inventive Activity: Economic
and Social Factors, pp. 609–625. Princeton, N.J.: Princeton University Press.
Atkinson, A. B. (2015). Inequality: What Can Be Done? Cambridge, MA: Harvard University Press.
Audretsch, D. & Feldman, M. (1996). R&D Spillovers and the Geography of Innovation and Production. The American Economic Review, 86, 630–640.
Nordic Economic Policy Review
Belderbos, R., Van Roy, V., Leten, B., & Thijs, B. (2014). Academic Research Strengths
and Multinational Firms’ Foreign R&D Location Decisions: Evidence from R&D Investments in European Regions. Environment and Planning A., 46, 9290–042.
Bessen, J. (2015). Learning by Doing: The Real Connection between Innovation, Wages, and Wealth. New Haven, CT: Yale University Press.
Bessen, J. & Maskin, E. (2009). Sequential Innovation, Patents and Imitation, RAND
Journal of Economics, 40, 611–635.
Bessen, J. & Meurer, M. J. (2008). Patent Failure: How Judges, Bureaucrats and Lawyers Put Innovators at Risk. Princeton, NJ: Princeton University Press.
Bloom, N., Draca, M., & Van Reenen, J. (2015). Trade Induced Technical Change: The
Impact of Chinese Imports on Innovation, Diffusion and Productivity. Review of
Economic Studies, forthcoming.
Boadway, R. & Keen, M, (2006). Financing and Taxing New Firms under Asymmetric
Information. FinanzArchiv, 62, 471–502.
Boldrin, M. & Levine, D. K. (2013). The Case against Patents. Journal of Economic
Perspectives 27: 3–22.
Bozkaya, A., & Kerr, W. R. (2013). Labor Regulations and European Venture Capital.
Journal of Economics & Management Strategy 23, 776–810.
Branstetter, L., Fisman, R., & Foley, F. (2006). Do Stronger Intellectual Property
Rights Increase International Technology Transfer? Empirical Evidence from U.S.
Firm-Level Panel Data. Quarterly Journal of Economics, 121, 321–349.
Branstetter, L., & Sakakibara, M. (2002). When Do Research Consortio Work Well
and Why? Evidence from Japanese Panel Data. American Economic Review, 92,
Breschi, S., Tarasconi, G., Catalini, C., Novella, L., Guatta, P., & Johnson, H. (2006).
Highly Cited Patents, Highly Cited Publications, and Research Networks, mimeo.
Brynjolfsson, E., & McAfee, A. (2015). The Second Machine Age: Work, Progress and
Prosperity in a Time of Brilliant Technologies.
Busom, I., B. Corchuelo, & Martinez Ros, E. (2012). Tax Incentives or Subsidies for
R&D. UNU-MERIT Working Paper Series 2012–056.
Carlino, G., & Kerr, W. R. (2015). Agglomeration and Innovation. In G. Duranton, V.
Henderson, and W. C. Strange (eds.), Handbook of Regional and Urban Economics,
5, 349–404. Amsterdam: Elsevier.
Chava, S., Oettl, A., Subramanian, A., & Subramanian, K. (2013). Banking Deregulation and Innovation. Journal of Financial Economics 109: 759–774.
Nordic Economic Policy Review
Czarnitzki, D., E. Huergo, M. Köhler, P. Mohnen, S. Pacher, T. Takalo, & Toivanen, O.
(2015). Welfare Effects of European R&D Support Policies. mimeo.
Cerqueiro, G. D. Hegde, M. F. Penas, & Seamans, R. C. (2014). Debtor Rights, Credit
Supply, and Innovation. TILEC Discussion Paper No. 2014–011.
Cerulli, G. (2010). Modelling and measuring the effect of public subsidies on business R&D: a critical review of the economic literature. Economic Record 86,
pp. 421–449.
Conti, A. (2015). Evaluating Government R&D Grants to Startups: The Case of a Small
Open Economy. mimeo.
Cowan, R. & Zinovyeva, N. (2013). University Effect on Regional Innovation.
Research Policy, 42, 788–800.
Dabla-Norris, E., Ho, G., & Kyobe, A. (2013). Reforms and Distance to Frontier. IMF
Technical Note to Staff Discussion Note 13/08. Washington, D.C.: International
Monetary Fund.
David, B. Hall, & Toole, A. (2000). Is public R&D a complement or a substitute for
private R&D? A review of the econometric evidence. Research Policy 29, pp. 497–
De Meza, D. & Webb, D. C. (1987). Too Much Investment: A Problem of Asymmetric
Information. Quarterly Journal of Economics, 102, 281–292.
Devereux, M. & Griffith, R. (1998). Taxes and the Location of Production: Evidence
from a Panel of US Multinationals. Journal of Public Economics, 68, 335–367.
Devereux, M., Lockwood, B., & Redoano, M. (2008). Do Countries Compete over Corporate Tax Rates? Journal of Public Economics, 92, 1210–1235.
Duso, T., Röller, L.-H., & Seldeslachts, J. (2014). Collusion through Joint R&D: An Empirical Assessment. Review of Economics and Statistics, 96, 349–370.
Edler, J. & Georghiou, L. (2007). Public Procurement and Innovation – Resurrecting
the Demand Side. Research Policy, 36, 949–963.
Encaoua, D. & Hollander, A. (2002). Competition Policy and Innovation. Oxford Review of Economic Policy, 18, 63–79.
European Commission (2013). A Study on R&D Tax Incentives. Final Report. TAXUD/2013/DE/315. FWC No. TAXUD/2010/CC/104.
Galasso, A. & Schankerman, M. (2015). Patents and Cumulative Innovation: Causal
Evidence from the Courts. Quarterly Journal of Economics, 130, 317–369.
Galasso, A., Schankerman, M., & Serrano, C. J. (2013). Trading and Enforcing Patent
Rights. RAND Journal of Economics, 44, 275–312.
Nordic Economic Policy Review
Galasso, A. & Schankerman, M. (2015). Patents and Cumulative Innovation: Causal
Evidence from the Courts. Quarterly Journal of Economics, 317–269.
Garciá-Quevedo, J. (2004). Do public subsidies complement business R&D? A metaanalysis of the econometric evidence. KYKLOS 57, pp. 87–102.
Goldfarb, B. & Henrekson, M. (2003). Bottom-up versus Top-down Policies towards
the Commercialization of University Intellectual Property. Research Policy, 32,
Goldfarb, A., Greenstein, S., & Tucker, C. (eds.), (2015). The Economic Analysis of the
Digital Economy. Chicago, IL: Chicago University Press. Edward Elgar Publishing.
Goolsbee, A. (1998) Does Government R&D Policy Mainly Benefit Scientists and Engineers? American Economic Review, 88, 298–302.
Greenstein, S., Goldfarb, A., & Tucker, C. (eds.), (2013) The Economics of Digitization.
Cheltenham, UK: Edward Elgar Publishing.
Griffith, R., Lee, S. & Van Reenen, J. (2011). Is Distance Dying at last? Falling Home
Bias in Fixed Effects Models of Patent Citations. Quantitative Economics 2, 211–
Griffith, R., & Macartney, G. (2014). Employment Protection Legislation, Multinational
Firms and Innovation. Review of Economics and Statistics, 96, 135–150.
Griffith, R., H. Miller & O’Connell, M. (2014). Ownership of intellectual property and
corporate taxation. Journal of Public Economics, vol. 112, pp. 12–23.
Hall, B. & Lerner, J. (2010). The Financing of R&D and Innovation. In B. Hall & N.
Rosenberg (eds.) Handbook of the Economics of Innovation, 609–639. Amsterdam,
Hall, B. & Van Reenen, J. (2000). How effective are fiscal incentives for R&D? A review of the evidence. Research Policy 29, pp. 449–469.
Hashmi, A. R. (2013). Competition and Innovation: The Inverted-U Relationship Revised. Review of Economics and Statistics, forthcoming.
Hausmann, N. (2012). University Innovation, Local Economic Growth, and Innovation.
US Census Bureau Center for Economics Studies Paper No. CES-WP-12-1.
Helland, E. & Sovinsky (Goeree) M. (2010). Do Research Joint Ventures Serve a Collusive Function, mimeo. UZH.
Nordic Economic Policy Review
Holmstrom, B. & Myerson, R. (1983). Efficient and Durable Decision Rules with Incomplete Information. Econometrica, 51, 1799–1820.
Hyytinen, A., Kuosa, I., & Takalo, T., (2003). Law or Finance: Evidence from Finland.
European Journal of Law and Economics, 16, 59–89.
Isenberg, D. (2013). Worthless, Impossible and Stupid: How Contrarian Entrepreneurs
Create and Capture Extraordinary Value. Harvard Business School Publishing,
Boston, MA.
Izhak, O., Saxell, T., & Takalo, T. (2015). Patent Duration, Breadth, and Costly
Imitation: Evidence from the US Pharmaceutical Market. mimeo.
Jaffe, A. (1989). Real Effects of Academic Research. American Economic Review, 79,
Jaffe, A. B. & Lerner, J. (2004). Innovations and Its Discontents: How Our Broken Patent System is Endangering Innovation and Progress, and What to Do About It.
Princeton, NJ: Princeton University Press.
Jones, C. (2005). Growth & Ideas, ch. 16 in Handbook of Economic Growth 1B, Aghion P. & S. Durlauf (eds.), Elsevier B.V.
Jones, C. & J. Williams, (1998). Measuring the Social Return to R&D. Quarterly Journal of Economics, 113, 1119–1135.
Kamien, M. I. & Schwartz, N. L. (1975). Market structure and innovation: A survey.
Journal of Economic Literature 13: 1–37.
Kaplan, S. N. & Lerner, J. (2010). It Ain’t Broke: The Past, Present, and Future of Venture Capital. Journal of Applied Corporate Finance, 22, 1–12.
Kaplan, S. N. & Strömberg, P. (2003). Financial Contracting Theory Meets the Real
World: An Empirical Analysis of Venture Capital Contracts. Review of Economic
Studies, 70, 281–315.
Kerr, W. R. (2007). Ethnic Scientific Communities and International Technology Diffusion. Review of Economics and Statistics, 90, 518–537.,
Kerr, W. R, & Nanda, R. (2014). Financing Innovation. Forthcoming in Annual Review
of Financial Economics.
Klette, T. J., Møen, J., & Griliches, Z. (2000). Do Subsidies to Commercial R&D Reduce
Market Failures? Microeconomic Evaluation Studies. Research Policy 29 (4–5):
Kleven, H., Landais, C., Saez. E. & Schultz, E. (2014). Migration and Wage Effects of
Taxing Top Earners: Evidence from the Foreigners’ Tax Scheme in Denmark. Quarterly Journal of Economics, 333–278.
Koskinen, Y., Korkeamäki, T., & Takalo, T. (2007). Phoenix Rising: Legal Reforms and
Changes in Valuations in Finland during the Economic Crisis. Journal of Financial
Stability 3, 33–58.
Nordic Economic Policy Review
Kremer, M. (1998). Patent buyouts: A Mechanism for Encouraging Innovation. Quarterly Journal of Economics 113, 1137–1167.
Lach, S., Neeman, Z., & Schankerman, M. (2015). Government Funding for R&D StartUps, mimeo.
Lanjouw, J. (1998). Patent Protection in the Shadow of Infringement: Simulation
Estimations of Patent Values. Review of Economic Studies, 65, 671–710.
Lerner, J. (2009a). Boulevard of Broken Dreams. Why Public Efforts to Boost Entrepreneurship and Venture Capital Have Failed – and What to Do about It? Princeton
University Press, Princeton, N.J.
Lerner, J. (2009b). The Empirical Impact of Intellectual Property Rights on Innovation: Puzzles and Clues. American Economic Review (Papers & Proceedings), 99,
Mazzucato, M. (2013). The Entrepreneurial State, Anthem Press.
Maurer, S.M. & Scotchmer, S. (2004). Procuring Knowledge. Advances in the Study of
Entrepreneurship, Innovation and Growth, 15, 1–31, edited by G. Libecap. Amsterdam: Elsevier.
Menell, P. S. & Scotchmer, S. (2007). Intellectual Property. In M. Polinsky and S.
Shavell (eds.), Handbook of Law and Economics. Amsterdam: Elsevier.
Mirrlees, J., Adam, S., Besley, T., Blundell, R., Bond, S., Choate, R., Gammie, M., Johnson, P., Myles, G., & Poterba, J. (2011). Tax by Design, Oxford University Press, ISBN
Mohnen, P. & Lokshin, B. (2010). What does it take for an R&D tax incentive policy to be
effective? In Ghosal, V. (ed.), Reforming Rules and Regulations, pp. 33–58. Cambridge,
MA: MIT Press.
Moretti, E. (2004). Workers’ Education, Spillovers, and Productivity: Evidence from
Plant-Level Production Functions. American Economic Review, Vol. 94, No. 3, 656–
Nelson, R.R. (2009). Building Effective “Innovation Systems” versus Dealing with
“Market Failures” as Ways of Thinking about Technology Policy. In Dominique
Foray (ed.), The New Economics of Innovation Policy, Edward Elgar: Cheltenham,
Nelson, R.R. (1959). The simple economics of basic scientific research. Journal of
Political Economy 49, pp. 297–306.
OECD, (2011). OECD Science, Technology and Industry Scoreboard 2011. OECD Publishing.
Pakes, A. (1986). Patents as Options: Some Estimates of the Value of Holding European Patent Stocks. Econometrica, 54, 755–784.
Rob, A. & Robinson, D. (2012). The Capital Structure Decisions of New Firms.
Review of Financial Studies, 27, 153–179.
Nordic Economic Policy Review
Röller, L.-H., Siebert, R., & Tombak, M. (2007). Why Firms Form (or Do not Form)
RJVs? Economic Journal, 117, 1122–1144.
Sampat, B. & Williams, H. (2015). How Do Patents Affect Follow on Innovation? Evidence from the Human Genome. NBER Working Paper No. 21666.
Schankerman, M. & Serrano, C. (2010). The Dynamics of the Transfer and Renewal
of Patents. RAND Journal of Economics, 41, 686–708.
Scotchmer, S. (2004a). The Political Economy of Intellectual Property Treaties.
Journal of Law, Economics, and Organizations 20, 415–437
Scotchmer, S. (2004b). Innovation and Incentives. Cambridge, MA: MIT Press.
Segal, I. & Whinston. M. D. (2007) Antitrust in Innovative Industries. American Economic Review, 97, 1703–1730.
Sobel, D. (1995). Longitude: The True Story of a Lone Genius Who Solved the Greatest
Scientific Problem of His Time. New York, NY: Walker.
Sveikauskas, L. (2007). R&D and Productivity Growth: A Review of the Literature.
Bureau of Labor Statistics Working Paper 408.
Shapiro, C. (2002). Competition Policy and Innovation. OECD Science, Technology
and Industry Working Papers, 2002/11, OECD Publishing.
Shavell, S. & T. van Ypserele (2001). Rewards versus Intellectual Property Rights.
Journal of Law and Economics, 44, 525–547.
Serrano, C. J. (2010). The Dynamics of the Transfer and Renewal of Patents.
RAND Journal of Economics, 41, 686–708.
Takalo, T. (2001). On the Optimal Patent Policy, Finnish Economic Papers, 14, 33–40.
Takalo, T., Tanayama, T. and Toivanen, O. (2013a). Estimating the benefits of targeted R&D subsidies, The Review of Economics and Statistics, 95, 55–272.
Takalo, T., Tanayama, T., & Toivanen, O. (2013b). Market Failures and the Additionality Effects of Public Support to Private R&D: Theory and Empirical Implications.
International Journal of Industrial Organization. 31, 634–642.
Takalo, T. & Toivanen, O. (2012). Entrepreneurship, Financiership, and Selection,
Scandinavian Journal of Economics, 114, 601–628.
Toivanen, O. & Väänänen, L. (2015). Education and Invention, Review of Economics
and Statistics, forthcoming.
Turner, J. L., Bessen, J., Neuhäusler, P., & Williams, J. W. (2013). The Costs and Benefits of the United States Patents, Boston University School of Law, Law and Economics Research Papers No. 13–2014.
Verhoeven, W., van Stel, A., & Timmermans, N. (2012). Evaluate WBSO 2006-2010:
Effecten. Doelgroepbereik en uitvoering.
Nordic Economic Policy Review
Williams, H. (2013). Intellectual Property Rights and Innovation: Evidence from the
Human Genome, Journal of Political Economy, 121, 1–27.
Wolff, G. B. & Reinthaler, V. (2008). The effectiveness of subsidies revisited: Accounting for wage and employment effects in business R&D, Research Policy, 37,
Zúñica-Vicente, J., Alonso-Borrego, C., Forcadell, F., & Galan, J. (2014). Assessing the
Effect of Public Subsidies on Firm R&D Investment: a Survey. Journal of Economic
Surveys, 18, 36–67.
Nordic Economic Policy Review
5. Taxing mobile capital and
profits: The Nordic Welfare
Guttorm Schjelderup, 1 Norwegian School of Economics and Norwegian
Center for Taxation, e-mail: guttorm.schjeld[email protected]
This paper discusses trends in capital taxation and the role of the corporate tax rate in a welfare state. It provides a summary of the tax competition literature with special application to capital taxation in small
versus large countries. A main finding from this literature is that small
countries set lower taxes on capital than large countries. In line with
this prediction the paper shows that the Nordic countries undertook tax
reforms in the 1990s, which lead to lower ratios of statutory corporate
to wage taxes than in most OECD countries. The second part of the paper is devoted to tax base erosion by multinationals and how to combat
it. Finally, the paper offers some concluding remarks on redistribution
and the pressures of tax competition.
1 I am grateful for comments by Rolf Aarberge, Jarle Møen, Jørn Rattsø and Dirk Schindler. Financial
support from the Norwegian Tax Administration and the Norwegian Research Council is gratefully
The Nordic countries have traditionally been characterized by an extensive welfare state, a homogenous population and labour force, and redistributive taxation. This has changed in recent years. Increased immigration, an aging population, and competition for capital among countries have put pressures on public finances and the welfare state. These
changes can be attributed to the globalization process whereby national
economies become more integrated. Economic integration takes place
in terms of increasing factor mobility, in particular mobility of capital,
and rising volumes of trade in goods and services. Globalization has
costs and benefits. On the one hand, globalization leads to a more efficient allocation of worldwide resources and thus to higher output and
growth. On the other hand, globalization and free capital mobility disrupts employment patterns, makes incomes more volatile, and threatens the government’s ability to redistribute income and to provide public services.
An argument frequently used by political lobby groups is that with
free capital mobility corporations shouldn’t be taxed at all and that taxing investment income is actually bad for workers. The argument is that
if you cut taxes on investment income, more investment is encouraged.
More investment means people have more equipment and technology
to work with, which should increase the productivity of labour and thus
wages and economic growth. Put differently, a tax on mobile capital
would lead to an outflow of capital that would cause wages to fall; effectively shifting the full burden of the tax on capital onto workers. It is
then better to tax workers directly and levy a zero tax on capital.
The argument above relies on strong assumptions, among them that
labour is immobile and cannot evade taxation, that there are no country
specific rents, and that domestic firms are not owned in part by foreigners. If domestic firms, say, are partly owned by foreigners, taxing capital
would imply that some of the tax burden is shifted onto foreigners and
that part of the welfare state is then financed by foreigners. This alone
may warrant a positive tax on investment capital (see Huizinga and
Nielsen 1997). Industrial agglomeration also modifies the zero-tax results. If industrial agglomeration is concentrated in one single country, a
Nordic Economic Policy Review
government may, through a positive source tax on capital, be able to
exploit the locational rent created by agglomeration forces and thus increase welfare. 2 The zero tax on capital result is also difficult to confirm
empirically. Yagan (2014), for example, studies the effects of the 2003
dividend tax cut in the US. He finds that it caused zero change in corporate investment in U.S. unlisted firms and that it had no impact on employee compensation. It did, however, have an immediate impact on
financial pay-outs to shareholders. 3 Alstadsæter et al. (2015) use Swedish panel data for unlisted firms and find that the Swedish 2006 dividend tax cut did not affect aggregate investment but that it affected the
allocation of corporate investment. In particular, they find that relative
to cash-rich firms, cash-constrained firms increased their investments
after the dividend tax cut. 4
In fact, there are good reasons to tax capital income at the corporate
level. An important reason is that the corporate tax plays an essential
withholding function, acting as a “backstop” to the personal income tax.
If a country abolished the corporate tax rate, wealthy individuals in particular would be given an incentive to reclassify their labour earnings as
corporate income, typically using offshore corporate structures and escape tax. The corporate tax might also be needed to avoid excessive income shifting between labour income and capital income. Finally, the
corporate tax also acts as a withholding tax on equity income earned by
non-resident shareholders, who might otherwise escape taxation in the
source country.
Countries throughout the world have reduced their corporate taxes
in an effort to attract or retain corporate investments. The Nordic countries have pioneered what is commonly known as the dual income tax
(DIT). It combines a flat tax rate on capital income with progressive taxation of labour income. One of the arguments in favor of the DIT is that
it allows policy makers to lower the corporate tax rate to reduce the
See Kind et al. (2000) on industrial clusters, economic rents and tax policy.
See also Serrato and Zidar (2014) for similar findings on corporate tax cuts.
4 Very little is known about the effect a dividend cut has on investments by listed firms. The new view of
dividend taxation assumes that investments are funded by retained earnings rather than new equity and
suggests that listed firms should not be affected by a dividend tax cut (see Auerbach and Hasset 2002).
Nordic Economic Policy Review
risk of capital flight, whilst at the same time tax distributed dividends to
personal shareholders.
In the continuation I discuss globalization and capital taxation. Tax
competition is putting pressure on capital taxes and makes redistribution more difficult and may affect the size and structure of the welfare
state and increase income inequality. If what the Nordic countries
looked like in the past was the reason for their success, then the future
may seem bleaker.
Globalization, tax competition and trends in
capital taxation
The term tax competition is used in the literature to describe how capital
taxes are set by independent governments that do not cooperate, and the
effect of tax setting on national tax bases. The early contributions consider a country with many identical regions each playing host to competitive
firms producing a single output by means of a nationally fixed stock of
mobile capital and an immobile factor fixed in supply. The latter could be
interpreted as land or labour, and may give rise to pure profits. It is assumed that each region’s supply of a public good is financed entirely by a
tax on capital employed within its borders (source tax). Tax policy affects
the distribution of the country’s (world) capital stock.
A fundamental insight is that a rise in the capital tax rate of one region benefits other regions by increasing their capital supplies and,
hence, their revenues. Put differently, a tax increase in one region causes a positive externality for other regions. However, the government in
each region neglects these externalities since it is only concerned with
the welfare of its own residents. The end result is that taxes are set too
low resulting in underprovision of public goods. An increase in all tax
rates at the same time by a small amount would increase public goods
supplies and welfare in all regions. 5
See e.g. Zodrow and Mieszkowski (1986) and Wilson (1986).
Nordic Economic Policy Review
Later amendments to the early theories of tax competition have for
example allowed countries to use expenditure levels of public input
goods, and multiple tax instruments as strategic variables. These expansions still show that there is still a negative externality from competition over capital that puts pressure on tax rates and the financing of the
welfare state. 6
If the literature on tax competition was correct in its prediction one
should observe falling tax rates on capital, since capital arguably is the
most mobile tax base. Indeed, statutory corporate tax rates in the OECD
have fallen substantially since the liberalization of foreign exchange regulations in the mid-80s. The average (unweighted) corporate tax rate in the
OECD fell from 48.2% in 1985 to 24.8% in 2015. The same drop is evident
in the EU countries where the EU average in 2015 is 22.2%. However, the
development in statutory corporate tax rates alone is not a good indicator of whether tax competition reduces tax revenue from the corporate
tax. A better indicator is the development in corporate tax revenue. As
seen from Figure 1, corporate tax revenue as a share of GDP does not
show a drop. There is country variation, but quite a number of OECD
countries show a weakly rising trend over time. This is not what theories of tax competition predict. Figure 1 also shows that the corporate
tax is a modest revenue raiser, which begs the question why one should
worry about corporate tax competition. One reason for why policymakers should worry is that the corporate tax rate is a backstop for the personal tax rate. The latter is a major revenue raiser in all OECD countries.
One reason for the rise in tax revenue is that corporate tax rate reductions have been accompanied by base broadening policies in most countries, for example, by limitations to interest tax deductibility through thin
capitalization rules, reduced investment credits and less favourable depreciation allowances. Furthermore, a growing degree of incorporation
may also explain part of the broadening of the corporate tax base. Finally,
6 See e.g. Wildasin (1989), Bucovetsky and Wilson (1991) and Bjorvatn and Schjelderup (2002). For a survey see Wilson (1999) and Wildasin and Wilson (2004).
Nordic Economic Policy Review
the reduction in corporate tax rates may have encouraged a shift of income from the personal towards the corporate tax base. 7
Figure 1: Corporate tax as share of GDP 1955–2013
% of GDP
OECD Average
Source: OECD Tax Database 2015.
Some of the findings in the tax competition literature are particularly relevant for small open economies such as the Nordic countries. The literature models differences in size by assuming that each household owns a
unit of capital, but that regions differ by population size (only). Per capita
levels are therefore the same in each region, and imply that capital will
not move between regions unless taxes differ. In this setting, the literature finds that the small region has an incentive to underbid the large
country. Since the large region is a large demander in the international
capital market, a reduction in its tax rate (t) will increase the after-tax
return to capital (r) substantially. The movement of capital across regions
depends on the cost of capital, that is, r+t, which means that a reduction
in t has a modest effect on the cost of capital. As a result, the large country
has weak incentives to bid for capital. In contrast, the small country canAlstadsæther and Thoresen (2009) study income shifting between personal and corporate tax bases in
Norway for the period after the 1992 tax reform. They find significant evidence for income shifting.
Nordic Economic Policy Review
not affect the after-tax return on capital and a reduction in its tax rate will
therefore lower the cost of capital by a large amount.
An interesting result in the tax competition literature is that if the
small country is sufficiently small, it sets a lower tax rate than the large
country and attracts an over proportional share of the total capital
stock. By doing so it “wins” the competition for capital in that it can obtain higher per capita utility in equilibrium (Bucovetsky (1991) and
Wilson (1991)). The result has interesting policy implications, as small
countries may do relatively well in an integrating world. It should be
noted, however, that the outcome is still inefficient in the sense that tax
rates would be too low and the provision of public goods less than in
the absence of competition.
There are other ways of analyzing country size than by population
size. Haufler and Wooton (1999) model a multinational firm that wants to
sell its goods in both a small and a large country. The large country has
more consumers. The firm faces a locational choice: it can only locate and
produce its good in one country. They show that firms prefer to locate in
large countries (large market size) in order to save transport costs on exports (it is better to export small amounts of goods than large amounts
when there are transport costs). As a consequence, the large country can
utilize its market size and set a higher capital tax rate. As in the models by
Bucovetsky (1991) and Wilson (1991), the outcome of tax competition is
that the large country sets a higher tax rate.
In order to put these theories to the test one can group countries according to their size. For example, the G7 countries had an average
(unweighted) statutory corporate tax rate of 38.7% in 2000, and this
rate had fallen to 31% in 2014. 8 In contrast, the Nordic countries had an
average statutory corporate tax rate of 29.4% in 2000 and 22.7% in
2014. The difference in corporate tax rates among these large and small
countries is quite telling. The Nordic countries have substantially lower
corporate tax rates than the G7 countries.
8 G7 countries are USA, Japan, Germany, UK, France, Australia and Italy. The Nordic countries are Denmark,
Iceland, Finland, Norway and Sweden.
Nordic Economic Policy Review
The Nordic countries implemented major tax reforms in the early
1990s partly in response to the pressures of globalization. Klemm et al.
(2009) study the relationship between wage taxes and corporate taxes.
They document that on average, European Union (EU) member states
have reduced their reliance on capital taxes and increased the share of
labour taxes in total tax revenues during the past 30 years. They also
find that the policy responses have been rather diverse. In 2004, the
classical welfare states in Scandinavia and continental Europe had lower ratios of statutory corporate to wage taxes than the Anglo-Saxon
countries (except Ireland). In 2004, the corporate tax rate was only
63% of the wage tax rate for an average worker in Sweden, but 171% of
the wage tax rate in the United States. Such differences are in striking
contrast to the common perception that social democratic governments
(as in Scandinavia and continental Europe) share a higher preference
for redistribution, as compared to more conservative and free market
oriented types of governments.
The political economy of tax competition
In the studies mentioned above an underlying assumption is that politicians conduct policies that are to the best for society as a whole by maximizing the sum of individuals’ utilities (Benevolence). A different perspective on tax competition is taken by the public choice literature,
which challenges the notion that competition to attract capital is harmful. The basic idea is that competition reduces the rent-seeking activities
of government officials and may force a more efficient use of public
funds. The literature can be divided into two categories.
The first category does not take into account electoral systems or
re-election concerns, but assumes that governments are partly benevolent and partly Leviathan. Hence, government officials are concerned in
part with maximizing the public sector by diverting some tax revenue
for own consumption. This strand of the literature finds that the outcome of tax competition on tax rates, public expenditures, and welfare
Nordic Economic Policy Review
depends on an assessment of the relative strength of Leviathan versus
Benevolence. 9 In the context of the Nordic countries, the discussion of
Leviathan versus Benevolence could be related to politicians’ receiving
campaign funding and retirement positions from wealthy individuals
and special interest groups in return for favorable policy.
The second part of the literature models tax competition in the
presence of voting and there are different approaches to how this is
done. Persson and Tabellini (1992) study a two-country model where
each government levies a source tax on mobile capital to finance government transfers. A fall in the cost of investing abroad (i.e. increasing
competition) puts downward pressure on tax rates. At the same time,
however, there is a second, political effect in place since policy is chosen
by a policymaker who represents the preferences of the median voter.
Tax competition is shown to make the median voter select a more leftist
government, whose distributional preferences call for higher taxes on
capital, and this partly mitigates the tendency of tax competition to
lower taxes on capital.
Biglaser and Mezzetti (1997) study how regions compete to attract
large firms. Their starting point is the observation that some US states
seem to offer “tax packages” to firms that often exceed the “economic
value” of the firm’s instate investment project. They assume that when
preparing a bid, legislators take into account both the public’s interest
and the bid’s impact on their probability of re-election. The competition
among regions follows the rules of an English auction. Since politicians
value their re-election, their bid for investments is distorted away from
the value of the project to voters and may result in an inefficient location of firms in the sense that legislators give away too much of the taxpayers’ money in order to attract firms.
Janeba and Schjelderup (2009) compare the outcomes of increasing
capital tax competition under presidential–congressional and parliamentary democracies, in a setting where politicians value rents and reelection to office. In their model, a presidential-congressional system
features shifting majorities in the legislature that are issue dependent
See, for example, Brennan and Buchanan (1980); Edwards and Keen (1996); Rauscher (1998).
Nordic Economic Policy Review
(here the revenue and expenditure sides of the government budget).
The majority that passes tax policy may differ from the majority passing
the expenditure allocation. Thus shifting majorities limit the possibility
of rent-seeking and increases accountability of elected policy makers.
By contrast, in a parliamentary democracy a cohesive majority passes
the entire budget in one vote. In a closed economy, the cohesive majority in a parliamentary regime tends to deliver more public goods than
under a presidential system because it appeals to voters from all supporting legislators’ districts. Yet, the system has also a negative consequence because the majority coalition is powerful and therefore tends
to extract more rents. They find that tax competition among presidential–congressional democracies is typically welfare improving, while
harmful among parliamentary democracies if under the latter, public
goods are sufficiently valued. The results hold when politicians seek reelection because of exogenous benefits of holding office. By contrast,
when politicians hold office only to extract rents, tax competition is
harmful if politicians are sufficiently patient.
Profit shifting and multinationals
One of the most pronounced characteristics the last 30 years is the
growth in foreign direct investments (FDI) and thus the rising importance of multinationals. Growth rates have been between 10 and
20% annually and an increasing share of trade worldwide is between
affiliates of multinational firms. The rising importance of multinationals
has gone hand in hand with industrialized countries reforming their
corporate income tax policies in order to attract investment. Statutory
rates in the OECD, for example, have fallen from an average of 50% in
the early 1980s to 25% in 2014. Despite falling tax rates, multinationals
have come under fire for siphoning off profits into tax havens. Corporations have responded by saying that their objective is to reduce their
worldwide taxes consistent with national laws in order to maximize
post-tax global profit. This has prompted governments around the
world to overhaul their tax systems and the OECD to launch its BEPS
project (OECD, 2013).
Nordic Economic Policy Review
The OECD (2013) report on Base erosion and profit shifting (BEPS)
identifies transfer pricing and debt shifting (thin capitalization) as major
reasons for the tax-revenue drain in high-tax countries. Both strategies
are regulated by the OECD’s arm’s length standard, which states that
transfer prices should reflect market prices chosen by unrelated parties
engaged in similar trades under similar circumstances (Eden, 1998;
OECD, 2010, art. 9). Such pricing, however, may be difficult to enforce because of the lack of market parallels, multinationals’ use of tax havens,
and lack of disclosure of either earnings worldwide or pricing methods.
Multinationals in effect report income by choosing prices on intrafirm trade. By selecting to over-invoice (under-invoice) sales to affiliates in high-tax (low-tax) countries, multinationals can shift profits to
low-tax countries and thus save taxes. For instance, royalties for using a
brand name or a patent do not have an obvious market parallel; hence,
multinationals have considerable discretion in setting prices on such
transactions. There is clearly a grey area between strictly legal tax planning and illegal tax evasion. Multinationals may voluntarily or involuntarily cross this line. Furthermore, in some cases, the deviation from the
true price of a good or service is so small that the tax authorities would
not bother with it, but if the transaction volume is large, substantial
amounts of profits can be shifted.
There is substantial evidence of profit shifting by multinational
across countries. Pak and Zdanowicz (2001) find that the volume of
profit shifting in U.S. multinationals was equal to 18% of total reported
corporate profits in 2000. Bartelsman and Beetsma (2003) study OECD
data and point out that 65% to 87% of the (potential) additional tax
revenue, stemming from a unilateral tax increase, is lost due to profit
shifting by transfer pricing. The literature on profit shifting by abusive
transfer prices indicates that it is differences in statutory tax rates that
provide profit shifting incentives. 10
10 Evidence for transfer pricing in the U.S. is given in Clausing (2003) and Bernard et al. (2006); for Norway
in Langli and Saudagaran (2004); for Germany in Weichenrieder (2009). Oyelere and Emmanuel (1998)
show that foreign-owned affiliates in the UK are characterized by lower profits but higher dividend distributions (than UK-controlled firms). Evidence for transfer pricing in European multinationals is given in Dharmapala and Riedel (2013).
Nordic Economic Policy Review
There are few empirical studies on Scandinavian data. Langli and
Saudagaran (2004) compare the profitability of Norwegian-owned and
foreign-owned companies in manufacturing and trade in the years 1993
to 1996. They find that foreign-owned enterprises have a profit margin
2.6 percentage points lower than Norwegian-owned enterprises. This is
consistent with a net shifting of profits out of Norway by foreign-owned
enterprises. Balsvik et al. (2009) expand the data set used by Langli and
Saudagaran (2004). They find that multinational corporations shift
profits both out of Norway and into Norway but that the net flow is out
of Norway. The loss in tax revenue is estimated to be in the order of
30% of the potential tax revenue from foreign multinational enterprises. Another finding in this study is that multinational enterprises in
Norway have a profit margin of 1.5 to 4 percentage points lower than
comparable domestic enterprises. Their findings, then, are consistent
with the findings of Langli and Saudagaran (2004).
The fact that multinationals pay less tax than national firms is not
only due to abusive transfer prices. Multinationals can also structure
their financing arrangements to minimize tax. The capitalization of a
company has an impact on the amount of profit a company reports for
tax purposes. Since interest is tax deductible in most countries, a high
level of debt, and thus the amount of interest it pays, reduces taxable
profit. Lending and borrowing arrangements can be structured so that
affiliates in high-tax countries have “too much debt” (thin capitalization) and where the set-up is that interest is received by an affiliate in a
jurisdiction that does not tax interest income.
Tax motivated profit shifting by multinationals erodes national tax
bases and constitutes a serious risk to tax revenues, tax sovereignty and
tax fairness. It also means that multinationals have a competitive edge
that has implications for competition in markets and in the long run, the
ownership structure in industries. It also implies that the multinationality
in the tax base rises and thus that the tax sensitivity of the corporate tax
base goes up. The latter may limit the scope for corporate taxes and increase the excess burden from taxation due to a narrower tax base. This
prompts the question of what countries can do to reduce this problem.
Nordic Economic Policy Review
Curbing base erosion
The main challenge for tax authorities is to figure out whether intrafirm transactions across borders satisfy the arm’s length principle
(ALP). This means leverage and the prices on intra-firm transactions
correspond to what two independent entities would have agreed on.
This is often difficult, since some goods especially intangibles have no
obvious market parallel. Loans are also firm and project specific, and it
can be a challenge to assess the terms a third party lender would be
willing to enter into. The OECD identifies five factors that determine
comparability: the functions performed by the parties, the contractual
terms, the economic circumstances, the characteristics of the property
or service transferred, and the business strategies pursued by the parties (OECD 2010). To carry through an evaluation based on these criteria is costly and very difficult. For such reasons some have proposed
abandoning the principle of separate accounting (SA) that most countries rely on to determine profits, but rather to consolidate all profits
into a worldwide singe measure and then apportion taxable income to
countries based on activity weights of each firm.
There is much to be said about a transition to formula apportionment
(FA), but it requires, among other things, political cohesion to agree on
uniform apportionment weights. Nielsen et al. (2010) develop a theoretical model that compares basic properties of FA to SA. The focal point of
their analysis is how changes in tax rates affect capital formation, input
choice, and transfer pricing, as well as on spillovers on tax revenue in
other countries. A significant difference between the two tax principles is
that the SA system is based on reported income whilst taxation under the
FA system is based on reported activity. They show that these fundamental characteristics introduce different tax spillovers across countries under the two tax systems, which makes it impossible to unambiguously
favour one system over the other. Nielsen et al. (2010) find that the relative strength of tax spillovers under the two regimes depends on how
costly it is for multinational enterprises to undertake transfer pricing, and
how much pure profit the MNEs generate. These considerations also determine whether SA or FA implies the higher level of tax in a nonNordic Economic Policy Review
cooperative equilibrium, and in the end which of the two schemes is preferable from an international perspective.
Under the SA system, the amount of interest that an affiliate of a
multinational can deduct is determined by the rate of interest applied to
its debt and the amount of its debt. Countries limit tax-induced income
shifting via the transfer price by auditing a firm’s transfer price to make
sure that the interest rate is in line with what ALP. This involves considering the specific attributes of the company in determining the amount
of debt that the company would be able to obtain from independent
lenders. There are clear disadvantages to using the ALP on the interest
rate, because it requires skills, resources, and specialization to establish
what a third party would lend. For such reasons many countries rely on
ratio approaches.
Under a ratio approach, also often referred to as a safe harbor rule
or a thin capitalization rule, the amount of debt on which interest may
be deducted for tax purposes is set by a pre-determined ratio. The exact
definitions of the debt measure in the numerator of the ratio and of assets or equity in its denominator vary across countries, but the most
common rule is either to use a ratio based on total debt-to-equity or internal (corporate group) debt-to-equity. The empirical literature on the
effect of different types of thin capitalization rules on the firm’s financial
structure encompasses both US and European multinationals. It concludes that thin capitalization rules have a substantial effect on both
internal and external leverage. 11
Despite that empirical studies show that the ratio rules have an impact on multinationals’ ability to shift profit by debt, there is a growing
perception that these rules are not effective. A small but growing group
of countries have therefore implemented what is called earnings stripping rules. 12 These rules operate to restrict interest deductions that ex11 Buettner et al. (2012) study foreign affiliates of German multinationals whereas Blouin et al. (2014) investigate how thin capitalization rules worldwide affect the capital structure of foreign affiliates of US multinational firms. Both find that thin capitalization rules affect the leverage in multinationals.
12 A handful of countries use both safe harbour rules and earnings stripping rules, either simultaneously or
they impose a marginal earnings stripping requirement that applies only if the safe harbour limit is exceeded. The countries that fall into the first category are Denmark and Japan. Bulgaria, France and the US impose
an earnings stripping rule only if the safe ratio rule is exceeded.
Nordic Economic Policy Review
ceed a certain threshold, such as a percentage of EBITDA or EBIT. 13 Finland, Germany, Italy, Norway, Portugal, and Spain have implemented
such rules. The Norwegian Tax Committee proposed in 2014 (NOU
2014:13) that both internal and external debt should be embedded in
the earnings stripping rule and that EBIT was a better measure. They
argued that external debt could be used to shift profit and that firms’
leverage was too high due to the deductibility of interest. The latter implies too high risk premiums and too little investments. 14
There is a literature that discusses which rule, ratio or earnings
stripping, should be preferred if the aim is to maximize national income.
In a recent paper this question is answered by Gresik et al. (2015b).
They develop a general equilibrium framework with both labour and
capital that allows them to analyze the variation in thin capitalization
rules observed in practice. Their model embeds thin capitalization and
transfer pricing behaviour of multinationals. They show that the policy
that maximizes the host country’s national income is an earnings stripping rule without a safe harbour rule.
FDI, tax havens and multinationals: A bane or a
boon for a host country?
An interesting question is whether attracting foreign direct investment
(FDI) from multinationals is a bane or a boon for a host country given
the ability multinationals have to shift income. Multinationals often use
tax haven conduit companies to shift income. Hines (2010) argues that
although the tax avoidance opportunities presented by tax havens may
reduce revenues in high-tax jurisdictions, they may have offsetting effects on FDI that are attractive to the same governments. If governments cannot distinguish between mobile and immobile investments,
tax havens permit governments to subject immobile investments to
higher taxes than mobile investments. Hong and Smart (2010) demon13 EBITDA= Earnings Before Interest, Tax, and Depreciation Allowances. EBIT = Earnings Before Interest
and Tax.
14 See Sørensen (2014) for an analysis.
Nordic Economic Policy Review
strate this effect in a model where multinationals shift profit by debt
from a tax haven affiliate. They show that providing a tax deduction for
interest payments on subsidiary debt allows host countries to maintain
or even increase high business tax rates, and to attract more mobile investments from multinationals because the tax deductibility of interest
reduces the firm’s after-tax cost of capital. The end outcome is higher
host welfare.
Slemrod and Wilson (2009) model tax havens that are “parasitic” on
the tax revenues of non-haven countries in that they sell concealment
services to taxpayers in non-havens. Non-haven countries must expend
real resources to prevent tax base erosion. They show that tax havens
increase the social costs that a country incurs when it increases its tax
on capital. This aggravates the tax competition problem and results in
lower welfare.
Gresik et al. (2015a) use the model by Hong and Smart but also include transfer pricing in the model. They allow the host country to decide on the corporate tax rate and thin capitalization rules (equity-debtratio) that may limit profit shifting by excessive interest deductions. In
their model the multinational firm has a financing subsidiary located in
a tax haven and an operational subsidiary in a high-tax country. The
multinational can shift profit to the tax haven affiliate by the level of internal debt and the interest rate (transfer price) it charges on intracompany loans.
Which countries benefit or lose from attracting FDI depends in general on country characteristics. Developed countries have better institutional quality than emerging or developing countries in the sense that
their tax systems make it more costly for multinationals to engage in
aggressive tax-induced transfer pricing. They also have a low cost of
capital, high rents for domestic entrepreneurs, and a moderate to high
capital share in multinational production relative to emerging and developing countries. Gresik et al. (2015a) show that these differences
matter. Developed countries can benefit from FDI and that a welfare
maximum exists with an optimal corporate tax rate and a thin capitalization rule that are largely in line with average current tax rates and
thin capitalization rules in the OECD. Developing countries, however, do
not stand to benefit from policies that attract positive FDI. While per106
Nordic Economic Policy Review
missive thin capitalization limits may be needed in developing countries
to attract FDI, the amount of debt financing allowed by such permissive
rules may facilitate aggressive transfer pricing that can result in lower
welfare. The optimal tax policy for developing countries is to effectively
eliminate the tax benefits of debt financing and only tax domestic firms.
Those who advocate the usefulness of tax havens as conduits lack a
fully convincing explanation for why governments need to use tax havens to discriminate between mobile and immobile tax bases, rather
than designing their tax systems to achieve this discrimination at lower
costs. 15 Is the inability to conduct rational tax policy due to failures in
the political system and/or lobbying, say? Schjelderup (2015) emphasizes the secrecy aspects of tax havens and argues that tax havens extend beyond just profit shifting activities and that to fully assess the
welfare effects of them we need to fully assess the full range of their activities. Nevertheless, the lesson from a policy perspective from these
studies is that the tax authorities must have resources and tools to secure tax compliance at their disposal. If not multinational investments
may be a bane.
Some concluding comments
In this paper I discuss the challenges of taxing capital for small open
economies. Although the corporate tax share of GDP in most countries is
only around 3–4%, it is an important tax because it acts as a “backstop”
for the personal tax rate. Wealthy individuals in particular would be
given an incentive to reclassify their labour income as corporate income
typically using offshore corporate structures to escape tax. The pressures of tax competition are exacerbated by tax planning and income
shifting to low tax countries by multinationals. Studies show that multinationals pay less tax than domestic firms and this may give them a
competitive edge over domestic firms. The long term effects may be
changes in ownership structure that affect competition in markets and
Wilson (2015) points this out in a survey of the literature on tax havens and tax competition.
Nordic Economic Policy Review
make the corporate tax base more tax sensitive. Profit shifting is undertaken through transfer pricing and thin capitalization (excessive debt).
Recent studies show that rules that restrict interest deductions that exceed a certain threshold such as a percentage of EBIT or EBITDA are the
best defence against the use of debt to shift profit. Source taxes on royalty payments that are another effective defence mechanism. Yet establishing arm’s length prices in transactions between affiliates of multinationals is a problem that will rid tax authorities also in the future.
I have not discussed taxes that fall on the capital stock such as property taxes, the wealth tax and the inheritance tax. Concerning the two
latter taxes, the Nordic welfare states set themselves apart from larger
countries. Table 1 provides and overview of the wealth and inheritance
tax in the Nordic countries versus some large countries.
Table 1: Wealth and inheritance taxes
Finland 16
United States*
United Kingdom*
Wealth Tax
Inheritance tax
* The U.S. inheritance tax has an exemption of USD 5,430,000 in 2015. This is considerably larger
than the exemptions in France (USD 105,945), Germany (USD 423,782), and the UK (USD 488,280).
Source: Center For Federal Tax Policy, 2015.
It is interesting to note that the Nordic countries seem to have gone further in terms of abolishing redistributive capital taxes than countries
traditionally associated with polices much less tuned to redistribution.
Aaberge and Atkinson (2010) shown how income inequality at the top
of the distribution has increased in Anglo-Saxon countries, whereas the
same rise in top income shares was not experienced by Continental Eu16
Finland had a wealth tax until 2006 and a temporary wealth tax reintroduced in 2010, for four years.
Nordic Economic Policy Review
ropean countries. They find that the Norwegian and Swedish experience
over the twentieth century is similar to the Anglo-Saxon countries in
that top shares, and the concentration among top incomes have first
fallen and then risen. Norway differs from Sweden in that that the top
shares rose more sharply in the period 1990–2006. Between 1980 and
2004, for example, the share of the top 1% more than doubled in Norway, but rose less than half in Sweden.
Several explanations have been put forward to explain why Norway
sets itself apart. The implementation of the 1992 tax reform abolished
the dividend tax and lead to a sharp increase in dividends and capital
gains among the richest in Norway. Capital taxation in Sweden was less
favourable. Substantial oil production in Norway started some 15 years
before the rise in inequality, but could still be an explanatory factor due
to constrained cash in this sector in the initial phase of production. Capital market reforms with liberalization of interest rates and an upturn in
business cycles are also important factors that are hard to disentangle,
but they certainly played a role.
Capital taxation also affects income mobility, and concerns about
rising inequality have often been countered by constant changes in the
composition of top income earners. If so, the rise in top incomes may
not translate into “economic power”. Aaberge et al. (2013) study who
enters and leaves the top income groups in Norway in the period 1967–
2011. Their main conclusion is that despite large changes in top income
mobility over the last four decades, the magnitude of the effect of the
changes in mobility on the income shares was moderate.
An interesting question is how voters will respond to rising inequality. Standard neoclassical theory predicts that inequality and the size of
behavioural responses determines redistributive preferences (Meltzer
& Richard, 1981). Following this literature one would expect voters to
respond to lower capital taxes and increased inequality by demanding
redistributive measures. A major concern, however, is that the public is
misinformed about income inequality (Bartels, 2005; Slemrod, 2006).
Such misconception may explain why there has been so little redistribution in the US, and that the political response to rising inequality in the
US has been to further decrease capital taxes by reducing the top marginal tax rate from 75% in 1970 to 35% in 2012 (see IRS, 2014).
Nordic Economic Policy Review
Gilens and Page (2014) study American politics and use data from
1981–2002 that has been collected with the purpose for estimating the
influence upon public policy of poor citizens, “affluent” citizens, and
those in the middle of the income distribution. A central message that
emerges from their study is that: “...economic elites and organized
groups representing business interests have substantial independent
impacts on U.S. government policy, while mass-based interest groups
and average citizens have little or no independent influence”. Their
study indicates that the majority does not rule, and that when the majority of citizens disagree with economic elites or their organized interest, they generally lose. In the Nordic countries, the economic elites and
commercial interests have been strong drivers in reducing capital tax
rates. In particular, this pertains to the abolishment of the wealth and
the inheritance tax. But it is not clear in these specific cases what the
preferences of the majority of voters were. A better example of conflict
between the majority of voters and the economic elite is the increase in
property taxes. In Norway, for example, “affluent voters” and organized
interest from the business community have been advocates of a transition from the wealth tax to higher property taxes. The median voter, in
contrast, is strongly against property taxation. Property taxes have certainly gone up through higher valuation of housing values, and the
wealth tax has been reduced.
Top income earners in the Scandinavian countries mainly derive their
income from capital. Competition among countries to attract mobile capital is a persistent phenomenon and will be a driver towards still lower
taxes on mobile capital. A major change from the past, then, is less ability
to redistribute, increasing income inequality, and rising immigration from
poor countries. In sum these forces may affect trust between members of
society. The level of trust is positively linked to economic growth. 17 Herein lies a major challenge for the Nordic welfare states
For a survey of the literature on trust and economic growth see Christian Bjørnskov (2012).
Nordic Economic Policy Review
Aaberge, R. & Atkinson, A.B. (2010). Top incomes in Norway, in Atkinson, A.B. and
Piketty, T. (Eds.): Top Incomes A Global Perspective, Oxford University Press, Oxford, 448–482, SSB.
Aaberge, R,. Atkinson, A.B., & Modalsli, J. (2013). The ins and outs of top income mobility. SSB Discussion paper No. 762.
Alstadsæther, A. & Thoresen, T.O. (2009). Shifts in organizational form under a dual
income tax system, International Tax and Public Finance 66(4), 384–418.
Alstadsæter, A., Jacob, M. & Michaely, R. (2015). Do dividend taxes affect corporate
investment. Journal of Public Economics (forthcoming).
Auerbach, Alan J., & Kevin A. Hassett (2002). On the marginal source of investment
funds. Journal of Public Economics, 87, 205–232.
Balsvik, R., Jensen, S., Møen, J. & Tropina, J. (2009). Kunnskapsstatus for hva
økonomisk forskning har avdekket om flernasjonale selskapers internprising i Norge.
SNF Rapport 11/09.
Bartels, L. M. (2005). Homer gets a tax cut: Inequality and public policy in the American mind. Perspectives on Politics, 3 (01), 15–31.
Bartelsman, E.J. & Beetsma, R.M.W.J. (2003). Why pay more? Corporate tax avoidance through transfer-pricing in OECD Countries. Journal of Public Economics 87,
Bernard, A.B., Jensen, J.B. & Schott, P.K. (2006). NBER Working Paper No. 12493.
Bjørnskov, C. (2012). How does social trust affect economic growth. Southern Economic Journal 78(4), 1346–1368.
Biglaser, G. & C. Mezzetti, (1997). Politicians’ decision making with re-election concerns. Journal of Public Economics 66, 425–447.
Blouin, J., Huizinga, H. P., Laeven, L. & Nicodeme, G. (2014). Thin Capitalization Rules
and Multinational Firm Capital Structure. CentER Discussion Paper 007, Tilburg.
Buettner, T., Overesch, M., Schreiber, U. & Wamser, G. (2012). The impact of thincapitalization rules on the capital structure of multinational firms, Journal of Public
Economics 96, 930–938.
Brennan, G. & Buchanan, J. (1980). The Power to Tax: Analytical Foundations of a
Fiscal Constitution. Cambridge University Press.
Bucovetsky, S. (1991). Asymmetric tax competition. Journal of Urban Economics 30,
Nordic Economic Policy Review
Bucovetsky, S. & Wilson, J. D. (1991). Tax competition with two instruments, Regional
Science and Urban Economics 96, 329–341.
Clausing, K. (2003). Tax-motivated transfer-pricing on Intra-firm Trade. Journal of
Public Economics 87, 2207–2223.
Dharmapala, D. & Riedel, N. (2013). Earnings shocks and tax-motivated incomeshifting: Evidence from European multinationals. Journal of Public Economics 97,
Eden, L. (1998). Taxing Multinationals: Transfer Pricing and Corporate Income Taxation in North America. University of Toronto Press, Toronto.
Edwards, J. & Keen, M. (1996). Tax Competition and Leviathan, European Economic
Review 44, 113–34.
Gilens, M. & Page B.I. (2014). Testing Theories of American Politics: Elites, Interest
Groups, and Average Citizens. Perspectives on Politics, Vol. 12(03), pp 564–581.
Gresik, T., Schindler, D. & Schjelderup, G. (2015a). The Effect of Tax Havens on Host
Country Welfare. CESifo Working Paper Series 5314, CESifo Group Munich.
Gresik, T., Schindler, D. & Schjelderup, G. (2015b). Immobilizing profit shifting: Is it
safe to strip in the harbor. Manuscript University of Notre Dame.
Haufler, A. & Wooton, I. (1999). Country size and tax competition for foreign direct
investment. Journal of Public Economics 71, 121–139.
Hines, J.R. (2010). Treasure Islands. Journal of Economic Perspectives 24, 103–126.
Hong, Q. & Smart, M. (2010). In praise of tax havens: International tax planning and
foreign direct investment. European Economic Review 54:82–95.
Huizinga, H. & Nielsen, S.B. (1997). Capital income and profit taxation with foreign
ownership of firms. Journal of International Economics, 1997, 88–106.
IRS. (2014). SOI Tax Stats – Historical Data Tables, Table 23. See
Janeba, E. & Schjelderup, G. (2009). The Welfare Effects of Tax Competition Reconsidered. Politicians and Political Institutions. Economic Journal 119, 1–19.
Keen, M. & Marchand, M. (1997). Fiscal competition and the pattern of public spending. Journal of Public Economics 63, 33–53.
Kind, H.J., Midelfart Knarvik, K.H., & Schjelderup, G. (2000). Competing for capital in
a lumpy world. Journal of Public Economics 78, 253–274.
Nordic Economic Policy Review
Klemm, A., Haufler, A. & Schjelderup, G. (2009). Economic integration and the relationship between profit and wage taxes. Public Choice 138, 423–446.
Langli J.C. & Saudagaran, S.M. (2004). Taxable Income Differences Between Foreign
and Domestic Controlled Corporations in Norway. European Accounting Review
13(4), 713–41.
Meltzer, A. H. & Richard, S. F. (1981). A Rational Theory of the Size of Government. The
Journal of Political Economy 89 (5), 914–927.
Nielsen, S.B., Raimondos-Møller, P., & Schjelderup, G. (2010). Company taxation and
tax spillovers: Separate Accounting versus Formula Apportionment. European
Economic Review 54, 221–232.
NOU (2014). 13. Kapitalbeskatning i en internasjonal økonomi. Norges Offentlige
Utredninger. Oslo.
OECD (2010). OECD Transfer Pricing Guidelines for Multinational Enterprises and Tax
Administrations 2010. OECD Publishing, Paris.
OECD (2013). Action Plan on Base Erosion and Profit Shifting. OECD Publishing, Paris.
Oyelere, P. B. & Emmanuel, C.R. (1998). International Transfer Pricing and Income
Shifting: Evidence from the UK. European Accounting Review 7, 623–635.
Pak, S.J. & Zdanowicz, J.S., (2001). Executive Summary of “US Trade with the World:
An Estimate of 2000 Lost US Federal Income Tax Revenues due to Over-Invoiced Imports and Under-Invoiced Exports”. Mimeo, Florida International University.
Persson, T. & G. Tabellini, (1992). The politics of 1992: Fiscal policy and European
integration. Review of Economic Studies 59, 689–701.
Rauscher, M. (1998). Leviathan and Competition Among Jurisdictions: The Case of
benefit taxation. Journal of Urban Economics 44, 59–67.
Schjelderup, G. (2015). Secrecy Jurisdictions, International Tax and Public Finance,
Serrato, J. C. S. & Zidar, O. (2014). Who benefits from state corporate tax cuts? A local
labor approach with heterogeneous firms. NBER Working paper 20289.
Slemrod, J. (2006). The role of misconceptions in support for regressive tax reform.
National Tax Journal, 57–75.
Slemrod, J. & Wilson, J.D. (2009). Tax Competition with Parasitic Tax Havens. Journal
of Public Economics 93, 1261–70.
Sørensen, P. B. (2014). Taxation and the Optimal Constraint on Corporate Debt Finance. ESifo Working Paper No. 5101.
Weichenrieder, A.J. (2009). Profit shifting in the EU: Evidence from Germany. International Tax and Public Finance 16, 281–297.
Nordic Economic Policy Review
Wildasin, D.E. (1989). Nash equilibria in models of fiscal competition. Journal of
Public Economics 35, 299–240.
Wildasin, D.E. (1991). Some rudimentary “duopolity” theory. Regional Science and
Urban Economics 21, 193–212.
Wildasin, D.E. & Wilson, J.D. (2004). Tax competition: bane or boon? Journal of Public Economics 88, 1065–1091.
Wilson, J.D. (1986). A theory of interregional tax competition, Journal of Urban Economics 19, 296–315.
Wilson, J.D. (1999). Theories of tax competition, National Tax Journal 52, 269–304.
Wilson, J.D. (1991). Tax competition with interregional differences in factor endowments. Regional Science and Urban Economics 21, 423–451.
Wilson J.D. (2015). Tax havens in a world of competing countries. CESifo Dice Report 12 (4). Zodrow, G. and Mieszkowski, P. (1986). Pigou, Tiebout, property taxation and the underprovision of local public goods. Journal of Urban Economics 19,
Yagan, D. (2001). Capital tax reform and the real economy: The effects of the 2003
dividend tax cut. NBER working paper 21003.
Nordic Economic Policy Review
6. Nordic family policy and
maternal employment
Julian V. Johnsen, Department of Economics, University of Bergen, email:
[email protected] and Katrine V. Løken, Department of Economics,
University of Bergen, email: [email protected]
We survey the quasi-experimental literature to evaluate the effectiveness
of family policies in terms of increasing maternal employment. The findings are mixed but mostly follow the theoretical predictions, although the
effects are smaller than expected. While parental leave policies do not
affect maternal employment, cash-for-care benefits decreases maternal
employment in the short and long run. Subsidized child care can increase
maternal employment, especially for mothers of younger children. When
evaluating the overall effect of the “package” of Nordic family policies using cross-country evidence, we find that the remaining gender gaps in the
labor market are concentrated at the top of the career ladder. To increase
the share of female managers and senior officials, the current system of
family policies should be made more flexible.
Because women tend to be the main caregivers for children, having
children is associated with an increase in the gender gap in the labor
market. The wage gap between women and men with children is much
larger than the wage gap between women and men without children
(Bertrand et al., 2010). In Sweden, the within-couple wage gap between
husband and wife more than doubles after the couple becomes parents
(Angelov et al., 2013). A recent paper provides causal evidence that in
Denmark having a child has a large and long-lasting negative effect on
female labor supply, especially in terms of employment rates (Lundborg
et al., 2014). Economists explain this pattern with comparative advantages leading to specialization in the household (Becker, 1981). If
women have a comparative advantage in terms of child care, having
children increases the intra-household specialization, leading to less
gender equality in the labor market.
In general, family policies aim to aid the reconciliation of work and
family life, making it easier for parents to work (increasing the maternal
labor supply) and for workers to become parents (increasing fertility). 1
This paper focuses on the former, the impact of family policies on maternal labor supply. This is not to say that the potential fertility effects of
family policies are unimportant. The effects are mainly left out because of
the limited scope of the paper and for the simple reason that there is little
hard evidence to review on the fertility effects of family policies. 2
Boosting female employment and gender equality has been a key
goal of Nordic family policies (Björnberg, 2013). Several comparative
studies have shown that the Nordic countries share key similarities in
family policies and that the Nordic approach to family policies differs
from that of other European countries (e.g., Ellingsæter, 2006; Bradshaw and Hatland, 2007). Two key policies that the Nordic countries
1 Another important goal of family policies, which is not discussed here, is to promote child welfare and
2 In theory, most family policies involve a transfer of resources from people without children to people with
children and thus could increase fertility by increasing the incentives to have children.
Nordic Economic Policy Review
share are a long and generous paid parental leave and high coverage
rates of subsidized pre-school child care.
To discuss the potential effects of Nordic family policies on maternal
employment, it is important to distinguish between policies that promote home care versus policies that promote the use of formal child
care. Parental leave programs and cash-for-care benefits promote home
care. In theory, home care leads to increased intra-household specialization and thus decreases maternal employment. 3 Promoting the use of
formal child care through subsidies should work in the other direction,
potentially increasing maternal employment by decreasing intrahousehold specialization.
In this paper, we have conducted a partial review of the effectiveness of the main Nordic family policies in increasing gender equality in
the labor market. At first glance, the “package” of family policies seems
to have been effective: Many of the family policies were implemented or
expanded in the 1970s, which is the same period that the Nordic countries saw the largest increase in female employment rates. However,
there is a basic “chicken and the egg” problem with this interpretation:
Do Nordic countries have high gender equality in the labor market because of their generous family policies, or has the high share of working
women resulted in increased political demand for generous family policies? 4 This question cannot be easily answered. In addition, the different policies should in theory have opposite effects on maternal employment depending on whether the policies promote home care or the
use of formal child care. To assess the causal effect of Nordic family policies, the literature review focuses on studies that used quasiexperimental designs. These designs used reforms and other “natural”
variations to identify the causal effects of one variable on another. The
benefit of quasi-experimental studies is that they deliver clear-cut results on the immediate effects of the policy reforms studied.
3 Most parental leave programs include job protection, allowing mothers to return to their previous job after
parental leave. Shorter parental leaves with job protection could increase maternal employment by providing job stability and the legal right to continuous employment.
4 Even if increased female labor force participation caused demand for family policies (instead of generous
family policies driving female labor force participation), figuring out the causal effect of the demanded family policy would remain an important policy question.
Nordic Economic Policy Review
The drawback of quasi-experimental studies is that they mainly
provide insights into the immediate individual effects of policy reforms.
Some policies might work in the long run by changing norms related to
family and work. Family policies could also have important general
equilibrium effects that are hard to study in a quasi-experimental setting. Policies promoting home care incentivize women to take longer
career breaks after child care, while promoting child care works in the
other direction. Because employers do not know which women will
have children, longer parental leaves create a disincentive to hire women if promoters dislike long career breaks. Promoting the use of formal
child care or reserving part of parental leave for the exclusive use of fathers alleviates these disincentives for employers. These potential general equilibrium effects cannot easily be tested in a causal framework,
and we are reduced to speculation based on cross-country evidence. In
general, countries with more generous family policies have higher maternal employment but fewer female managers and senior officials. This
has led to speculation that Nordic family policies have created a glass
ceiling, incentivizing women to enter the labor market but stopping
them from climbing the career ladder (Datta Gupta et al., 2008).
The rest of the paper proceeds as follows. Section 1 presents the details of Nordic family policies, first for parental leave and cash-for-care
and second for subsidized child care. Section 2 provides a partial review
of the literature on the causal effect of Nordic family policies on maternal employment, distinguishing between policies that promote home
care and policies that promote child care. Section 3 summarizes the
conclusions from the literature review and provides suggestions for
how future policies can close the remaining gender gaps in the Nordic
labor markets. This section also draws on cross-country evidence to fill
the gaps unanswered by the quasi-experimental evidence.
Nordic Economic Policy Review
Nordic family policies
This section presents details on the Nordic family policies of parental
leave, cash-for-care, and subsidized child care. The Nordic approach to
family policies, combining generous parental leave with extensive provision of subsidized child care, is expensive. The Nordic countries outspend all other countries when it comes to parental leave and child care
support. These expenditures contribute to the high tax pressure in the
Nordic countries. According to the Organization for Economic Cooperation and Development (OECD), in 2011 the total tax revenue as a
percentage of the gross domestic product (GDP) was 47.7% in Denmark
and 44.2% in Sweden, which makes them the countries with the highest
tax pressure in the world. Finland follows with 43.7%, then Norway
with 42.5%, and finally Iceland with 36.0%. The unweighted average for
the OECD countries is 34.1%.
Parental leave programs and cash-for-care
Parental leave policies provide parents with a job leave of a certain
length, in order for them to take care of infant children. Parental leave
can be assigned to the mother (maternity leave) or the father (paternity
leave) or be available to either parent (parental leave). Often, these policies include a form of job protection, in that the parent on leave is entitled to return to her previous job after her leave expires. Parental leave
can either be paid or unpaid. Paid parental leave provides either a certain percentage of income replacement or a universal flat rate of benefits. Parental leave policies of some form exist in all OECD countries.
In addition to parental leave, several Nordic countries have implemented cash-for-care programs. This policy is aimed at smoothing the
transition between parental leave and formal child care, giving families
the opportunity to stay home after the parental leave expires. Cash-forcare policies are in reality extensions of parental leave, although with a
fixed benefit for staying at home instead of replacing a certain percentage of income. This makes cash-for-care benefits more beneficial for
parents with lower incomes. The take-up of these programs is relatively
low, except for Finland. Column 1 of Table 1 shows the total weeks of
Nordic Economic Policy Review
paid leave available to mothers in Nordic countries and selected comparison countries. The numbers include maternity leave, parental leave,
and total weeks of eligibility for cash-for-care benefits. The numbers are
scaled by the replacement rates, meaning that the numbers show the
number of full-rate equivalent weeks available to mothers.
The take-up of parental leave and cash-for-care benefits is gendered, with many countries providing mostly maternity leave. In general, replacing maternity leave with paternity leave (available to either
parent) does not change the gendered take-up (Datta Gupta et al.,
2008). However, the introduction of paternity leave quotas has induced
more men to stay at home during the child’s first year. The Nordic countries, with the exception of Denmark, have been the forerunners in implementing paternity leave quota policies and have paternity leave quotas larger than the OECD average. The number of full-rate equivalent
weeks of leave available to fathers in the Nordic countries and selected
comparison countries can be seen in column 2 of Table 1.
The combination of long parental leave periods and high replacement rates means that the Nordic countries spend more on parental
leave per child than the average OECD country, as can be seen in column
3 of Table 1. The Nordic system of paid parental leave includes job protection. Although there are some small differences in the length and replacement rates across the Nordic countries, most countries provide
more generous programs than the OECD average. Evidence from one
country is therefore likely to be externally valid for all Nordic countries.
Subsidized child care
The Nordic countries, with the exception of Finland, have high formal
child care enrollment rates. For 3- to 5-year-olds, Denmark, Iceland,
Norway, and Sweden have close to 100% enrollment rates (Finland has
74%) compared to the OECD average of 82%, as seen in column 5 of
Table 1. Finland’s lower enrollment rates are probably driven by the
cash-for-care program that can be used until the child is 3 years old. The
enrollment rates for 0- to 2-year-olds are lower in the Nordic countries,
largely because of the long parental leave. Still, the Danish, Icelandic,
Norwegian, and Swedish enrollment rates for 0- to 2 year olds are high120
Nordic Economic Policy Review
er than the OECD average, as shown in column 4 of Table 1. The Nordic
child care model is based on day care, meaning that children are cared
for only during standard weekday work hours. As a result, publicly provided child care is typically inflexible and does not cater to child care
needs at irregular hours.
The Nordic countries have high enrollment rates in formal child
care because of extensive subsidies. In terms of public expenditure per
child on child care and pre-primary education, the Nordic countries far
outspend the OECD average and the comparison countries, as seen in
column 6 of Table 1. Even though Finland has lower child care enrollment rates compared to the other Nordic countries, Finland still spend
similar amounts on child care support. This is because the measure of
child care support includes cash-for-care benefits, which in Finland are
generous and can be claimed up until the child is 3 years old. 5
The causal effects of Nordic family policy on
gender equality
In this section, we review studies on the effects of paid parental leave,
cash-for-care programs, and subsidized child care on maternal employment. The review is based on selected studies using Nordic data
and quasi-experimental methods aiming at identifying causal effects.
Effects of paid parental leave and cash-for-care
Even in the absence of a formal right to parental leave and job protection,
mothers are likely to take leaves around childbirth and the first weeks or
months after. The existence of a certain amount of parental leave combined with job protection could in theory increase maternal employment,
because the leave smoothens the career break of the mother around
5 With the exception of Finland, when evaluating the effects of subsidized child care on the gender gap, the
other Nordic countries have very similar systems, and the external validity of single country studies are
likely to be high for the other Nordic countries.
Nordic Economic Policy Review
childbirth and allows her to return to her previous employment. Several
studies show this to be the case: Job protection can increase the job continuity and labor force participation of women after childbirth (e.g., Waldfogel, 1998). The right to maternity leave and legal protection against job
loss ensure that women can return to employment after childbirth and
enhance gender equality in the labor market. 6
However, policies that promote longer periods of home-based care
can in theory have a negative impact on maternal labor supply. The
longer the parental leave, the longer the mother’s job disruption. In addition, longer leave periods might increase the degree of intrahousehold specialization, with the mother’s household work increasing.
The Nordic countries provide long periods of job-protected parental
leave, and because the replacement rates are high, most women are incentivized to use the entire length of the leave. If mothers have long periods outside the labor market, this can negatively impact their careers
in the long run.
Despite these theoretical predictions, the empirical results show that
the length of parental leave has no causal effect on maternal labor supply.
Carneiro et al. (2015) and Dahl et al. (2015) show that the different reforms that increased parental leave in Norway had no effect on long-run
maternal labor supply. Similarly, Liu and Skans (2010) find no effect on
mothers’ earnings after an increase in parental leave in Sweden. These
results are confirmed in studies from non-Nordic countries, such as Austria (e.g., Lalive et al., 2014) and Germany (Schönberg and Ludsteck,
2014). It seems as if you already participate in the labor market you are
not hurt by taking a long leave. This could be because the take-up rates of
parental leave are close to 100% for Nordic mothers, a sign that there is
little stigma against women taking long parental leaves.
Similar to parental leave, cash-for-care benefits are a family policy
promoting the use of home care. Because cash-for-care benefits last for
a much longer period than parental leave (in Finland, up until the child
is 3 years old), the potential negative effect on maternal employment is
6 In theory, a minimum amount of parental leave with job protection could increase fertility, because working women can have children without fear of losing their employment. We do not know of any studies that
test this prediction with a causal design.
Nordic Economic Policy Review
larger. In addition, because the take-up rates are much lower than for
parental leave, employers could discriminate against mothers who take
up cash-for-care benefits. For these policies, empirical studies have confirmed the negative theoretical effects. Cash-for-care delay mothers’ reentry in the labor market (e.g., Schøne, 2004, for Norway and Kosonen,
2014, for Finland). In addition, Drange and Rege (2013) find that mothers have lower labor market earnings in the years after the cash-forcare program runs out. However, when the child starts school at age 6,
the effects are no longer present. The reason is that cash-for-care programs do not make mothers exit the labor market altogether. The negative earnings effect in the medium run is driven by affected mothers
working part time instead of full time. When looking at sub-groups, the
negative effect of cash-for-care policies on maternal employment is
much larger for non-Western immigrant mothers (Naz, 2010). In summary, the literature consistently shows that cash-for-care programs
have a clear negative effect on the gender gap in the labor market.
A potentially important effect of policies that promote home-based
care is related to incentives for employers. If employers dislike longer
job leave periods, then generous parental leave policies can affect the
employment opportunities of not just mothers but also all women of
child-bearing age. If employers cannot know up front which women are
planning to have children, longer parental leaves create incentives
against hiring all women of childbearing age. 7 This disincentive could
decrease women’s chances of landing a first job or being promoted, especially in certain occupations. Therefore, although long periods of maternity leave can enable women to combine work and family life, long
leaves might harm women’s labor market opportunities in the long run,
by reinforcing the incentives for women to specialize in the household
sector. Such general equilibrium effects are potentially important, but it
is difficult to verify their existence empirically.
A theoretical solution to the problem of parental leave creating disincentives to hire women is to reserve parts of parental leave for fa7 This is an example of statistical discrimination, where employers discriminate against women of childbearing age because they more often have longer periods away from work. Statistical discrimination is normally
illegal, but in reality, such laws are hard to enforce.
Nordic Economic Policy Review
thers. If parental leave is split into two equal quotas, one for the exclusive use of the mother and one for the exclusive use of the father, then
there would be no disincentive effect for employers to hire women. In
addition, paternity leave quotas can potentially affect fatherhood norms
and increase fathers’ long-term involvement in the child’s upbringing.
Both theoretical predictions are unfortunately difficult to test in a causal framework.
In contrast to many other countries, the Nordic paternity leave quotas have been effective in inducing men to take parental leave. The full
effect of the policy takes time to accumulate and is reinforced by peer
effects (Dahl et al., 2014), perhaps suggestive that paternity leave quotas have slowly changed underlying gender norms. Despite the success
of paternity leave quotas in inducing men to provide home care, there is
scant causal evidence that paternity leave quotas have long-run impacts
on the employment of mothers and fathers. Paternity leave quotas do
not affect intra-household specialization and gender gaps in wages and
employment in Sweden (Ekberg et al., 2013). In Norway, the findings
are more mixed, with one study confirming the Swedish results (Cools
et al., 2015), and another finding that paternity leave decreased fathers’
long-run earnings, possibly by increasing fathers’ long-term involvement in the child’s upbringing (Rege and Solli, 2013). 8
The general finding that paternity leave quotas had little effect on
gender equality in the labor market is perhaps not surprising in light of
the literature that has found no effect of extending the length of parental leave for mothers: If there is no immediate harm to women’s employment by extending parental leave, then there are no short-run
harmful employment effects to alleviate with paternity leave quotas.
However, paternity leave quotas could have general equilibrium effects
by decreasing the incentives for employers to statistically discriminate
against women of childbearing age. This effect is theoretically reasonable but hard to measure in a causal framework.
The study measures long-run earnings up until the child is 5 years old.
Nordic Economic Policy Review
Effects of subsidized child care
Without affordable child care, some (predominantly) women face a
choice between having children and taking care of them instead of
working, or working and not having children. In theory, subsidizing
child care could therefore increase fertility (of women who had previously decided to work and remain childless), or the maternal labor supply (of mothers who previously stayed at home with her children), or
both. On the individual level, subsidizing the use of formal child care
should in theory increase maternal labor supply, by making it more attractive for mothers to work when their children are pre-school age.
Additionally, the disincentives for employers to hire women decrease
because increased use of child care instead of home care will shorten
the length of mothers’ career interruptions after childbirth.
The empirical findings on the effect of subsidized child care on maternal employment indicate that the institutional setting plays an important role. An often-cited study by Havnes and Mogstad (2011) finds
that the expansion of formal child care in Norway in the 1970s did not
increase maternal employment. Instead, the expansion led to a crowding
out of the use of informal child care, 9 suggesting a significant net cost of
the child care subsidies. However, there are important differences between the institutional setting of the 1970s and today. Most importantly,
there has been a reduction in the availability of informal child care and an
increase in child care slots for the youngest children. A more recent study
by Andresen and Havnes (2015) finds that an expansion of subsidized
child care for 1 to 2 year olds during the 2000s in Norway increased the
employment of the affected mothers. The findings from studies on nonNordic countries are mixed, although many studies have shown positive
effects (e.g., Bauernschuster and Schlotter, 2015).
Regarding the cost of formal child care, the findings are also mixed.
Black et al. (2014) find that reduced child care costs in Norway did not
increase the labor supply of affected mothers. However, the study involved only the mothers of 5 year olds due to data limitations. Similarly,
9 Informal child care arrangements are often performed by grandparents but can also involve the use of
nannies or babysitters, although these latter arrangements are uncommon in Nordic countries today.
Nordic Economic Policy Review
Lundin et al. (2008) find no effect of reduced child care costs on the maternal labor supply in Sweden. However, Simonsen (2010) shows that
in Denmark increases in the cost of child care have a significant negative
effect on the maternal labor supply. However, the effect is only in the
short run during the first 12 months after childbirth. Denmark differs
from the other Nordic countries in that the take-up of child care during
the first 12 months is significantly higher. This again shows that the institutional setting seems to matter for the effectiveness of subsidizing
child care in increasing the maternal labor supply.
Summary and suggestions for future policy 10
The Nordic countries have been at the forefront when it comes to implementing family policies that are often seen as enabling the reconciliation
of work and family life (Datta Gupta et al., 2008). The combination of
working and having children is relatively common in the Nordic countries. Family policies could be important for keeping fertility levels up,
and the Nordic countries have avoided the dramatic fall in fertility rates
that has occurred in Southern Europe during the last few decades. 11
A clear goal of the Nordic family policies has been to promote gender equality in the labor market. However, theoretically, policies that
promote the use of home care could decrease maternal employment
while policies that promote the use of child care could increase maternal employment. The empirical literature tends to confirm the theoreti10 Our policy suggestions are aimed at increasing gender equality through increasing maternal employment.
As mentioned, family policies also have other important goals such as increasing fertility and promoting
child welfare and development. The effect of family policies on these outcomes, not discussed here, should
also be considered for policy reforms.
11 Whether the higher fertility rates of the Nordic countries are caused by their family policies is hard to
answer. An ideal experiment would be to compare the fertility choices of women who live in an institutional
setting with no parental leave to the fertility choices of women who live in an institutional setting with
generous parental leave. This can be done only by comparing women who live in different countries or in
different time periods, making it nearly impossible to separate the effect of generous parental leave from
other variations across time periods and countries. A counter-argument to family policies being the main
driver of high fertility in the Nordic countries is that the fertility rate in the US, a setting without generous
family policies, is as high as in the Nordic countries. More research is needed to understand the role of family policies for fertility decisions.
Nordic Economic Policy Review
cal predictions, although the estimated effects are perhaps smaller than
expected. The findings from the literature focusing on the causal effects
of pro-home-care policies can be summarized as follows: First, job protection around childbirth promotes job stability and maternal employment. Second, expanding the length of maternal parental leave has not
decreased the long-term maternal labor supply. Third, the introduction
of paternity leave quotas has induced men to take parental leave, but
there is little causal evidence that paternity leave has changed intrahousehold specialization. Finally, cash-for-care policies are harmful to
gender equality in the labor market. In summary, dropping cash-forcare programs seems the obvious method for changing home care policies to promote more gender equality in the labor market.
The effectiveness of subsidized child care in increasing maternal
employment depends on the institutional setting. Policies that promote
the use of child care can be effective when little informal child care is
available. This is especially the case when child care is provided to
groups who previously had little access to formal and informal child
care. For Norway and Sweden, at least, mothers of 1 to 2 year olds are
“caught in between” the parental leave that expires after approximately
1 year and the fact many children do not get a spot in formal child care
when they turn 1. Expanding child care to these mothers seems to be
effective in increasing their labor supply.
Moving on from the causal evidence available, when looking at the
cross-country evidence, the Nordic countries in general do well in terms
of gender equality in the labor market. Since the 1970s and 1980s, the
employment rates for Nordic women have approached those of men,
and the Nordic countries are close to full gender equality when it comes
to the labor force participation rates of men and women. The employment rate for women with young children is far higher in the Nordic
countries compared to the OECD average, as shown in columns 7 and 8
of Table 1.
Another interesting point to note from columns 7 and 8 of Table 1 is
that Finland has low employment rates for mothers of children aged
less than 3, a pattern that fits the consistent causal evidence that cashfor-care benefits decrease maternal employment. Dropping the cashfor-care policies in favor of earlier access to subsidized child care will
Nordic Economic Policy Review
increase gender equality in the labor market. Finland has most to gain
from this policy since this country has the most generous cash-for-care
program, least generous subsidized child care, and lowest employment
rates for mothers with children younger than 3 years old. The other
Nordic countries have less to gain from changing current family policies
for the overall gender employment gap; however, the employment rates
for mothers with 1 to 2 year olds could likely be slightly increased by
dropping cash-for-care policies in favor of earlier access to subsidized
child care. 12
The remaining gender gap in the Nordic countries seems to be concentrated at the top: In terms of the gender gap among senior managers
and officials, the World Economic Forum (2013) ranked Denmark 72nd,
Finland 68th, Iceland 22nd, Norway 58th, and Sweden 44th. Arulampalam et al. (2007) show that countries with more generous family policies have larger gender wage gaps at the higher end of the income distribution. This has led to the theory that generous family policies might
create a glass ceiling, meaning that very few women enter management
jobs. Albrecht et al. (2015) argue that what is different about the Nordic
countries are the constraints that women face. The authors show relationships in Sweden between the glass ceiling and the take-up of parental leave that are consistent with generous parental leave being an explanation. Datta Gupta et al. (2008) argue that longer job leaves are
more harmful to the careers of highly educated women, and women in
the private sector. However, it is very hard to distinguish the separate
effect of generous parental leave from other potential explanations. 13
To close the remaining gender gap at the top of the career ladder,
we argue that the focus should be on making the existing family policies
more flexible. Public child care typically offers care only during regular
work hours, which is probably insufficient to cater to the needs of
12 Take-up of cash-for-care programs is higher among non-Western immigrants, and the negative effect on
maternal employment is higher for this group (Naz, 2010). Dropping cash-for-care programs in favor of
increased access to subsidized child care could therefore also be seen as a tool to promote social integration
of non-Western immigrant women.
13 Other potential reasons for the glass ceiling effects could be statistical discrimination, gender differences
in career orientation, or a limited market for household services (that women in top jobs may need to succeed) because of the compressed wage structure in the Nordic countries.
Nordic Economic Policy Review
women in jobs that demand regular overtime. The statistics seem to indicate that this is a potential problem: Women in Nordic countries select part-time work and work less overtime, fewer long hours, and fewer evenings compared to men (Plantenga and Remery, 2010). There is
also suggestive evidence that providing more flexible child care arrangements can boost the full-time labor supply of women. Johnsen
(2015) finds that grandparents’ retirement can increase adult daughters’ labor supply through providing child care. If grandparents’ child
care can increase maternal labor supply in Norway, a country with high
coverage rates of formal child care, the effect is likely to be driven by
child care during irregular hours. To achieve more flexibility in the system, potential policy reforms include admitting children to formal child
care twice a year instead of once a year, and increasing the options for
women to substitute parts of the parental leave for subsidized child
care. Another option is to subsidize informal child care at irregular
hours (nannies at home, pick-up-services at day care), or other domestic services. Hallden and Stenberg (2013) evaluate a policy in Sweden
that gives a 50% tax deduction on domestic services. They find that the
policy increased the labor supply of married women. The authors cannot isolate the effect on career-oriented women; however, if anything,
the increase in labor supply was similar for women of various socioeconomic backgrounds. Thus, this type of policy might help reduce the
overall gender gap in the labor market (for example, pushing mothers
from part-time to full-time work), as well as increase the share of female managers and top earners.
Albrecht, J., Skogman Thoursie, P., & Vroman, S. (2015). Parental leave and the glass
ceiling in Sweden. Research in Labor Economics, 41, 89–114.
Andresen, M.E., & Havnes, T. (2014). Women and children first? Labor market effects
of universal child care for toddlers. Working Paper.
Angelov, N., Johansson, P., & Lindahl, E. (2013). Is the persistent gender gap in income and wages due to unequal family responsibilities? IZA Discussion Papers 7181.
Nordic Economic Policy Review
Arulampalam, W., Booth, A.L., & Bryan, M.L. (2007). Is there a glass ceiling over Europe? Exploring the gender pay gap across the wage distribution. Industrial and
Labor Relations Review, 60, 163–186.
Bauernschuster, S., & Schlotter, M. (2015). Public child care and mothers’ labor supply
– Evidence from two quasi-experiments. Journal of Public Economics, 123, 1–16.
Becker, G. (1981). A Treatise on the Family, Harvard University Press, Cambridge, Mass.
Bertrand, M., Goldin, C., & Katz, L.F. (2010). Dynamics of the gender gap for young
professionals in the financial and corporate sectors. American Economic Journal:
Applied Economics, 2, 228–255.
Björnberg, U. (2013). Introduction. In U. Björnberg and M.H. Ottosen (eds.), Challenges for Future Family Policies in the Nordic Countries, SFI, Copenhagen.
Black, S.E., Devereux, P.J., Løken, K.V., & Salvanes, K.G. (2014). Care or cash? The
effect of child care subsidies on student performance. Review of Economics and
Statistics, 96, 824–837.
Bradshaw, J., & Hatland A. (2007). Social Policy, Employment and Family Change in
Comparative Perspective, Edgar Elgar, London.
Carneiro, P., Løken, K. V., & Salvanes, K. (2015). A flying start? Maternity leave and
long run outcomes of children. Journal of Political Economy, 123, 365–412.
Cools, S., Fiva, J.H., & Kirkebøen, L.J. (2015). Causal effects of paternity leave on children and parents. Scandinavian Journal of Economics, 117, 801–828.
Dahl, G., Løken, K.V., & Mogstad, M. (2014). Peer effects in program participation.
American Economic Review, 104, 2049–2074.
Dahl, G., Løken, K.V., Mogstad, M., & Salvanes, K. (2015). What is the case for paid
maternity leave? Review of Economics and Statistics, forthcoming.
Datta Gupta, N., Smith, N., & Verner, M. (2008). The impact of Nordic countries’ family friendly policies on employment, wages, and children. Review of Economics of
the Household, 6, 65–89.
Drange, N., & Rege, M. (2013). Trapped at home: The effects of mothers’ temporary
labor market exits on their subsequent work career, Labour Economics, 24, 125–136.
Ekberg, J., Eriksson, R., & Friebel, G. (2013). Parental leave – A policy evaluation of
the Swedish “Daddy-Month” reform. Journal of Public Economics, 97, 131–143.
Ellingsæter, A.L. (2006). The Norwegian childcare regime and its paradoxes, in A.L.
Ellingsæter and A. Leira (eds.), Politicising Parenthood in Scandinavia. Gender Relations in Welfare States, The Policy Press, Bristol.
Halldén, K., & Stenberg, A. (2013). The relationship between hours of domestic services and female earnings: Panel register data evidence from a reform. IZA Discussion Papers 7649.
Havnes, T., & Mogstad, M. (2011). Money for nothing? Universal child care and maternal Employment. Journal of Public Economics, 95, 1455–1465.
Johnsen, J.V. (2015). Work and Families: Four Essays in Applied Microeconomics. University of Bergen, Bergen.
Nordic Economic Policy Review
Kosonen, T. (2014). To work or not to work? The effect of childcare subsidies on the
labour supply of parents. BE Journal of Economic Analysis and & Policy, 14, 817–848.
Lalive, R., Schlosser, A., Steinhauer, A., & Zweimüller, J. (2014). Parental leave and
mothers’ careers: The relative importance of job protection and cash benefits. Review of Economic Studies, 81, 219–265.
Liu, Q., & Skans, O.N. (2010). The duration of paid parental leave and children’s
scholastic performance, BE Journal of Economic Analysis & Policy, 10.
Lundborg, P., Plug, E., & Rasmussen, A.W. (2014). Fertility effects on female labor
supply: IV evidence from IVF treatments. Working Paper.
Lundin, D., Mörk, E., & Öckert, B. (2008). How far can reduced childcare prices push
female labour supply? Labour Economics, 15, 647–659.
Naz, G. (2010). Effect of a family policy reform on immigrants’ labor supply and
earnings. Labour, 24, 74–92.
Nielsen, H.S., Simonsen, M. & Verner, M. (2004). Does the gap in family friendly policies drive the family gap? Scandinavian Journal of Economics, 106, 721–744.
Plantenga, F., & Remery, C. (2010). Flexible Working Time Arrangements and Gender
Equality. Office for Official Publications of the European Communities, Luxembourg.
Rege, M., & Solli, I.F. (2013). The impact of paternity leave on fathers’ future earnings. Demography, 50, 2255–2277.
Schönberg, U., & Ludsteck, J. (2014). Expansions in maternity leave coverage and
mothers’ labor market outcomes after childbirth. Journal of Labor Economics, 32,
Schøne, P. (2004). Labour supply effects of a cash-for-care subsidy. Journal of Population Economics, 17, 703–727.
Simonsen, M. (2010). Price of high-quality daycare and female employment, Scandinavian Journal of Economics, 112, 570–594.
Waldfogel, J. (1998). The family gap for young women in the United States and Britain:
Can maternity leave make a difference? Journal of Labor Economics, 16, 505–545.
World Economic Forum (2013). The Global Gender Gap Report 2013, World Economic Forum, Geneva.
Nordic Economic Policy Review
Table 1: Indicators of family policies and gender equality in the Nordic countries compared to the OECD average and a selection of comparison countries
Full-rate equivalent total
paid weeks of parental
Public ex- Pre-school enrollment rates
penditure on
maternity and
leaves per 0–2 year olds 3–5 year olds
child born in
Expenditure on
childcare support
and pre-primary
education per
child in USD (PPP
Maternal employment
Female share
of managers
divided by
female share
of total employment
Available to
Reserved for
OECD average
* Data from 2009.
** Employment rate for all women aged 25–39.
*** At least one child aged 0–14, data from 2002.
Source: All data from the OECD online family database:
Nordic Economic Policy Review
child 0–2
Gender gap
in full-time
Youngest rates (15–64)
child 3–5
7. Education and equality of
What have we learned from
educational reforms1
Helena Holmlund, Institute for Evaluation of Labour Market and Education Policy (IFAU)
Equality of opportunity has been one of the central ideas governing education policy in the Nordic welfare state. This paper takes its starting
point in the shared history of educational reform in the Nordic countries,
and presents evidence that the comprehensive school reforms that implied
a shift from selective two-tier schooling systems to unified compulsory
schools were beneficial for equality of opportunity. This evidence is compared to a choice and voucher reform that in the 1990s introduced pedagogical as well as organizational variety in the education system in Sweden. The Swedish choice reform is unique in an international perspective,
and has reshaped the education sector dramatically as a growing number
of pupils attend non-public independent schools. The current education
debate shows a widespread concern that the introduction of choice has
led to a backlash for equality of opportunity. However, recent research
finds no indication that parental background has become more important
1 I am grateful to Anders Björklund, Bertil Holmlund, Kristian Koerselman, Jesper Roine and conference
participants at the NEPR conference in Helsinki, October 2015, for valuable comments.
in explaining pupil outcomes. The Swedish education system nevertheless faces a number of challenges if it is to level the playing field and create equal opportunities for all pupils: schools are becoming increasingly
more segregated, much as a consequence of immigration, and disadvantaged pupils are less likely to exercise school choice compared to their
more advantaged peers. This development calls for a debate on policies
that aim to limit school segregation.
There is a widely shared belief that all children, regardless of family
background, should face equal life chances and have equal opportunities to succeed in life. Family background, in a broad sense, can be referred to as background factors not chosen by the individual, and inequalities in outcomes that are related to family background are therefore considered unfair. Despite this belief, parental background remains
a very strong predictor for children’s educational attainment and for
success in the labour market in general. For example, in PISA (Programme for International Student Assessment), all surveyed countries
show a strong socioeconomic gradient with respect to students’ results
(National Agency for Education, 2013) and correlations between parents’ and children’s years of schooling range from about 0.20 in the
Nordic countries to about 0.5 in the U.S. (Holmlund et al., 2011). Because of these strong correlations, policy makers often turn to the education system as a tool to level the playing field and create equal opportunities for children from different backgrounds.
The Nordic countries have a shared history of educational reform,
including the expansion of publicly subsidized pre-school and the
school reforms that gradually expanded and re-modelled compulsory
education. One of the explicit aims of these reforms was to enhance
equality of opportunity and increase social mobility, and as such, these
policies played an important role in the construction of the welfare
state. Research shows that in the Nordic countries, the relationship between parents’ and children’s socio-economic status is weaker than in
for example the U.S. (Björklund and Jäntti, 1997; Björklund et al., 2002).
Nordic Economic Policy Review
Studies in this field typically estimate the association between parents’
and children’s education or earnings, and the stronger the association,
the more important is family background for an individual’s life outcomes, and the lower is social (or intergenerational) mobility. There are
many different reasons why social mobility has been higher in the Nordic countries compared to many other nations, but the strong emphasis
on equality of opportunity in education is one potential candidate that
may explain this pattern. 2
From a theoretical perspective, there are several reasons for which
the design of the education system can be important for the intergenerational persistence in, e.g., educational outcomes. Public investment in
education may help poor families to overcome credit constraints, and
give all children access to education throughout the education system.
Public policy can also affect intergenerational persistence through its
interplay with educational choices (Björklund and Salvanes, 2011). If
families of different socio-economic status hold different information
about school quality and future returns to education, or have different
preferences for their children’s education, elements of choice through
early tracking or choice of school may imply a stronger intergenerational relationship compared to a comprehensive system with limited
options to choose between different tracks, pedagogical orientations or
schools. In the latter case, the degree of parental influence over educational choices will be limited.
The Nordic countries all reformed their compulsory school systems
throughout the 1950s–1970s. Common features were to extend the
length of compulsory education, and to postpone the differentiation of
pupils into different educational tracks (typically vocational and academic tracks) to a higher age. Since the early 1990s however, school
reforms in Sweden have introduced new elements to the education system, which distinctly differentiates Sweden from most other countries. 3
2 In this paper the term “social mobility” is used broadly, as a synonym to “intergenerational mobility”, and
refers to mobility in income and educational outcomes. In the sociology literature, “social mobility” relates
more closely to the concept of social class (see Erikson and Goldthorpe, 1992).
3 Sweden is one of few countries in the world with a generalized voucher system and independent voucher
schools face relatively loose regulations. In the Nordic countries, Denmark has a long tradition of independent schools (around 13% of all compulsory school pupils attend an independent school), which are often
Nordic Economic Policy Review
In the 1980s, critics had started to question the long-lasting socialdemocratic influence over the school and education system, and in particular advocated more diversity in educational provision, and the right
for parents to exert influence over their children’s education (National
Agency for Education, 2003). This debate eventually led to the passing
of a series of educational reforms in Sweden. In the early 1990s, a
choice and voucher reform implied that private schools were granted
public funding, and greater opportunities for pupils to opt out of their
assigned local public schools to either attend a voucher-funded private
(or “independent”) school, or a public school outside of their catchment
area. During the same time period, the public school system was decentralized and the previous state control was replaced by a municipality
(i.e. local authority) maintained system, with the purpose to let local
priorities play a larger role in e.g., resource allocation.
Since the introduction of universal vouchers, many new independent voucher-funded schools have opened in Sweden. As of 2014, about
14% of all compulsory school pupils attend an independent school. 4
These schools represent various pedagogical ideas and religious affiliations, but it is also common that for-profit corporations run schools
with a general profile.
There were two main arguments for school choice and diversity in
educational provision put forward in the debate: parents should have a
right to choose their child’s school, and competition between schools
and different pedagogical ideas should increase efficiency in the school
system and improve educational outcomes (Government bill
1991/92:95). Opponents of the reforms, on the other hand, argued that
a choice-based system with independent schools would lead to segregation and inequality, and that the reforms were at odds with the ideals of
an inclusive education system providing equal education to all pupils,
regardless of family background (Opposition bill 1991/92: Ub62). In the
run by parent co-ops with religious or pedagogical profiles. Ownership regulations are different than in the
Swedish system and imply that schools cannot be run as for-profit corporations. In Finland and Norway,
independent schools are rare and only about 2% of compulsory school level pupils attend such a school
(SOU 2013:56).
4 Calculation based on statistics available at
Nordic Economic Policy Review
current education debate, 25 years after the passing of the reform, equity concerns are again on the policy agenda: between-school inequality is
rising both in terms of test scores and parental background (Holmlund
et al., 2014). In tandem, Sweden’s scores in PISA have declined sharply
and this empirical correlation has brought additional light on the topic
of choice, segregation and educational performance.
The purpose of this paper is twofold. The first purpose is to summarize the existing evidence on the role of educational reforms for intergenerational mobility in the Nordic countries. Specifically, the paper
will compare comprehensive schooling reforms that took place in the
1950s–1970s, to the choice and voucher reform in Sweden in the 1990s. 5
Importantly, the latter reform has affected more recent cohorts that
have not yet completed their education and entered the labour market.
Research on intergenerational mobility in relation to this reform has
therefore used age 16 school performance to proxy for intergenerational mobility. That is, using intermediate outcomes that are highly correlated with long-run education and labour market outcomes, it is possible to shed light on more recent trends in mobility.
The second purpose is to identify and discuss future challenges to
the ideal of equality of opportunity in education, in light of recent developments in Swedish society. Reconciling the literature, the main conclusions of the paper are that the comprehensive school reforms likely
have improved social mobility, while the more recent school choice reform in Sweden does not seem to have affected short-run outcomes that
proxy for social mobility. The Swedish education system nevertheless
faces a number of challenges if it is to level the playing field and create
equal opportunities for all pupils: school segregation has been on the
rising for a long time, much as a consequence of immigration, and disadvantaged pupils are less likely to exercise school choice compared to
their more advantaged peers.
The remainder of the paper is structured as follows: section 2 discusses some theoretical predictions relating to intergenerational mobil5 This paper will not cover the literature on the expansion of subsidized pre-school education in the Nordic
countries. Evidence from Norway shows that provision of subsidized pre-school has increased intergenerational mobility (Havnes and Mogstad, 2011).
Nordic Economic Policy Review
ity and the workings of the school system; section 3 summarizes the
evidence from the compulsory schooling reforms in the Nordic countries; section 4 moves on to more recent evidence on the choice and
voucher reform in Sweden and section 5 outlines future challenges to
equality of opportunity in the Swedish school system. Finally, section 6
concludes with a short summary and discussion.
Theoretical considerations related to the intergenerational transmission of human capital
Economic models of human capital take their starting point in a model
of investment, originating from Becker and Tomes (1979). In such a
model, a child’s human capital, defined broadly as for example education, and cognitive and non-cognitive skills, will be determined by parental inputs (both in terms of nature and nurture) and by public investment in education. In this framework, one would expect a positive
correlation between child’s and parent’s human capital (or income), in
part because of the genetic transmission of abilities, and in part because
credit constraints might imply that well-off families can afford to invest
more in their offspring’s education than poor families. Parental investment in human capital should be interpreted not only as formal education, but also as parents’ own time spent with the child, the quality of
time spent together, and their parenting skills.
The optimal parental investment will depend on factors such as the
return to human capital, the degree of public investment, and the institutional set-up of the school system – and therefore these factors may in
turn affect the strength of the intergenerational associations. As an example, a model by Solon (2004) shows that intergenerational income
mobility is decreasing in the returns to human capital and increasing in
the degree of public investment in education.
Public investment in education is clearly a policy that relaxes credit
constraints for poor families and the theoretical prediction is therefore
that it should lower intergenerational persistence in economic outcomes. But how can institutional arrangements such as ability tracking,
Nordic Economic Policy Review
comprehensive vs. selective systems, and school choice affect the intergenerational link and social mobility? 6
First, consider the use of ability tracking within a school, i.e., pupils
are sorted into different classes based on their previous school performance. As suggested by Betts (2011), in a system without tracking, affluent parents might invest in private tutoring on top of public education, in order to obtain good results and gain access to the best universities. In a system with ability tracking, on the other hand, well-off parents might see less need for extra tutoring if they consider placement of
their children in a high track as a substitute for private investment.
Second, how does parental background influence educational choice
in a selective two-tier system compared to a comprehensive system?
Early differentiation is argued to increase the importance of parental
characteristics for educational choice. The earlier the choice is made,
the less accurate is information about pupil ability, which tends to put
weight on parental socioeconomic background rather than pupil ability
in the educational decision (Björklund and Salvanes, 2011). Moreover,
the younger is the child, the stronger is the influence of parental, rather
than the pupils’ own, preferences for education. The prediction is therefore that postponing differentiation into academic and vocational tracks
to a later age will limit the importance of family background for educational choice, and therefore be beneficial for social mobility.
Third, school choice policies may also be of importance for intergenerational mobility. On the one hand, if parental preferences for education vary by family background, or if families have different access to
information about school quality and different abilities to interpret such
information, it is possible that choice leads to a segregated school system where the type or quality of the school attended is correlated with
family background. On the other hand, it can be argued that school
choice provides an opportunity for disadvantaged children to “escape”
low-performing schools in poor neighbourhoods and the link between
school quality and family background should therefore weaken.
6 This question highlights the potential trade-off between efficiency and inequality, where proponents of
tracking argue that tracking increases efficiency, while opponents point to increased inequality as a result of
separating pupils into different tracks or schools (Betts, 2011).
Nordic Economic Policy Review
Most empirical studies that focus on the decision of opting out from
assigned schools find that advantaged families are more prone to opting
out (see Böhlmark et al., 2015 for a summary), which implies that increased sorting seems to be the dominating factor. In such systems,
children with a favourable background will access the best schools,
which will strengthen the association between parents’ and offspring’s
outcomes. School choice within the publicly funded school system can
thus lead to sorting, but it can also limit the demand for highly selective
tuition-charging private schools (Epple et al., 2002). High demand for
such schools could potentially lead to sorting between the public and
private sectors, and imply very strong correlations between family
background and children’s educational and labour market outcomes.
All in all, summing up the theoretical predictions above, it can be argued that a comprehensive system without the possibility of opting out
of assigned schools will tend to limit the influence of parental background for educational outcomes. However, in such a system, parental
preferences for high-quality selective education or for a specific peer
group composition might open up for a market of tuition charging private schools to which only affluent families have access. This scenario
predicts that family background will have a strong influence on educational success. Which theoretical mechanism that dominates will ultimately be an empirical question.
Compulsory schooling reforms, tracking and
intergenerational persistence
The compulsory schooling reforms
In the 1940s and 1950s, children in the Nordic countries commonly
started school at the age of 7 and went to a common primary school for
4–6 years, after which they were split into two tracks: remaining in
Nordic Economic Policy Review
primary school or attending a general lower secondary school. 7 Pupils
who remained in primary school completed 7–8 years of schooling,
which at the time was the compulsory minimum. Admission to lower
secondary school was typically based on school grades, teacher assessments and entrance examinations. Attending lower secondary school
implied leaving school after 9–10 years, if not continuing further at the
upper secondary level. This two-track selective system differentiated
pupils at the age of 10–12, into paths that were distinctly different: the
secondary school path had a more academically oriented curriculum
and prepared for future academic studies, while the primary school
path was followed by vocational training and did not grant access to
academic studies.
This educational model was reformed and replaced by a comprehensive school with a similar curriculum for all pupils from 1st to 9th
grade. 8 One explicit motivation for these reforms in the Nordic countries was to give all children, irrespective of family background, the
same basic education, that is, to provide equal educational opportunities for all. Early differentiation was considered a disadvantage for children from low-educated households who disproportionally remained in
primary school, and the reforms aimed at raising these children to a
higher academic level. 9
One common feature of the reforms was that they were rolled out
gradually across different regions. In Sweden, the reform started at a
small scale in 1949 and was implemented throughout the 1950s. In
Norway, the reform period spanned 1960–1972. In Finland, the comprehensive school was introduced in the 1970s, but at this point in time
the length of schooling had already been harmonized for different
7 This section builds on Meghir and Palme (2005), Arendt (2005), Pekkarinen et al. (2009) and Aakvik et al.
8 In Sweden, some tracking initially remained within the comprehensive system, but was abolished in 1969
(National Agency for Education, 1969).
9 Equality of opportunity was an explicit motivation for the reforms, but also other motivations are worth
highlighting. In Sweden for example, demand for education beyond the compulsory level was growing
among the baby boom cohorts born in the 1940s, and the education level was increasing in the population
already before the reform was rolled out. The compulsory school reform was therefore also a means to meet
increased demand for education in the population (see Holmlund, 2007).
Nordic Economic Policy Review
tracks, and as such the reform served mainly to merge the two-tier system into a comprehensive school system. The gradual implementation
has been of great importance to researchers, who have been able to
evaluate the reforms by comparing individuals who went to school in
different regions at different points in time. The exception is Denmark,
where reforms were implemented throughout the country at a single
point in time: In 1958 early sorting was limited, and in 1975 the two
track-system was fully abandoned and the years of compulsory education increased from 7 to 9. The next section summarizes some of the
findings from this research.
Causal effects of compulsory schooling reforms
on intergenerational mobility
Although there are some theoretical predictions that relate features of
the education system, such as ability tracking, to social mobility, it is a
challenging task to provide empirical evidence on this topic. As already
mentioned, comparisons of intergenerational income mobility in different countries have shown that mobility is higher in the Nordic countries
than in continental Europe and in the U.S. and the U.K. (Björklund and
Jäntti, 1997, Blanden, 2013). This pattern is often interpreted as a success of the Nordic welfare state, but cross-country comparisons are not
enough to understand the underlying mechanisms of specific policies,
and may also be explained by other cross-country differences. When it
comes to education policy and ability tracking specifically, most crosscountry comparisons show that tracking increases inequalities and the
role of family background for educational outcomes (Amermüller, 2005,
Hanushek and Wossman, 2006, Brunello and Checci, 2007). 10 These
studies all share the limitations of a cross-country comparison: it is not
obvious that the analysis can account for all the “other” relevant factors
that explain differences in outcomes between countries.
See Waldinger (2007) for an exception.
Nordic Economic Policy Review
More recently, experimental designs have been used to compare the
outcomes of ability-grouped vs. mixed-group pupils. Evidence from
primary school children in Kenya (Duflo et al., 2011) and university
students in the Netherlands (Booij et al., 2015) shows that tracking can
have positive effects on outcomes both for low and high ability students,
and that the positive tracking effects for low ability students can be
higher than the positive peer effects that would be incurred from a
mixed peer group.
While the cross-country variation is potentially unsuitable for causal inference, the recent experiments are more compelling when it comes
to internal validity of the estimates. Nevertheless, their results are not
always generalizeable to a system-wide context and they provide limited evidence on the effects of reforms at the institutional level. Instead,
a large body of research based on the comprehensive school reforms in
the Nordic countries has provided evidence on the causal effects of a
system-wide change to the education system on a range of outcomes.
These within-country comparisons exploit the gradual roll-out and are
therefore able to isolate the effects of the reforms in a setting where it is
unlikely that other factors confound the results.
The first paper to exploit the gradual implementation in order to
study the effects of a comprehensive school reform in the Nordics was
Meghir and Palme’s (2005) influential study on the Swedish reform. 11
The main findings from their paper are that the reform increased years
of schooling and lifetime earnings for children with low-skilled fathers,
while there is a tendency to find negative earnings effects for children
whose fathers were high-skilled. These results imply that the reform
contributed to reducing inequalities in labour market outcomes by
family background, and as such, one of Meghir and Palme’s conclusions
is that the reform increased intergenerational mobility.
The Finnish reform, explored in two papers by Pekkarinen et al.
(2009; 2013) provides evidence in line with the study of the Swedish
case. Pekkarinen et al. (2009) estimate the intergenerational income
11 Erikson (1996) had previously studied equality of opportunity in the light of the Swedish reform, and
found that the introduction of the comprehensive school coincided with reduced inequality in education.
Nordic Economic Policy Review
elasticity, which is the regression coefficient from a regression of offspring’s log life time income on parents’ log lifetime income. A high value indicates strong income persistence between generations (i.e., low
social mobility) and a low value indicates high social mobility. They find
that the reform reduced the intergenerational income elasticity by 23%,
from 0.30 to 0.23. This is a sizeable effect and is backed up further by
the 2013 study which finds that the reform improved the cognitive
skills of boys with low educated parents, while not affecting boys whose
parents had some education above compulsory level. 12
Estimates for Norway are in line with the evidence presented above,
and lend further support to the conclusion that the comprehensive
school reforms contributed to increasing intergenerational mobility.
Aavik et al. (2010) find that the effect of family background on educational attainment was lowered as a result of the Norwegian reform. The
Danish reform, which was instituted throughout the country at a single
point in time, does not allow for an evaluation design similar to that of
the studies described above. As a consequence there is no published
research that can complete the picture with estimates from Denmark.
Before concluding this section, it is worth noting that these studies
build on regression models that relate children’s education and earnings to their parents’ education and earnings. While these observed
measures of socio-economic status are highly relevant, alternative empirical methods provide additional evidence to support the findings
presented above. Björklund et al. (2009) make use of “sibling correlations” to capture the importance of family and community background
for children’s labour market success. The sibling correlation estimates
how much of the variation in outcomes, in this case life time income,
that can be attributed to the family in a broad sense. If siblings are very
similar (the sibling correlation is high), shared family background factors, such as socio-economic status, marital stability and parenting
skills, as well as shared community factors including the school and the
neighbourhood, are important for children’s outcomes. If the sibling
12 Similar estimations of the reform effect on the intergenerational income elasticity in Sweden can be found
in Holmlund (2008), who finds that the Swedish reform reduced the elasticity by 12%.
Nordic Economic Policy Review
correlation is low and siblings are not alike in terms of outcomes, family
and community background seems to be a less important explanation.
The sibling correlation, by virtue of capturing all sorts of shared background factors – also factors that cannot be observed in the data – is
considered an important complement to the more traditional regression-based measures that link parents to their children. Figure 1 shows
the brother correlation in income, as estimated in Björklund et al.
(2009). 13 The figure displays a sharp decline in the brother correlation,
from 0.49 for cohorts born 1932–1938 to 0.32 for cohorts born around
1950, thus suggesting that family and other shared background factors
became less important determinants of long-run income in adulthood.
In additional analyses, it is shown that the decline can be explained by
changes in the distribution of years of education. While the early decline
pre-dates the comprehensive school reform, these results may reflect
earlier expansions of compulsory education (see Fisher et al., 2013 for a
description of earlier reforms) and are suggestive of the importance of
the education system for intergenerational mobility.
13 Figure 1 is based on full biological brothers born at most seven years apart. One concern with comparing
the sibling correlation over time is that it might be sensitive to trends in child spacing and family structure.
Björklund and Jäntti (2012) find that sibling similarities are not much affected by age differences of siblings
in the sample and they therefore conclude that shared permanent factors are most important in explaining
the sibling correlation. The sibling correlations in Figure 3 are based on siblings with the same mother (i.e.
also half siblings on the maternal side) born at least three years apart. The conclusions from Figure 3 are
robust also to alternative specifications including only full biological siblings.
Nordic Economic Policy Review
Figure 1: Brother correlations in income in Sweden
Source: Björklund et al. (2009).
In the next section, indicators of social mobility will be related to more
recent reforms to the education system in Sweden.
Parental choice and the importance of family
background for pupil achievement
The research presented in the previous section shows that the shift
from a selective to a comprehensive system, as manifested by the reforms in the Nordic countries in the 1950s–1970s, reduced inequalities
in educational attainment and in life time income by family background.
In this section, these results will be contrasted with a more recent re-
Nordic Economic Policy Review
form in Sweden that in the early 1990s introduced universal school
vouchers and generalized school choice. 14
The choice and voucher reform
In 1992, Sweden introduced a nation-wide voucher system, which facilitated public funding of private schools, and introduced school choice. In
the pre-reform period, with institutions dating back to the comprehensive system described above, almost all pupils attended the local school
in their catchment area; less than one percent of all pupils attended a
private school. The reform was based on two main elements. First, privately maintained “independent” schools receive public funding
through school vouchers after having gained approval by the Swedish
National Agency of Education, NAE. Pupils’ home municipalities have to
provide independent schools with a grant, based on the average perpupil expenditure in the public school system, for each pupil who
chooses to enrol in an independent school. This new law has given rise
to a large number of new schools, whose existence depends solely on
funding through vouchers. The number of independent schools (at the
compulsory level) has increased, from about 170 registered independent schools in 1993 to 800 in 2014. In 2014, about 14% of all compulsory school pupils attended an independent school. Independent schools
are open to all pupils: by law they are not allowed to charge tuition fees
on top of the voucher, nor can they select pupils by ability or family
background. If an independent school is oversubscribed, three selection
criteria for admission are allowed: proximity to the school; waiting list
(by date of application); and priority for children whose older siblings
are already enrolled in the school. 15 Second, the voucher reform also
For more details on the reform and the current institutional setting, see SOU 2013:56.
Independent schools were initially allowed to charge moderate tuition fees, but the right to charge fees
was abolished in 1997. The funding rules have varied over time: at the outset of the reform, the voucher
should amount to 85% of per-pupil expenditure in the public schools in the municipality. After 1997, the
voucher to independent schools should be determined “on the same grounds as to public schools” (Government bill 1994/95:157). Independent schools can have different types of governing bodies, for example
non-profit foundations and for-profit corporations. A majority of pupils attend schools with a general profile, belonging to a for-profit corporation (Swedish Association of Independent Schools 2015).
Nordic Economic Policy Review
introduced choice between public schools, although maintaining priority for pupils residing close to the public school. Slots are first allocated
to pupils within the public schools’ catchment areas, after which pupils
from other areas can be granted admission.
School choice and competition cannot explain the sharp decline in
Swedish PISA scores (Böhlmark and Lindahl, 2015, Wondratschek et al.,
2013). The current Swedish education debate is nevertheless strongly
concerned that inequalities in educational outcomes are on the rising,
and it is often believed that the importance of family background for
educational success has increased as a result of the choice and voucher
reform (National Agency for Education, 2009). The rationale for these
concerns is that the possibility to exert school choice and the probability of gaining access to the preferred school might be related to a pupils’
family background, as discussed in Section 2. Even though the institutional set-up is designed with the purpose of giving equal access to children from different family backgrounds, socio-economically advantaged
families might have better information about school quality and about
the waiting list principle at oversubscribed schools, which in turn implies a higher probability of accessing a popular, high-quality, school.
While research based on the Nordic comprehensive school reforms
has focused on the relationship between family background and adult
outcomes, such as completed education (years of schooling), earnings
or income, recent studies of the Swedish choice and voucher reform
have limited the analysis to intermediate outcomes such as grade point
averages (GPA) and test scores at age 16. The rationale for this is that
the cohorts that are included in the analyses have been too young for
long-term outcomes to be relevant (cohorts born 1972–1993). Since
school results at age 16 are good predictors of long-term labour market
outcomes, the results can nevertheless serve as a proxy for intergenerational mobility in the cohorts affected by the choice reform. The next
section presents evidence that relates indicators of intergenerational
mobility to the choice and voucher reform.
Nordic Economic Policy Review
Parental background and school performance
The first study to assess the family background gradient in school performance, in the wake of the choice reform, was Björklund et al. (2003).
Since then, a number of studies have followed suit and reported results
on this topic (see for example Gustafsson and Yang-Hansen, 2009, Fredriksson and Vlachos, 2011, National Agency for Education, 2012, Böhlmark and Holmlund, 2012). More recently, Holmlund et al. (2014) provide a range of results, based on various data sources, which describe
how the importance of family background for school performance at
age 16 has evolved in the period 1988–2009. The main results are presented in Figures 2 and 3.
Figure 2 shows the GPA difference, expressed in standard deviations,
between pupils characterized by high and low parental education and
income, respectively. The vertical line in 1988 reflects a grading reform.
First, the graph reveals that there are sizeable differences in terms of
school performance for pupils of different background. The difference
between children of high and low educated parents is about 0.7 standard
deviations; slightly below 0.6 comparing high and low income, and the
differences remain stable until 1998. The new grading system implied a
shift towards smaller differences between groups: family background has
a lower explanatory power of school performance in the new system. The
GPA difference by parental education turns out to be relatively stable
over time also after the new grading system was introduced: if anything,
the gap between pupils with high and low educated parents is shrinking.
On the contrary, differences by parental income are increasing throughout 1998–2009. This pattern can be the result of an increasing role for
parental income in determining school performance, but it cannot be excluded that the result is explained by compositional effects that are linked
to increased income inequality in the population. The group defined as
high income (the top 25% of the income distribution) is gradually becoming richer during the study period, which can in itself affect the GPA difference presented in Figure 2.
Nordic Economic Policy Review
Figure 2: Differences in age-16 GPA by family background
Parental education
Parental income
Notes: The figure shows differences in the GPA (based on Swedish, English and mathematics) between pupils with different family background. The lines show the differences between pupils characterized by high/low parental education or income. Highly educated parents are defined as at least
one parent belonging to the top 25% of the education distribution (by parents’ birth year and gender); the rest are defined as low educated. High income parents are defined as at least one parent
belonging to the top 25% of the income distribution (by parents’ birth year and gender), when income is measured in the age range 35–45. Remaining parents are defined as low income parents.
The vertical line indicates a change in the grading system.
Source: Holmlund et al. (2014).
Next, Figure 3 presents sibling correlations in GPA at age 16. The correlations reveal that family and community background factors shared by
siblings can explain about 50% of the variation in school performance,
as measured by the GPA. Thus, siblings’ shared background explains a
large fraction of the variation, especially compared to the school which
Nordic Economic Policy Review
explained 11% of the variation in 2009. 16 As in the previous graph, the
shift to a new grading regime is demonstrated as a break in the time
series. Importantly, there appears to be no trend in the sibling correlation over time, which means that family background factors have become neither more nor less important in shaping pupils’ age-16 GPA.
Figure 3: Sibling correlations in age-16 GPA
Notes: The figure shows sibling correlations (fraction of variance explained by the family) in 9th
grade GPA. Calculations are based on siblings born at most three years apart. The vertical line indicates a change in the grading system.
Source: Holmlund et al. (2014).
The key findings from Figures 2 and 3 are that family background factors are very important for school performance – but their importance
is remarkably stable over time. While there is some evidence indicating
that differences are increasing by parental income, the other measures
16 See Holmlund et al. (2014) for variance decompositions of GPA at the municipality, school and family
Nordic Economic Policy Review
adopted do not confirm the notion that family background has become a
stronger determinant of educational success in Sweden. 17 Based on
these time-series from 1988–2009, it seems like the choice-based system introduced in the 1990s has not affected intergenerational mobility,
at least not in the short run. Admittedly, this conclusion is based on the
observation of a time series, and does not rely on a solid identification
strategy as was the case for the comprehensive school reforms. It cannot be ruled out that introducing choice has increased the social gradient in school performance, but other changes in society may have had
effects going in the opposite direction and therefore cancelled out the
effects of choice (see Björklund et al., 2003).
The Swedish model: lessons learned and challenges for the future
The previous section provided evidence indicating that the growing sector of independent schools and the possibility to opt out of the local public school, have not strengthened the link between parental background
and children’s outcomes among cohorts that left compulsory education
between 1988 and 2009. There are however some striking empirical features, related to the choice and voucher reform, but also to other changes
to the Swedish society, that constitute challenges to the future of the
school system and to equality of opportunity. In this section, a few of
these challenges and their related policies will be discussed.
17 Holmlund et al. (2014) show that these conclusions are robust also to using alternative sources of data.
Analyses based on PISA data are somewhat inconclusive as to whether parental background has become
more important for pupils’ results in Sweden. The PISA index for socio-economic status has become a
stronger predictor for reading performance between 2000 and 2009, while the same index has not become
more important for mathematics between 2003 and 2012 (National Agency for Education 2013).
Nordic Economic Policy Review
Choice and sorting to schools
Swedish independent schools cannot charge tuition fees, nor can they
cream-skim the best pupils. 18 However, there is scope for independent
schools to influence the pool of applicants indirectly; either by locating in advantaged areas, or by targeting information to specific groups
of parents. In addition, anecdotal evidence indicates that independent
schools do not always abide by the rules and reject low-performing
students with disruptive behaviour (SVT 2013). On the demand side,
preferences for schooling and access to information about school quality might differ by family background and affect the decision to opt out
of the local public school. The waiting list principle to admit pupils to
oversubscribed schools might also discriminate against groups of pupils with less information about how the school choice system works.
These mechanisms typically imply that school choice might increase
school segregation.
Who are the pupils that opt out of their local public school to attend
an independent school? Table 1 shows that native children, as well as
children with high-income and highly educated parents are more likely
to attend an independent school, compared to immigrants and children
whose parents have lower income and education.
Independent schools cannot charge tuition fees, but schools are allowed to receive donations from parents.
Nordic Economic Policy Review
Table 1: Fraction of 9th grade pupils in different demographic groups attending an independent
school in 2009
Immigrant background
Swedish background
High parental education
Low parental education
High parental income
Low parental income
Note: High parental education refers to at least one parent with a three year university degree. Low
parental education is defined as both parents holding only compulsory education. High vs. low parental income is defined as having family income (defined during school age) belonging to the top
20 or bottom 20 percentiles of the income distribution.
Source: Own calculations based on the 9th grade register matched with parental background
Figure 4 (adapted from Holmlund et al., 2014) adds to this picture by
presenting odds ratios of the probability of attending an independent
school, for pupils of different family background, living in different
types of neighbourhoods. Neighbourhoods (defined as catchment areas)
are characterized as advantaged or disadvantaged based on an index of
family background that takes into account parental education, parental
income, and migration history. Odds ratios are calculated by comparing
only pupils living in the same catchment area, who should be expected
to face the same supply of schools to choose from at a reasonable distance. An odds ratio equal to one means that two groups of students,
e.g., immigrants and natives, have the same probability to attend an independent school, while an odds ratio below (above) one implies a lower (higher) probability. The figure shows that foreign background pupils have a much lower probability (odds ratio 0.55) than natives to attend an independent school, if they live in a disadvantaged catchment
area. That is, children with Swedish background are more likely than
immigrant background children to attend independent schools if they
live in poor neighbourhoods – and strikingly, the figure also shows that
this is not the case in catchment areas with a more favourable demographic composition, where pupils of Swedish background have a much
Nordic Economic Policy Review
lower probability to attend independent schools. Similarly, children
with highly educated parents have a much higher probability (odds ratio 1.83) of attending an independent school compared to children
whose parents have lower education, if they live in a catchment area
with high proportions of disadvantaged background peers. Instead, the
probabilities are reversed for children of different educational background in advantaged catchment areas.
How can these sorting patterns be interpreted? The demographic
composition of the catchment area can be seen as a proxy for the expected peer group in the local public school. In disadvantaged areas,
pupils with Swedish background or with highly educated parents are
more likely to leave the public school to attend an independent school
than pupils with immigrant background and low educated parents. In
contrast, in socially strong areas, the same pupils have a lower likelihood of attending an independent school than disadvantaged pupils.
Advantaged pupils who live in neighbourhoods with similar peers remain in the public school and even attend the public school to a larger
extent than more disadvantaged children do.
To sum up, the odds ratios in Figure 4 show that pupils with Swedish background and pupils whose parents have high education sort
themselves into school types (public or independent) depending on the
demographic composition of the catchment area which they belong to. If
the demographic composition is favourable, they are likely to remain in
the public school, while they are more likely to opt out and attend an
independent school if they expect to find a more disadvantaged peer
group in the local public school. To some extent, these sorting patterns
reveal that peer group considerations are important when school choice
is exercised. 19
See also Andersson et al. (2012) for similar findings.
Nordic Economic Policy Review
Figure 4: Probabilities of attending an independent school, by parental background and
neighbourhood demographics
High educated parents
Immigrant background
Odds ratio
Disadvantaged neighbourhood
Advantaged neighbourhood
Notes: Neighbourhoods are defined as advantaged (disadvantaged) if they are above (below) the
median in the pupil-weighted distribution of the family background index. The family background
index is a measure of expected GPA given by parental background, and is calculated by predicting
GPA using the regression coefficients from a regression of GPA on family background factors such as
parental education, parental income and immigrant status. High education parents are defined as at
least one parent with a three-year university degree. Pupils with foreign background are born
abroad to two foreign-born parents, or born in Sweden to two foreign-born parents. Odds ratios are
computed by comparing pupils living in the same catchment area.
Source: Holmlund et al. (2014).
The sorting to independent schools as displayed in Table 1 and Figure 4
suggests that school choice is related to school segregation. The next
section discusses the evolution of school segregation and its causes in
more detail.
Nordic Economic Policy Review
School segregation
Starting in the mid-1990s, school segregation has increased gradually in
Sweden. Figure 5 illustrates this development with one indicator of segregation: the between-school variation in various measures of family
background. 20 First, the solid line shows the between-school variation
in the family background index “expected GPA”, that is, the GPA predicted by a set of family background characteristics such as parental education, parental earnings and migration history. It is clear from the graph
that in terms of this combined measure of family background factors,
schools have become more segregated throughout the period 1988–
2009. Next, by studying segregation by different family background
characteristics separately, it turns out that segregation between pupils
of Swedish or immigrant background has increased the most, while the
increase in segregation by education background is less dramatic.
This pattern is robust also using alternative segregation indices; see for example Böhlmark et al. (2015).
Nordic Economic Policy Review
Figure 5: Between-school variation in family background
Predicted GPA
Parental education
Immigrant bakgrund
Note: The between-school variation is calculated annually using ANOVA. Parental education refers
to the parental average years of schooling. Pupils with immigrant background are born abroad to
two foreign-born parents, or born in Sweden to two foreign-born parents. GPA is predicted by parental background factors: education, income and migration background.
Source: Adapted from Holmlund et al. (2014).
Most children attend a school in their local catchment area, which
means that school segregation is likely to be explained by residential
segregation. However, segregation between schools took off in the early
1990s and has continued to increase as the number of pupils in independent schools has been growing. This empirical correlation has
brought attention to the question of whether choice exacerbates school
segregation over and above the segregation that is given by residential
segregation. To bring clarity on this topic, Böhlmark et al. (2015) and
Holmlund et al. (2014) study the association between the fraction of
pupils opting out to independent schools, and various indicators of
school segregation at the municipality level in Sweden. Their key finding is that the main contributor to school segregation is residential seg158
Nordic Economic Policy Review
regation, but segregation between schools has increased more than
what is predicted by changes in residential sorting in municipalities
where choice has become more prevalent. Figure 6 illustrates this result
by graphing the between-school and between-neighbourhood (defined
as catchment areas) variation in the family background index “expected
GPA” over time. Around 1990, before the choice reform when almost all
pupils attended the local public school, segregation between schools
and residential neighbourhoods was virtually identical. Over time, both
residential segregation and school segregation has increased, but as the
number of pupils enrolled in independent schools is increasing, segregation between schools is increasing more than segregation between
neighbourhoods. Nevertheless, residential segregation remains the
main explanation to the development over time, and as seen in Figure 5,
segregation between children with different migration histories has increased the most. Holmlund et al. (2014) show that municipalities and
residential neighbourhoods with a large immigrant population host
disproportionate numbers of new immigrants, as the share of immigrants in the population is growing. This has undoubtedly increased
residential segregation in the country, and has naturally also affected
school segregation.
Nordic Economic Policy Review
Figure 6: Between-school and between-catchment area variation in family background
(predicted GPA)
Catchment area
Note: The between-school variation is calculated annually using ANOVA. GPA is predicted by parental background factors: education, income and migration background.
Source: Holmlund et al. (2014)
Policy considerations of choice and sorting
This section discusses policy considerations related to the social gradient in enrolment at independent schools and to increasing school segregation, in relation to equality of opportunity. Its purpose is to highlight policy areas that are relevant for the future development of the
school system with regards to segregation and sorting.
As Section 5.1 has shown, the choice to opt out is not equally distributed across different socioeconomic groups. This has not implied
large changes to school segregation at the aggregate level, but the social
gradient in independent school attendance is nevertheless relevant
from the perspective of equality of opportunity. This social gradient can
be the result of differences in preferences, differences in access to in160
Nordic Economic Policy Review
formation (and abilities to interpret information), and may also result
from the waiting list principle which benefits parents who are forwardlooking, well informed and do not move home. To date, there is no research using data on parents’ revealed preferences for schools – most
research builds on the school attended by the pupils. As a consequence,
we know little about whether the sorting to independent schools is due
to families ranking school differently (i.e., differences in preferences
and/or information), or due to the assignment mechanism (i.e., families
rank schools similarly, but the waiting list leads to sorting). However,
from the perspective of equal life chances – and the notion that children
do not choose their parents – none of the above explanations for the
social gradient are justifiable. Even in the extreme case where there are
no quality differences between schools and the sorting of pupils has no
consequences for pupils’ learning, one might still argue that families
derive utility from the consumption value of school choice, and that the
allocation of this consumption value across families is a policy issue.
Should policy makers be worried about increasing school segregation? To answer this question, it is important to consider what consequences segregation may have for pupil achievement – on average and
for different groups of pupils – and for society at large. Taken together,
the evidence in Böhlmark and Lindahl (2015) and Böhlmark et al.
(2015) indicate that if segregation induced by choice has had negative
effects on pupil performance on average – these have been offset by
positive effects from school competition, indicating that the net effect of
the choice reform on pupil performance is slightly positive. But because
of residential sorting, school segregation has increased much more than
what can be explained by the choice reform, and it is therefore important to consider its potential effects for the education system, and
for society at large.
First, how does growing up in a segregated neighbourhood affect
children’s life chances? How does a segregated school affect the learning
environment, how does it affect inequality in pupils’ outcomes and are
there effects on other life outcomes, such as criminal involvement? As
already mentioned in Section 3.2, recent experimental evidence has
shown that sorting pupils by ability can improve outcomes, and that the
positive effect of tracking can be larger than the positive peer-effect for
Nordic Economic Policy Review
low-ability students in a mixed peer group. However, it is unclear to
what extent the evidence from these studies can be generalised to the
Swedish compulsory education system and to a setting where pupils are
segregated by family background, into different schools. Grouping pupils by ability is not the same as grouping by socioeconomic background, and within-school ability sorting is different than betweenschool segregation (see Betts (2011) for a discussion on the differences
between tracking within and between schools).
Research focusing on segregation per se is potentially more informative for the Swedish case. Billings et al. (2014) present evidence
that increased racial segregation in a U.S. school district, as a result of
ending a de-segregating busing scheme, led to lower test scores for both
white and minority pupils who were assigned to schools with larger
numbers of minority pupils. In addition, the study finds that while compensatory resource allocation was able to efficiently remedy the negative effects on test scores, crime increased substantially among minority
males who were assigned to schools with large proportions of other
minority students. These findings align with earlier studies on peer racial composition and school segregation (see for example Deming,
2011). The key insights from these studies is that high concentrations of
disadvantaged youth in segregated schools might increase test score
inequality between children of different family backgrounds, and affect
behaviour also in other dimensions such as crime. The conclusions are
further corroborated by evidence from the Moving to opportunityprogramme (MTO) that offered randomly selected families, living in
disadvantaged neighbourhoods in the U.S., financial help to move to a
better neighbourhood. Children who were able to move out of poor
neighbourhoods at a young age turn out to be more likely to attend college and to live in better residential areas as adults, compared to children who were not able to move (Chetty et al., forthcoming).
The conclusions from the U.S. literature on racial segregation are
suggestive of the consequences of segregation also in other settings, but
the Swedish case of segregation between refugee immigrants and natives also has its unique features. Refugee families might have gone
through traumatic events in their home countries; they come from
many different nations, speak different languages, have varying educa162
Nordic Economic Policy Review
tional histories and have arrived in Sweden at different times. Research
on the outcomes of refugee children in Sweden shows that growing up
in a neighbourhood with a large immigrant population increases the
probability of engaging in crime, while there are no negative effects on
school performance at age 16 (Grönqvist et al., 2015). As such, the findings of the effects of segregation in Sweden are in line with those of
Billings et al. (2014). Neighbourhood exposure to refugee peers does
not seem to be negative in all respects, however. Åslund et al. (2011)
find that a larger ethnic community, that is, a larger community of individuals from the same country (or region) of origin, is beneficial for
school performance. In addition, the positive effect of the ethnic community is increasing the more highly skilled are its members. One way
of interpreting these results is therefore that high concentrations of the
most disadvantaged groups in general is detrimental and implies increased risks for criminal activity, while sorting specifically by ethnic
groups might be beneficial for refugees. 21
Second, moving beyond the effects of segregation on individual outcomes, there is also a general concern that social cohesion in society is
adversely affected if interactions between children from different backgrounds are limited (Levin 1998). This argument is discussed further in
Blomqvist and Rothstein (2000) who argue that that integration of different groups is beneficial to foster tolerance and solidarity, and to discourage discrimination, inequality and violence in society.
To sum up, the research presented above has shown that sorting
of pupils by family background, with large concentrations of the disadvantaged (minority or immigrant) group in some schools or neighbourhoods increases crime. There is thus cause for concern that a high
level of segregation can have detrimental effects for individuals as
well as for society.
If the policy aim is to provide equal access to the independent
school sector for all children regardless of their family background, and
21 Segregation and differentiation by ability may also affect what pupils learn in school through other
mechanisms. For example, if achievement targets and teachers’ expectations are adjusted to the group
average, differentiation may increase inequality in educational performance (Figlio and Stone 2004;
Bonnesrönning 2008).
Nordic Economic Policy Review
to limit school segregation in order to avoid highly disadvantaged clusters, what are the consequences for policy?
To begin with, it is a challenge to design a school choice mechanism
that allows for parental choice but at the same time limits sorting and
quality differences across schools. The Swedish model has explicitly
aimed at equal access for all pupils: independent schools cannot charge
tuition fees, nor can they cherry-pick the best pupils. However, an international outlook shows that there are other models than using date
of application, as in Sweden, to allocate pupils to oversubscribed
schools. In this regard, there is room for a policy debate that opens up
to alternatives. International examples show that school choice can be
combined with re-distributive vouchers, such as in the Netherlands, and
that the mechanism to assign pupils to schools can include minority
quotas or lotteries, such as in the U.S.
Next, the overall trends in residential segregation show that school
segregation must be tackled not only by providing equal access to independent schools. Examples of de-segregation policies in the U.S. that
aim to counterbalance residential segregation include busing and minority quotas. Recent experiences of failing schools in disadvantaged
areas in Sweden have led to similar actions: the closing of a school in
Rosengård (Malmö) and busing of its pupils to a number of other
schools in the city. Ultimately, residential segregation, refugee placements and catchment area boundaries constitute policy areas that are
all relevant for school segregation. It should be noted that if the policy
objective is to limit sorting and to create more heterogeneity by parental background, alternatives for admitting pupils to public schools must
also be discussed.
Nordic Economic Policy Review
This paper has summarised evidence from the Nordic countries on the
role of the education system for equality of opportunity. The comprehensive school reforms that were rolled out in the Nordic countries in the
1950s–1970s aimed to reduce inequalities in educational outcomes, and
the research evidence shows that the reforms were successful in this regard: parental background became less important for labour market outcomes for the cohorts that went to the comprehensive school.
More recently, Sweden has undertaken reforms to introduce parental choice and a universal voucher scheme, and as a result many new
independent voucher-funded schools have opened. These schools have
pedagogical, religious or general profiles, and can be run as a non-profit
foundation or a for-profit corporation. While the comprehensive school
reforms aimed to unify the curriculum and the length of compulsory
schooling, the choice and voucher reform can be considered as a step
towards more heterogeneity in educational provision. Although this
heterogeneity does not resemble the selective, early tracking system
pre-dating the comprehensive school, it will potentially imply that pupils are exposed to different “types” of schools and to differences in
pedagogical profiles and school governance. It has been hypothesised
that this educational variety, in combination with parental choice, provides a setting where parental background will become more important
for school performance. The recent studies on this topic show that there
is little evidence to support this argument: family background factors
remain strong predictors for school performance, but have not become
more important as school choice has grown more common. Using
school performance as a proxy for future labour market outcomes, there
is no indication that social mobility is declining for the cohorts that left
compulsory education up to 15 years after the choice reform was introduced. In addition, there is no evidence supporting the notion that
school choice and competition can explain the plummeting Swedish PISA scores (Böhlmark and Lindahl, 2015, Wondratschek et al., 2013).
Education policy with the aim to promote equality of opportunity is
nevertheless facing challenges for the future. The Swedish example
shows that segregation has been increasing for a long time, and current
Nordic Economic Policy Review
migration flows are highlighting the importance of policies that limit
high concentrations of socioeconomically disadvantaged or foreignborn pupils in schools. It is also evident that access to independent
schools is not equal for pupils of different family background. There is a
social gradient in school choice. The comprehensive school abandoned
ability tracking, and today’s independent schools are not allowed to
admit pupils based on ability tests. Selection on ability is thus considered as an unfair mechanism that creates inequality. From an equality
of opportunity perspective however, one can argue that (de facto) selection on parental background is equally unfair, and it is not clear why
one type of selection is considered acceptable and the other one not.
Aakvik, A., Salvanes, K. G. & Vaage, K. (2010). Measuring heterogeneity in the returns to education using an education reform. European Economic Review, 54, pp.
Amermüller, A. (2005). Educational opportunities and the role of institutions.
Centre for European Economic Research, Discussion paper 05–44.
Andersson, E., Malmberg, B., & Östh, J. (2012). Travel-to-school distances in Sweden
2000–2006: Changing school geography with equality implications.
Journal of Transport Geography, 23, pp. 35–13.
Arendt, J. N. (2005). Does education cause better health? A panel data analysis using
school reforms for identification. Economics of Education Review, 25, pp. 149–160.
Åslund, O., Edin, P.-A., Fredriksson, P. & Grönqvist, H. (2011). Peers, neighbourhoods,
and immigrant student achievement: Evidence from a placement policy. American Economic Journal: Applied Economics, 3, pp. 67–95.
Becker, G. S. & Tomes, N. (1979). An equilibrium theory of the distribution of income and intergenerational mobility. Journal of Political Economy, 87(6),
pp. 1153–1189.
Betts, J. R. (2011). The economics of tracking in education. In E. A. Hanushek,
S. Machin and L. Woessmann (eds.), Handbook of the Economics of Education, vol.
3, Elseveier.
Billings, S. B., Deming, D. J., & Rockoff, J. E. (2014). School segregation, educational attainment and crime: Evidence from the end of busing in Charlotte-Mecklenburg. Quarterly Journal of Economics, 129(1), pp. 435–476.
Nordic Economic Policy Review
Björklund, A. & Jäntti, M. (1997). Intergenerational income mobility in Sweden
compared to the United States. American Economic Review, 87(5), pp. 1009–1018.
Björklund, A., Eriksson, T., Jäntti, M., Raaum, O., & Österbacka, E. (2002). Brother
correlations in earnings in Denmark, Finland, Norway and Sweden compared to
the United States. Journal of Population Economics, 15, pp. 757–772.
Björklund, A., Lindahl M., & Sund, K. (2003). Family background and school performance during a turbulent era of school reforms. Swedish Economic Policy Review,
10, pp. 111–136.
Björklund, A. Jäntti, M., & Lindquist, M. J. (2009). Family background and income
during the rise of the welfare state: Brother correlations in income for Swedish
men born 1932–1968. Journal of Public Economics, 93, pp. 671–680.
Björklund, A. & Salvanes, K. G. (2011). Education and family background: Mechanisms and policies. In Hanushek E. A., S. Machin and L. Woessmann (eds.), Handbook of the Economics of Education, vol. 3, Elseveier.
Björklund, A. & Jäntti, M. (2012). How important is family background for laboureconomic outcomes? Labour Economics 19, pp. 465–474.
Blanden, J. (2013). Cross-country rankings in intergenerational mobility: A comparison of approaches from Economics and Sociology. Journal of Economic Surveys,
27(1). pp. 38–73.
Blomqvist, P. & B. Rothstein (2000). Välfärdsstatens nya ansikte – demokrati och
marknadsreformer inom den offentliga sektorn. Agora förlag, Stockholm.
Bonesrönning, H. (2008). The effect of grading practices on gender differences in
academic performance. Bulletin of Economic Research, 60(3), pp. 245–264.
Booij, A. S., Leuven, E., & Oosterbeek, H. (2015). Ability peer effects in university:
Evidence from a randomized experiment. IZA Working paper No. 8769.
Brunello, G. & Checchi, D. D. (2007). Does school tracking affect equality of opportunity? New international evidence. Economic Policy, 22(52), pp. 781–861.
Böhlmark, A. & Holmlund, H. (2012). Lika möjligheter? Familjebakgrund och skolprestationer 1988–2010. IFAU report 2012:14.
Böhlmark, A., H. Holmlund & Lindahl, M. (2015). School choice and segregation: Evidence from Sweden. IFAU working paper 2015:8.
Böhlmark, A. & Lindahl, M. (2015). Independent schools and long-run educational
outcomes – evidence from Sweden’s large scale voucher reform. Economica 82,
pp. 508–551.
Chetty, R., Hendren, N., & Katz, L. (forthcoming). The effects of exposure to better
neighbourhoods on children: New evidence from the moving to opportunity experiment. American Economic Review.
Nordic Economic Policy Review
Deming, D. J. (2011). Better schools, less crime? Quarterly Journal of Economics, 126,
pp. 2063–2115.
Duflo, E., P. Dupas & Kremer, M. (2011). Peer effects, teacher incentives and the impact of tracking: Evidence from a randomized evaluation in Kenya. American Economic Review, 101, pp. 1739–1774.
Epple, D., Newlon, E. & Romano, R. (2002). Ability tracking, school competition, and
the distribution of educational benefits. Journal of Public Economics, 83, pp. 1–48.
Erikson, R. & Goldthorpe, J.H. (1992). The constant flux – A study of class mobility in
industrial societies, Clarendon, Oxford, England.
Erikson, R. (1996). Explaining Change in Educational Inequality – Economic Security
and School Reforms. In Robert Erikson and Jan O. Jonsson (eds.), Can Education Be
Equalized?, Westview Press.
Figlio, D. N. & Lucas, M. E., (2004). Do high grading standards affect student performance? Journal of Public Economics, 88(9–10), pp. 1815–1834.
Fisher, M., Karlsson, M., & Nilsson, T. (2013). Effects of compulsory schooling on
mortality: Evidence from Sweden. International Journal of Environmental Research
and Public Health, 10, pp. 3596–3618.
Fredriksson, P. & Vlachos, J. (2011). Reformer och resultat: Kommer regeringens utbildningsreformer att ha någon betydelse? Report to the Swedish Fiscal Policy
Council 2013/3.
Government Bill 1991/92:95. Om valfrihet och fristående skolor.
Government Bill 1994/95:157. Ändring av bidragsbestämmelserna för fristående
skolor godkända för vanlig skolplikt m.m.
Grönqvist, H., Niknami, S., & Robling, P.-O. (2015). Childhood exposure to segregation and long-run criminal involvement. SOFI Working paper 1/2015.
Gustafsson, J. E. & Yang-Hansen, K. (2009). Resultatförändringar i svensk grundskola.
Vad påverkar resultaten i svensk grundskola? Kunskapsöversikt om betydelsen av
olika faktorer. Stockholm, Skolverket.
Hanushek, E. A. and Woessmann, L. (2006). Does educational tracking affect performance and inequality? Differences-in-differences evidence across countries.
Economic Journal, 116, pp. C63–C76.
Havnes, T. & Mogstad, M. (2011). No child left behind: Subsidized child care and
children’s long run outcomes. American Economic Journal: Economic Policy 3, pp.
Holmlund, H. (2007). A researcher’s guide to the Swedish compulsory school reform.
SOFI Working paper 9/2007.
Holmlund, H. (2008). Intergenerational mobility and assortative mating: Effects of an
educational reform. CEE Discussion paper 91.
Nordic Economic Policy Review
Holmlund, H., M. Lindahl and E. Plug (2011). The causal effect of parents’ schooling
on children’s schooling: A comparison of estimation methods. Journal of Economic
Literature, 49(3). pp. 615–651.
Holmlund, H., Hägglund, J., Lindahl, E., Martinson, S., Sjögren, A., Vikman, U., &
Öckert, B. (2014). Decentralisering, skolval och fristående skolor: resultat och likvärdighet i svensk skola. IFAU report 2014:25.
Levin, H. (1988). Educational vouchers: Effectiveness, choice and costs. Journal of
Policy Analysis and Management, 17(3), pp. 373–392.;2-D
Meghir, C. & Palme, M. (2005). Educational reform, ability and family background.
American Economic Review, 95:1, pp. 414–424.
National Agency for Education (Skolöverstyrelsen) (1969). Läroplan för grundskolan Lgr69. Skolöverstyrelsen, Stockholm.
National Agency for Education (Skolverket) (2003). Valfrihet och dess effekter inom
skolområdet. Report 230.
National Agency for Education (Skolverket) (2009). Vad påverkar resultaten i svensk
grundskola? Skolverket, Stockholm.
National Agency for Education (Skolverket) (2012). Likvärdig utbildning i svensk
grundskola? – En kvantitativ analys av likvärdighet över tid. Report 374.
National Agency for Education (Skolverket) (2013). PISA 2012. 15-åringars kunskaper i matematik, läsförståelse och naturvetenskap. Report 398.
Opposition bill 1991/92: Ub62. Motion med anledning av prop. 1991/92:95 Om valfrihet och fristående skolor.
Pekkarinen, T., Uusitalo, R. & Kerr, S. (2009). School tracking and intergenerational
income mobility: Evidence from the Finnish comprehensive school reform. Journal
of Public Economics, 93, pp. 965–973.
Pekkarinen, T., Uusitalo, R. & Kerr, S. (2013). School tracking and the development
of cognitive skills. Journal of Labor Economics, 31:3, pp. 577–602.
Solon, G. (2004). A model of intergenerational mobility variation over time and
place. In M. Corak (ed.), Generational income mobility in North America and Europe.
Cambridge University Press, Cambridge.
SOU 2013:56. Friskolorna i samhället. Fritzes, Stockholm.
SVT (2013). Friskolor väljer bort besvärliga elever. Downloaded from Published
28 October 2013. Downloaded 28 January 2013.
Swedish Association of Independent Schools (Friskolornas riksförbund) (2015).
Fakta om friskolor.
Nordic Economic Policy Review
Waldinger, F. (2007). Does ability tracking exacerbate the role of family background
for students’ test scores? Unpublished manuscript.
Wondratschek, V., Edmark, K., & Frölich, M. (2013). The short and long-term effects
of school choice on student outcomes – Evidence from a school choice reform in
Sweden. Annals of Economics and Statistics. 111–112, pp. 71–102.
Nordic Economic Policy Review
8. Retirement and Health in the
Nordic Welfare State1
Nabanita Datta Gupta, Department of Economics and Business Economics,
Aarhus University, [email protected] and Bent Jesper Christensen,
Department of Economics and Business Economics, Aarhus University,
[email protected]
In Nordic welfare state countries, a multitude of exit options with relatively high replacement rates are available for workers approaching
retirement age. We review the recent literature on retirement and
health in Nordic countries, with main focus on Denmark, over a period
characterized by reforms of the welfare system, paying special attention to eligibility to early retirement programs, disability pension, and
wage-subsidized jobs. We collect evidence on two main questions: 1)
Are there substantial moral hazard effects (ex ante and ex post) in
claiming various retirement and health related social benefits in the
Nordic Welfare State? 2) Can early retirement lead to (mental) health
deterioration? The analyses reported on make use of comprehensive
register data on individuals, linking their labor market characteristics
to medical diagnoses.
1 The authors gratefully acknowledge comments and suggestions from Tarmo Valkonen (the discussant)
and seminar participants at the 2015 Nordic Economic Policy Review Conference in Helsinki, October, 2015,
and from an anonymous referee.
In Nordic welfare state countries, a multitude of retirement options exist for older workers which potentially reinforce the effect of health deterioration on retirement. The existence of various early retirement and
disability exit options coupled with universal health care regime and
low out-of-pocket medical expenditures may imply that older workers
when faced with a health shock have a strong incentive to reduce their
labor force participation. Even if an individual remains able to work,
labor market exit may be a tempting choice if benefits are sufficiently
generous relative to earnings from continued work, particularly when
figuring in utility from increased leisure time. A moral hazard problem
may arise, with the worker exaggerating the severity of the health condition in order to obtain benefits, failing to report on recovery from an
earlier condition leading to benefits, or before health problems arise
engaging in risky behavior and life style choices increasing the risk of a
health shock, knowing the generosity of the system in place. Further,
moral hazard may be affected by social norms (when is it acceptable to
opt out?) and justification effects (a health issue may be claimed as an
excuse for exit).
Fuelled by concerns about such over-exploitation of the system,
Nordic governments over the past two decades have strengthened incentives to retain older workers, made disability screening procedure
more stringent, expanded wage subsidized jobs substantially, and started gradually shortening the early retirement period. Some have viewed
the reforms as problematic, but if the moral hazard problem is substantial, the reforms are probably justified, schemes having been on the
generous side in the past. Therefore, the first aim of this paper is to shed
light on whether there are significant moral hazard effects in claiming
various retirement and health related social benefits in the Nordic Welfare Model.
An additional rationale for retaining older workers is the continuing
increase in longevity. As the population ages, dealing with mental deterioration will add substantially to health care costs. Indeed, it may be
that withdrawal from work can lead to deterioration of mental health. If
staying longer on the labor market is a way to stave off cognitive de172
Nordic Economic Policy Review
cline, there are additional benefits to reforms increasing statutory retirement ages, aside from the need to cope with aging and demographic
change per se. Therefore, the second aim of this paper is to investigate
whether there are negative health effects of early retirement, in terms
of deterioration of cognitive capacity.
We review the literature on health and retirement in the Nordic
welfare state, with main focus on Denmark, over a time period characterized by reforms of the welfare system. We pay special attention to
eligibility to early exit programs, disability pension, and wage subsidized jobs. The analyses reported on in this paper make use of comprehensive register data on individuals, linking their labor market characteristics to medical diagnoses, in contrast to many studies in this area
which are limited to using self-reported health and survey data.
In what follows, we describe in Section 1 the numerous Nordic welfare state institutions relevant for retirement and health. In Section 2,
we review the evidence of moral hazard induced retirement in Nordic
countries, with a particular focus on Denmark. In Section 3, we use Danish register data to conduct new analyses of the effect of retirement on
diagnosed cognitive decline. In Section 4, we report on recent policy
initiatives undertaken in Sweden and Denmark to encourage individuals with some remaining working capacity to remain on the labor market. Finally, in Section 5, we offer a brief conclusion.
Welfare state institutions that insure against
health-related loss of working capacity
The Nordic welfare states share in common an elaborate tax-financed
social safety net in the event of a health shock. When a worker suffers
health-related loss of working capacity, depending on whether the
loss is temporary or permanent, the relevant compensatory institutions are sickness absence, wage subsidy programs, disability pension,
and, to some extent, early retirement. Furthermore, health services
are provided in a setting of universal health insurance and largely
public health care delivery decentralized at the level of the region, but
Nordic Economic Policy Review
moving towards greater consolidation following the economic crisis
(Saltman et al., 2012).
Sickness absence programs compensate employers for the period
the worker is unable to perform due to injury/illness. All Nordic countries offer quite generous sick leave with maximum period of benefit
varying between a year (Norway, Finland, and Denmark) to 2.5 years
(Sweden) (Thorsen et al., 2015). There is some co-insurance element in
that employers are responsible for the compensation for a part of this
period while the state and local authorities cover the rest. It is usually
the municipalities within the Nordic system that are responsible for
administering benefit programs, following up on health assessments,
and initiating various return-to-work (RTW) measures. In Denmark, the
municipalities are responsible for vocational rehabilitation programs
including workplace accommodation, working aids, counselling, reeducation/re-training offers, etc. (see Bach et al., 2007). Harmonized
data from the European Union Labor Force Surveys show that longterm sickness absence is high in Sweden and Norway, but low in Denmark and Iceland, with Finland in between, while the pattern is reversed for short-term sickness absence (Thorsen et al., 2015).
In recent years, all the Nordics have implemented a series of
measures to strengthen RTW of the sick-listed, including workability
assessments, more frequent follow-up assessments, and partial sick
leave. Particularly the latter seems to be related to increased workability of persons with disability (Thorsen et al., 2015). In Denmark, if ordinary return to work is not possible, workers can return to wagesubsidized jobs (Flexjobs) involving flexible working conditions, shorter
hours, etc., where employers are compensated 50–67% of the wage depending on the extent of working capacity loss. The introduction of
Flexjob in Denmark has been shown to improve the employability of
disabled workers in the 35–44 years age group by 10.5–12.5 percentage
points (Datta Gupta and Larsen, 2010a).
The increasing efforts on getting sick-listed workers back to work as
quickly as possible may not be optimal from a welfare point of view if it
means that more and more workers report to work while still sick or in
pain, as this increases the disutility from work. On the other hand, some
degree of presenteeism (reporting to work while sick or in pain) could
Nordic Economic Policy Review
be beneficial. Randomized trials from Finland and Norway have shown
that low back pain, which is the most prevalent reason for sickness absence, may actually heal faster with light activity. Markussen et al.
(2012) show based on Norwegian data that sick-listed employees who
are issued graded (partial) absence certificates from their physicians
actually experience shorter absence spells and higher employment
rates two years after than those issued full certificates. A similar therapeutic effect of work may be present for some less serious mental disorders (Waddell and Burton, 2006).
When workers suffer a permanent injury or illness they are eligible
for disability pension. Like most other Western industrialized nations,
the Nordics also have relatively high rates of disability pensioning (seen
in relation to objective measures of health), with about 5–10% of the
16/18–64/66 population receiving disability pension in 2000–2013 (as
for sickness absence, the rate for women exceeds that for men) (Thorsen et al., 2015). The high disability rates are also remarkable in relation to the low unemployment rates of these countries. Together with
the Netherlands, the Nordics rank top six in terms of disability shares
out of 28 countries with comparable data (OECD, 2009b). Remarkably,
there has been little change in this share over the period for most Nordics, the exceptions being rising disability in Iceland after 2009 and falling disability in Sweden after 2007. One study shows that for Denmark,
take-up of disability seems to be on its own track and quite unrelated to
a number of health indicators (Bingley et al., 2014). The compensation
under disability in most cases is a flat-rate benefit, although in Sweden
it is a function of previous earnings. The implication of a flat-rate benefit in disability and other pension programs in most Nordic countries is
that the replacement rate is comparatively higher for the less-educated.
Most Nordic countries have redefined eligibility for disability pension in
recent years in terms of strict medical criteria and have expanded partial disability provisions, but have not altered the generosity of pension
benefits (Andersen, 2011). Still, these reforms have most likely kept
disability from rising even further, and have seemingly reversed the
trend in Sweden even as the population continues to age.
Early retirement before the age of 65 in Nordic countries is prevalent in Denmark and Finland due to the existence of comprehensive
Nordic Economic Policy Review
public early retirement schemes. In 1979, a voluntary early retirement
program (Efterløn) available at age 60 to workers who had contributed
to an unemployment insurance (UI) fund for a sufficient length of time
was introduced in Denmark. The rationale given at the time for its introduction was the need to ameliorate the youth unemployment problem and to give workers in physically demanding jobs an exit route out
of the labor market without having to satisfy formal medical criteria.
One year after, in 1980, the labor force participation rate of men aged
60–64 dropped 20 percentage points and remained at this lower level
for the next two decades (Bingley et al., 2012). Following the Retirement Reform in 2011, the program has been restricted substantially. In
the future, the duration of benefits is shortened from 5 to 3 years, along
with considerable implied deductions in pension assets, implying the de
facto cancellation of the program in due course (Andersen, 2011).
In Finland, too, access to early retirement is no longer possible for
cohorts born after 1943 (Finnish Pension Reform of 2005). While Denmark and Finland have historically had the highest rates of early retirement, Iceland has had the lowest rate among the Nordics until the
financial crisis hit, after which the share has become more variable.
Sweden and Norway fall somewhere in between. There are larger differences in retirement patterns than in sickness absence and disability
patterns due to somewhat different structuring of the pension systems
and differing relative importance of occupational pensions. Nonetheless, all the Nordic countries have now embraced automatic indexation
of pensions to longevity, which ought to improve the future sustainability of the system (Andersen, 2011).
As funded pension plans (occupational and private) grow in importance in the Nordic countries, a question for the future is whether
statutory retirement programs will become irrelevant as people begin
to retire when they have built up sufficient pension funds, and what the
resulting implications of such behavior will be for labor supply and public finances. A study from Denmark shows that individuals are increasingly beginning to combine work and retirement, i.e., partial retirement
(Larsen and Pedersen, 2012). Earnings-related pensions systems are
also undergoing reforms currently. The 2005 reform of the Finnish private-sector earnings-related pension system curbed the increases in the
Nordic Economic Policy Review
contribution rate without hurting replacement rates and simulations
show the reform will lead to a postponing of retirement (Lassila and
Valkonen, 2006).
Is there evidence of moral hazard induced
Not all retirement options are freely available to a worker at any point
in time. For example, disability pension is reserved for cases of permanent injury or illness. As already alluded to, this implies an incentive on
the part of the worker to misrepresent the true health condition, i.e., a
moral hazard problem (Rothschild and Stiglitz, 1976, Diamond and
Mirrlees, 1978). This may not be a case of consciously trying to cheat
the system, in the sense of insurance fraud involving a completely false
claim by a healthy individual with purely financial motives, but could be
a case of exaggerating an actual health problem, in order to increase the
probability of obtaining the desired pension and leisure. Further, the
moral hazard problem has to be viewed in the context of evolving social
norms regarding how to use the system in place. Thus, the decision
about when it is acceptable to exit may depend on the local participation rate, an idea pushed in the literature on reference points in the labor decision (Lindbeck et al., 1999). Similarly, in the case of sickness
absence, Lindbeck et al. (2011) present evidence from Swedish full
population panel data of geographical differences consistent with local
variation in social norms, but unexplained by socioeconomic differences. A justification issue may arise, as well, with individuals overstating their health problem after receiving the benefit, as an explanation or
excuse. Nevertheless, regardless the exact motives for the potential
over-exploitation of the system, the phenomenon has caused authorities
to implement increasingly rigorous eligibility tests, e.g., requiring specific medical conditions to be met for disability, to prevent free-riding.
While false or overstated reports of injury, illness, or other medical
conditions for the desired retirement program may be classified as the
results of ex post moral hazard (i.e., the claim is that the condition for
admission to the program has already been met), another mechanism
Nordic Economic Policy Review
known as ex ante moral hazard may be in play, as well. According to
this, individuals may be less careful at work and engage in more risky
activities, simply because of the existence of the retirement program
insuring their continued income in the event of a loss of work capacity
caused by the hazardous behavior. In this case, the strict medical criterion, e.g., for disability retirement, may truly be met at the time of the
claim, i.e., there is no misreporting, but it is debatable to what extent
admission to the program is deserved or socially desirable.
The usual solution to the ex ante moral hazard problem is partial insurance, e.g., a lower replacement ratio. This is no longer optimal when
taking into account the simultaneous presence of ex ante and ex post
moral hazard (Mookherjee and Png, 1989, Bond and Crocker, 1997). A
main problem is costly state verification, i.e., the government is unable
to catch misreports in a cheap manner. Indeed, the more difficult it is to
detect the true health condition, the greater the incentive to misreport.
The upshot is overinsurance for small losses and underinsurance for
large, i.e., most people pay too much to keep the system in place, and
those who really need it receive insufficient benefits. These issues have
been studied, e.g., in the case of workers’ compensation and workplace
injuries in the U.S. (Butler et al., 1996) and Canada (Bolduc et al., 2002).
Leth-Petersen and Rotger (2009) study whiplash claimants in Denmark.
Diagnosing lasting disorders associated with whiplash (e.g., from a rearend car collision) is difficult, and hence the incentive to misreport the
condition should be high. The results show that claimants who are
compensated fail to regain their previous earnings levels, thus suggesting little misreporting. On the other hand, about half the claims are
turned down, and these claimants do in fact on average regain their preinjury earnings levels. This indicates that misreporting and thus ex post
moral hazard exists in this group, and that the claims system is capable
of revealing this. Ex ante moral hazard may be present, too, in that those
compensated used more health care in the year prior to the injury, consistent with poorer general health and perhaps risky behavior. The similar issues apply in the retirement case.
Different forms of insurance against health-related loss of work capacity are subject to different screening mechanisms, and different degrees of moral hazard problems. Sickness absence programs involve
Nordic Economic Policy Review
follow-up health assessments, and disability programs include strict
medical eligibility criteria. On the other hand, early retirement eligibility criteria typically do not involve health conditions. Furthermore, an
additional exit route may present itself to the individual strongly determined to leave work in favor of paid leisure, namely, unemployment
accompanied by the associated UI benefits. Thus, the issue arises
whether unemployment may in some cases serve as a voluntary route
to retirement, as well, and, if so, whether take-up in this case is induced
by deteriorating health.
Some evidence on this is presented by Christensen and KallestrupLamb (2012). This study is based on merged register data from Denmark on individual objective medical ICD diagnosis codes upon hospitalization and labor market and socioeconomic variables. Workers are
followed from age 50 until labor market exit through either of a number
of distinct routes. Duration models are fit to explain the time elapsed
until exit, controlling for time-varying conditioning variables including
gender, marital status, education, experience, occupation, financial variables such as income and wealth, and a host of diagnosis-based health
shock indicators, including diseases of the circulatory, respiratory, musculoskeletal, nervous, digestive, and genitourinary systems, cancer, nutritional and metabolic diseases, mental disorders, and injuries. The five
exit routes considered are disability (DI); early retirement (ER) immediately preceded by work; unemployment (UI) followed by ER; UI followed by other retirement states; and other schemes (e.g., civil service
pension). Durations are censored at the old age pension (OAP) age, 67
for the cohort under study, aged 50 in 1985. For this age group, many UI
spells are followed by one of the retirement states (DI, ER, OAP, or other), and hence this sort of terminal unemployment is considered a form
of labor market exit.
During the study period, UI benefits could generally extend up to a
maximum of 4 years. The ER age was 60, contingent on eligibility criteria including having paid into a UI fund for sufficiently long. Of particular interest for the moral hazard question is a rule that was in place during the period and that waived the 4 year maximum for those unemployed workers between ages 55 and 59 who would become eligible for
early retirement by age 60. In effect, individuals wishing to exit the laNordic Economic Policy Review
bor market could do so already by age 51 by becoming voluntarily unemployed, getting the waiver from the 4-year limit to UI benefits at age
55 and thus continuing on benefits until transiting to (official) ER at age
60. This route was clearly illegal, as unemployed workers must remain
available to the labor market, but it would appear attractive to individuals wishing to exit, hence posing a moral hazard problem. Again, this
should be seen in the context of social norms, i.e., unemployment may
be a social norm (Clark, 2003), in that the decision whether to pursue
this combined UI-ER exit route may depend on the behavior of peers
that the worker uses as reference points.
Whether individuals observed to follow the UI-ER route in question
initially became unemployed by chance, or with a plan to remain unemployed until ER, is of course not directly observable. Instead, in a competing risk framework, with the five exit routes (DI, ER, UI-ER, UI-other,
other) representing the competing risks, Christensen and KallestrupLamb (2012) compare the routes in terms of the effects of explanatory
variables, including health shocks. In this sense, the UI-ER route is similar to the ER route per se, whereas UI-other is not (actually, it is more
similar to DI). This suggests that there are differences between UI
spells, even these terminal spells leading into ER or other retirement,
with some of the unemployment possibly being planned and voluntary
(that leading to ER), and some rather being associated with the type of
health shocks that could also lead to disability.
In terms of the detailed results, increased wealth reduces exits
through both the DI and UI-other routes, but has only an insignificant
effect (in the opposite direction) on UI-ER exits, and indeed induces ER.
Greater household income strongly increases DI and UI-other, i.e., the
income effect dominates the substitution effect, but not so for ER or UIER. Naturally, DI is induced by many of the serious health conditions,
such as diseases of the circulatory, respiratory, musculoskeletal, and
nervous systems, cancer, nutritional and metabolic diseases, mental
disorders, and injuries, but not by diseases of the genitourinary systems, which include, e.g., kidney stone, renal failure, and other diseases
of the urinary and genital organs. In contrast, the latter group of diseases strongly increases exits through the UI-ER route, perhaps suggesting
that selection of this combined route may not purely reflect suspicious
Nordic Economic Policy Review
behavior, but also indicate health conditions that in fact lead to a loss of
work capacity, without being sufficiently recognized by disability criteria in place. Interestingly, UI-ER exits are furthermore induced by endocrine, nutritional and metabolic diseases, significantly and more strongly so than the other exit routes. These diagnoses include life style diseases such as diabetes, obesity, etc. The results are consistent with risky
behavior (ex ante moral hazard) leading to life style diseases and prolonged voluntary unemployment followed by early retirement.
Many other studies investigate issues of risk, insurance, and health
from different perspectives. Datta Gupta, Kleinjans and Larsen (2015)
compare the effects of health shocks on the probability of not working
across elderly workers in Denmark and the U.S., using Danish registers
and the Health and Retirement Survey (HRS), respectively. With universal insurance and nationalized health care in Denmark, compared to the
job lock effect of employer-based health insurance and the stronger income effect stemming from higher out-of-pocket medical expenditures
in the U.S., it is expected that Danish workers are more likely than their
U.S. counterparts to exit the labor force following health shocks. However, the results show little difference across the two countries in this
respect. This is shown to be a result of differences in mortality, baseline
health, and the way health care is provided. Datta Gupta and Larsen
(2010b) offer a further comparison across register and survey data of
health effects on planned labor supply, here with both data sources
from Denmark. This allows a direct measurement of the bias arising due
to the survey responses being self-reported and thus potentially involving misreports. This could reflect moral hazard, or a justification bias
may exist, as the severity of a health conditions is overstated to provide
an excuse for early labor market exit. The results indicate that men’s
self-reports on myalgia and back problems and women’s on osteoarthritis yield biased estimates of the impact on planned retirement age. This
is consistent with strategic misreporting, i.e., the existence of insurance
covering the retired state makes it worthwhile to misreport.
Further evidence on the economic incentives for misreporting of
own health is offered by Datta Gupta et al. (2010), demonstrating in a
multi-country comparison that it is exactly in welfare state countries
with significant social transfers that workers are most likely to act opNordic Economic Policy Review
portunistically and falsely self-report disability, whereas this phenomenon is less marked in Southern European countries. Sweden represents the only deviation from this pattern, possibly due to the strictness of vocational assessments in disability cases in Sweden compared
to other SHARE (Survey of Health and Retirement in Europe) countries (Börsch-Supan, 2007). This indicates a case of a successful stateverification mechanism.
Although welfare state countries clearly offer incentives to report
that conditions for retirement are met, it is difficult to associate the labor supply response to health shocks with specific welfare state institutions. Datta Gupta and Larsen (2007) use Danish register data to show
that the increase in retirement rate following a health shock is unaffected by eligibility to early exit programs (eligibility dummies are insignificant), by the long duration of sickness benefits (results are unchanged
when including receipt of sickness benefits in the definition of the retirement state), and by the promotion of corporate social responsibility
(CSR) since 1994 and subsidized employment (Flexjob) since 1998 (no
significant difference between pre- and post-promotion estimates). This
complicates policy choices. Thus, although increasing longevity calls for
policies to retain workers longer, the best method for identifying such
policies remains elusive, based on these results. Nevertheless, that there
is scope for longer work lives in the future is established by Bingley et
al. (2015) using SHARE data for Denmark. The health capacity to work
longer is estimated by comparing how much older individuals work
now with how much those (younger individuals) with similar mortality
rates worked in the past, and with how much younger individuals with
similar self-reports of health work now. Any bias due to moral hazard
and misreporting may thus cancel when constructing the estimate
based on this comparison or matching procedure.
Nordic Economic Policy Review
Does early retirement bring about cognitive
As the population in Nordic countries continues to age, dealing with the
effects of mental deterioration will add substantially to health care
costs. The Nordic Council estimates there will be a doubling of the number of people living with dementia in the Nordic countries over the next
35 years (Nordic Council, 2010). In Denmark alone, the Alzheimer’s Association estimates that annual costs on treatment and care (direct plus
indirect) exceed 15 billion kroner. Longer life expectancy coupled with
early retirement programs tends to add to years spent out of the labor
market. On the other hand, the employment rate in especially the 60–64
age group has been rising in most OCED countries, including the Nordics, and the recent spate of reforms may intensify this trend.
Can raising the retirement age ward off cognitive decline? A wave of
studies has looked at this question, but employing survey-based tests of
cognitive functioning as their outcome (Coe & Zamarro, 2008, Rohwedder and Willis, 2008, Bonsang et al., 2012, Mazzona and Peracchi, 2012,
and Bingley and Martinello, 2013). We bring novel evidence from the
comprehensive Danish registers using as outcome a more reliable
measure of mental decline than unincentivized tests of word recall,
namely, diagnosis of Alzheimer’s disease (AD)/dementia made in the
hospital. In the Danish system, specialist visits are only possible by GP
referral and a diagnosis of AD/dementia usually requires scanning and
tests performed by a specialist. Medical diagnoses, however, may be unrelated to functioning on the labor market, especially if the disease progresses slowly. Furthermore, getting a diagnosis may vary systematically with education or work status. Those who are retired have lower
time costs of investing in their health (getting diagnosed). At the same
time, they also face less of a monetary incentive to invest in their own
health, so effects may cancel out.
We use data on the population of older males 2 observed in the labor
force in 1998 and followed annually up to 2007. Information on the di2
For brevity we only show results for males. Results for females actually do not differ much.
Nordic Economic Policy Review
agnosis of AD/dementia at the time of hospitalization is obtained from
the National Patient Registry and linked to labor market registers. 3 We
condition on absence of dementia/AD in 1998 and on being in the work
force in 1998 (employed or unemployed). We control for age and education (completed basic schooling or not). We limit the sample to those
individuals who in each year are observed to be either in the labor force
or retired. That is, transitions to disability or other types of benefit
schemes are excluded from the sample. The resulting sample consists of
3,449,037 person-year observations. Unlike much of the previous literature, we distinguish between early retirement (ER) and normal old-age
pension (OAP). Basic descriptive statistics reveal that average sample
age is 58, varying from 47 to 74. Dementia/AD incidence is 1.4%. This
definition is purely based on hospital diagnoses and therefore does not
include any diagnosis made in a doctor’s office and not necessitating a
hospital visit. Omitting the non-retired, retirement duration is on average 3.7 years for ER and 3 years for OAP. 57% of the sample has completed their basic schooling.
Figure 1 shows the incidence of dementia/AD according to retirement duration in the sample. The incidence obviously rises with retirement duration as people get older. ER retirees are a much larger subsample and therefore resemble the overall sample while OAP retirees
have both higher incidence of dementia/AD and a sharper rise in this
when going from work to retirement. By definition, OAP retirees are
older when they retire, compared to ER retirees, and the diagram does
not control for age. It is also important to be able to disentangle whether the abrupt rise among OAP pensioners when retiring indeed is the
causal effect of going on retirement or whether it reflects reverse causation, i.e., people going on OAP when diagnosed with dementia/AD.
3 In practice, the broader ICD category we access includes other mental and behavioral disorders, and nervous system disease, but many of these are rare in the population and do not exhibit increasing incidence
with age.
Nordic Economic Policy Review
Figure 1: Dementia incidence by retirement duration, males aged 47–74, 1998–2007
Dementia incidence in %
Retirement duration in years
We estimate a fixed effects regression of the probability of obtaining a
diagnosis of dementia/AD in the sample period as a function of retirement, a polynomial in age, 4 and education, including an individual timeconstant effect, and an idiosyncratic error term. Comparison within individuals allows us to see whether the individual is more likely to obtain a diagnosis for dementia/AD after retirement. The fixed effects
model removes all time-constant heterogeneity from the model: Indeed,
education is most likely superfluous, as it rarely changes at older ages,
although adult education is fairly prevalent in Denmark.
ER is available from age 60 in the sample period, although a financial incentive is in place to defer it to age 62, and OAP for most of the
period from age 67, as in Section 3 (changed to 65 starting in 2004).
Since the decision to retire is obviously endogenous, we instrument ER
by being 60 or above, and OAP by being 65 or above, as clear spikes
have been shown in the retirement hazard in Denmark over this time
period at ages 60, 62, and 65 (Bingley et al., 2014).
4 A quadratic in age has been found to capture the age dependence of cognitive decline quite well, see Bonsang et al. (2012).
Nordic Economic Policy Review
Table 1: Effect of retirement & retirement duration on dementia/ad diagnosis
Has begun early retirement
Has begun normal OAP retirement
Log early retirement duration
Log normal retirement duration
Education above basic school
Note: FE is fixed effects. FE IV is fixed effects with instrumental variables estimation.
Sample includes all workers and all ER retirees, 1998–2007, 47 ≤ age < 74. Has begun ER retirement instrumented by Age ≥ 60.
Sample includes all workers and all OAP retirees (not including ER retirees), 1998–2007,
47 ≤ age ≤ 74. Has begun OAP instrumented by Age ≥ 65. F tests show that both instruments are
positive and very strong. F tests of individual effects show joint significance in all cases.
Sample includes all ER retirees only with positive retirement duration, 1998–2007.
Sample includes all OAP retirees only with positive retirement duration, 1998–2007.
Results of the regression analysis appear in Table 1 (standard errors in
parentheses). In columns 1 and 2 we see that the rise in dementia incidence at retirement occurs mainly among OAP retirees. The column 1
sample includes all workers and all ER retirees. The column 2 sample
includes all workers and OAP retirees only in each year (i.e., excluding
ER retirees who would automatically transit to OAP at age 65). Thus,
working individuals contribute to both samples. The size of the effect on
Nordic Economic Policy Review
the cognitive decline of OAP retirees is 0.37 percentage points on a
mean overall incidence of 3.3%, about 11% of baseline.
As argued in Rohwedder and Willis (2008), the effect of non-work
on cognitive decline may appear only with a lag. Thus, we next identify
the effect of retirement duration on dementia diagnosis given that one
is retired, i.e., for the sample of retirees only. Retirement duration is
measured in logs as in the previous literature (see Bonsang et al, 2012).
Retirement duration is not instrumented for in this analysis, and endogeneity only controlled for via fixed effects. Arguably, the more severe selection is encountered for the decision of when to retire, and not
for the duration of retirement, which is mainly driven by age and program structure.
Results appear in columns 3 and 4 of Table 1. Longer retirement duration is associated with cognitive decline, but again only evident in the
sample of OAP retirees. The effect size is a 2.7 percentage point increase
in dementia incidence for every log point increase in retirement duration. Since mean OAP duration is 3 years, for every 3.6 months increase
in retirement duration, dementia incidence rises approximately 0.27
percentage points, equalling 8.6% of baseline, i.e., a strong effect.
To the extent that the policy variables which we use to instrument for
retirement and the fixed effects specification together eliminate the endogeneity bias, the results show both an immediate increase in dementia
incidence when going on normal (OAP) retirement, which is not just reverse causation, and a further increase as more time is spent on OAP.
For early retirees, the results actually show positive (protective) effects, both of going on ER, and of ER duration (although the effect size of
duration is an order of magnitude smaller than for OAP). This result
may be interpreted in a number of ways. First, ER retirees are younger
when retiring, and so their cognitive reserves more intact. With time,
ER retirees may well begin to show the same patterns of cognitive decline as normal retirees. It could also be that those diagnosed with dementia in the ER group transit to disability (DI) and hence disappear
from the sample, leaving behind a healthier group. In a robustness test,
individuals in the relevant age-interval who were on DI and were observed sometime during the period in the ER state were pooled with the
sample in column 1, Table 3. This increased the sample size slightly, but
Nordic Economic Policy Review
did not change the coefficient appreciably. It is possible that ER is not
health related, whereas OAP is, meaning that people who work beyond
ER age have stronger work preferences, and probably more health
knowledge, and only stop working when they are diagnosed with mental decline. We also saw that the effect of retirement duration given that
one was retired was stronger among OAP retirees, controlling for age
and its square. Greater health knowledge might mean that OAP retirees
visit the doctor more frequently, although some of this should be captured by the fixed effect. A further explanation can be that ER retirees
are healthier when they retire, and are able to stave off cognitive decline by pursuing a more active and healthy lifestyle during retirement,
whereas those in the OAP age group are not able to, because of poorer
health status and comorbidities when they retire (we control for age,
but not health). Relief from work-related stress and strain and more
frequent exercise have been found to be the primary mechanisms
through which retirement has a positive effect on health (Eibich, 2015).
A final explanation for the findings could be that the nature of work is
different for ER and OAP retirees – if the latter are giving up mentally
demanding work, this could bring about faster cognitive decline for
them during retirement. This is also an area that can be studied further.
Recent policy changes targeting disabled with
remaining working capacity
Here, we consider recent policy initiatives undertaken in Sweden and
Denmark to encourage individuals with some remaining working capacity to remain on the labor market.
Sickness absence and disability policy changes in
Sweden since 2006
Sweden has traditionally had very high rates of sickness absence (it is
the world leader in terms of long-term sickness absence). It is also one
of the countries with the highest rates of inflow to disability from sickness absence. This has been due to relatively lenient eligibility criteria
Nordic Economic Policy Review
and easy access to generous benefits (also temporary benefits), compared to the other Nordic countries. Since 2007, however, the trend in
disability has been reversed (except among the youngest age group),
and sickness absence has fallen (OECD, 2009, and Thorsen et al., 2015).
A series of reforms have been implemented by the Swedish government, establishing a “rehabilitation chain” with fixed time limits and
regular reassessments. Among other things, the sickness absence period
was reduced to one year, employers’ copayment was increased, and
permanent benefits replaced by temporary benefits coupled with work
incentives for young and prime-aged workers. Effectively, disability
pension would henceforth only be granted to those with a permanent
reduction in working capacity. An important element of the reforms
was to centralize the Social Insurance Agency (SIA), to increase its contact and cooperation with the Public Employment Service (PES), and to
give county councils and regions financial incentives to improve the
quality and efficiency of the sickness certification process (the so-called
“Swedish sick leave billion”, see Thorsen et al., 2015). In 2008, a rehabilitation guarantee targeted workers with mental disorders such as depression, anxiety, or stress, providing them with free cognitive behavioral therapies and employing multimodal rehabilitation efforts.
How much of the reduction in sickness absence and disability in recent times in Sweden can be attributed to the government’s reforms?
Hartman (2011) looks at the evolution of sickness absence and disability
since the 1990s and points out that they began trending downward even
before the current government came into office. Although evaluation
studies of these reforms are still in the making, the review of a first batch
of studies in Hartman (2011) indicates that the rehabilitation chain appears to have had a substantial impact, and that the transfer to the PES
during the first year of sick leave appears to be functioning smoothly.
However, in her opinion, the government moved too hastily to implement
these reforms, and in the process, groups that were already in the system
according to the old rules clearly lost out. The eligibility criteria for permanent disability may have become too stringent and thereby difficult for
the oldest disabled to fulfill, and it is worrisome that the use of disability
benefits among the young is increasing in Sweden, in parallel with growNordic Economic Policy Review
ing academic requirements from school-leaving and vocational programs
(Hägglund and Skogman Thoursie, 2010).
4.2. Partial sick leave and wage subsidy schemes in
Partial sick leave has been prevalent in many Nordic countries for decades – it offers sick-listed workers the possibility to return to work on a
part-time basis. Finland was a relative latecomer, instituting partial sick
leave in 2007 (Thorsen et al., 2015). The partial sick leave policy in
Denmark has been evaluated by Høgelund, Holm and McIntosh (2010).
Using both Danish register and survey data, this study examined the
return to normal working hours of sick-listed workers covered by a national graded RTW program. The program was designed such that during work hours the workers receive the normal wage, and for the hours
off work spent sick they receive sickness pay. The approach takes into
account unobservable differences between 265 program participants
and 669 non-participants, all of whom were sick-listed for at least eight
weeks. The results show that graded RTW significantly increases the
likelihood of returning to regular working hours. Similar evidence of the
success of partial sick leave programs is found for other Nordic countries, too, by Andrén and Svensson (2012) for Sweden, Kausto et al.
(2014) for Finland, and by Markussen et al. (2012) for the full Norwegian population.
While partial sick leave policies appear to increase the probability of
RTW of the sick-listed, wage subsidy programs are potentially powerful
for reducing unemployment of the disabled, by replacing passive social
insurance payments by employer subsidies to hire disabled workers
and improve their attachment to the labor force. The Danish Flexjob
program is a unique subsidy program for the disabled that has been cited as a good example of a support scheme by the OECD, but only recently has it been reliably evaluated. Datta Gupta, Larsen and Thomsen
(2015) use Danish register data and exploit a 2002 change in the reimbursement to government units (but not to municipal or regional units)
within a difference-in-difference framework. A lowering of the subsidy
from full to partial for governmental units only changed their hiring
Nordic Economic Policy Review
practices to favor insiders over outsiders. This suggests that an increase
in wage subsidies can reduce rehiring of previous employees who are
disabled and increase the hiring of non-employed disabled.
The Nordic welfare states share in common an elaborate tax-financed
system of income compensation in the event of a loss of working capacity. Benefits are generous and the coverage period long in an international context. The criteria for eligibility for such programs were also
relatively lax in the past, but have recently been tightened up. Although
this policy shift has been viewed as problematic by some, it may have
been justified by the existence of moral hazard problems. Indeed, several studies reviewed here point to the existence of moral hazard in the
Nordic context, which may manifest itself in an increased tendency to
exaggerate health problems for the purpose of claiming disability (ex
post moral hazard), and in risky behavior (ex ante moral hazard), e.g.,
leading to life style diseases and prolonged voluntary unemployment
followed by early retirement. Clearly, misreports of health generate biases in estimates of the true effect of health on retirement. On the other
hand, when health is more objectively measured, such as via hospital
diagnoses, the effects on retirement are expected to be real.
Nordic countries are also experiencing a rise in longevity, and recent
results show that there is scope for having people work longer in the future. We provide novel analyses based on comprehensive register data
from Denmark, showing that retaining individuals on the labor market
beyond age 65 has a causal effect on reducing their incidence of dementia/Alzheimer’s disease. Interestingly, however, no such effect is present
for early retirees. A number of possible reasons for this are considered,
including that early retirees are younger when retiring, and their cognitive reserves more intact. With time, they may well begin to show the
same patterns as normal retirees. Thus, beside coping with aging, there
may be additional benefits to reforms retaining older workers on the labor market, in terms of reductions in cognitive decline.
Nordic Economic Policy Review
One of the issues not touched upon in this paper is the interdependence of couples’ retirement decisions. Any policy affecting retirement or
disability may have spousal multiplier effects beyond the affected individual worker, and this may extend to concerns of moral hazard. For
instance, Johnsen and Vaage (2015) found that the effect of the husband
being eligible for ER is a 2.9 percentage points increase in the likelihood
of the wife receiving disability pension. Similarly, spouses’ retirement
decisions, shared leisure, and lifestyle choices will affect both partners’
physical and mental functioning.
Some of the most successful initiatives undertaken in recent times by
Nordic governments to try to improve labor market attachment and return-to-work of the sick and disabled appear to be partial sick leave policies and wage subsidy programs, coupled with a tightening of disability
criteria, shortening of the sickness absence period, frequent reassessments, employer co-payments, and close cooperation between disability
councils and employment services. The evidence provided in this paper
suggests that these changes have been necessary for curbing the outflow
of workers with remaining stocks of working capacity from the labor
market, thereby improving dependency ratios, labor market shortages,
and public finances, while also impacting individual own health.
Andersen, J. G. (2011). Ageing and welfare reform in the Nordic Countries, 1990–
2010. Multipillar pensions and “activation” of social and labour market policies.
Aalborg University Working Paper.
Andrén, D. & Svensson, M. (2012). Part-time sick leave as a treatment method for
individuals with musculoskeletal disorders. Journal of Occupational Rehabilitation
22, 418–426.
Bach, H., Datta Gupta, N. & Høgelund, J. (2007). Employment effects of educational
measures for work-injured People. IZA discussion paper 2657.
Bingley, P., Datta Gupta, N., Jørgensen, M. & Pedersen, P.J. (2014). Health, disability
insurance and retirement in Denmark. In D. A. Wise (ed.), Social Security Programs
and Retirement Around the World: Disability Insurance Programs and Retirement.
Nordic Economic Policy Review
Bingley, P., Datta Gupta, N. & Pedersen, P.J. (2012). Disability programs, health, and
retirement in Denmark since 1960. In D. A. Wise (ed.), Social Security Programs
and Retirement around the World: Historical Trends in Mortality and Health, Employment, and Disability Insurance Participation and Reforms. University of Chicago
Press, Chicago.
Bingley, P., Datta Gupta, N. & Pedersen, P.J. (2015). Health capacity to work at older
ages in Denmark. Aarhus University Working Paper.
Bingley, P. & Martinello, A. (2013). Mental retirement and schooling. European Economic
Review 63, 292–298.
Bolduc, D., Fortin, B., Labrecque, F. & Lanoie, P. (2002). Workers’ compensation,
moral hazard and the composition of workplace injuries. Journal of Human Resources 37, 623–652.
Bond, E. W. & Crocker, K.J. (1997). Hardball and the soft touch: The economics of
optimal insurance contracts with costly state verification and endogenous monitoring costs. Journal of Public Economics 63, 239–264.
Bonsang, E., Adam, S. & Perelman, S. (2012). Does retirement affect cognitive functioning? Journal of Health Economics 31, 490–501.
Börsch-Supan, A. (2007). Work disability, health and incentive effects. Sonder Forschungs Bereich 504, 07–23.
Butler, R. J., Durbin, D.L. & Helvacian, N.M. (1996). Increasing claims for soft tissue
injuries in workers’ compensation: Cost shifting and moral hazard. Journal of Risk
and Uncertainty 13, 73–87.
Christensen, B. J. & Kallestrup-Lamb, M. (2012). The impact of health changes on
labor supply: Evidence from merged data on individual objective medical diagnosis codes and early retirement behaviour. Health Economics 21, 56–100.
Clark, A. (2003). Unemployment as a social norm: Psychological evidence from panel data. Journal of Labor Economics 21, 323–351.
Coe, N. B. & Zamarro, G. (2011). Retirement effects on health in Europe, Journal of Health
Economics 30, 77–86.,
Datta Gupta, N., Kleinjans, K.J. & Larsen, M. (2015). The effect of a severe health shock on
work behavior: Evidence from different health care regimes. Social Science & Medicine
136–137, 44–51.
Datta Gupta, N., Kristensen, N. & Pozzoli, D. (2010). External validation of the use of
vignettes in cross-country health studies. Economic Modelling 27, 854–865.
Datta Gupta, N. & Larsen, M. (2007). Health shocks and retirement: The role of welfare state institutions. European Journal of Ageing 4, 183–190.
Nordic Economic Policy Review
Datta Gupta, N. & Larsen, M. (2010a). Evaluating labour market effects of wage subsidies for the disabled. The Danish Flexjob scheme. SFI Working Paper 07:2010.
Datta Gupta, N. & M. Larsen (2010b). The impact of health on individual retirement
plans: Self-reported versus diagnostic measures. Health Economics 19, 792–813.
Datta Gupta, N., Larsen, M. & Thomsen, L.S. (2015). Do wage subsidies for disabled
workers reduce their non-employment? – Evidence from the Danish Flexjob
Scheme. IZA Journal of Labor Policy 4:10,
Diamond, P. A., & Mirrlees, J. A. (1978). A model of social insurance with variable
retirement. Journal of Public Economics 10, 295–336.
Eibich, P. (2015). Understanding the effect of retirement on health: Mechanisms and
heterogeneity. Journal of Health Economics, 43, 1–12.
Hartman, L. (2011). Swedish policies for the disabled. What do we know about the
effects? Uppsala University Working Paper.
Hägglund, P., & Skogman Thoursie, P. (2010). De senaste reformerna inom
sjukförsäkringen: En diskussion om deras förväntade effekter (The latest reforms
within sickness insurance: A discussion of the expected effects). Studier i Finanspolitik 2010/5, Finanspolitiska Rådet.
Høgelund, J., Holm, A. & McIntosh, J. (2010). Does graded return-to-work improve
sick-listed workers’ chance of returning to regular working hours? Journal of
Health Economics 29, 158–169.
Johnsen, J.V. & Vaage, K. (2015). Spouses’ retirement and the take-up of disability
pension. University of Bergen Working Papers in Economics, No. 3/15.
Kausto, J., Viikari-Juntura, E., Virta, L.J., Gould, R., Koskinen, A. & Solovieva, S. (2014).
Effectiveness of new legislation on partial sickness benefit on work participation: A
quasi-experiment in Finland. BMJ Open 2014;4:e006685.
Larsen, M. & Pedersen, P. J. (2012). Paid work after retirement: Recent trends in
Denmark. IZA Discussion Paper No. 6537.
Lassila, J & Valkonen T. (2006). The Finnish pension reform of 2005, The Research
Institute of the Finnish Economy (ETLA). Discussion Papers, No. 1000.
Leth-Petersen, S., & Rotger, G.P. (2009). Long-term labour market performance of
whiplash claimants. Journal of Health Economics 28, 996–1011.
Lindbeck, A., Nyberg, S., & Weibull, J. (1999). Social norms and economics incentives
in the welfare state. Quarterly Journal of Economics 114, 1–35.
Lindbeck, A., Palme, M., & Persson, M. (2011). Sickness absence and local benefit cultures. IZA Discussion Paper No. 6245.
Nordic Economic Policy Review
Markussen, S., Mykletun, A. & Roed, K. (2012). The case for presenteeism – evidence
from Norway’s sickness insurance program. Journal of Public Economics 96, 959–
Mazzona, F., & Peracchi, F. (2012). Aging, cognitive abilities and retirement. European
Economic Review 56, 691–710.
Mookherjee, D., & Png, I. (1989). Optimal auditing, insurance and redistribution.
Quarterly Journal of Economics 104, 205–228.
NOMESCO (2014). Health Statistics in the Nordic Countries in 2014. Nordic MedicoStatistical Committee (NOMESCO), 102:2014, Copenhagen.
Nordic Council (2010). Crack the dementia code.
OECD (2009a). Sickness, disability and work: Breaking the barriers. Sweden: Will the Recent Reforms Make It? OECD iLibrary.
OECD (2009b). Sickness, disability and work – keeping on track in the economic
downturn, background paper. OECD.
Rohwedder, S. & Willis, R.J. (2010). Mental retirement, Journal of Economic Perspectives 24, 119–138.
Rothschild, M., & Stiglitz, J. E. (1976). Equilibrium in competitive insurance markets:
An essay on the economics of imperfect information. Quarterly Journal of Economics
90, 629–649.
Saltman, R. B., Vrangbaek, K., Lehto, J. & Winblad, U. (2012). Consolidating national authority in Nordic health systems. Eurohealth incorporating Euro Observer 18, 21–24.
Thorsen, S.V., Friborg, C., Lundstrøm, B., Kausto, J., Örnelius, K., Sundell, T., Kalstø, Å.
M., Thune, O., Gross, B.-O., Petersen, H. & Haram, Ö. (2015). Sickness absence in the
Nordic countries. Nordic Social Statistical Committee (NOSOSCO), Copenhagen.
Waddell, G. & Burton, A. K. (2006). Is work good for your health and well-being? Report for Department for Work and Pensions, TSO (The Stationery Office), London.
Nordic Economic Policy Review
9. The future of Welfare
How worried should we be about
Wagner, Baumol and Ageing?1
Andreas Bergh, Research Institute of Industrial Economics (IFN),
Box 55665, 102 15 Stockholm, Sweden
Welfare services are an important part of the Nordic welfare states both
financially and for welfare state redistribution. Baumol’s cost disease,
Wagner’s law, and population ageing are often said to bring challenges
for the future provision of welfare services. While none of the three
poses an immediate threat against the financial sustainability of the
welfare state, they have important implications for distribution and for
the political support for the welfare state. The combination demographic change, a higher relative price of welfare services and increasing demand for welfare services may force politicians to make a difficult
choice between increasing taxes, allowing people to top up publicly financed services with additional private financing, or risk eroding support for the welfare state.
1 Financial support from the Swedish Research Council and Torsten Söderberg Foundation is gratefully
acknowledged. The author also thanks Sebastian Jävervall for research assistance.
In welfare state research, a distinction is often made between cash benefits and benefits in kind. The former consists of income transfers and
social insurance schemes; the latter largely consists of a set of welfare
services (such as schooling, health care and elder care) that are mainly
or entirely tax financed and available for citizens at low or no monetary
costs. The discussion about role of benefits in kind in the welfare state
goes far back. Based on traditional welfare economics it can be argued
that in-kind redistribution is inefficient, because the potential beneficiary could typically do better (and never worse) if he were given the
cash equivalent of the in-kind subsidy. One might thus ask why not all
redistribution is done using cash transfers. There are different types of
answer to that question. First, the traditional welfare economic argument that it is sufficient to redistribute income and to rely a markets for
allocative efficiency rests on assumptions such as markets being sufficiently competitive (Arrow, 1963) and that equilibria are unique (Foldes, 1967). A second more fundamental objection is that the traditional
welfare economic approach implicitly assumes that tax payers care only
about the utility of beneficiaries, which may not be true. As noted by
Buchanan (1968) taxpayers may well aim to support only “specific
spending patterns” (p. 189), in which case re-distribution in kind is not
obviously inefficient. 2
Regardless of how welfare services are motivated, they are an important part of the welfare state and they matter a lot for welfare state
redistribution. Compared to other countries, welfare services are particularly extensive in the Nordic welfare states. In the literature on welfare state classifications (Titmuss, 1974, Esping-Andersen, 1990), welfare states are typically divided into three categories: the “marginal”
(typical for Anglo-Saxon countries), the “industrial achievement” (typical for Central European countries) and the “institutional” (typical for
Scandinavia). As noted by both Bambra (2005) and Jensen (2008), the
research on welfare regimes has focused on cash transfers, paying less
See Garfinkel (1970) for an analysis that accounts for both beneficiary and tax payer preferences.
Nordic Economic Policy Review
attention to the role of welfare services. The contributions of Bambra
and Jensen confirm however that the Nordic countries are different also
when welfare services are accounted for. 3
Compared to the OECD average, the Nordic countries actually differ
more from the OECD average in the spending on welfare services than
they differ in the spending on cash transfers. As shown in Figure 1,
spending on welfare services in the Nordic countries are now financially
as important as cash benefits, with both at 14% of GDP in 2014.
Figure 1: Services and transfers (% of GDP), average for Sweden, Denmark, Finland, Norway
compared to OECD average 1980–2014
Nordic services
Nordic transfers
OECD services
OECD transfers
Source: OECD.
Welfare services and redistribution
Though sometimes neglected, the provision of welfare services has important consequences for welfare state redistribution. Rothstein (2001)
provides a simple model of how universal programs aimed at the entire
3 The one exception is health care expenditure, which according to Jennsen (2008) is characterized by uniform levels of expenditure across countries.
Nordic Economic Policy Review
population have a redistributive effect even when they are financed using proportional taxes, illustrated in figure 2.
Figure 2: Redistribution in a stylized welfare state
Source: Based on Rothstein 2001.
Because cash transfers and welfare state services both matter for welfare state redistribution, inequality measures based only on the distribution of household income will not fully capture all political efforts to
increase equality. It is far from obvious, however, how the redistributive effect of welfare services should be accounted for.
Welfare services contribute to equality both directly and indirectly.
Figure 3 illustrates an attempt by Verbist and Matsaganis (2012) to approximate the extra redistribution created by welfare services by simply adding the value of welfare services to household disposable income,
and calculating Gini coefficients that include the value of welfare services. The OECD average Gini coefficient in 2008 decreases from 0.29 to
0.23 when the value of welfare services is included. The Nordic countries are similar to the average, with Sweden falling from 0.25 to 0.19,
and Denmark falling from 0.22 to 0.17.
Nordic Economic Policy Review
Figure 3: Gini coefficient 2008 for disposable household income including and excluding the value
of tax financed services
United States
United Kingdom
With services
Without services
Source: Verbist and Matsaganis (2012).
The calculations in Figure 3 are problematic in several ways. Most importantly, they are based on the assumption that receiving welfare services worth a certain amount is similar to receiving the same amount in
cash. That is typically not the case. When the content of publicly provided services differs more from what households would buy if given cash
instead, there is less extra direct redistribution associated with welfare
Nordic Economic Policy Review
services. The estimates in Figure 3 should therefore be seen as an upper
bound on the extra redistribution created by welfare services.
A second problem is that the degree to which the content of the welfare services provided matches what households would buy with the
cash equivalent is likely to differ between countries because political
systems differ. As a result, the cross country ranking shown in figure 3
is not entirely reliable.
Both problems are related to the discussion above on benefits in kind
versus cash benefits. Should the value of welfare services and their impact on distribution be calculated using the preferences of the beneficiaries or the preferences of the tax payers, or possibly using some other set
of social preferences? The welfare services provided by the welfare state
may differ not only from what beneficiaries would buy with the cash
equivalent, but also from what tax payers would want to provide.
A third problem with the data in Figure 3 is that they do not capture
the indirect or dynamic channels by which welfare services can affect
the income distribution. For example, if publicly provided schooling
contribute to a more egalitarian distribution of human capital in the
population, one should expect that incomes are also more equally distributed. The main reason why primary schooling contributes to a more
equal income distribution is not that households receive schooling
worth a certain amount of money yearly, but rather the way in which
schooling affects the distribution of human capital in the population. A
substantial and significant equalizing effect of primary education has
been identified for Sweden by Meghir and Palme (2005) and a survey
by Abdullah et al. (2015) demonstrates that a large literature has found
a similar link between education and income equality.
To summarize: The provision of welfare services is financially important. As shown in Figure 1, the Nordic countries have during the recent two decades become more similar to the OECD countries when it
comes to expenditure on cash transfers, but they have actually become
more different in spending on welfare services. Welfare services have
thus become an increasingly important characteristic of for the Nordic
welfare states. Welfare services are also an important mechanism by
which the welfare states affects the income distribution and promotes
equality more generally.
Nordic Economic Policy Review
Against this background, it is motivated to ask if there are threats or
challenges for the future provision of welfare services. There is no
shortage of analyses that point to potential problems for the welfare
state connected to the provision of welfare services. In particular, problems related to Baumol’s cost disease, Wagner’s law and the population
aging are often mentioned. These three are discussed in the following.
The Baumol effect
The Baumol effect, also known as Baumol’s cost disease, can be traced
back to Baumol (1967) where the explicit premise is that economic activities can be grouped into two types: “technologically progressive activities in which innovations, capital accumulation, and economies of
large scale all make for a cumulative rise in output per man hour” and
“activities which, by their very nature, permit only sporadic increases in
productivity” (p. 415–6). More generally, activities differ in the relative
importance of man hours as a production factor. For some activities,
typically those that require face to face human interaction, productivity
increases are rare and difficult to achieve. While technological progress
has increased output per hour worked substantially in the manufacturing industry, the time needed to produce many services, for example the
time it takes to help an old man eat a meal, has remained more or less
constant. More generally, if wage increases tend to be uniform across
the labour market, the relative price of services will increase.
The implications of Baumol’s analysis is sometimes said to have dramatic consequences for the public sector, and in particular for the provision of welfare services. As recently noted by Andersen and Kreiner
(2015), both the IMF (2012) and the European Commission (2013) refer
to Baumol’s cost disease as an important expenditure driver for the public sector. But under what circumstances must will higher productivity in
some parts of the economy lead to cost increases or other problems for
the public sector? First, two clarifications must be made.
First, the effect noted by Baumol (1967) applies to all services
where manual labor time and human interaction are important inputs
with no or very imperfect substitutes, regardless of whether these are
Nordic Economic Policy Review
part of the public or the private sector. Baumol (2012) discusses health
care and education but also the performing arts. The Baumol effect has
also been used to explain the decline of the big bands and the growth of
rock ‘n’ roll (DeBoer, 1985). While it may well be the case, especially in
universal welfare states, that many of the activities where productivity
increases are rare and difficult to achieve occur within the public sector,
the Baumol effect itself is independent of the public-private dimension.
Second, it is far from clear how productivity should be measured in
the public sector. Despite these measurement problems, there are many
welfare services where both potential and actual productivity increases
are possible, among other things because of labor saving technologies
and improved management (see, for example OECD, 2006, Productivity
Commission, 2005, Carter et al., 2011 and Arnek, et al. 2013).
When it applies, Baumol’s disease implies that the tax rate is determined by the fraction of total labor employed in the public sector
(Lindbeck, 2006, Andersen and Kreiner, 2015). It is a necessity that taxes must be increased when the fraction of the labor force employed in
the public sector increases, but for a given fraction, the Baumol effect
does not imply any upward pressure on expenditure and thus not on
taxes. A higher relative price of services caused by increasing productivity in manufacturing is in itself is no expenditure driver. It does, however, increase the opportunity cost of services, and policymakers that
wish to maintain or even increase public financing of welfare services
may find it harder to motivate this spending pattern.
It bears emphasizing that the driving force for the Baumol effect is a
fundamentally desirable development, namely increasing productivity
in some parts of the economy, leading to rising real incomes. The challenges come from the fact that these productivity increases do not occur
uniformly. For any given average rate of productivity growth in an
economy, it is only to be expected that in some sectors it will be higher
and in others it will be lower. If large parts of the economy experiences
high productivity growth, average income will increase and demand
patterns will change as a result of income effects. The result may well be
Nordic Economic Policy Review
increasing demand for services where productivity has increased less. 4
If this is the case, it is actually be a sign of progress that low-productive
sectors of the economy are expanding.
It is thus incorrect to describe Baumol’s disease as an expenditure
driver, and it does not suggest that the welfare state will run into problems due to rapidly increasing expenditure on public services. Andersen
and Kreiner (2015) formally shows that the Baumol effect does not imply
that the welfare state is unsustainable. In fact they show that, under certain assumptions, the Baumol effect is compatible with constant principles for the supply of services, a constant distribution of well-being and a
balanced budget, and also leaves room for Pareto improvements.
While the challenges for the welfare state caused by the Baumol effect are less about financial sustainability, they are probably more about
political sustainability and distribution. As noted above, welfare services play an important role in welfare state redistribution. When welfare services become relatively more expensive, it can be seen as increasing the relative price of redistribution. As possible response is to
make publicly financed welfare services less redistributive. Policymakers who wish to maintaining public support for redistributive welfare
services may thus find it increasingly difficult to do so.
The Wagner effect
Adolph Wagner (1893) proposed that there is a positive relationship
between the level of economic development and the size of the public
sector. This general proposition is compatible with several exact meanings. Henrekson (1993) discusses various interpretations and concludes
that it is a reasonable interpretation is that Wagner claimed that public
spending as a share of GDP is positively related to real GDP per capita.
As noted by Durevall and Henrekson (2011) Wagner’s view that the
public sector’s share of GDP tends to grow as real per capita GDP increas4 It should be noted that in many cases, productivity is notoriously hard to measure because products and
services, as well as the quality with which they are associated, are hard to define and measure.
Nordic Economic Policy Review
es was an accepted fact in public economics among influential observers
such as Atkinson and Stiglitz (1980) and Easterly and Rebelo (1993), and
also in mainstream textbooks (such as Hindriks and Myles (2006)). Until
the early 1990s, Wagner’s proposition had received strong empirical
support (see, for example Musgrave 1969, Mann 1980 and Abizadeh and
Gray, 1985) and is still today often referred to as a law.
Surveying the more recent literature, Durevall and Henrekson
(2011) categorize 40 studies published after 1990 and show that about
35% fail to find evidence for Wagner’s Law, 30% find support when
controlling for other variables or focusing on specific types of expenditures, and 35% obtain unqualified support for the hypothesis. Their
own study of the long run development within Sweden and the UK since
1800 suggest that Wagner’s Law does not hold in the long run, but that
data are consistent with Wagner’s Law for the periods 1860–1913 and
Can it be said that Wagner’s law is now falsified? Strictly speaking,
any observation where it clearly does not hold is arguably sufficient to
change the label from law to regularity. Given how rare laws are in the
social sciences, the positive correlation between the public sector’s
share of GDP and real GDP per capita remains an important regularity.
An important explanation for the Wagner regularity is that the long
run income elasticities are well above 1 for many welfare services that
in many countries are provided mainly by the state. An often cited
source is Fogel (1999), who calculated long run income elasticities for
various consumption categories in USA as shown in Table 1.
Table 1: Long-term income elasticities for USA (1875–1995)
Source: Fogel (1999).
Nordic Economic Policy Review
Health Care
According to Fogel (1999) the long run income elasticity for welfare
services such as health care and education is well above unity. 5 Hirsch
(1961) calculates an income elasticity at 1.1 for the US over the period
1900–1958 for public education expenditure only. High income elasticities for health care in other OECD countries are also documented by
OECD (1985) for the period 1960–1983 period. More recent studies
based on micro data (e.g. Parkin et al. (1987)) have found lower elasticities, but according to Getzen (2000) the diversity of findings reflects the
fact that individual income elasticities are typically near zero, while national health expenditure elasticities are commonly greater than 1.0.
As a result, we expect expenditure on health care and education to
expand as a share of total expenditure as countries grow richer. This is
also the trend documented by OECD (2006) for both public and private
expenditure in OECD over the period 1970–2005. The fact that countries that grow richer tend to spend a larger share of the GDP on these
services does not necessarily mean that public expenditure on these
areas must also increase their GDP share. While lots of historical data
are in line with the Wagner regularity, the crucial question with regard
to the future of the welfare state is if further income growth will imply
further increases in public sector size. The answer ultimately depends
on political decisions, discussed further in section 5.
Demographic change is an often mentioned challenge for the welfare
state. The basic idea is that a relatively older population increases expenditure via both the pension system and increasing demand for welfare services such as health care and elder care. For example, Morrow
and Roeger (1999) estimated that the ratio of the population above 65
to the population in labor active groups will increase from 24 to 49%
within the EU15 countries between 2000 and 2040. Similar predictions
5 Note also the high income elasticity reported by Fogel for leisure. It is natural that people want to expand
non-material activities as incomes increase. That trend poses yet another challenge for the public sector
because labour income is taxed whereas leisure is not.
Nordic Economic Policy Review
are made in more recent analyses: According to the 2015 Ageing Report
from the European Commission, the ratio is projected to increase from
27.8% today to 50.1% in 2060 for the EU as a whole.
Based on such population forecasts it is easy to predict a dismal future for ageing societies, in particular where the welfare state plays an
important role in the income smoothing over the life cycle. Again, one
must be careful not to overstate the problems.
While the demographic changes are often mentioned as a challenge
for the welfare state, the demographic situation is actually less alarming
in the Nordic countries compared to for example Greece, Poland and
Portugal (European Commission, 2015), and a likely explanation is that
the family and labor market policies of the welfare state have contributed to keeping fertility rates (Rovny, 2011). 6
Italy, possibly as a result of child and parent friendly welfare state
policies (see for example OECD, 2006). Most importantly, the long run
problems associated with population ageing are smaller if people can be
expected to work roughly a constant share of their lifespan. Still, it must
be stressed that substantial challenges during the 2020s and 2030s remain even if the average retirement age increases. To see this, a simple
calculation based on the official population forecasts in the Nordic
countries is sufficient.
Figure 5a shows how the working age share of the Swedish population (15 to 64 years old) is expected to decrease from 63% today to
60% in the 2030s, and falling further during the 2050s to reach 58%
around 2060. Over the same time period, the share aged 75 or older is
predicted to increase from 8.5% today to 11% in the 2020s and to reach
13% in the 2040s. Based on numbers like these it seems safe to predict
an increasing demand for elder care and health care that coincides with
a shrinking tax base made up of those in working age. Such predictions,
however, ignore the counter effect that the definition of “working age”
is likely to change as life expectancy increases.
6 Interestingly, Rovny also find that the presence of employment protection legislation — rules concerning
hiring and firing – hinders the growth of fertility rates.
Nordic Economic Policy Review
Looking back one decade in Sweden reveals that the employment
rate among 65 to 74 year olds has been steadily increasing, from 9.9%
in 2005 to 16.5% in 2014. This corresponds to a yearly increase of 0.73
percentage unit in the employment rate. Moreover, the increasing employment trend for 65 to 74 year olds was not visibly affected by the
financial crisis, in contrast to the employment for the entire adult population. The trend towards increasing employment rates among 65 to 74
year olds in Sweden thus seems to be relatively robust. Adding employed 65 to 74 year olds to those who are counted as working age,
changes the prediction for the future substantially if we are willing to
assume that employment among 65 to 74 years olds continues to increase the way it has the most recent decade. As a more conservative
scenario, assume instead that the future yearly increase is 70% of the
yearly increase from 2005 to 2014 (which for Sweden means a yearly
increase of 0.7*0.73 = 0.51 percentage units). The difference between
the two scenarios is illustrated by the fact that employment among 65
to 74 year olds will have risen to 40% in 2060 in the conservative scenario, and to 50% in the optimistic scenario.
Figure 5a–d shows the result of doing these calculations using the
official population forecast for Sweden, Denmark, Norway and Finland. 7
As can be seen in Figure 5, even the conservative scenario improves the
picture substantially in all countries except Denmark (where employment for 65 to 74 years olds fell from 10% in 2005 to 9.1% in 2009, and
then increased to 11.8 in 2014 (corresponding to a yearly increase of
0.2 percentage units over the latest decade). 8 On the other hand, even in
the optimistic scenario demographic balance will still worsen in all
countries at least until the 2030s.
Figure 5a–d. Population share in working age 16 to 64 years
(dashed) and including an increasing share of 65–74 year olds.
7 Note that only Sweden and Norway have forecasts that reach the year 2110, whereas Denmark’s stops at
2050 and Finland’s at 2060.
8 In Finland the trend in employment for 65 to 74 years olds is similar to the trend in Sweden, with employment increasing from 5.1% in 2005 to 10.2& in 2014, corresponding to a yearly increase of 0.56 percentage units. Norway is also similar, with an increase in employment for 65 to 74 years during the same
period from 12.8 to 19.3%, or 0.72 percentage units yearly.
Nordic Economic Policy Review
Upper solid: optimistic scenario with future increase based on trend
2005 to 2014. Lower solid: conservative scenario based on trend 2005
to 2014 multiplied by 0.7). Source: Author’s calculations based on official population forecasts.
Figure 5a: Sweden
Figure 5b: Denmark
Nordic Economic Policy Review
Figure 5c: Norway
Figure 5d: Finland
Nordic Economic Policy Review
As shown by Bengtsson and Scott (2010) the main driver behind the
demographic changes is the transition from high fertility to low fertility,
and only a small part is explained by increased longevity. A schematic
explanation is given by Bergh (2010), reproduced in Figure 6.
Figure 6: Schematic population pyramids during before, during and after the transition from high
to low fertility
Source: Bergh (2010).
When fertility and mortality is high, the demographic structure is akin
to the left triangle in Figure 6. When fertility drops, the population pyramid is transformed to the well-known mushroom shape, with a relatively large share of old in the population. The mushroom shape is the
reason why the demographic balance in Figure 5a to 5d will deteriorate
for most countries during the next decades, but the situation should improve once the cylinder stage is reached.
Finally, it is motivated to as how increased longevity will affect
health care expenditure. If we live longer, will we also need more health
care? The answer depends on the extent to which increasing life expectancy adds healthy years to our lives, or years when we need expensive
health care to stay alive. This is a well-researched topic, and there is
agreement in the literature that health care costs increase by increased
proximity to death, suggesting that longer lives should only have small
effects on total expenditure. Christiansen et al. (2012) summarize the
state of research and confirm that ageing as such can be expected to
cause only a modest increase in health care expenditure per capita in
the future. Their own empirical study, based on 15 EU countries, support this conclusion. They also give examples of cross-country studies
Nordic Economic Policy Review
where demography loses significance when institutional characteristics
are controlled for factors. OECD (2006) reaches a similar conclusion:
rising health care expenditure is mainly explained by rising incomes,
and only to some extent by population aging.
Taxes, topping up or paying twice – can the
welfare state muddle through?
For the welfare state, the combination of the Baumol effect, the high income elasticity of welfare services and the aging population creates a
challenge that must be acknowledged and discussed. As productivity
increases, society grows richer. With higher income, people are likely to
demand more welfare services. But some incomes grow faster than others. When increasing average incomes are combined with increasing
income dispersion, and the income elasticity for welfare services is
above unity, the difference between the level and quality of welfare services demanded by high income earners and low income earners will
grow. Furthermore, welfare states services have traditionally played an
important role in making universal redistributive towards low-income
earners. As the Baumol effect means that they become relatively more
expensive, they are increasingly demanded by high income earners.
When more is demanded from services that are to be produced by the
public sector, higher taxes are required. Raising taxes may be problematic for efficiency reasons, especially in countries where average taxes
are already relatively high. Moreover, the need to raising taxes in order
to improve welfare services in ways demanded by high income earners
adds a potentially problematic political dimension. 9
If tax increases are ruled out, two remaining options are the socalled topping up strategy and the paying twice strategy. In this context,
topping up entails that those who desire to do so may add private financing on top of the publicly financed welfare service level in order to
9 However, see Blomquist et al. (2010) for an interesting idea involving increased tax progressivity as a way
to make high income earners to pay more for publicly provided services.
Nordic Economic Policy Review
achieve the desired level, quantity or quality of the service. If topping up
is not allowed, high income earners still have the option to pay twice, in
the sense that after paying taxes to a welfare state where the provision
levels are insufficient by their standards, they buy private insurance
arrangements on the market for privately provided welfare services.
In the short run, the difference between the “topping up” and the
“paying twice” strategies may seem small, but the long run political dynamic is fundamentally different. Allowing topping up can be increase
the political support for public provision (Epple and Romano, 1996,
Gouveia, 1997), while having high income earners paying twice is likely
to erode their willingness to pay taxes to finance the public system. Empirical support for the latter mechanism has been provided by Hall and
Preston (1998) who showed that people who opt out from publicly provided health care and pay for private health insurance support less
spending on the public system.
To put it harshly, policy makers are facing an unpleasant choice between three alternatives: Increasing taxes, facilitating topping up and
accepting increasing inequality of access to welfare services, or having
high income earners paying twice and risk eroding welfare state support. In practice, the three strategies are not mutually exlusive. As discussed by Bergh (2008) there are signs that Swedish policy makers implement at least the incremental changes that are necessary to secure
majority support for a high tax welfare state. In many cases, the changes
are likely to imply that the vertical income distribution of the welfare
state decreases, as many welfare services are changed according to the
preferences of the middleclass. The tendency for welfare state programs to adjusted to the interests of the middle class for political reasons is well-documented (Goodin and Le Grand, 1987)
If the strategy continues, policy makers in the future may well opt
for a mix of higher taxes, topping up and having some groups paying
twice, resulting in the Nordic welfare states muddling through the challenges ahead. In any case it can be concluded that while the problems
caused by the Baumol effect for financial sustainability have been
somewhat exaggerated, the political problems related to vertical income redistribution may well be underestimated.
Nordic Economic Policy Review
Abdullah, A., Doucouliagos, H., & Manning, E. (2015). Does education reduce income
inequality? A meta-regression analysis. Journal of Economic Surveys 29, 301–316.
Abizadeh, S. & Gray, J. (1985). Wagner’s Law: A Pooled Time-Series, Cross-Section
Comparison. National Tax Journal 38:209–218.
Andersen, T. M., & Kreiner, C. T. (2015). Baumol’s cost disease and the sustainability
of the welfare state. EPRU Working Paper Series 2015-02.
Arnek, M., Bergman, M., Edquist, H., Hultkrantz, L., Lundberg, S. & Svensson, M.
(2013). Den offentliga sektorn – en antologi om att mäta produktivitet och prestationer. Rapport till expertgruppen för studier i offentlig ekonomi 2013:8.
Arrow, K. J. (1963). Uncertainty and the Economics of Medical Care. American Economic Review 55, 942–943.
Atkinson, A. B. & Joseph E. Stiglitz. (1980). Lectures on Public Economics. New York:
Baumol, W. J. (1967). Macroeconomics of unbalanced growth: the anatomy of urban
crisis. American Economic Review 57, 415–426.
Bengtsson, T. (2010). Population Ageing, a Threat to the Welfare State? – The Case of
Sweden. Springer.
Bergh, A. (2010). Towards a New Swedish Model? In T. Bengtsson (ed.), Population
Ageing – A Threat to the Welfare State? pp. 109–118. Springer.
Buchanan, J. M. (1968). What Kind of Redistribution Do We Want? p. 185.
Carter, K, Dorgan, S. & Layton, D. (2011). Why hospital management matters. Health
International 11.
Christiansen, T., Lauridsen, J., & Bech, M. (2012). Ageing populations: More care or
just later care? Nordic Economic Policy Review, 23–53.
Durevall, D. & Henrekson, M. (2011). The futile quest for a grand explanation of
long-run government expenditure. Journal of Public Economics 95:708–722.
Easterly, W., & Rebelo, S. (1993). Fiscal policy and economic growth. Journal of Monetary Economics 32, 417–458.
Epple, D. & Romano, R. E. (1996). Public Provision of Private Goods. Journal of Political Economy 104, 57–84.
European Commission (2013). Report on Public Finances in the EMU. European
Economy 4-2013.
European Commission(2015). The 2015 Ageing ReportEconomic and budgetary projections for the 28 EU Member States (2013–2060). European Economy 3/2015.
Nordic Economic Policy Review
Fogel, R. W. (1999). Catching up with the economy. American Economic Review 89,
Foldes, L. (1967). Income Redistribution in Money and in Kind. Economica 34, 30–
Garfinkel, I. (1970). Is in-kind redistribution efficient? Institute for Research on Poverty Discussion Paper 74-70.
Getzen, T. E. (2000). Health care is an individual necessity and a national luxury:
applying multilevel decision models to the analysis of health care expenditures.
J Health Econ 19, 259–70.
Goodin, R. E., & Grand, J. L., (1987). Not Only the Poor: The Middle Classes and the
Welfare State. London: Allen and Unwin.
Gouveia, M. (1997). Majority rule and the public provision of a private good. Public
Choice 93, 221–244.
Henrekson, M. (1993). Wagner’s Law – A Spurious Relationship? Public Finance 48,
Hindriks, J. & Myles, G. M. (2006). Intermediate Public Economics. Cambridge, MA.:
MIT Press.
Hirsch, W. Z. (1961). Income Elasticity of Public Education. International Economic
Review 2, 330–339.
Horngren, L. (2001). Pension reform: The Swedish case. Journal of Pensions Management 7, 131–138.
IMF. (2012). The Economics of Health Care Reform in Advanced and Emerging
Lindbeck, A. (2006). Sustainable social spending. International Tax & Public Finance
13, 303–324.
Mann, A. J. (1980). Wagner’s Law: An Econometric Test for Mexico, 1925–1976. National Tax Journal 33, 189–201.
Musgrave, R. A. (1969). Fiscal Systems. New Haven and London: Yale University Press.
OECD (1985). Measuring health care 1960–1983. Expenditure, costs and performance. OECD Social Policy Studies No. 2.
Parkin, D., McGuire, A., & Yule, B. (1987). Aggregate health care expenditures and
national income. Is health care a luxury good? J Health Econ 6, 109–27.
Productivity Commission. (2005). Impacts of Medical Technology in Australia (Progress Report). Melbourne, April 2005.
Rovny, A. E. (2011). Welfare state policy determinants of fertility level: A comparative analysis. Journal of European Social Policy, 21(4). 335–347.
The Cost Disease: Why Computers Get Cheaper and Health Care Doesn’t (2012).. USA:
Yale University Press.
Nordic Economic Policy Review
The drivers of public expenditure on health and long-term care: An integrated approach. (2006). OECD Economic Studies No. 43, 2006/2.
Wagner, A. (1893). Grundlegung der politischen Ökonomie (3rd edition). Leipzig.
Verbist, G. & Matsaganis, M. (2012). The Redistributive Capacity of Services in the
EU. Gini Discussion Paper 53.
Nordic Economic Policy Review
10. Ethnic fractionalization
and the demand for
– Potential implications for the
Nordic model
Johanna Mollerstrom 1
A distinctive feature of the Nordic model is that economic resources are
redistributed between citizens to a high degree. Historically, the Nordic
voters have expressed a higher demand for such redistribution than
people in other parts of the world. This paper considers the factors that
determine individual preferences for redistribution, giving special attention to heterogeneity in the form of ethnic fractionalization. Such
heterogeneity is generally linked to reductions in demand for redistribution. The paper takes as its starting point the fact that the populations in all the Nordic countries are, to varying degrees, becoming increasingly heterogeneous. Potential mechanisms for why this may have
a negative impact on demand for redistribution, and potential consequences for the Nordic model are discussed.
1 Interdisciplinary Center for Economic Science, George Mason University, Fairfax, VA, USA. Research Institute for Industrial Economics, Stockholm, Sweden. Email: [email protected] I am grateful to Henrik Jordahl, Magnus Henrekson, Bjørn-Atle Reme, Jesper Roine and the participants in the 2015 NEPR Conference
for very useful comments and suggestions. Thank you also to Swan Lee for excellent research assistance.
10.1 Introduction
The Nordic model is a combination of a free market economy and an
extensive welfare state, and a system that all the Nordic countries
(Denmark, Finland, Norway, Iceland, and Sweden) exhibit. The model is
intensely debated, admired by many (e.g. Logue, 1979; Popenoe, 1994;
Sachs, 2006; Hilson, 2008; and Brandal et al, 2013), but often also
heavily criticized (see Sanandaji, 2015 for an overview). This paper focuses on one particular aspect of the Nordic model: extensive economic
redistribution between citizens.
Economic redistribution means that economic resources are reallocated between citizens in such a way that the distribution becomes
more equal. Such redistribution can be done through the public sector
or through charities (this distinction is discussed in e.g. Alesina and
Glaeser, 2004), with many people having strong opinions about which
entity is most well suited to conduct the redistribution. The extensive
economic redistribution in the Nordic countries is almost exclusively
conducted through the public sector.
Basically all countries in the world redistribute resources between
citizens to some extent, and policies with redistributive components,
such as social security, universal health insurance, and progressive taxation, have increased in importance over time (Alesina et al. 2004). The
fact that within country inequality has increased in most Western democracies over the last decade(s) is hence primarily reflecting that economic redistribution has not kept up with the pace of the increase in the
pre-redistribution inequality, rather than mirroring decreases in public
redistribution (Wang et al., 2014).
Historically, the Nordic countries have redistributed resources between citizens to a much larger extent than what is the case in other
Western democracies (Alesina and Glaeser, 2004). This is still the case,
but there are tendencies of growing inequality also in the Nordic countries (Atkinson and Morelli, 2014; Morelli et al., 2014; and Wang et al.,
2014). Not only the implemented redistribution but also the preferences of the population are different in the Nordic countries with Danes,
Finns, Norwegians, Icelanders and Swedes being more politically sup220
Nordic Economic Policy Review
portive of economic redistribution than people in other parts of Europe
or in the US.
Preferences for redistribution are not, however, exogenously given
but something that change over time. This paper is discussing how to
understand the political support for redistribution in general. As many
of the Nordic countries are currently experiencing increasing ethnic
heterogeneity, in particular through immigration, the paper specifically
focuses on how this may impact demand for redistribution.
The paper continues as follows: Section 1 discusses cross-country
differences in demand for redistribution from the empirical perspective
and goes on to give an overview of the vast literature on which factors
that determine demand for redistribution. Section 2 focuses more narrowly on the research that relates demand for redistribution to heterogeneity and ethnic fractionalization. Section 3 looks specifically at the
empirics of immigration in the Nordic countries, and discusses historical and current trends. Section 4 concludes.
10.2 The demand for redistribution
10.2.1 Cross-country differences
To measure exactly how much a country redistributes between its citizens is challenging as systems for taxation and welfare spending look
very different in different countries. Direct cross-country comparisons
are therefore hard to make. There are also other policies, such as labor
market regulations (which also tend to differ a lot in structure between
countries), that are designed to benefit lower income groups in particular. Despite these measurement difficulties there is vast agreement on
the general patterns of redistribution in the Western world: European
countries has much more of it than the US, and in Europe the Nordic
countries top the redistribution league.
There are several ways to illustrate this. The gini coefficient is one
of the most commonly used measures of how equal incomes are distributed between people in a country (but many other measures exist
and using them instead yields very similar results). The gini coefficient
Nordic Economic Policy Review
is a measure of statistical dispersion, with 0 indicating perfect equality
and 1 indicating perfect inequality. Table shows gini coefficients for
primary (pre-redistribution) and disposable (post-redistribution) income, for four Nordic countries (Denmark, Finland, Norway and Sweden), three other European countries (France, Germany, UK) and the
United States. By definition, the difference between the gini coefficient
for primary and disposable income is a measure of fiscal redistribution.
It is important to note, however, that the shape of the welfare state will
influence not only post- but also pre-redistribution inequality, so this
difference is not the only relevant redistribution measure.
Some patterns are immediately obvious from Table 1. First, we see
a trend of increasing inequality over time which can be observed in
most countries. This is especially pronounced in the distribution of
primary income.
Second, compared to the US, the Nordic countries have both a
more equal distribution of primary income and a (much) more equal
distribution of disposal income. This indicates that, compared to the
US (and to some extent the UK), the higher equality in the Nordic
countries stems both from structural factors that make preredistribution income more equal, and from the fact that the public
sector is larger and more redistributive.
Third, if we instead compare the Nordic countries to the three other
European countries in the table, France, Germany and the UK (but
again, in some regards the UK has more in common with the US than
with most European countries), we see a different pattern. The gini coefficient for primary income indicates more equality in the Nordic countries, but it is not immediately obvious that the public sector in other
European countries is much smaller or less redistributive than in Denmark, Finland, Norway and Sweden.
Table 2 shows other indicators of the size of the public sector in the
form of tax revenue, and public expenses, in percentage of GDP. To
some extent this reveals information about the degree of redistribution
(since a lot of redistribution is done through taxes and expenses) but
there is of course, also significant public spending which is not redistributive (military spending is one, but not the only example). We see,
again, the sharp divide between Europe and the US, with both taxes and
Nordic Economic Policy Review
public expenses making up a significantly higher share of GDP on the
European side of the Atlantic. In terms of taxation, the Nordic countries
display a relatively larger public sector than the other European countries, but the difference is not anywhere as large as the difference between Europe and the US. For public expenses, the difference between
the Nordic countries and the other European countries in the table is
less pronounced.
Table 1a: Trends in income inequality and redistribution, 1985–2005 (Girni primary income)
Girni primary income
United States
Source: Morelli et al. (2014). Data from the OECD inequality data base.
Table 1b: Trends in income inequality and redistribution, 1985–2005 (Gini disposable income)
Gini disposable income
United States
Source: Morelli et al. (2014). Data from the OECD inequality data base.
Nordic Economic Policy Review
Table 1c: Trends in income inequality and redistribution, 1985–2005 (Redistribution)
United States
Source: Morelli et al. (2014). Data from the OECD inequality data base.
Table 2: Tax revenues and public expenses as % of GDP
Tax revenue
in % of GDP
Public expenses
in % of GDP
Public expenses in %
of GDP, excluding military
Source: The World Bank. Data from 2012.
Lastly, we consider expressed demand for redistribution on behalf of
the population. The World Value Survey, a global research project which
investigates people’s values and beliefs and how these change over
time, ask a question about to what extent the respondent would like incomes in society to be made more equal. Answers are given on a scale
between 1 and 10, where 1 is defined as “Incomes should be made more
equal” and 10 as “We need larger income differences as incentives for
individual effort”. This data is only available for a subset of the countries that we are interested in, and as the World Value Survey is conducted continuously, it can differ quite substantially when the latest da224
Nordic Economic Policy Review
ta point for a country is dated. People in the US tend to give a higher answer to this question, on average, with the latest data point being 5.58.
For Finland, Sweden, France, Germany and the UK, the corresponding
figures are 4.97, 4.88, 5.12, 4.08, 5.36 (World Value Survey, 2015 – data
from 2005–2014). Again we see that the Nordic countries are among
those who demand most redistribution but that for example Germany is
not lagging far behind.
Another way to look at cross country differences in demand for redistribution is to compare political party preferences. Given that the US
has never had an influential Social Democratic party, there is a clear difference between the two sides of the Atlantic in this regard. Within Europe, the Nordic countries have historically had stronger Social Democratic parties than many other European countries. This can be seen as
an indicator of a strong Nordic demand for redistribution, but it should
be noted that party structures are impacted not only by political preferences but also by institutional factors, such as the degree of federalism
and whether the country has proportional or majoritarian representation (Alesina and Glaeser, 2004).
10.2.2 Underlying factors that shape demand for
The theoretical start of the quest to understand how individual demand
for redistribution is shaped started with seminal contributions by Romer
(1975), Roberts (1977), and Meltzer and Richard (1981). They present
simple models that suggest that a person’s relative income or wealth is
decisive for her demand for redistribution: since a relatively richer person benefits less, in monetary terms, from redistribution she should demand less of it. A relatively poor person, on the other hand, has more to
gain from redistribution and should therefore demand more.
These early theoretical predictions about individuals who are relatively rich, compared to others in society, wanting less redistribution
has found empirical support: People with a higher income generally
want less redistribution (see, e.g., Pelzman, 1985; Alesina and Giuliano,
2010; Margalit, 2013; Durante et al., 2014; Powdthavee and Oswald,
2014; and Elinder et al., 2015). Other studies utilize the fact that people
Nordic Economic Policy Review
often have misperceptions about their own position in the income distribution. Mollerstrom et al. (2015) conduct a survey experiment in
Sweden and inform a treatment group about their true relative position,
thereby exogenously (but truthfully) manipulating perceptions of relative income. They show that this causes changes in individual demand
for redistribution and that the direction of these changes indicate that a
substantial fraction of people do exhibit the theoretically proposed negative correlation between (perceived) relative income and demand for
redistribution (see also Cruces et al., 2013, for evidence corroborating
this conclusion).
Income and wealth are not the only individual heterogeneities that
generate differences in demand for redistribution and understanding the
role that relative income, or the perceptions of relative income, play for
the demand for redistribution is made more difficult by the fact that other
underlying variables may also cause a correlation between income and
political preferences. Mollerstrom and Seim (2014) find, for example, that
high-IQ individuals in Sweden favor less redistribution. This could reflect
that high ability individuals, who more easily succeed economically, lean
toward a more individualistic, right-wing world view. In general, there is
ample evidence, however, that also other factors, that have less to do with
pure monetary self-interest, matter for a person’s demand for redistribution. For example, people are often found to care also about the consumption of others (see e.g. Fehr and Schmidt, 1999; and Bolton and Ockenfels, 2000) and such other regarding preferences tend to be positively
correlated with the demand for redistribution (Sears and Funk, 1990;
Fong, 2001; and Alesina and Giulino, 2010).
Individual beliefs about the income generating process have also
been studied theoretically (Piketty, 1995; and Benabou and Tirole,
2006) with special emphasis put on beliefs about the extent to which
individual economic success can be attributed to effort, rather than to
luck, and how such beliefs can be self-fulfilling. Empirically, these beliefs have been found to be a stronger determinant of preferences for
redistribution than income itself (Fong, 2001) with people who believe
that individual economic success can primarily be attributed to individual effort wanting significantly less redistribution in society than those
who perceive luck as being the most important underlying factor.
Nordic Economic Policy Review
A third factor which has been studied theoretically in order to understand how individual demand for redistribution is shaped is risk
aversion. The provision of insurance against future negative economic
shocks have been suggested to be part of the attraction of redistribution. Harsanyi (1953) was one of the first to illustrate how the insurance motive links the demand for redistribution to risk aversion: in his
model, individuals are asked to state their preferences for redistributive
policy behind a veil of ignorance, before their position in the income
distribution is determined by a lottery. In such a situation, demand for
redistribution should be increasing in a person’s degree of risk aversion. Benabou and Ok (2001) extends and enriches a version of the
Meltzer and Richards (1981) model and show that when individuals
care not only about their current position in the income or wealth distribution, but also about their future position, insurance motives, and
hence risk aversion, play a role for demand for redistribution. Empirically, the proposed positive relation between individual risk aversion
and demand for redistribution has been shown to hold (see, e.g., Rainer
and Siedler, 2008; Gaertner et al., 2015). 2
In addition to theoretically founded discussions between demand
for redistribution and its correlates there is also evidence that other
factors matter. There is, for example, consistent, cross-country evidence
that women want more redistribution than men, that those who are unemployed want more redistribution than the employed, that married
people prefer less redistribution than singles and that having more education is negatively correlated with demand for redistribution (Alesina
and Guiliano, 2010). In the US context, blacks have been shown to want
more redistribution than whites (Alesina and Guiliano, 2010).
2 The question of risk exposure and the size of the welfare state has mostly been analyzed from the perspective of individuals, but Rodrik (1998) discusses this also from an institutional perspective.
Nordic Economic Policy Review
10.3 Ethnic heterogeneity and preferences for
In their seminal work seeking to understand why European countries
exhibit much more extensive redistribution than the United States,
Alesina and Glaeser (2004, see also Alesina et al., 2001) argue that two
factors are key to understanding the difference across the Atlantic: political institutions and ethnic heterogeneity.
Regarding political institutions, Alesina and Glaeser argue that the
fact that almost all European countries (with the notable exceptions of
the UK and France) have proportional representation systems have contributed strongly to the redistributive differences between the US and
Europe. According to them, this system facilitated the growth of communist and social democratic parties that support increased redistribution and a large welfare state. The majoritarian American system, on the
other hand, made it difficult for such parties (which at the beginning
where very small) to get any representatives elected. The American
federalism and decentralization, which is stronger than what all European countries except Switzerland display, may also have contributed.
Alesina and Glaeser (2004) conclude that institutional differences can
explain about half of the difference in social spending between Europe
and the US.
The main explanatory factor for the remaining half of the difference in
redistribution between Europe and the US is, according to Alesina and
Glaeser (2004), differences in racial and ethnic fractionalization. The US
has historically been a much more racially and ethnically fragmented society than European countries. Such fractionalization appears, in turn, to
make people demand less redistribution, especially when poverty is concentrated to minority groups. Given the changes in the direction of more
ethnic fractionalization that the Nordic countries (to a varying degree)
are experiencing, it is important to understand what impact this may
have on the Nordic model with its extensive redistribution.
Nordic Economic Policy Review
10.3.1 Ingroup/outgroup bias
A person’s social identity is often partly defined as the sense of self that
she derives from perceived membership in a social group. The seminal
work of Billig and Tajfel (1973, see also Tajfel 1982 and Tajfel et al.,
1971) was the starting point for the empirical research on intergroup
behavior in laboratory settings. They showed that people have a tendency to put both themselves and others into categories and that this
categorization gives rise to favorable treatment of the people in the
same social group as oneself (the ingroup) compared to those in other
groups (the outgroup), i.e. people exhibit an ingroup bias. The way the
groups are created has been shown to matter for the strength of the ingroup bias, but also completely random formation of groups give rise to
the effect (Billig and Tajfel, 1973; and Locksley et al., 1980).
The first experiments investigating the ingroup/outgroup-effect
were done with participants dividing a valuable asset (usually money or
lottery tickets) between another member of the ingroup and a member
of the outgroup, i.e. they could not allocate anything to themselves (see
Tajfel and Turner, 1986; and MacDermott, 2009 for surveys). Experimental economists have also considered situations, both in laboratory
and in field settings, where the individual herself has money at stake in
the decision, for example in the form of a dictator game or a prisoners’
dilemma. Bernard et al. (2006), Goette et al. (2006) and Chen and Li
(2009) show that an ingroup bias arises in such situations as well, with
people behaving more altruistically and cooperatively towards people
in their ingroup (see also Mollerstrom, 2015).
10.3.2 Evidence on the importance of racial/ethnical heterogeneity for demand for redistribution
There is significant evidence that feelings of altruism and willingness to
redistribute in general weaken across racial and ethnic lines. Especially
when those with lower income consist disproportionately of people of
racial or ethnical minorities, the majority prefers less redistribution
(Alesina and Giuliano, 2010). We may not like it, but it is widely observed that individuals are more generous towards others who are
Nordic Economic Policy Review
similar to themselves racially, ethnically and linguistically (see also
Alesina et al., 1999; Alesina et al., 2004; Luttmer, 2001; and Fong and
Luttmer, 2011).
The data in Table 3 come from Alesina and Glaeser (2004) and show
evidence on racial, ethnic, linguistic and religious fractionalization
across Europe and the US. Fractionalization is measured on a range between 0 and 1, and a lower value indicates more homogeneity. For all
measures, except linguistic fractionalization, the US exhibits significantly more fractionalization than what the European countries included in
the table do on average. The three Nordic countries included in the table, Denmark, Norway and Sweden, are even less fractionalized. 3
Alesina and Glaeser (2004, see also Alesina et al., 2001) go on to
show that the correlation between racial fractionalization and social
spending as a share of GDP is statistically significant and strongly negative for a global sample of 52 countries. This indicates that the more
fractionalized a country is, the smaller is the share of GDP that is allocated to social welfare spending. The results is equally strong and statistically significant if only high-income countries are included. Racial
fractionalization is the best predictor of social welfare spending, but the
results for other types of fractionalization are similar. Alesina and Glaeser (2004) also present similar results regarding the relation between
fractionalization and social spending from the 52 US states. 4
The evidence presented above is to a large extent correlational
(with the exception of the laboratory studies which show a causal relation between heterogeneity and demand for redistribution). There are,
however, also field studies that attempt to study the causal relation between ethnic fractionalization and preferences for redistribution. One
influential study, conducted in Sweden, is Dahlberg et al. (2012). They
use a Sweden-wide program for placing refugees in municipalities to
3 As discussed below, these patterns are changing. More recent data for the Nordic countries are considered
in Table 5.
4 Lindqvist and Östling (2010) have in a related paper showed that there is a negative correlation between
size of government and political polarization. Their paper does not study ethnic fractionalization directly,
but the results are of interest to this context as there is a positive relation between ethnic fractionalization
and political polarization.
Nordic Economic Policy Review
generate the exogenous variation needed to study the relationship between increased immigration and demand for redistribution.
Table 3: Fractionalization indices
United Kingdom
United States
Source: Alesina and Glaeser (2004). Data from 1990–2000.
The starting point of Dahlberg et al. (2012) is the fact that immigration
of workers and refugees to European countries in general (and some of
the Nordic countries in particular) have increased substantially over
time. They ask how this affects native citizens’ views on redistribution
and the size of the welfare state. The main contribution of their paper is
that they, in contrast to the (predominantly negative) correlational evidence that several other studies have documented between immigration and demand for redistribution (in addition to Alesina et al., 2001
and Alesina and Glaeser, 2004, see also Shayo, 2009; Stichnoth and Van
der Straeten, 2013; Harmon, 2014), 5 claim to be able to study the causal
During the years 1985 to 1994, a refugee placement program was in
place in Sweden to achieve a more even distribution of refugees over
5 Note that even though the majority of studies find a negative correlation between immigration and the
demand for redistribution, there are some paper which find no correlation (e.g. Gerdes and Wadensjö, 2008;
Brady and Finnigan, 2013) or a conditional negative one (e.g. that the negative correlation only holds when
immigrants are overrepresented among those with lower income, e.g. Finseraas, 2012).
Nordic Economic Policy Review
the country. 277 out of the 286 Swedish municipalities participated,
neither the refugees nor the municipalities could affect the placement
and the total arrival of refugees in the relevant years exceed 200,000.
Given these circumstances the identifying assumption of Dahlberg et al.
(2012), that the placement of refugees was exogenous with respect to
the inhabitants of the municipalities’ preferences for redistribution,
seem justified.
The authors match the municipal-level data on refugee placements,
immigration shares, and other municipal covariates, with individual
level survey information. The survey data come from the Swedish National Election Studies Program which is a rotating panel. The measure
of individual preferences for redistribution was extracted from a survey
question regarding if a person would be “in favor of decreasing the level
of social benefits.”
Using the refugee placement program as an instrument for the share
of immigrants living a particular municipality, Dahlberg et al. (2012)
show that there is a negative and statistically significant relation between increasing the share of immigrants in a municipality and how
much social benefits people in that municipality prefer to see. Moreover, the effect is especially pronounced among white-collar, highincome earners, meaning that the respondents who contribute more extensively to the redistribution scheme are those whose support for redistribution is reduced as the group of potential recipients become
more ethnically diverse (Dahlberg et al., 2012, p. 69). 6
10.3.3 Potential underlying mechanisms
Having concluded that most research indicates that there is a negative
correlation between the increased ethnic fractionalization caused by
immigration and support for redistributive policies, and that this relation may well be causal, the question arises what the underlying mecha6 It should be noted that the results of Dahlberg et al. (2012) have also invoked criticism. See e.g. Nekby and
Pettersson-Lidbom (forthcoming), who claim that the measure of demand for redistribution used in Dahlberg et al. (2012) is suboptimal, that the results are not generalizable to the full Swedish population, and
that the measurement of the refugee placement program is flawed.
Nordic Economic Policy Review
nisms are. Existing research suggests a number of potential mechanism,
but very few studies have attempted to disentangle them. Here we will
discuss three potential mechanism.
First, it may be the case that altruism is lower when directed towards
people of different race or ethnicity as compared to when it is directed to
people who are more similar to the person offering the help. This is suggested and confirmed in, for example, Fong and Luttmer (2011), where it
is showed, in a setting with charitable giving, that people perceive those
who are more similar to themselves as more worthy of help.
A second hypothesis is that it is not altruism, but rather trust and
reciprocity that is impacted by ethnic fractionalization. Trust and reciprocity are both positively related to willingness to redistribution. The
former because you need to trust that people are not taking unfair advantage of the redistributive system in order to demand extensive redistribution, and the latter because you have to be convinced that if you
were the one in need of help, others would provide you with it, just as
you would if you are in the position to help. In general, differences between people, both ethnic, racial and other, seem to generate distrust
(see e.g. Glaeser et al., 2000; Zak and Knack, 2001; Alesina and La Ferrara, 2002; Knack and Zak, 2002, Dienesen and Sønderskov, 2015. The
work of Alesina and Zhuravskaya (2011) is also related although it has a
slightly different focus). Gustavsson and Jordahl (2008) consider this in
a Nordic setting (Sweden) and find a similar negative association between proportion of people born in a foreign country on the one hand
and trust and reciprocity on the other hand.
Third, it is also possible that larger interpersonal differences, for example in the form of ethnic or racial difference, impacts how people
view themselves and the likelihood that they may be on the receiving
end of the redistributive system in the future. As concluded by Benabou
and Ok (2001), people generally do not care only about their current
position in the income distribution when their demand for redistribution is formed – their beliefs about their future position also matters.
Finseraas (2012) find evidence that it is especially the people with a
higher income who demand less redistribution when fractionalization is
high. He presents evidence that suggests that this does not happen because of a decrease in altruism or in trust. Instead, people with a higher
Nordic Economic Policy Review
income are less concerned with downward income mobility, and believe
that the likelihood that this will happen to them, is lower when ethnic
fractionalization is higher.
These hypotheses are not necessarily exclusive and all three mechanisms may be at play when the negative relation between immigration
and demand for redistribution arises. For those interested in affecting
this relation, the policy implications of the various mechanisms are
however very different. Hence, research is most likely going to continue
to try to disentangle these and other underlying mechanisms in order to
gain a better understanding of exactly why we observe a negative relation between immigration and ethnic fractionalization, and demand for
10.4 The Nordic countries: moving towards more
Up until World War II, Denmark, Finland, Norway and Sweden were all
very ethnically homogenous in comparison with most other European
nations. Since then, the paths have been diverging. The differentiation
began during the war, when Sweden took in a substantial number of
refugees. After World War II, immigration was to a large extent intraNordic, but at the beginning of the 1960s demand for labor increased in
Sweden, Norway and Denmark which led to substantial immigration
from other parts of Europe. Finland also delivered a substantial number
of emigrants to Denmark, Norway and Finland (Yousfi, 2010; Kouvo and
Lockmer, 2012).
Thereafter, immigration to Sweden, Norway and Denmark has consisted of refugee migration and family reunification. After the fall of the
Soviet Union, and after joining the European Union, migration flows to
Finland have increased rapidly with migrants predominantly coming
from the former Soviet area. (Kouvo and Lockmer, 2012.)
Regarding regulations, the Nordic countries also exhibit considerable differences. Sweden has been, by far the most permissive whereas
Finland and Norway, and especially Denmark, are more regulated. Taking more types of policies into account, including the allowance of dual
Nordic Economic Policy Review
citizen-ships and the funding of ethnic group organizations or activities,
Kymlicka and Banting (2006) classify 21 OECD countries according to
their multiculturalism. Sweden is the only Nordic country having modest multiculturalism policies, whereas Denmark, Finland and Norway
are categorized as having weak multiculturalism policies.
The Nordic countries are also different in the sense that the antiimmigration political movement grow strong earlier in Denmark and
Norway as compared to Finland and Sweden (Kouvo and Lockmer,
2012). Currently, there is substantial anti-immigration sentiment in all
four countries, however, and this is currently gaining more strength.
Despite recent tightenings, Sweden still has the most generous policies
on immigration. Considering recent trends, all Nordic countries are experiencing a rise in the inflow of immigrants in general and refugees in
particular and the trend for the proportion of the population that is foreign, or foreign-born, is positive in most of the countries. These trends
are illustrated in Table 4 below.
Table 4: Recent trends in immigration for the Nordic countries
A. Immigration, per 1,000 citizens
B. Inflow of asylum seekers, per 1,000 citizens
C. Foreign-born, percentage of population
D. Foreign, percentage of population
Source: OCED (2015).
Nordic Economic Policy Review
As panel A of Table 4 shows, Norway has the highest immigration
among the four Nordic countries. However, this consists to a large extent of temporary workers, often from the other Nordic countries. Panels A and B together paints the picture of significantly more asylum
seekers in Sweden than in any of the other Nordic countries. All countries except Finland are experiencing a rising trend, however.
As for the stock of immigrants, the percentage of foreign-born
(panel C) is highest in Sweden, followed by Norway. The trend is positive in all four countries. The fact that Sweden is more generous regarding naturalization as compared the other three countries is evident from the data in panel D in relation to panel C: fewer people remain foreign in Sweden as it is easier to eventually become a citizen.
Again, the trend is that of a rising share of foreign population in all the
four countries.
Taken together, these data support the view of Sweden as the most
immigration-friendly Nordic country, in particular when it comes to
immigrants with a different ethnic background (which is more common
among refugees than among other types of immigrants in general). All
four Nordic countries are experiencing an increase in immigration and
the population is becoming more diverse. This ethnic fractionalization is
most and least manifest in Sweden and Finland, respectively.
10.5 Concluding remarks
What the Nordic model, with its extensive redistribution of economic
resources between citizens, will look like in the future depends on the
preferences of the voters in the Nordic countries. Historically, they have
demanded more redistribution than voters in almost all other parts of
the world. Preferences for redistribution are not fixed however but
change over time.
The academic quest for understanding how individual demand for
redistribution is shaped has provided ample insights. We know that
more inequality in itself generally leads to a higher demand for redistribution, that women want more redistribution than men, that people
with higher income, more education and higher cognitive ability tend to
Nordic Economic Policy Review
demand less redistribution and that a person who is very risk-averse is
generally going to want more redistribution than a person who is more
willing to take on risks.
We also know that the heterogeneity of a country’s population matter for how much redistributions its citizens want to see. More heterogeneous groups have been shown to be less generous and cooperative
in laboratory settings – also when the heterogeneity is artificially created. This holds empirically also when considering political preferences
within countries. More racial, ethnic and literacy fractionalization, for
example due to historical factors and to immigration, is generally found
to be negatively associated with demand for redistribution, and there
are reasons to believe that this relation is causal.
The Nordic countries are, to varying degrees, experiencing a rise in
ethnic heterogeneity. This may lead the Nordic model to change in the
direction of less redistribution, simply because citizens are likely to demand less of it. The underlying mechanism is not clear, but it is probable
that feelings of altruism, trust and reciprocity are involved. It is also likely
that especially the part of the native population with a higher income regard it as more unlikely that they will be on the receiving end of the redistributive system and hence demand less economic redistribution.
A striking example (which in itself is, of course, only illustrative and
does not provide a proof in any way) is the fact that at the same time as
Denmark has employed restrictive immigration policies it has overtaken Sweden as the Nordic country with the highest taxes. Sweden, on the
other hand, which has a very generous immigration policy, has witnessed a more pronounced trend towards lower taxes and a less extensive welfare state.
Needless to say, there are also other trends that are impacting demand for redistribution in the Nordic countries. For example, income inequality in general is increasing which in itself may lead to higher demand for redistribution and thus a mitigation of the effect discussed
above. Also, as immigrants become eligible to vote, it is unclear in which
direction this will impact demand for redistribution. What the net effect
will be is hence impossible to say. Given the rise in anti-immigration sentiment in the Nordic countries, we can however conclude that it is a curNordic Economic Policy Review
rently an extremely salient topic; the likelihood that it has a substantial
impact on preferences for redistribution is therefore likely to be high.
We may not like that more heterogeneity reduces the willingness to
redistribute – but this is what is observed empirically. Some would like
to see a situation where substantial heterogeneity is combined with extensive redistribution, and there are of course many individuals who
hold those preferences. But research indicates that the majority of people behave differently. And hence the question that Alesina and Glaeser
(2004, p. 181) is more topical than ever: “We shall see whether the generous [Nordic] welfare state can really survive in a heterogeneous society.” If current trends continue, it is, in light of the research presented
here, unlikely that the exceptionally high levels of redistribution that
have historically been demanded in the Nordic countries will persist.
To what extent these changes in demand for redistribution are perceived as problematic depend, of course, on individual views on the optimal size of the welfare state. For a person who is generally positive to
immigration and at the same time wishes to see a smaller welfare state,
the negative relation between ethnic heterogeneity and demand for redistribution does not necessarily pose a problem.
On the other hand, for a person who is likewise positive to immigration but wants the welfare state to remain large, there is a challenge.
Importantly and as discussed above, research tells us that the negative
correlation between immigration and redistributive preferences is especially pronounced when ethnic minorities are overrepresented
among those who are dependent on the welfare state. This implies that
the integration of immigrants is crucial as it, as is making it possible for
them to be successful in the labor market. Doing this successfully may
weaken the negative link between ethnic heterogeneity and demand for
redistribution, and could hence make it easier to successfully combine
generous immigration policies with a continued high demand for an extensive welfare state. 7
7 Unfortunately the experiences of immigrants on the Nordic labor markets is not particularly positive,
something which is further discussed by Bratsberg and Røed (2015) in this volume. They present facts
regarding labor market participation and welfare dependency for immigrants to the Nordic countries, and
discuss policy changes that could make the welfare state more robust to handling extensive immigration.
Nordic Economic Policy Review
Alesina, A., & Giuliano, P. (2010). Preferences for redistribution. In Jess Benhabib, Matthew O. Jackson and Alberto Bisin (eds.), Handbook of Social Economics, 1A:93–131.
Alesina, A., & La Ferrara, E. (2002). Who trusts others? Journal of Public Economics,
85(2): 207–234.
Alesina, A., Baqir, R, & Hoxby, C. (2004). Political Jurisdictions in Heterogeneous
Communities. Journal of Political Economy, 112(2): 348–396.
Alesina, A., Baqir, R., & Easterly, W. (1999). Public Goods and Ethnic Divisions. Quarterly Journal of Economics, 114(4): 1243–84.
Alesina, A., E. Glaeser, & B. Sacerdote (2001). Why Doesn’t the United States Have a
European-Style Welfare State? Brookings Papers Econ. Activity, no. 2: 187–254.
Alesina, A., Tella, R. D., & MacCulloch, R. (2004). Inequality and happiness: Are Europeans and Americans different? Journal of Public Economics, 88(9–10): 2009–
Alesina. A., & Glaeser, E. (2004). Fighting Poverty in the US and Europe – A World of
Difference. Oxford University Press, Oxford, UK.
Alesina, A., & Zhurayskaya, E. (2011). Segregation and the Quality of Government in
a Cross Section of Countries. American Economic Review, 101(5): 1872–1911.
Atkinson, A.B., & Morelli, S. (2014). The Chartbook of Economic Inequality. Accessed on 11 September 2015.
Benabou, R., & Ok, E.A. (2001). Social Mobility and the demand for redistribution.
Quarterly Journal of Economics, 116(2): 447–487.
Benabou, R., & Tirole, J. (2006). Belief in a just world and redistributive politics.
Quarterly Journal of Economics, 121(2): 699–746.
Bernard, H., Fehr, E., & Fischbacher, U. (2006). Group Affiliation and Altruistic Norm
Enforcement. American Economic Review, 96(2): 217–221.
Billig, M., & Tajfel, H. (1973). Social Categorization and Similarity in Intergroup Behavior. Journal of Social Psychology, 3(1): 27–52.
Bolton, G. E., & Ockenfels, A. (2000). ERC: A theory of equity, reciprocity, and competition. American Economic Review, 90(1): 166–193.
Nordic Economic Policy Review
Brady, D., & Finnigan, R. (2014). Does Immigration Undermine Public Support for
Social Policy? American Sociological Review, 79(1): 17–42.
Brandal, N., Bratberg, Ø., & Thorsen, D.E. (2013). The Nordic Model of Social Democracy. Palgrave MacMillan, New York, NY, USA.
Bratsberg, B. & Røed, K. (2015). The Nordic Welfare Model in an Open European
Labor Market. Nordic Economic Policy Review: this volume.
Chen, Y., & Xin Li, S. (2009). Group Identity and Social Preferences. American Economic Review, 99(1): 431–457.
Cruces, G., Perez-Truglia, R., & Tetaz, M. (2013). Biased perceptions of income distribution and preferences for redistribution: Evidence from a survey experiment. Journal of
Public Economics, 98: 100–112.
Dahlberg, M., Edmark, K., & Lundqvist, H. (2012). Ethnic Diversity and Preferences
for Redistribution. Journal of Political Economy, 120(1): 41–76.
Dinesen, P. T., & Sønderskov, K. M. (2015). Ethnic Diversity and Social Trust: Evidence from the Micro-Context. American Sociological Review, 80(3): 550–573.
Durante, R., Putterman, L., & van der Weele, J. (2014). Preferences for redistribution
and perception of fairness: An experimental study. Journal of the European Economic Association, 12(4): 1059–1086.
Elinder, M., Jordahl, H., & Poutvaara, P. (2015). Promises, policies and pocketbook
voting. European Economic Review, 75(April): 177–194.
Fehr, E., & Schmidt, K. M. (1999). A theory of fairness, competition, and cooperation.
Quarterly Journal of Economics, 114(3): 817–868.
Finseraas, H. (2012). Poverty, ethnic minorities among the poor, and preferences for
redistribution in European regions. Journal of European Social Policy, 22(2): 164–
Fong, C. (2001). Social preferences, self-interest, and the demand for redistribution.
Journal of Public Economics, 82(2): 225–246.
Fong, C., & Luttmer, E. (2011). Do Race and Fairness Matter in Generosity? Evidence
from a Nationally Representative Charity Experiment. Journal of Public Economics,
95(5–6): 327–394.
Gaertner, M., Mollerstrom, J., & Seim, D. (2015) Risk and redistribution. Mimeo,
Stockholm University.
Gerdes, C., & Wadensjö, E (2008). The Impact of Immigration on Election Outcomes in
Danish Municipalities. IZA Working Paper 3586.
Glaeser, E. Laibson, D., Sheinkman, J., Soutter, C. (2000). Measuring Trust. Quarterly Journal of Economics, 115(3): 811–846.
Nordic Economic Policy Review
Goette, L., Huffman, D., & Meier, S. (2006). The Impact of Group Membership on Cooperation and Norm Enforcement: Evidence Using Random Assignment to Real
Social Groups. American Economic Review, 96(2): 212–216.
Gustavsson, M., & Jordahl, H. (2008). Inequality and trust in Sweden: Some inequalities are more harmful than others. Journal of Public Economics, 92: 348–365.
Harmon, N. (2015). Immigration, Ethnic Diversity and Political Outcomes: Evidence
from Denmark. Mimeo.
Harsanyi, J.C. (1953). Cardinal utility in welfare economics and in the theory of risktaking. Journal of Political Economy, 61: 434–435.
Hilson, M. (2008). The Nordic Model. Reaktion Books, London, UK.
Knack, S., & Zak, P.J. (2002). Building trust: public policy, interpersonal trust, and
economic development. Supreme Court Economic Review, 10: 91–107.
Kouvo, A., & Lockmer, C. (2012). Imagine all the Neighbours: Perceived Neighbourhood Ethnicity, Interethnic Friendship Ties and Perceived Ethnic Threat in Four
Nordic Countries. Urban Studies, 50(16): 3305–3322.
Kymlicka, W., & Banting, K. (2006). Immigration, multiculturalism, and the welfare
state. Ethics & International Affairs, 20(3): 281–304.
Lindqvist, E., & Östling, R. (2010). Political Polarization and the Size of Government.
American Political Science Review, 104(3): 543–565.
Locksley, A., Ortiz, V., & Hepburn, C. (1980). Social Categorization and Discriminatory Behavior: Extinguishing the Minimal Intergroup Discrimination Effect. Journal
of Personality and Social Psychology, 39(5): 773–783.
Logue, J. (1979). The Welfare State: Victim of its Success. Daedalus, 108(4): 69–78.
Luttmer, E. (2001). Group Loyalty and the Taste for Redistribution. Journal of Political Economy, 109: 500–528.
MacDermott, R. (2009). Psychological Approaches to Identity: Definitions, Measurement and Experimentation. In Abdelal, R., Y. Herrera, A. Johnson, & R. McDermott (eds.), Measuring Identity: A Guide for Social Scientists,. Cambridge, UK: Cambridge University Press.
Margalit, Y. (2013). Explaining social policy preferences: Evidence from the great
recession. American Political Science Review, 107(01): 80–103.
Meltzer, A. H., & Richard, S. F. (1981) A rational theory of the size of government.
Journal of Political Economy, 89(5):914–927.
Nordic Economic Policy Review
Mollerstrom, J. (2015). Favoritism reduces cooperation. Mimeo, George Mason University.
Mollerstrom, J., & Seim, D. (2014). Cognitive ability and the demand for redistribution. PloS one, 9(10): e109955.
Morelli, S., Smeeding, T., & Thompson, J. (2014). Post-1970 Trends in Within-Country
Inequality and Poverty: Rich and Middle Income Countries. IRP Discussion Paper,
No. 1419-14.
Mounir, K., Mollerstrom, J., & Seim, D. (2014). Richer (and Holier) than Thou? The
Effect of Relative Income Improvements on Demand for Redistribution. IFN Working
Paper No. 1042.
Nekby, L. & Pettersson-Lidbom, P. (forthcoming). Revisiting the Relationship between Ethnic Diversity and Preferences for Redistribution. Scandinavian Journal of
OCED (2015). International Migration Outlook 2015. OECD Publishing, Paris, France.
Pelzman, S. (1985). An economic interpretation of the history of congressional voting in the twentieth century. American Economic Review, 75(4): 656–678.
Piketty, T. (1995). Social mobility and redistributive politics. The Quarterly Journal
of Economics, 110(3): 551–584.
Popenoe, D. (1994). Scandinavian Welfare. Society, 31(6): 78–81.
Powdthavee, N., & Oswald, A. J. (2014). Does money make people right-wing and inegalitarian? A longitudinal study of lottery winners. Warwick University Economics
Working Paper 1039.
Rainer, H., & Siedler, T. (2008). Subjective income and employment expectations
and preferences for redistribution. Economics Letters, 99: 449–453.
Roberts, K. (1977). Voting over Income Tax Schedules. Journal of Public Economics,
8(3): 329–340.
Rodrik, D. (1998). Why do More Open Economies have Bigger Governments? Journal
of Political Economy, 106(5): 997–1032.
Romer, T. (1975). Individual welfare, majority voting, and the properties of a linear
income tax. Journal of Public Economics, 4(2): 163–185.
Sachs, J.D. (2006). The Social Welfare State, Beyond Ideology. Scientific American,
16 November.
Sanandaji, N. (2015). Scandinavian Unexceptionalism. The Institute of Economic Affairs, London, UK.
Sears, D. O., & Funk, C. L. (1990). The limited effect of economic self-interest on the
political attitudes of the mass public. Journal of Behavioral Economics, 19(3): 247–
Shayo, M. (2009). A Model of Social Identity with an Application to Political Economy: Nation, Class, and Redistribution. American Political Science Review, 103: 147–
Nordic Economic Policy Review
Stichnoth, H., & Van der Straeten, K. (2013). Ethnic Diversity, Public Spending, and
Individual Support for the Welfare State: A Review of the Empirical Literature.
Journal of Economic Surveys, 27(2): 364–389.
Tajfel H., Billig M., Bundy R., & Flament C. (1971). Social categorization and intergroup
behaviour. Eur J Soc Psychol 1:149–178.
Tajfel, H. (1982). Social Psychology of Intergroup Relations. Annual Review of Psychology, 33: 1–39.
Tajfel, H., & Turner, J. (1986). The Social Identity Theory of Intergroup Behavior. In (eds.
Stephen Worchel, and William Austin) The Psychology of Intergroup Relations, 7–24.
Chicago: Nelson-Hall.
Wang, C., Caminada, K., & Goudswaard, K. (2014). Income Redistribution in 20
Countries over Time. International Journal of Social Welfare, 23: 262–275.
World Value Survey (2015). Accessed on
15 September 2015.
Yousfi, S. (2010). The development of immigration. In: R. Andersson, H. Dhalmann, E.
Holmqvist, T. M. Kauppinen et al. (eds.) Immigration, Housing and Segregation in the
Nordic Welfare States, pp. 271–274. University of Helsinki Press, Helsinki, Finland.
Zak, P.J., & Knack, S. (2001). Trust and growth. Economic Journal, 111(470): 291–
Nordic Economic Policy Review
11. The Social Upper Class under
Social Democracy
Kalle Moene, Department of Economics, University of Oslo 1
In Capital in the 21st Century Thomas Piketty explores the history of the
distribution of income and wealth. Among other things he demonstrates
how top income shares have soured recently. Although top income
shares are somewhat lower in the Nordic countries, the recent
developments follow a similar pattern as elsewhere (Roine and
Waldenström, 2015). So also in countries characterized as social
democratic the rich obtain a higher share of a growing pie.
These trends in the distribution of income and wealth may have
political consequences. Piketty rightly insists “that economic and
political changes are inextricably intertwined and must be studied
together”, and those who have a lot of wealth, he continues, “never fail
to defend their interests” politically (p. 577). The concentration of
income and wealth in the hands of the rich, may therefore give them
more political clout as well.
Yet without discussion Piketty seems to assert that the distribution
of political power is linked directly to the distribution of top income
shares. I agree that the prosperous rich are the political powerful in
most countries, but I am less confident that the share of income to the
richest x percent is a good measure of the political power of the upper
class. In this essay I therefore emphasize not only how rich the rich are
1 I’m grateful for discussion with Rolf Aaberge, Halvor Mehlum, Debraj Ray and Gaute Torsvik. I have benefited from useful comments by Torben Andersen and efficient assistance by Kristian Harald Myklatun. This
paper is part of the research activities at ESOP, a research center funded as a center of excellence by the
Norwegian Research Council.
relative to the rest of society, but also how many rich people there are
relative to the size of the population. Accounting for the number of rich
– for instance the share of people with wealth above y – gives us
another picture of the wealth concentration in the Nordic countries
than what is normally associated with social democracy. It gives us a
picture of a larger upper class, at least in Sweden and Norway,
compared to other countries.
The upper class is the class composed of the wealthiest members of
society who also wield the greatest political power, Wikipedia rightly
points out. The social democratic upper class is not a class of the
wealthiest social democrats, but the class of the richest people created
by social democracy. Many of its members are critical of welfare
spending and small income differentials even though these social
democratic institutions may have laid the foundation for their affluence
and for the size of the upper class.
11.1 Prosperity, power and poverty
All too many of us have a great respect for people with money. The
social habit is not new. Writing in 1759, Adam Smith was worried about
“the corruption of our moral sentiments, which is occasioned by this
disposition to admire the rich and the great, and to despise or neglect
persons of poor and mean condition” (Moral Sentiments, p. 84). He also
pointed to social and political consequences. In my reading, the respect
for money includes the respect for the wealthy, and therefore also for
the number of wealthy people.
The number of wealthy people is even more important in
democratic societies than in the autocratic ones that Adam Smith
considered. The division into groups and classes is still important, and
so is how much resources each of them has. Yet a person who owns
everything, has less power than a group that owns everything.
The wealthy Norwegian family Løvenskiold is a case in point. The
family owns the huge forest and recreational area, Nordmarka outside
Oslo, a local public good. Obviously the family could make a lot of
money by developing the area commercially for housing, shopping
Nordic Economic Policy Review
centers and businesses. It is equally clear, however, that if the family
tried, it would most likely never get the permission to develop the area.
Democratic institutions can easily control one family. But if the same
area had thousands of wealthy owners, they would have a much easier
time getting permission to develop the area. Numbers count under
democratic rule.
The influence of the rich is well illustrated in a study by Martin
Gilens and Benjamin Page (2014). They look at the connection between
public opinion polls and political decisions in the US, based on 1779
cases or propositions in the period 1981 to 2002. The study shows that
whether ordinary people support or oppose a proposition has little
influence for whether the proposition is adopted. If the richest tenth of
the population support the proposition, however, it has a higher
probability of being adopted.
How many wealthy people that support the proposition plays a
significant role. A proposition supported by one in five wealthy people,
has only 18% chance of being adopted, while a proposition supported by
four out of five wealthy people, has a 45% chance of being adopted. These
correlations are worth noting even though we don’t quite know whether
the propositions were adopted because the rich supported them, or not.
Interpreted with the same caveat, the paper also shows that the rich have
an effective veto right over politics. If they are against a proposition, the
chances that the proposition is adopted are very small.
All in all the study reminds us that “votes count, but resources
decide”, as Stein Rokkan said in 1966. The influence of the rich seems to
increase when they are many, and when there is a large gap between
the rich and the rest (see for example Bartels, 2008, Gilens, 2005 and
2009, Peters and Ensik, 2014).
The interests of a high number of wealthy people count both
because they are wealthy, and because they are many. This is important
for how to measure their potential power. It is not only their share of
the national income that matters. Measuring richness, I suggest, is
similar to measuring poverty. In the low end of the income distribution
we could also look at income shares, for instance the share of the
national income that goes to the z poorest percent of society. Such a
measure of poverty, however, tells us nothing about how many people
Nordic Economic Policy Review
who live in poverty – and thus how long the hunger march can become.
To get an approximation of the magnitude of the problem, it is better to
look at the so-called head count measure of poverty. This tells us the
number of people with a lower income than a certain threshold, the
poverty line, and who therefore are not able to achieve the socially
decided minimum level of material well being – the lowest level
required to appear in public without shame, as Adam Smith said.
In the same way, we should also look at the number of extremely
wealthy people, the number of people with a wealth beyond a certain
threshold which we could call the richness line – perhaps the highest
level tolerated to appear in public without shame. To measure wealth as
the relative number of wealthy people in the population – the head
count measure of richness – should be as natural as measuring the
relative number of poor people. In practice, of course, it is not obvious
where the richness line should be set, just as it is not obvious where the
poverty line should be. There is no objective distinguishing line, neither
for extreme poverty nor for extreme richness.
11.1.1 Dollar billionaires
We can use the Forbes list of billionaires as an illustration. It uses an
implicit richness line of wealth equal to USD 1 billion, which is the
threshold wealth to be included on the list. By using the Forbes list of
the world’s billionaires, we can examine which countries have the
highest number of billionaires, relative to the population. Is US ahead of
the Nordic countries? The answer is not obvious. We know that the
wealthiest persons in the US are wealthier than the wealthiest persons
in the Nordic countries, For example, the 0.1% richest in the US get 11%
of the national income, while the 0.1% richest in Norway get 2.5% of the
national income. So according to these numbers social democracy can
be viewed as effective in abolishing extreme wealth.
Nordic Economic Policy Review
Wealth concentration represented by the head count measure of
richness, however, provides an entirely different picture. While the US
has 1.7 billionaires per million inhabitants, the corresponding numbers
for the Nordic countries are:
Denmark 0.9.
Finland 0.9.
Norway 2.0.
Sweden 2.4.
Iceland 3.1.
In comparison the head count measure of billionaires in other European
countries is in the range of 0.5 to 1.3 per million inhabitants. For
instance, Spain has 0.4, Italy 0.6, France, 0.7, UK 0.8 and Germany 1.2.
So, compared to the rest of Europe there are more billionaires relative
to the population in the Nordic countries.
As seen, Norway, Sweden and Iceland have even more billionaires
relative to the population than United States. Measured by the head
count measure of richness, the social democratic model in Norway and
Sweden creates more wealthy people than the American model.
Studying the Forbes list, we find that only countries that are pure tax
havens, or would like to become tax havens – Iceland, Singapore,
Switzerland Cyprus, Hong Kong, St. Kitts and Nevis and Monaco – have a
higher share of billionaires than Norway and Sweden.
In sum, all this indicates that the social democratic upper class can
be larger, relatively speaking, than the upper class in most countries,
and in particular it can be larger than the American upper class. But
Forbes operates with a very high richness line. The question is whether
we find the same pattern when we use a lower richness line of wealth,
and when we account for differences in national income per capita.
Being in the start of a research project on these issues, my plan is to
explore several data sets to compare countries and the development
over time. At this stage I can offer preliminary illustrations. In addition
to the observations above based on the Forbes list, I can add more
Nordic Economic Policy Review
detailed observations based on another data source. I focus on the
comparison between Norway and the US.
11.1.2 An example: the super rich in Norway and in the US
In this illustration I use numbers from The Wealth Report 2014,
published by the international consultancy Knight Frank. Statistics on
wealth are often incomplete, and usually build on rough
approximations. Like Forbes list, the Wealth Report 2014 is no
exception. It is also difficult to check the quality of the numbers, but the
report has a good reputation. 2
The Wealth Report provides numbers on the wealth distribution for
a many countries, including the US and Norway, concentrating on
wealthy individuals. In the discussion below I use two richness lines:
the wealth levels of USD 30 million and USD 100 million. Those with
higher wealth than 30 million are called rich, those with wealth higher
than 100 million are called super rich. The two richness lines are not
corrected for purchasing power. A purchasing power adjustment would
only show how much the wealth is worth when it is exclusively used to
purchase goods and services in the country of its owner.
Per million inhabitants in 2013 we have:
124.6 rich and 36.5 super rich individuals in the US.
472.8 rich and 63.6 super rich individuals in Norway.
In other words, Norway has almost four times as many rich individuals
per capita compared to the US, and almost two times as many super rich.
When we measure richness as the relative number of rich people in
the population, it does not look like social democracy limits the size of
the upper class. On the contrary, a social democratic system seems to
increase the recruitment to this class. Relative to the population, social
2 I have started checking the numbers, comparing them to information from other sources. There is a clear
spread in the reported numbers (surveys) that should make us less confident. Yet, the basic pattern that I
focus on in this essay seems to prevail in the data I’ve checked so far.
Nordic Economic Policy Review
democracy creates more rich and super rich than the American model
does. But are these differences only a reflection of the fact that Norway
is richer per capita compared to the US? And, should we not therefore
correct for this fact?
It is reasonable to assume that richer countries have more rich and
super rich members than not quite so rich countries. However, we do
not correct the head count measures of poverty because some countries
have a higher level of income than others. It is far from obvious that we
should do this when measuring and evaluating the head count meausre
of richness.
Yet, it is still interesting to know whether the US would have the
same relative number of rich and super rich if the US had the same
income per capita as Norway. Does the fact that Norway is richer than
the US explain why Norway has more multimillionaires than the US?
11.1.3 If the US was as rich as Norway
Norway has an income per capita of USD 65,461, while the US has
USD 53,042, such that Norway is 23.5% richer than the US. We will now
try to estimate how many rich and super rich individuals would have
been lifted above the richness line if the US was 23.5% richer than what
the country is today. Since the wealth distribution has a thin tail for the
very wealthy, the relative increase in the number of multimillionaires, is
usually disproportionately large compared to the increase in the income
per capita. A ten percent increase in income would give a larger than
ten percent increase in the number of multimillionaires.
By how much does the share of rich and super rich individuals
increase as income goes up? The answer depends on three things:
Firstly, we have to know how large the income growth will be for
the Americans that are almost rich, and almost super rich, as the per
capita income in the US increases by 23.5%. I emphasize three
alternatives where the income of the rich and the super rich increase by
a percentage which is a) equal to the growth of national income, b)
twice as high as the national income growth, c) triple the size of the
national income growth.
Nordic Economic Policy Review
Secondly, we have to know how much higher wealth this increase
in income gives us. As an approximation I assume that the wealth is
proportional to income, such that wealth increases at the same rate
as income.
Thirdly, we have to know how fat the tail of the wealth distribution
is, as this determines how many almost rich and almost super rich that
are lifted above the richness line. Here we can utilize that the Paretodistribution can give a good approximation of the upper part of the
wealth distribution. The Pareto-distribution is determined by two
parameters, one that controls how fat the tail is, so we can find out how
many almost rich and almost super rich that pass the richness line when
their incomes increase. In the simple calibration I perform I use the
shares of the population with wealth of at least USD 30 million, and with
a wealth of at least USD 100 million (see the appendix for details).
Following this recipe, the results should be compared to the case of
Norway with 472.8 rich, and 63.6 super rich per million inhabitants. I
get the following hypothetical upper class in the US:
If the income growth of all wealthy individuals was equal to the
growth of national income, the US would have: 154.5 rich, and 45.2
super rich per million inhabitants.
If the income growth of all wealthy individuals was twice the
growth in national income, the US would have: 184.5 rich, and 54
super rich per million inhabitants.
If the income growth of all wealthy individuals was three times the
growth of national income, the US would have: 215.3 rich, and 63.1
super rich per million inhabitants.
Even if the income of the rich grows at three times the rate of the rest, the
US would not have a higher share of rich or super rich than Norway. So if
we choose US as the bench-mark for the link between income and wealth,
we would have to conclude that Norway has a higher share of rich and
super rich individuals than the income per capita predicts. In comparison
with the US, Norwegian social democracy creates a higher share of people
Nordic Economic Policy Review
with wealth around NOK 700–800 million, and in particular a higher
share of people with wealths around NOK 200–300 million.
I have also tried other methods, including calibrating the
parameters of the Pareto-distribution to the inequality of the entire
wealth distribution expressed by the Gini coefficient – and I obtain
similar results (see the appendix). The Gini of the wealth distribution in
the US is slightly above 0.8. The large wealth inequality that such a high
Gini coefficient indicates, means that there is a low number of rich and
super rich individuals that would be lifted over the richness line when
their incomes increase. Norway has a lower Gini coefficient around 0.6
for the wealth distribution. This indicates that Norway has a higher
density of rich and super rich individuals who would be lifted upwards
when their incomes grow.
The higher shares of rich and super rich in Norway does not mean
that Norway has more concentration of wealth than the US. Both in the
US and in Norway the population share with wealth over 30 million USD
possesses a considerable share of the entire wealth in the nation.
Norway has more wealth equality among the rich than the US, such that
Norway has a higher share of rich and super rich individuals in the
population than the US. In other words, the US has fewer super rich
than Norway, but they own a greater share of the national wealth.
To be clear, if one person owned the entire wealth of the country,
the Gini coefficient would be equal to 1. The head count measure of
richness would then be approximately equal to zero, and increasing the
income would not raise the share of rich individuals. Of course, the
concentration of wealth is not that high in the US. But the example
illustrates that the US can have a so high concentration of wealth that
income growth only lifts rather few people above the richness line.
11.2 Why so many wealthy?
Is the wealth of the social democratic upper class old or new, inherited
or recently accumulated? A check of the family background of the
members of the richest groups in Norway (done together with the
newspaper Dagens Næringsliv, indicates that the upper class to a large
Nordic Economic Policy Review
extent, but not entirely, is produced under social democracy. Going
through the background of the 100 richest persons according to the
Norwegian tax register in 2014, reveals that less than one third of them
had inherited their wealth, while more than two thirds of them had built
up their wealth under social democratic rule.
If this is true more generally, it remains to be explained how social
democracy produces so many wealthy persons. In my view it follows
naturally from the social democratic development strategy of
combining social protection with capitalist dynamics. Contrary to what
most people believe, the strategy does not represent a compromise
between socialist and capitalist ideals that leads to neither. Rather the
two ideals are complementary in the sense that we get more out of each
by combining them. 3 This complementarity can also give more
multimillionaires and a larger upper class once we incorporate how
wage equality can raise the sum of asset values to capital, and how
social mobility can expand the numbers of capital owners.
Over time, new technologies and new organizations replace the old
and outdated. The goal of the Scandinavian union movement has for
long been to speed up this process of creative destruction, as Joseph
Schumpeter called it, under a policy of maintaining full employment.
The strategy could also be seen as an attempt to raise long term real
wages by wage restraint in the short run.
Unions have followed a wage policy that raises the profitability of
investments and modernization by holding back the highest wages in
the name of solidarity. The goal has also been to make the export sector
more competitive. Both goals were reached by taking the wage
determination out of market competition and placing it in a system of
collective decision making. The result was a highly compressed exportlead wage structure, eliminating employment rents in high productivity
jobs and raising the pay in low productivity jobs. Over time the new
wage structure has also altered the composition of high and low
productivity jobs.
3 Together with co-authors I have for long argued this. See for instance Moene and Wallerstein (1993) and
Barth, Wallerstein and Moene (2003), and Barth, Moene and Willumsen (2014).
Nordic Economic Policy Review
The isolated effect of wage restraint in high-skilled jobs is to raise
profits and thus to induce more investments and in turn higher demand
for all types of labor. The unions could therefore raise the lowest wages
without increasing unemployment. Aiming for small wage differentials
and full employment lead to more equal wages. It also lead to higher
profits, increasing the asset values of capital owners. The social
democratic growth strategy therefore raised the total wealth of the
upper class.
The high number of wealthy people is in part explained by this
expansion of wealth and in part by the access to free higher education for
all and the corresponding social mobility that enabled more people to
take advantage of the possibilities to become rich. Individuals without
inherited wealth simply has a higher probability of succeeding
economically in the Nordic countries than in the US. The Nordic countries
have more equal opportunities. For instance, the chances that children of
manufacturing workers in the 1960–70s end up in a higher class of
income than their parents are considerably higher than in the US. The
probability that a son with a father in the bottom fifth of the income
distribution actually ends up in the top fifth of the income distribution, is
8% in the US, and more than 12% in Norway (Jänti et al., 2006).
Some might argue that the mobility in the Nordic countries is higher
than in other countries because income equality is more pronounced.
Thus, the differences between high and low income is relatively small,
and the distance between classes is low, which enables more people to
move upwards socially and financially. This is not a counter argument,
however. It is part of the explanation for what we observe – that the
upper class is largest where the differences among workers are smallest.
11.3 Policy implications or implied policy?
Political actors may be worried or enthusiastic about the size of the
upper class. Both groups may nevertheless consider to adjust wealth
taxes, inheritance taxes and other re-distributive means to achieve their
political goals – if they can. Policy implications, however, are never as
straight forward as most economists assert. Economists easily
Nordic Economic Policy Review
exaggerate the independence of policy – and pay too little attention to
implied policy by endogenous forces in society.
In our case one relevant issue is what it really means to ask for the
policy implications if the distribution of wealth to some extent
determines policies – which again may affect the ability of the rich to
accumulate wealth. When money gives power, a higher gap between the
wealthy and the rest may imply more unequal power and influence over
policy. The reinforcing development can be strong: the political
influence of the rich may affect the economy; the economic changes may
make it easier to become rich.
If this is right, more inequality might lure us into a trap where those
who should be the object of regulation, becomes those who in fact
regulate – and where the new multi-millionaires influence the economic
policies in their own favor. Wealth may bring power, and power may
bring more wealth to the wealthy – such that the rich grow both in
number and in wealth.
To illustrate the possible consequences of having a big upper class,
let us return to the paper by Gilens and Pages and take their numbers
literally: The fraction of wealthy people supporting or opposing a
political proposition, is imperative to whether it passes or not. The
numbers suggest that a quadrupling of the support by the richest 10%
of Americans, increases the chances of the proposition passing, by a
factor of 2.5. If we use this factor on the numbers we have calculated
for the relative size of the upper classes in the US and Norway, we find
that the rich in Norway potentially has 2.4 times the influence of the
rich in the US.
Of course this approximation is too simpleminded. There are also
opposing forces as can be seen by the fact that almost equally rich
countries have very different income distributions – one hundredth of
the population gets almost 25% of the income in the US, but less than
8% in Norway (Aaberge and Atkinson, 2011). One reason for the
difference may be that labor unions represent a countervailing power to
the political and economic influence that the rich otherwise would have.
The logic of this countervailing power is simple: When workers are
poorly organized, it is much easier to become exploited by a big capitalist
Nordic Economic Policy Review
than by many smaller ones. When workers are well organized, in contrast,
it is easier for them to tame a few big capitalists than many small.
Some central union leaders, at least in Norway and Sweden, may
think of the capitalists as their saving machines, and the large retained
profits as the savings of the union members. For this reason, they have
not found it beneficial to squeeze the capitalists too hard. As mentioned
they may have thought that some wage restraint of the best paid
workers would benefit the great majority of union members via the
investments that the retained profits generated.
In this context capitalist consumption is a cost. Few and big savings
machines may thus also be viewed as more cost efficient than many and
small savings machines. Union leaders might also for that reason have
preferred to have few and large capitalists and thus more concentrated
The comparison between the US and Norway shows that countries
do not always get the upper class that fits the best. Norway, with its
strong union associations has a large upper class with more dispersed
wealth. The US, in contrast, has a more concentrated and top-heavy
upper class, but almost no unions in the private sector. In the US, where
it is easier to buy influence through gifts to political campaigns, political
advertising and otherwise, the concentrated wealths are a threat to
democracy. The inequality in Norway could also become a similar
threat, especially when unions are in decline.
The richest capitalist in Sweden has long been the Wallenberg
family. At times, the family has been seen as a social democratic
capitalist light, acting as the saving machine for the Swedish union
members. The most famous capitalist in the US, in contrast, John D.
Rockefeller, was perceived more as a threat to democracy than as a
saving machine of the country. Even when he started to donate parts of
his fortune, it could be seen as an unreasonable concentration of power
over what should be financed.
Perhaps the clearest policy implications for those who are
concerned with the low level inequality of the Nordic model, is to
emphasis the importance of a well organized labor market – both on the
workers’ side and on the employers’ side. A strong union movement is
decisive to reduce the power that a large upper class otherwise would
Nordic Economic Policy Review
imply. A strong employers’ association is important to voice long term
producer interests as opposed to short term distributional share-owner
interests. A strong employers’ association is also important as a threat
against sheltered unions that otherwise might prefer to operate alone
and not coordinate with the unions in traded goods industries.
The irony is, perhaps, that the structure of organized interests that
can prevent the bad political consequences of a large upper class also
may lay the foundation for its expansion. Similarly, while the goal of
social democracy as a political movement used to be to work for
socialist egalitarian ideals, the simple empirics in this essay indicate
that the social democratic combination of worker security and capitalist
dynamics may generate not only high growth and small wage
differentials, but also a large upper class. Having a larger upper class
than the US teach us that it cannot be particularly difficult to become
rich under social democracy in Norway and Sweden. A large upper class
may also be a sign of financial success.
Yet, it may undermine democratic ideals. In practice, of course,
democracy always entails a combination of one person one vote and
one dollar one vote. The weights on each depend on the social
organization and on the wealth distribution. When many individuals
become wealthy, the political system may put less weight on the
average vote of citizens. In this way social democracy in Norway and
Sweden can be a victim of its own success.
Aaberge, R. & Anthony B. Atkinson (2010). Top Incomes in Norway. In A.B. Atkinson
and T. Piketty (eds.), Top Incomes A Global Perspective, Oxford: Oxford University
Press, 448–482.
Bartels L.M (2008). Unequal Democracy: the political Economy of the New Gilded Age.
Princeton: Princeton University Press.
Barth E., Moene, K., & Wallerstein, M. (2003). Likhet under press: utfordringer for den
skandinaviske fordelingsmodellen. Oslo: Gyldendal Akademisk.
Forbes list. The World’s Billionaires.
15 November 2015).
Nordic Economic Policy Review
Gilens, M. (2005). Inequality and Democratic Responsiveness. Public Opinion
Quarterly, 69(5):778–896.
Gilens, M. (2009). Preference Gaps and Inequality in Representation. PS: Political Science
and Politics, 42(2): 335–341.
Gilens, M. & Benjamin, I. P. (2014). Testing Theories of American Politics: Elites,
Interest Groups, and Average Citizens. Perspectives on Politics, 12(03), 564–581.
Jänti, M., Bratsberg, B., Røed, K., Raaum, O., Naylor, R., Österbacka, E., Björklund, A.,
& Eriksson, T. (2006). American Exceptionalism in a New Light: A Comparison of
Intergenerational Earnings Mobility in the Nordic Countries, the United Kingdom
and the United States. IZA Discussion Paper no. 1938.
Moene, K. & Wallerstein, M. (1993). What is Wrong with Social Democracy. In Pranab
Bardhan and John E. Roemer (eds.), Market Socialism. The present debate.
Cambridge, Cambridge University Press.
Barth, E., Moene, K., & Willumsen, F. (2014). The Scandinavian Model – an interpretation.
Journal of Public Economics.
Ensik S.J. & Peters, Y. (2014). Differential Responsiveness in Europe. The Effects of
Preference Difference and Electoral Participation. West European politics.
Piketty, T. (2014). Capital in the Twenty-First Century. Cambridge, MA: Harvard
University Press. Statistical series and technical appendix. capital21c.
Roine, J. & Waldenström, D. (2015). Long-Run Trends in the Distribution of Income
and Wealth. In A.B. Atkinson and F. Bourguignon, F. (eds.), Handbook of Income
Distribution, vol. 2A. Amsterdam: North-Holland.
Smith, A. (1759). The Theory of Moral Sentiments Library of Economics and Liberty.
26 November 2015.
The Wealth Report 2014. (2014). London: Knight Frank.
Nordic Economic Policy Review
To see the impact of a higher national income (in the US) on the share of
rich and super rich people we ustilize that the Pareto-distribution is a
good approximation to the wealth distribution.
We first use how the Pareto-distribution fits around the two
richness lines of 30 and 100 millions USD, but not necessary for the
entire wealth distribution. Let the share of people with wealth higher
than x millions USD (in the US) be r(x ) given by
r( x ) = Ax −α
where α > 0 and A > 0 . With the numbers we have for the US we can
then calculate
r(30) =
= 0,0000365
= 0,000125 andr(100) =
Using (1) and (2) we have two equations to “determine” A og α . This
calibration yields
A = 0,004 and α = 1.02
We also assume that wealth is proportional to income y implying that a
λ -doubling of the national income raises the income of the rich and
super rich by the factor λ .
Clearly if the income increases by a factor 1.235, people with wealth
between x/1.235 and x would be lifted up to a wealth higher than x .
If the income growth of all wealthy individuals was equal to the
growth of national income of 23.5%, r(30/1.235) = 0.0001544 rich, and
r(100/1,235) = 0,0000452 super rich.
Nordic Economic Policy Review
If the income growth of all wealthy individuals was twice the
growth in national income, i.e. 47%, r(30/1.47) = 0.0001845 rich, and
r(100/1.47) = 0.000054 super rich.
If the income growth of all wealthy individuals was three times the
growth of national income, i.e. 71%, r(30/1.71) = 0.0002153 rich, and
r(100/1.71)= 0.0000631 super rich.
Alterantively we could calibrate α in the Pareto-distribution such
that it fits to the rough estimates of the Gini coefficient for welath equal
to 0.8 in the US, assuming that the Pareto-distribution fits fits for the
entire wealth distribution.
The Gini for the wealth distribution can be written
2α − 1
If G ≈ 0.8 we have α ≈ 1.1 , (not far from our calibrated value of 1.02.
Taking expectations
Ex =
α −1
Utilizing that the average wealth is proportional to average income, i.e.
Ex = kEy , we obtain
 (α − 1) kEy 
r( x , Ex ) = 
x 
 α
We see that a λ -doubling of Ey yields
r( x , λEx ) = λα r( x , Ex )
Nordic Economic Policy Review
The calculation shows that if the US increased its income by 23.5%: the
US would have had 163.8 rich and 47.8 super rich per million
Calibrating the Norwegian wealth distribution in the same manner,
using NBER’s estimate of G = 0.633 for Norway, we get α = 1.29 . an
increase in the mean income in Norway with 23.5% would raise the
share of rich and super rich by a factor (1.235)1.29 = 1.31 . this is far from
enough to explain that Norway has a much higher share of rich and
super rich than the US.
By calibrating Ax −α locally so that it fits for r(30 ) and r(100) for
Norway, we get α = 1.67 og A = 0.037 . The high value of α here shows
that Norway has a much thicker tail in the wealth distribution implying
that a rise in income relative to the US leads to more rich and super rich
in Norway compared to the US.
Nordic Economic Policy Review
TN2016503 omslag.indd 2
09-02-2016 07:49:39
Ved Stranden 18
DK-1061 Copenhagen K
TemaNord 2016:503
Nordic Council of Ministers
Nordic Economic Policy Review
TemaNord 2016:503
ISBN 978-92-893-4450-0 (PRINT)
ISBN 978-92-893-4452-4 (PDF)
ISBN 978-92-893-4451-7 (EPUB)
ISSN 0908-6692
NUMBER 2 / 2015
The review appears once a year. It is published
electronically on the website of the Nordic Council of
Ministers: On that website, you can
also order paper copies of the Review (enter the name
of the Review in the search field, and you will find all the
information you need).
The Nordic Economic Policy Review is published by the
Nordic Council of Ministers. This year’s issue is part of the
Danish presidency programme for the Nordic Council of
Ministers in 2015. The review addresses policy issues in
a way that is useful for informed non-specialists as well
as for professional economists. All articles are
commissioned from leading professional economists and
are subject to peer review prior to publication.
Torben M. Andersen and Jesper Roine
The Nordic welfare model in an open European labor market
Bernt Bratsberg and Knut Roed
Future Pathways for Labour Market Policy: Including the Excluded
Michael Svarer and Michael Rosholm
Economics of Innovation Policy
Tuomas Takalo and Otto Toivanen
Taxing mobile capital and profits: The Nordic Welfare States
Guttorm Schjelderup
Nordic family policy and maternal employment
Julian V. Johnsen and Katrine V. Løken
Education and equality of opportunity: What have we learned from educational reforms?
Helena Holmlund
Retirement and Health in the Nordic Welfare State
Nabanita Datta Gupta and Bent Jesper Christensen
The future of Welfare services: How worried should we be about Wagner, Baumol and Ageing?
Andreas Bergh
Ethnic fractionalization and the demand for redistribution – Potential implications for the Nordic model
Johanna Mollerstrom
The social upper classes under Social Democracy
Kalle Moene
NUMBER 2 / 2015
TN2016503 omslag.indd 1
09-02-2016 07:49:39
Was this manual useful for you? yes no
Thank you for your participation!

* Your assessment is very important for improving the work of artificial intelligence, which forms the content of this project

Download PDF