Causal factors behind household expenditure leakage

Causal factors behind household expenditure leakage
Causal factors behind household expenditure leakage
and its effect on community resource dependence in Quebec
D. H. Kuhnke and W. A. White
To order publications on-line, visit the Canadian Forest Service Bookstore at:
Information Report NOR-X-418
bookstore.c fs.nrcan.gc.ca
NORTHERN FORESTRY CENTRE
CANADIAN FOREST SERVICE
EDMONTON, ALBERTA
QUEBEC
N
Bas-Saint-Laurent
Model Forest
E
W
S
The Northern Forestry Centre is one of five centres of the Canadian Forest Service, which has its headquarters in
Ottawa, Ontario. This centre undertakes the regional delivery of national projects.
The Canadian Forest Service’s main objective is research in support of improved forest management for economic,
social, and environmental benefits to all Canadians.
Le Centre de foresterie du Nord constitue l’un des cinq établissements du Service canadien des forêts, dont
l’administration centrale est à Ottawa (Ontario). Le Centre entreprend la réalisation régionale de projets nationaux.
Le Service canadien des forêts s’intéresse surtout à la recherche en vue d’améliorer l’aménagement forestier afin
que tous les Canadiens puissent en profiter aux points de vue économique, social et environnemental.
and
Causal Factors Behind Household Expenditure Leakage
its Effect on Community Resource Dependence in Quebec
D. H. Kuhnke and W. A. White
INFORMATION REPORT NOR-X-418
Canadian Forest Service
Northern Forestry Centre
2010
© Her Majesty the Queen in Right of Canada, 2010
Natural Resources Canada
Canadian Forest Service
Northern Forestry Centre
5320–122 Street
Edmonton, Alberta T6H 3S5
Catalogue No. Fo133-1/418E-PDF
ISBN 978-1-100-15038-3
ISSN 0831-8247
For an electronic version of this report, visit the Canadian Forest Service
Bookstore at http://bookstore.cfs.nrcan.gc.ca/
TTY: 613-996-4397 (Teletype for the hearing-impaired)
ATS: 613-996-4397 (appareil de télécommunication pour sourds)
Library and Archives Canada Cataloguing in Publication
Kuhnke, D.H.
Causal factors behind household expenditure leakage and its effect on
community resource dependence in Québec [electronic resource] /
D.H. Kuhnke and W.A. White.
(Information report ; NOR-X-418)
Electronic monograph in PDF format.
Includes bibliographical references.
ISBN 978-1-100-15038-3
Cat. no.: Fo133-1/418E-PDF
1.
2.
3.
I.
II.
III.
IV.
Consumption (Economics)--Québec (Province)--Bas-Saint-Laurent.
Bas-Saint-Laurent (Québec)--Economic conditions.
Resource-based communities--Economic aspects--Québec (Province)
-- Bas-Saint-Laurent.
White, W.A.
Northern Forestry Centre (Canada).
Title.
Series: Information report (Northern Forestry Centre (Canada) :
Online) NOR-X-418.
SD146 Q4 K83 2010
339.4’70971476
ii
C2010-980066-4
NOR-X-418
Kuhnke, D.H.; White, W.A. 2010. Causal factors behind household expenditure leakage and
its effect on community resource dependence in Quebec. Nat. Resour. Can.,
Can. For. Serv., North. For. Cent., Edmonton, AB. Inf. Rep. NOR-X-418.
ABSTRACT
The term “resource-dependent community” denotes a rural community
dependent on one or several resource-based industries for most of its
economic livelihood. However, the concept of resource dependence can be
broadened to describe a community’s degree of dependence on all base and
nonbase sectors that make up its economy. In this study, sector dependence
indices based on location quotients were used to examine the effect of
household expenditure within and outside of communities associated with
the Bas-Saint-Laurent Model Forest in Quebec on traditional interpretations
of community resource dependence. The causal factors behind household
expenditure leakage were also investigated through ordinary least-squares
modeling to determine linkages between dependence and characteristics of
both the community and its households. The study revealed that traditional
sector dependence indices based on employment and income are biased
when dependence is examined in a larger context. The sector of employment
also had little bearing on leakage except through income. Family structure,
home community population, travel distances, and education were the main
causal factors driving the degree of household expenditure leakage.
RÉSUMÉ
L’expression « collectivité dépendante des ressources » désigne une
collectivité rurale dont la survie économique dépend principalement d’une
ou plusieurs industries liées à l’exploitation d’une ressource naturelle.
Toutefois, le concept de « dépendance » peut être élargi pour comprendre la
mesure dans laquelle une collectivité dépend de l’ensemble des secteurs qui
composent son économie. Dans la présente étude, les indices de dépendance
à l’égard des secteurs basés sur les coefficients de localisation ont été utilisés
pour examiner l’effet des dépenses des ménages au sein et à l’extérieur des
collectivités de la Forêt modèle du Bas‑Saint‑Laurent sur les interprétations
traditionnelles de la dépendance des collectivités par rapport aux ressources.
Les facteurs agissant sur la fuite des dépenses des ménages ont également
été examinés à l’aide de la méthode des moindres carrés ordinaires en vue
de déterminer les liens entre la dépendance et les caractéristiques de la
collectivité et de ses ménages. L’étude a révélé que les indices traditionnels
de dépendance à l’égard des secteurs sont faussés lorsque la dépendance est
examinée dans un contexte élargi. Le secteur d’emploi a peu d’incidence sur
la fuite, sauf dans le cas des revenus. La structure familiale, la population
de la collectivité, les distances de déplacement et l’éducation étaient les
principaux facteurs influant sur le degré de fuite des dépenses des ménages.
iii
NOR-X-418
iv
NOR-X-418
Contents
INTRODUCTION . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
COMMUNITY DEPENDENCE AND ITS MEASUREMENT . . . . . . . . . . . 2
RESEARCH QUESTIONS . . . . . . . . . . . . . . . . . . . . . . . . . 5
CANADIAN FOREST SERVICE SURVEY OF HOUSEHOLD EXPENDITURES
IN MODEL FOREST COMMUNITIES . . . . . . . . . . . . . . . . . . 5
Survey Methods . . . . . . . . . . . . . . . . . . . . . . . . . . 5
Survey Results . . . . . . . . . . . . . . . . . . . . . . . . . . 6
ANSWERING THE RESEARCH QUESTIONS: LEAKAGE MODELING
AND DEPENDENCE INDICES . . . . . . . . . . . . . . . . . . . .
15
Leakage Modeling . . . . . . . . . . . . . . . . . . . . . . . . 15
Methods . . . . . . . . . . . . . . . . . . . . . . . . . . .
15
Results and Discussion . . . . . . . . . . . . . . . . . . . .
16
Dependence Indices . . . . . . . . . . . . . . . . . . . . . . . 21
Methods . . . . . . . . . . . . . . . . . . . . . . . . . . .
21
Results and Discussion . . . . . . . . . . . . . . . . . . . .
25
SUMMARY AND CONCLUSIONS . . . . . . . . . . . . . . . . . . . .
31
ACKNOWLEDGMENTS . . . . . . . . . . . . . . . . . . . . . . . . . 33
LITERATURE CITED . . . . . . . . . . . . . . . . . . . . . . . . . . 33
Figure
1.
Bas-Saint-Laurent Model Forest areas and study-associated
communities. . . . . . . . . . . . . . . . . . . . . . . . . . . 8
Tables
1.
Communities and their populations in the Canadian Forest Service
survey of household expenditures for the Bas-Saint-Laurent Model
Forest . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
2.
Distribution of usual location of purchases for households in the
Bas-Saint-Laurent Model Forest, by product class, based on the
Canadian Forest Service household expenditure survey . . . . . . 9
3.
Data for surveyed and derived variables in the Canadian Forest
Service household expenditure survey for the Bas-Saint-Laurent
Model Forest . . . . . . . . . . . . . . . . . . . . . . . . . .
v
10
NOR-X-418
4.
Distribution of family type in the Bas-Saint-Laurent Model
Forest, based on the Canadian Forest Service household
expenditure survey . . . . . . . . . . . . . . . . . . . . . . .
10
5.
Distribution of primary sector of occupation in the
Bas-Saint-Laurent Model Forest, based on the Canadian
Forest Service household expenditure survey . . . . . . . . . . . 11
6.
Distribution of highest level of education achieved in the Bas-SaintLaurent Model Forest, based on the Canadian Forest Service
household expenditure survey . . . . . . . . . . . . . . . . . .
11
Expenditures and percent leakage by product class for the
Bas-Saint-Laurent Model Forest, based on the Canadian Forest
Service household expenditure survey . . . . . . . . . . . . . .
13
7.
8.
Percent household expenditure leackage by community for the
Bas-Saint-Laurent Model Forest, based on the Canadian Forest
Service household expenditure survey . . . . . . . . . . . . . . 14
9.
Main reasons for decisions about where to purchase goods
and services . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
10. Descriptions of basic and derived variables used in the model
of percent household expenditure leakage for all goods
and services . . . . . . . . . . . . . . . . . . . . . . . . . .
16
11. Results of ordinary least squares regression model of percent
household expenditure leakage for all goods and services . . . . .
17
12. Descriptions of basic and derived variables used in the model
of percent household expenditure leakage for durable goods . . . . 20
13. Results of ordinary least squares regression model of percent
household expenditure leakage for durable goods . . . . . . . . . 21
14. Raw and basic employment, by sector and community, in the
Bas-Saint-Laurent Model Forest . . . . . . . . . . . . . . . . . . 23
15. Sector dependence indices for communities in the
Bas-Saint-Laurent Model Forest, based on three methods
of calculating dependence . . . . . . . . . . . . . . . . . . . . 27
16. Summary of ordinal ranking of sector dependence indices
by economic sector . . . . . . . . . . . . . . . . . . . . . . .
30
Appendixes
1.
Telephone Questionnaire Script – Bas- St.-Laurent
Model Forest . . . . . . . . . . . . . . . . . . . . . . . . . 35
2.
An example of a calculation of household expenditure
for a product class using the modifier calculation . . .
vi
45
NOR-X-418
INTRODUCTION
Hundreds of rural communities across Canada
depend on one or more industries associated
with natural resources for most or all of their
economic livelihood. Because of the importance
of natural resources to the Canadian economy,
these resource-dependent communities have
been the subject of many studies investigating
the relationship between community well-being
and dependence on the resource sector. In
recent years, however, the widespread adoption
of sustainable development as a guiding ethic
in resource management has led to questions
about what dependence really means. This has in
turn resulted in a logically broader interpretation
of dependence to mean the degree to which a
community depends on all of the sectors that
make up its economy, not just the base sectors
(where base sectors are the sectors that draw
in employment or income from outside the
community or region, usually through the export
of goods).
Although there is no universally accepted method
for determining the degree to which resourcedependent communities actually depend on the
resource sector, this broader view of dependence
has given rise to a need to identify the level and
type of sector dependence. Studies by Fletcher
et al. (1991), Horne and Penner (1992), Horne
and Robson (1993), Jagger et al. (1998),
Korber et al. (1998), Williamson et al. (1999),
and Stedman et al. (2005) are examples of the
many Canadian contributions to the literature
on community dependence. Employment data,
which were used by Stedman et al. (2005), for
example, are readily available from the Census
of Canada, but employment figures alone do not
account for income differences between various
sectors. In particular, income-based measures of
dependence could be prone to misinterpretation
because of the common assumption that income
earned in a community is spent within that
community (Jagger et al. 1998). Jagger et al.
(1998) investigated the utility of household
expenditure leakage as an indicator of the
severity of effects that an economic shock
is likely to have on a community. A high level
of leakage may mean that the effect of an
economic shock will be felt to a greater extent
outside the community than would be the case
for a community where most purchases are
made locally.
In this report, we briefly review the literature on
community dependence and its measurement
and then investigate the interrelationships among
three methods of dependence measurement
within an economic base modeling framework.
This analysis is intended to demonstrate that
traditional measures of dependence, which
are based on income and employment alone,
are biased or inaccurate because they do not
account for the effects of income leakage. We
contend that accounting for leakage within an
economic base modeling framework portrays
a more complete description of community
dependence. This work is based on a survey of
households from communities within the BasSaint-Laurent Model Forest, a region of Quebec
that is believed to be heavily dependent on
resource-based industries such as forestry and
agriculture, including maple syrup production.
COMMUNITY DEPENDENCE AND ITS MEASUREMENT
Harold Innis
(1894–1952),
the
country’s
generous but far-flung and difficult-to-access
endowments of natural resources led to a largely
resource-based economy that now supplies
raw or semiprocessed materials to established
industrial centers in central Canada, the United
States, Europe, and Japan. As a result, many
rural communities became established across
Canada to provide the social and economic
nexus for the harvest or extraction of one or
more resources.
The dependence of communities on extraction
or harvests of natural resources has been the
subject of much study by numerous scholars
and researchers, especially in the United States,
including pioneering early works such as that
by Kaufman and Kaufman (1946). Much of this
study has been spurred by academic interest in
the paradox of poverty in the midst of resource
abundance, sometimes referred to as the
resource curse (Ross 1999). It has long been
noted that resource-dependent communities
suffer from a range of social ills, including poverty
(however defined), high unemployment, low
income, and low human capital (Stedman et al.
2004; Leake et al. 2006), although considerable
variation, according to the type of resource
(forests, minerals, etc.), the region, and other
factors, has been reported (Stedman et al.
2004). Much of the research in this area has
therefore investigated the relationship between
a community’s natural resource dependence and
its well-being or relative lack thereof.
Traditional measures of dependence have
concentrated on readily available (and relatively
inexpensive) economic data, chiefly data for
employment and income, which are arguably
important contributors to community and
individual well-being. The distribution of income
within or between communities is also often used
as an indicator of well-being. Economists tend
to favor the use of income rather than numbers
of people employed, as the income associated
with various jobs is not uniform (Stedman et al.
2007). Higher proportions of well-paying jobs
may indicate greater well-being. These economic
data also have an important temporal aspect, as
they are collected periodically by state agencies
like Statistics Canada and the US Census Bureau.
Comparisons of the proportion of employment
or income that an industry represents within a
community relative to the proportions for other
industries (if there are any) within the same
community, or relative to the proportions in
other communities or regions, often form the
basis for assessing dependence and well-being
in resource-dependent communities.
Some might question the rationale for an
interest in the well-being of resource-dependent
communities and, by extension, the well-being of
their residents. A recent estimate put the number
of Canadian forest-dependent communities in
2001 at 324, but this figure refers specifically
to communities with at least 50% of total base
income coming from forestry (Stedman et al.
2007). In addition, almost 893 000 Canadians
were directly or indirectly employed in 1999 in
the forestry sector (Natural Resources Canada
2009). Forestry and its related industries, as
well as other natural resource-based industries
such as mining and fishing and their related
industries, provide employment and livelihoods
for many Canadians. The social and economic
well-being of Canadians employed in these
resource-based industries and living in the many
associated resource-dependent communities
cannot therefore go unnoticed in a nation heavily
dependent on natural resources for its overall
economic well-being. This heavy dependence is
perhaps best explained by staples theory, the
economic logic generally thought to best explain
the manner in which the nation’s economy
developed. According to staples theory, first
espoused by Canadian economic historian
An
enhancement
representing
a
more
sophisticated approach to measuring dependence
using employment or income data is the economic
base model. Economic base theory was conceived
in the 1920s and 1930s by various analysts
and urban planners who required a method
for estimating the total effect on a community
caused by introduction or expansion of a base
industry (Andrews 1953). The core of the model
is the division of local economic activity into two
sectors, the base sector and the nonbase sector.
The base sector consists of activity that brings in
income from outside sources or that generates
2
NOR-X-418
prevent leakage (Williams 1996). Others have
noted that the spatial scale used in examining
dependence is important, for there exists a
regional level of industrial diversification and a
more complex network of economic and social
relations that is less apparent when communities
are studied in isolation (Randall and Ironside
1996). Although the effect of leakage is
consistent with the economic base hypothesis, it
may not be adequately accounted for in practical
implementation (Robertson 2003), something
that the current work is intended to address. In
addition, a number of factors that contribute to
the propensity of residents to purchase goods
and services outside of their community are
investigated here.
employment from outside income, such as
income resulting from goods exported from the
community or region or money brought into
the community or region by tourists or through
transfer income. Horne and Robson (1993), in
their analysis of British Columbia communities,
identified the need to include nonemployment
income (sourced from outside the community)
in an examination of resource dependence, as
they found that dependence on the resource
sector was lower than had been estimated in
an earlier study (Horne and Penner 1992). For
example, in a pulp mill town the incomes of the
mill workers are considered base incomes. The
nonbase sector consists of the suppliers of goods
and services to the pulp mill workers, namely,
retail and grocery stores and the like, whose
incomes come mostly from the spending of the
pulp mill employees.
The economic base model can be actualized
through a number of methods, including the
minimum requirements approach and the
location quotient technique, the latter of which
was employed for the work reported here.
The location quotient technique has a number
of shortcomings, chief among them its high
sensitivity to the level of sector aggregation and
the absence of accounting for the existence of
cross-trading or cross-hauling, which occurs
in situations where communities concurrently
import and export similar goods and services,
as outlined by Robertson (2003). Some of these
shortcomings have been addressed through a
modification of the location quotient technique
that accounts for imports and exports as discussed
by Fletcher (1991), Korber et al. (1998), and
White and Watson (White, W.; Watson, D. 2004.
Natural resource based communities in Canada:
an analysis based on the 1996 Canada Census.
Internal report produced for the Winning In The
21st Century initiative of Natural Resources
Canada. Can. For. Serv., North. For. Cent.,
Edmonton, AB). The level of sector aggregation
is important because per capita consumption
differs between regions because of varying
incomes; as such, smaller sectors are generally
preferred for this type of analysis, as noted, for
example, by Schwartz (1982), who argued that
errors arising from differences in consumption
and productivity can be reduced if provincial
rather than national employment levels are
used. Another way to increase confidence in the
interpretation of a location quotient is to use
more than one reference economy for sector
aggregation (Persky et al. 1993).
The economic base model should not be
interpreted to mean that exports (or outside
income) constitute the only factor determining
the overall level of local economic activity (Power
1996). Overall local economic activity is also
heavily influenced by the structure and character
of the local economy itself, because of its role
in determining the level of income leakage. In
particular, not all of the income earned in the
base sector is spent in the local nonbase sector,
nor do those employed in the nonbase sector
spend all of their disposable income locally.
Income leakage forms the basis for determining
impact multiplier, the ratio of nonbase to base
income within a community or region, which are
used in the economic base model to determine
the extent to which a change in the base sector,
experienced as either a positive or negative
economic shock, causes a commensurate
change in the nonbase sector. The more quickly
injected income leaks out of the local economy,
the smaller the multiplier (Power 1996). If there
is income leakage in a small community, the
effects of changes in employment and income
may actually appear outside of the community
(Robertson 2003). Williams (1996) observed
that many local economies have substantial
leakage because of development policies that
cultivate the base sector of an economy with
little regard for the extent to which leakage of
income is taking place. Local consumer services
can function as base activities by drawing income
into the economy from outside, thus acting to
3
NOR-X-418
has become increasingly mobile and may not be
as beneficial in the long run as it is in the short
run (Overdevest and Green 1995). Peripheral
firms often supply raw materials to core firms;
the relationship between logging companies
and pulp and paper firms is a classic example.
However, it is not always the case that peripheral
firms are associated with lower well-being. For
example, in British Columbia, the logging and
lumber sectors, along with the pulp and paper
sector, have a positive association with wellbeing, because of a number of factors, including
the nature of the resource and high rates of
unionization (Parkins et al. 2003).
The number of forest-dependent communities
across Canada was reported earlier in this
document as 324, but other researchers have
arrived at different figures, depending not
only on the measurement method but also on
the rationale, if one was used or provided, for
delineating dependence from nondependence.
One of the earlier Canadian efforts on
community resource dependence (DREE 1979)
used a two-stage approach in which dependence
varied with community population. White et al.
(1986) expanded on this approach by adding
economic diversity criteria for communities in
British Columbia. Pharand (1988) described
the demographic characteristics of forestdependent communities across Canada using
an approach to defining dependence similar to
that used by DREE (1979). In the United States,
20% of total employment has traditionally been
used as the cutoff for high levels of resource
dependence in any particular sector; however,
many researchers use a 10% criterion, because
the 20% cutoff often results in too few cases
for regional analyses (Stedman et al. 2004).
Application of the 10% criterion yielded a total
of 918 forest-dependent communities across
Canada (Stedman et al. 2005). Randall and
Ironside (1996) found a direct relationship
between the degree of dominance of a resource
sector and distance to the nearest Canadian
metropolis, although there was considerable
variation between resource sectors and, in the
case of forestry, considerable variation within
the sector.
Aside from any discussion of measurement
methods or market segmentation, the literature
reveals that resource-dependent communities
share several fundamental characteristics. The
bulk of their economic livelihood stems from one
or several industries engaged primarily in the
extraction or harvest of natural resources; they
tend to have smaller populations than larger
urban centers; they are removed, though not
necessarily isolated, from larger urban centers;
and they suffer a range of social ills that are more
pronounced than those of larger urban centers
or the nation as a whole. Innis (1950) used a
meteorological metaphor, cyclones, to represent
the whirlwind frenzy of capitalist accumulation at
extraction sites and the equally frenetic decline
and destruction that follow. In this setting, it is
the vicissitudes of boom-and-bust cycles that
are generally the source of economic shocks,
which affect resource-dependent communities
more than they do the broader nation. For
resource-dependent communities in particular,
sustainability hinges on the ability to deal with
change, to reconfigure available resources,
and to recombine financial capital, local skills,
and natural resources in ways that create
sustainable livelihoods (Beckley et al. 2002).
The point of describing all these methods and
decision criteria for community dependence
and well-being is to emphasize that the method
and criteria chosen are primarily a function of
the researcher’s intentions, as influenced by
available data, research budgets, and the works
of previous researchers. Each method has pros
and cons, and this work seeks to address an
often overlooked aspect of one of them.
One source of variation within the forestry sector
is its segmented nature. This sector is made up
of several subsectors, such as logging, lumber,
and pulp and paper, which generally confer
different levels of well-being to their employees
depending on whether they are core or peripheral
industries. Core industries are represented by
large, well-capitalized firms that may enjoy
oligopolistic or oligopsonistic status within their
industries and may dominate their product
markets (Overdevest and Green 1995). These
industries contribute to enhanced community
well-being through the provision of stable, yearround employment, higher incomes, and fringe
benefits to their employees, which peripheral
firms cannot provide; however, the core sector
4
NOR-X-418
RESEARCH QUESTIONS
that some accounting of income leakage be
undertaken to obtain a more accurate picture of
resource dependence. Given the importance of
accounting for income leakage in economic base
models, one of the research questions for this
work was “What drives the degree of income
leakage?” Also, given that estimates of income
leakage are already available, what effect do
these leakages have on measures of community
dependence that are based on base income and
employment? In this study, the interrelationships
among three methods of measuring dependence
within an economic base modeling framework
were investigated to highlight the potential
biases inherent in measures of dependence
based on income and employment alone. First,
details are provided concerning the source of
data for this work. The models used to address
the research questions are then described, and
the findings presented. The report ends with a
summary and conclusions section.
Estimates of sector dependence in some of the
studies mentioned above relied on economic
base models rather than simple percentages
of income or employment. The advantage of
economic base models is their basic tenet that
communities depend on base employment, not
on total employment (since total employment
includes nonbase employment, which is
irrelevant to dependence). The percentage of
a resource industry’s base employment relative
to total base employment may be a more
accurate measure of resource dependence
than percentages of total employment. Another
advantage of economic base models, from
the viewpoint of community development or
planning, is that they can generate an estimate
or indication of the economic impacts on a
region or community that will result from policy
decisions or exogenous economic shocks.
Economic base models can also be constructed
using income instead of or in addition to
employment; however, it seems essential
CANADIAN FOREST SERVICE SURVEY OF HOUSEHOLD
EXPENDITURES IN MODEL FOREST COMMUNITIES
Survey Methods
data from 2 of these 18 classes were additionally
assigned to 2 more product classes created by
the authors. These assignments were made
because information available before the survey
began suggested that particular types of goods
within each of the two Statistics Canada classes
were often purchased at locations outside the
community. A response category for purchases
made using the Internet was included for eight
classes of nondurable and durable goods and
services; these purchases were considered to
have been sourced from outside the general
area of the model forest. Nondiscretionary goods
and services, such as rent, mortgage payments,
taxes, and utilities, were omitted from the survey,
because respondents did not have a choice about
where to make such expenditures. The survey
also included socioeconomic questions about
household income, respondent’s age and level of
education, family characteristics, the sector of
employment (forestry, mining, service industry,
etc.) of each adult within the household, and the
In 1998 and 1999, the Canadian Forest Service
(CFS) conducted nationwide surveys that
sampled households in communities close to or
within all 11 model forests across the country.
The chief aim of these surveys was to gather
baseline data on residents’ expenditures within
and outside their respective communities, to
allow examination of community dependence
from the perspective of leakage expenditures.
These analyses were intended to complement
traditional employment and income data. The
surveys were conducted by telephone, on the
basis of randomly selected telephone numbers
for residents in the model forest communities.
The respondents were asked where they usually
purchased durable and nondurable goods and
services. The list of goods and services used in the
survey was based mostly on Statistics Canada’s
standard 18 classes of household durable and
nondurable goods and services. Expenditure
5
NOR-X-418
French or English was spoken in the household.
Some communities were later dropped from
the study because they had too few eligible
households to represent a sufficient sample
size for determination of a sector dependence
index, as described later. This is partly why the
18 communities shown in Table 1 are not the
same as those listed in Appendix 1. In addition,
some of the communities listed in Appendix 1
were subsequently found to be part of larger
census subdivision (CSD) and therefore crucial
pieces of information for modeling purposes
for these communities were indistinguishable
from the CSDs. Other respondents had to be
dropped from the sample because too many
ambiguous destination communities were given
among the various product classes to permit
leakage modeling. Finally, the survey included
respondents living in larger centers such as
Rimouski, which were later deemed to lie outside
of the model forest area; those respondents
were also dropped from the final sample. As a
result of these exclusions, the final sample size
for this work was 499.
number of unemployed adults in the household.
Respondents were also asked about their
motivation for making purchases in different
locations. The survey instrument appears in
Appendix 1.
An underlying feature of these household
expenditure surveys was the assumption that,
for respondents who stated that the home
community was where they usually purchased
items in a particular product class (e.g., food from
grocery stores), 100% of items in that product
class were purchased within the home community.
In reality, the respondent might purchase 80%
of groceries in the home community and 20%
outside the community. Asking respondents to
estimate the split in spending between their home
community and various destination communities
would have added a great deal of subjectivity
to the responses, which might in turn have led
to estimated dollar values that were no closer
to the actual values than those achieved with
the existing method. Perhaps more importantly,
asking respondents for an estimated split for all
product classes would have made the telephone
interview longer than most respondents would
have found acceptable. The implicit assumption
behind the all-or-nothing survey questions was
that the large sample size would minimize any
bias introduced by this approach.
The 18 communities that were considered to be
associated with the BSLMF in the final sample
either were within one of the three areas of the
model forest or appeared to form part of a distinct
cluster of communities in close proximity to one
of the three areas. Destination communities
(communities cited by survey respondents as
sources of goods and services) were classified as
part of the BSLMF study area (which could be, but
were not limited to, one or more of the remaining
17 sampled communities) or outside of the study
area (Table 1). Edmundston (in New Brunswick),
Rimouski, and Rivière-du-Loup were the urban
centers outside of the BSLMF that were visited
most frequently by survey respondents for the
purchase of various goods and services. The
city of Québec was the major metropolitan area
(> 100 000 residents) closest to the study area,
but some respondents traveled as far as Montréal
to purchase various goods and services. Fourteen
of the communities in the survey sample were
small, with a population under 700, while a few
larger communities lead to the sample median
of 1658.The number and spatial concentration
of communities associated with the model forest
was higher than for most other Canadian model
forests. Information about communities, such
as their populations, was based on Statistics
The work reported here is based on the survey
conducted in the Bas-Saint-Laurent Model Forest
(BSLMF) of Quebec. The BSLMF differed from the
other model forests in important ways. All of its
three sections (Fig. 1), covering a total area of
113 200 ha, were made up of private woodlots or
tenant farms leased from a corporate landowner
(Abitibi-Consolidated). The BSLMF lay within the
Great Lakes – St. Lawrence forest region, which
is dominated by stands of maple (Acer spp.),
balsam fir (Abies balsamea (L.), and yellow birch
(Betula alleghaniensis Britt.).
Survey Results
The BSLMF survey sample totaled 2 082
households, or 13.4% of the total number of
households among the sampled communities.
For various reasons, calls to 509 of the telephone
numbers did not lead to interviews: the call
produced a busy signal; the number was for a
fax machine, modem, or pager; the number was
for a business rather than a household; or no
6
NOR-X-418
7
NOR-X-418
619
295
657
La Trinité-des-Monts
Lac-des-Aigles
281
Saint-Médard
13 397
434
323
Saint-Zénon-du-Lac-Humqui
Sainte-Irène
Total population
862
Saint-Valérien
1 009
657
Saint-Juste-du-Lac
Saint-Narcisse-de-Rimouski
106
474
Saint-Eugène-de-Ladrière
Saint-Guy
322
380
Saint-Charles-Garnier
Saint-Cléophas
2 152
536
La Rédemption
Notre-Dame-du-Lac
453
Esprit-Saint
3 317
Biencourt
Dégelis
520
Population
Auclair
Communities
surveyed in BSLMF
Sainte-Jeanne-d’Arc
Sainte-Angèle-de-Mérici
Saint-Michel-du-Squatec
Saint-Léon-le-Grand
Saint-Gabriel
Saint-Fabien
Les Hauteurs
Lejeune
Le Bic
Cabano
Destination communities
in BSLMF (not surveyed)
16 318
1 128
1 066
1 332
1 114
2 775
1 848
589
381
2 872
3 213
Population
1 999
Sayabec
Val-Brillant
Trois-Pistoles
1 409 544
997
3 635
36 786
2 218
Sainte-Blandine
Sorel
3 643
469
1 768
3 692
3 395
38 739
17 772
1 477
35 561
169 076
3 097
1 039 534
334
5 886
11 635
1 351
17 373
2 634
6 473
Population
Saint-Pascal
Saint-Marc-du-Lac-Long
Saint-Jean-de-Dieu
Saint-Jacques
Saint-Antonin
Saint-Hyacinthe
Rivière-du-Loup
Rivière-Bleue
Rimouski
Québec
Pohénégamook
Montréal
Mont-Lebel
Mont-Joli
Matane
Luceville
Edmundston
Causapscal
Amqui
Destination communities
not in BSLMF
(not surveyed)
Table 1. Communities and their populations in the Canadian Forest Service survey of household expenditures for the Bas-Saint-Laurent Model Forest (BSLMF)
(gas, diesel, and propane; tobacco and alcohol)
were there more households usually purchasing
within the home community than outside of it.
Also noteworthy is the fact that almost half of
the survey respondents reported never having
purchased computers, toys, or games or motor
homes and trailers (Table 2).
Canada’s CSDs. The locations of most of these
communities are shown in Figure 1.
Descriptive information on the frequency
distribution of survey households with respect to
home community and product class appears in
Table 2, which also includes information on the
number of households that never purchased a
particular product. For only two product classes
QUEBEC
Matane
N
Bas-Saint-Laurent
Model Forest
r
ce
St.
en
wr
Mont-Joli
e
Riv
Luceville
Rimouski
La
Le Bic
St.-Gabriel
Ste.-Blandine
St.-Valérien
St.-Fabien
St.-Eugènede-Ladrière
Trois-Pistoles
Les-Hauteurs
Mont-Lebel
St.-Charles
-Garnier
St.-Narcissede-Rimouski
St.-Micheldu-Squatec
Causapscal
St.-Zénon-duLac Humqui
Biencourt
Model Forest surveyed communities
Model Forest destination communities
Non-Model Forest destination communities
Nicolas Riou Model Forest area
Lac-Metis Model Forest area
Est du Lac Temiscouata Model Forest area
Municipalities
Auclair
Cabaon
St.-Juste-du-Lac
Notre-Dame-du-Lac
St.-Pascal
Degelis
Rivière-Bleue
0
St.-Marc-du-Lac-Long
S
Esprit-Saint
Lejeune
St.-Antonin
Figure 1.
St.-Léonle-Grand
Lac-des-Aigles
St.-Jean-de-Dieu
Pohénégamook
Ste.-Irène
E
La-Trinité-des-Monts
St. Guy
St. Médard
Rivière-du-Loup
W
Ste.-AngèleSayabec
de-Mérici
Val-Brillant
Ste.-Jéanned’Arc
St.-Cleophas
Amqui
La Redemption
16 km
Bas-Saint-Laurent Model Forest areas and study-associated communities.
8
NOR-X-418
Table 2. Distribution of usual location of purchases for households in the Bas-Saint-Laurent Model Forest, by product
class, based on the Canadian Forest Service household expenditure survey
Usual location of purchases;
no. (%) of respondents stating
usual location of purchasesa
Product class
Within home
community
Food from grocery stores
208
(41.7)
Outside of home
community
No. (%) of
respondents
who never made
purchasesb
291
(58.3)
0
(0.0)
Food from restaurants
153
(32.5)
318
(67.5)
28
(5.6)
Household supplies
190
(38.5)
304
(61.5)
5
(1.0)
16
(3.2)
481
(96.8)
2
(0.4)
Gas, diesel and propane
242
(50.8)
234
(49.2)
23
(4.6)
Dental and optical products
244
(48.9)
255
(51.1)
0
(0.0)
Medicine and pharmacy products
178
(35.8)
319
(64.2)
2
(0.4)
61
(15.3)
337
(84.7)
101
(20.2)
Clothing
Spectator and entertainment purchases
Computers, toys, and games
21
(7.9)
244
(92.1)
234
(46.9)
Tobacco and alcohol
201
(50.9)
194
(49.1)
104
(20.8)
Reading material
108
(24.1)
341
(75.9)
50
(10.0)
54
(11.1)
432
(88.9)
13
(2.6)
Small gifts and accessories
Furniture and appliances
113
(23.2)
374
(76.8)
12
(2.4)
Home entertainment
96
(19.8)
390
(80.2)
13
(2.6)
Sporting and recreation
61
(14.0)
374
(86.0)
64
(12.8)
Recreational vehicles
67
(17.3)
321
(82.7)
111
(22.2)
New cars and trucks
44
(10.0)
396
(90.0)
59
(11.8)
Used cars and trucks
58
(13.3)
377
(86.7)
64
(12.8)
Motor homes and trailers
9
(3.3)
263
(96.7)
227
(45.5)
Vacations
0
(0.0)
403 (100.0)
96
(19.2)
aIn
these two columns, the numbers in parentheses are percentages with respect to the total number of respondents who
made purchases (499 minus value in last column of table).
this column, the numbers in parentheses are percentages with respect to the number of survey respondents (499).
bIn
size of the destination communities relative to
that of the respondent’s home community and
road quality, may complicate this assumption.
Rimouski is the largest community in the BSLMF
area (Table 1), and distances between this center
and the surveyed communities varied widely. For
example, Saint-Valérien was only 20 km away,
whereas Dégelis was 139 km away. A geographic
information system was used to determine road
distances among the surveyed communities
and between the surveyed communities and
destination communities; these values were
included in percent expenditure leakage models.
Descriptive statistics about distances and other
Travel distance is an important variable
influencing
spending
behavior
(Yanagida
et al.1991; Olfert and Stadler 1994). The
individual trip information that would be
necessary for a travel cost model was not
available from the household expenditure survey,
so travel distances were deemed a reasonable
proxy of the travel costs that consumers had to
bear. The general assumption is that destination
communities located further away from the
community of residence than other destination
communities will have lower visitation because
of the higher travel costs involved; however,
numerous other factors, such as the population
9
NOR-X-418
had no quantitative meaning, i.e., they are not
expressed in terms of a physical or quantifiable
unit of measure. For example, education was
measured by type of highest level of education
achieved, rather than number of years of
education completed.
continuous and ordinal variables used to model
household expenditure leakage are shown
in Table 3. The categorical variables family
type, sector employment, and education are
presented in Tables 4 to 6, respectively. These
variables were disaggregated into dummy
variables for modeling purposes because they
Table 3. Data for surveyed and derived variables in the Canadian Forest Service household expenditure survey for the
Bas-Saint-Laurent Model Forest
Survey variable
Mean
Standard
deviation
Minimum
Maximum
Age classa
3.4
1.3
1
6
Income classb
3.4
1.7
1
8
No. of unemployed adults in household
0.8
0.8
0
4
1 375
1 204
106
3 317
Population of surveyed communities
Population of destination
communitiesc
400
Distance to destination communities (km)e
Sum of distances to destination communities
(km)f
000d
58 366
0
800 000
63.8
46.4
0
479.42
787.4
690.3
0
7 191.29
Per-household percent leakage for durable goods
86.3
26.6
0
100.0
Per-household percent leakage for all goods
72.1
27.4
0
100.0
aRespondents
were asked to specify their age within one of six age classes. See Appendix 1.
were asked to specify their total household income within 1 of 13 income classes (Appendix 1); the data were
subsequently condensed to 8 income classes because few respondents reported incomes over $80 000.
cBased on the average population of all destination communities across 19 product classes.
dValue shown is median of population of destination communities. Mean was 32 111.
eBased on distance traveled across 19 product classes.
fSum of the distances to all destination communities, even if more than one destination was the same across more than one
product class.
bRespondents
Table 4. Distribution of family type in the Bas-Saint-Laurent Model Forest, based on the Canadian Forest Service
household expenditure survey
Model
variable
names
Description of family type
FT1
Single man or woman < 45 years
No. (%) of
households
19
(3.8)
FT2
Married couple, no children
183
(36.7)
FT3
Married couple with husband < 45 years, 1 child
46
(9.2)
FT4
Married couple with husband < 45 years, 2 children
63
(12.6)
FT5
Married couple with husband < 45 years, ≥ 3 children
46
(9.2)
FT6
Lone-parent, any number of children
18
(3.6)
FT7
Three adults, no children
53
(10.6)
FT8
Three adults, any number of children
32
(6.4)
FT9
Four adults, no children
21
(4.2)
FT10
Four adults, any number of children
18
(3.6)
10
NOR-X-418
Table 5. Distribution of primary sector of occupation in the Bas-SaintLaurent Model Forest, based on the Canadian Forest Service
household expenditure survey
Model
variable
names
Employment sector
AG
Agriculture
FOR
Forestry
CONS
No. (%) of
households
36
(7.2)
107
(21.4)
Construction
19
(3.8)
FINPROF
Financial or professional services
37
(7.4)
OILGAS
Oil and gas (energy)
2
(0.4)
GOVT
Government
50
(10.0)
MINING
Mining industry
16
(3.2)
SERV
Service industry
34
(6.8)
TRAP
Transportation
30
(6.0)
TRAF
Transfersa
119
(23.8)
OTHER
Other sectors
49
(9.8)
aCanada
income.
Pension Plan benefits, social assistance payments, and investment
Table 6. Distribution of highest level of education achieved in the
Bas-Saint-Laurent Model Forest, based on the Canadian Forest
Service household expenditure survey
Model
variable
names
Description of education level
ED1
Never attended school
ED2
Completed grade school
ED3
Some high school
ED4
High school graduate
114
(22.8)
ED5
Technical school
52
(10.4)
ED6
Some college or university
64
(12.8)
ED7
Undergraduate university degree
44
(8.8)
ED8
Graduate university degree
16
(3.2)
No. (%) of
households
2
(0.4)
67
(13.4)
140
(28.1)
11
NOR-X-418
Expenditures and expenditure leakage by
product class and other characteristics are shown
in Table 7, whereas overall expenditure leakage
data for 17 of the 18 communities in the BSLMF
(excluding Saint-Guy, because no respondents
from this community were included in the final
survey sample) and for the whole model forest
are presented in Table 8. The figures for total
percent expenditure leakage in Table 7 are very
similar to those in Table 2 (see “Outside of home
community” column), the minor differences
being attributable to proportional variation by
income and family type. Households with higher
incomes may spend proportionately more on
vacations, for example. Table 7 also provides a
breakdown of expenditure leakage outside of the
home community between communities inside
and outside the BSLMF. The bulk of spending
outside of the home community (about 90%)
occurred outside of the model forest.
It would have been unrealistic to expect
respondents to provide dollar figures for each
product class over the telephone, so expenditure
data were drawn instead from Statistics
Canada’s survey of household expenditures
(Statistics Canada 1998). The direct estimates
of household expenditure from the Statistics
Canada survey did not concurrently incorporate
three important characteristics: expenditures
reflecting rural nonfarm households in Quebec
(which is how households in this area of
the province are classified by the Statistics
Canada census), level of household income,
and demographic structure. Therefore, before
total household expenditures per product class
could be determined for each respondent, it
was necessary to account for the effect of these
three elements on respondent expenditures for
each product class. Ten demographic groups
identified by family structure and 9 levels of
household income were identified in the survey
of household expenditures (Statistics Canada
1998). The following equation was used to
calculate a modifier for estimating the household
expenditure for product class i:
In general, as absolute spending outside the
community of residence increases, so too
does the proportion of total spending that
takes place outside the home community (the
percent expenditure leakage). Spending outside
of the home community for several product
classes (food from grocery stores; food from
restaurants; household supplies; gas, diesel,
and propane; dental and optical products;
medicine and pharmacy products; and tobacco
and alcohol), collectively accounting for about
51% of total purchases (Table 7), increased
more rapidly as percent expenditure leakage for
these 7 products increased than did increases
in spending as percent expenditure leakage
increased in the remainder of the product
classes. Percent expenditure leakage reached a
plateau at a total out-of-community expenditure
for all product classes of roughly $12 000,
after which percent expenditure leakage varied
between 84% and 100%.
(1)
where R is the average expenditure of a rural
Quebec household for product class i, QU is
the average expenditure across all Quebec
households for product class i, INC is the average
expenditure on product class i by income
category, and DEM is the average expenditure on
product class i by demographic group in Quebec.
Average expenditure (EXP) by product class i
and family structure j can be estimated with the
following equation:
(2)
The results of the modification yielded estimated
household expenditures for each of the 18 classes
of household goods and services (including
vacations) according to household income, family
structure, and the rural nature of the economy
(see Appendix 2 for an example). The modified
Statistics Canada expenditure data were then
twinned with data for individual participants in
the BSMLF telephone survey, according to the
respondent’s family type and income class.
Respondents spent more on food from grocery
stores than on items from any other product
class (24.2% of total expenditure), followed by
clothing (10.6%) and new and used cars and
trucks at 9.4% and 9.3%, respectively. Perhaps
the most telling feature evident in Table 7 is that
the percent leakage for nondurable products and
services was 65.4%, similar to the percentage
12
NOR-X-418
13
NOR-X-418
6.3
2.4
1.7
2.1
9.4
9.3
1.4
2.6
35.2
100.0
657
563
866
700
237
067
503
814
581
217
154
188
866
849
131
236
3 226 407
9 185 304
2.0
0.7
0.8
4.9
1.2
0.4
64.8
045
838
669
961
300
373
897
184
64
72
451
112
41
5 958
24.2
5.2
5.6
10.6
7.3
1.9
601
985
980
316
695
134
2 220
478
510
976
667
177
% of spending
on all product
classes
13.2
7.7
10.0
9.0
4.0
7.9
18.7
17.8
6.4
3.2c
12.0
5.5
7.7
7.3
3.1
1.2
4.6
3.4
6.0
6.1
1.6
7.3
4.9
% leakage to
model forest
communities
73.3
65.1
67.0
71.7
82.6
74.5
71.4
68.5
90.5
96.8
52.2
80.2
84.3
42.2
73.3
88.2
60.8
55.2
61.8
55.6
95.4
42.2
46.5
% leakage to
communities outside
model forest
86.5
72.8
77.0
80.8
86.6
82.4
90.1
86.3
96.9
100.0
64.2
85.7
92.0
49.5
76.4
89.4
65.4
58.6
67.8
61.7
97.0
49.5
51.3
Total %
leakage
617
723
061
474
683
860
472
814
201
554
881
788
838
004
338
213
737
702
969
561
890
2 790 704
6 689 042
447
175
134
155
780
732
127
236
118
55
66
223
85
37
3 898
1 302
324
315
946
330
90
Total
leakage ($)
such as all-terrain vehicles, boats, snowmobiles, and dirt bikes. A single expenditure figure from Statistics Canada was applied to both the survey’s recreational
vehicles category and its motor homes and trailers category for the purposes of modeling percent leakage. The difference between expenditures for the survey’s recreational
vehicles category and its motor homes and trailers category is attributable to respondents who never purchased one or the other product.
bStatistics Canada groups new and used cars and trucks into one category. The Statistics Canada expenditure figure was applied to both the survey’s new and used cars and
trucks categories for the purposes of modeling percent leakage. The difference between total expenditures for the new cars and trucks category and the used cars and trucks
category is attributable to respondents who never purchased one or the other product.
cSome expenditure associated with this percentage may have been made within the home community. There was no specific response category in the survey for at-home
vacations, on the assumption that most respondents would spend a vacation outside of their home community. All spending was therefore assumed to have occurred in other
communities in the Bas Saint-Laurent Model Forest.
dVacations are considered durable goods because their infrequent occurrence and relatively high cost are more akin to the characteristics of durable goods than of nondurable
goods and services.
aItems
Spectator and entertainment
Computers, toys, and games
Tobacco and alcohol
Reading material
Small gifts and accessories
Subtotal, nondurables
Durable goods and services
Furniture and appliances
Home entertainment
Sporting and recreation
Recreational vehiclesa
New cars and trucksb
Used cars and trucks
Motor homes and trailers
Vacations
Subtotal, durablesd
Overall total, all product classes
Totals for all product classes
Product class
Nondurable goods and services
Food from grocery stores
Food from restaurants
Household supplies
Clothing
Gas, diesel, and propane
Dental and optical products
Medicine and pharmacy products
Total
expenditure ($)
Table 7. Expenditures and percent leakage by product class for the Bas-Saint-Laurent Model Forest, based on the Canadian Forest Service household expenditure
survey
Table 8. Percent household expenditure leackage by community for the Bas-Saint-Laurent Model Forest, based on the
Canadian Forest Service household expenditure survey
Community
% leakage
Community
Auclair
60.9
Saint-Cléophas
Biencourt
65.3
Saint-Eugène-de-Ladrière
94.0
Dégelis
47.5
Saint-Irène
97.1
Esprit-Saint
96.2
Saint-Juste-du-Lac
92.7
La Rédemption
93.1
Saint-Médard
63.7
La Trinité-des-Monts
89.8
Saint-Narcisse-de-Rimouski
95.7
Lac-des-Aigles
74.0
Saint-Valérien
90.1
Notre-Dame-du-Lac
50.0
Saint-Zénon-du-Lac-Humqui
74.1
Saint-Charles-Garnier
95.7
Bas-Saint-Laurent Model Forest
72.8
100.0
destination communities. Some respondents
visited no destination communities for their
consumer purchases, whereas others traveled
to as many as 11 destination communities to
satisfy their consumer needs.
of total spending on these products (64.8%). In
contrast, 86.5% of spending on durable products
occurred outside the community of residence.
This suggests that overall percent expenditure
leakage was strongly influenced by spending on
durable products. Percent leakage for clothing
(97.0%) was higher than for any other product
class, except vacations, for which all spending
was assumed to occur outside the community
of residence. Percent leakage was slightly lower
(96.9%) for motor homes and trailers and
was 92.0% for computers, toys, and games.
Purchases made over the Internet amounted to
0.6% of total expenditures made outside of the
home community.
The household expenditure survey also posed
questions about the reasons for decisions
about where to purchase nondurable and
durable goods. For both nondurable goods
and services and durable goods, location was
cited as the dominant reason by the largest
proportion of respondents (39.7% and 36.1%,
respectively) (Table 9). This probably reflects the
inconvenience and cost involved in traveling to
another community to make similar purchases.
Also, 21.2% of the respondents indicated
that they made local purchases of nondurable
goods and services to support local businesses.
Support for local businesses was only 12.2% for
purchases of durable goods, such as automobiles
and furniture, reflecting the greater percent
leakage for this category of purchases relative
to nondurable goods. Price was the second most
prominent reason for choice of where to buy
durable goods, likely reflecting the higher cost
of durables in relation to nondurables. For both
nondurable goods and services, price was the
third most important reason, after location and
support for local businesses.
Percentage expenditure leakage by community
varied between 47% and 100% (Table 8),
with leakage closely tied to the community’s
population. The larger communities of Dégelis
and Notre-Dame-du-Lac exhibited the lowest
percent leakage values (47.5% and 50.0%,
respectively), whereas the smaller communities
of Saint-Cléophas and Sainte-Irène exhibited the
highest percent leakages (100.0% and 97.1%,
respectively).
The
most
frequently
visited
destination
community was Rimouski, followed by Rivière-duLoup, Edmundston, Cabano, Amqui, and Dégelis.
Survey respondents traveled on average to three
destination communities to meet their consumer
needs, purchasing items from an average of 12
and a maximum of 19 product classes in the
% leakage
14
NOR-X-418
Table 9. Main reasons for decisions about where to purchase goods and services
% of respondents
Reason for purchase
Nondurable goods
Durable goods
Price
17.2
20.2
Location
39.7
36.1
Selection
16.4
19.2
Support local businesses
21.2
12.2
5.4
12.2
Other
ANSWERING THE RESEARCH QUESTIONS:
LEAKAGE MODELING AND DEPENDENCE INDICES
Leakage Modeling
were expected to influence percent leakage:
“age,” the age of the respondent, “edu,” the
respondent’s highest level of education; “income,”
the total household income; “famtype,” 1 of 10
family types from the Statistics Canada survey
of household expenditures (Statistics Canada
1998); “unemp,” the number of unemployed
adults in the household; “hpop,” the population
of a respondent’s home community; “occup,” the
sector of occupation providing the majority of
the household’s income; “dpop,” the population
of destination communities; and “distance,”
the distance between the respondent’s home
community
and
the
various
destination
communities.
Methods
The general form of the relationship between
percent expenditure leakage and the explanatory
variables examined in the survey is described
below. The independent variables in the
household expenditure survey for model forests
were a mixture of ordinal, categorical, and
continuous variables, whereas the dependent
variable (experc, percent leakage for all goods)
was continuous, ranging between 0% and
100% (Table 3). A percent leakage value near
100% indicates that the majority of household
expenditures are made outside the home
community. The following independent variables
experc = β0 + β1edu + β2age + β3income + β4famtype + β5unemp + β6 hpop + (3)
β7occup + β8dpop + β9distance + ε
where
ß = estimated parameters or coefficients, and
ε = the model’s stochastic or random component.
15
NOR-X-418
Results and Discussion
significant in explaining percent expenditure
leakage. This is not surprising, given the need to
travel greater distances to purchase household
goods and services and consumers’ awareness of
the high travel costs associated with purchases
made in distant locations. As such, distance was
expected to dampen percent household leakage,
as indicated by the negative sign of the DISTKM2
variable. The marginal effect of the DISTKM
and DISTKM2 variables on the dependent
variable was to decrease the rate of increase
in percent expenditure leakage. Although an
increase in travel distance dampened percent
expenditure leakage in this study, the reverse
may be true elsewhere in Canada, where only
a limited number of rather distant destination
communities are reasonably available; in that
situation, residents might travel to the most
distant but largest destination community
because its choice of price and selection options
outweighs the higher travel costs.
The descriptions of basic and derived variables
used in the model of percent household
expenditure leakage for all goods and services are
reported in Table 10. The results of the ordinary
least squares regression model for percent
expenditure leakage are shown in Table 11. The
independent variables listed in Table 11 have
been scaled, as described in Table 10, so that
the model’s estimated coefficients are the same
order of magnitude. Because leakage spending
outside of the model forest far exceeded leakage
spending in other model forest communities,
and the explanatory variables would therefore
be generally more applicable to spending outside
of the region, the model was applied to total
percent leakage (experc) only.
The two distance variables (DISTKM, which is the
sum of all distances traveled to make purchases
across the product classes, and DISTKM2, which
is the square of the DISTKM variable) were
Table 10.
Descriptions of basic and derived variables used in the model of percent household expenditure leakage
for all goods and services
Variable name
and units
Variable description
Mean
Standard
deviation
DISTKM (km)
Sum of distances/1000a
0.787
0.691
DISTKM2
Square of DISTKM/10
2.58b
0.363
AVDIST (km)
Average of expenditure-weighted distances/100a
0.620
0.451
7.85c
AVDIST2
Square of AVDIST
HPOP
Population of home community/100
HPOP2
Square of HPOP/100
5.50d
4.45
INCOME
Total household income
3.42
1.69
SPENDPOP ($)
Average of nonlocal spending weighted by population of
destination communities/1000e
0.934
0.326
SPENDPOP2
Square of SPENDPOP
0.979
0.709
ED2–ED8
Dummy variables for education levels 2 to 8
NAf
NA
FT2–FT10
Dummy variables for family types 2 to 10
NA
NA
EDL
Cluster dummy variable for Est du Lac Témiscouata model
forest area
NA
NA
NR
Cluster dummy variable for Nicolas Riou model forest area
NA
NA
13.75
1.30
12.05
aDistances
include duplicate distances.
value; mean was 0.11.
value; mean was 0.59.
dMedian value; mean was 3.34.
eCommunities include duplicate communities.
fNA = not applicable.
bMedian
cMedian
16
NOR-X-418
Table 11.
Results of ordinary least squares regression model of percent household expenditure leakage for all goods
and servicesa
Variable nameb
Estimated
coefficient
Standard error
DISTKM
DISTKM2
AVDIST
AVDIST2
HPOP
HPOP2
INCOME
SPENDPOP
SPENDPOP2
ED2
ED3
ED4
ED5
ED6
ED7
ED8
FT2
FT3
FT4
FT5
FT6
FT7
FT8
FT9
FT10
EDL
NR
CONSTANT
64.600
–67.341
–72.898
12.072
–1.329
2.560
–1.809
63.063
–14.784
–18.167
–22.900
–19.482
–21.856
–23.258
–24.644
–26.215
–6.355
–3.849
–6.011
–3.249
–2.171
–6.962
–5.898
–4.506
11.165
–9.423
4.818
69.160
3.016
5.537
5.528
1.873
0.304
0.781
0.392
9.073
3.936
6.166
6.052
6.007
6.154
6.134
6.158
6.636
3.154
2.486
2.629
2.847
3.393
3.628
2.818
3.480
4.521
2.158
1.623
8.575
t ratio
(df = 471)
p value
21.420
–12.160
–13.190
6.445
–4.369
3.329
–4.616
6.951
–3.756
–2.946
–3.784
–3.243
–3.552
–3.792
–4.002
–3.950
–2.015
–1.548
–2.286
–1.141
–0.640
–1.919
–2.093
–1.295
2.469
–4.366
2.968
8.065
<0.001
<0.001
<0.001
<0.001
<0.001
0.001
<0.001
<0.001
<0.001
0.003
<0.001
0.001
<0.001
<0.001
<0.001
<0.001
0.044
0.122
0.023
0.254
0.523
0.056
0.037
0.196
0.014
<0.001
0.003
<0.001
R2 (adjusted) = 0.81. Log-likelihood = –1 933.63.
a
bVariables
described in Table 10.
DISTKM2 variables when used by themselves
in the model. However, they worked in concert
with the DISTKM and DISTKM2 variables (with
opposite signs) to greatly increase the model’s
ability to explain the variation in the dependent
variable (experc). The combined use of these
two sets of distance variables resulted in a better
model than those that resulted from using either
set on its own. This suggests that the effect of
distance on percent expenditure leakage was
driven in part by where most expenditures were
made, but this effect was modulated by the
distances traveled to all destination communities
to make purchases across all of the product
classes.
The AVDIST variable was the average of all
distances weighted by expenditures. AVDIST and
its square, AVDIST2, had an interesting effect on
percent expenditure leakage, with the destination
communities where respondents made the
greatest expenditures, such as Rimouski,
influencing this variable. In fact, Rimouski
accounted for 38% of all household expenditures
made outside of the home community, and the
top three destination communities (Rimouski,
Rivière-du-Loup, and Edmundston) accounted
for over 53% of all expenditures made outside
of the home community. The AVDIST and
AVDIST2 variables had the same effect on
percent expenditure leakage as the DISTKM and
17
NOR-X-418
The distances used in the regression model
included the use of duplicate distances,
which would occur when two or more product
classes were usually purchased in the same
destination community. Duplicate distances for
different product classes were common among
respondents to the household expenditure survey.
In contrast, using nonduplicate distances would
mean using only one distance per combination of
home and destination community. The total and
average nonduplicate distances to destination
communities were, interestingly, insignificant
in the ordinary least squares regression model,
which suggested that DISTKM, itself a sort of
weighted distance, may reflect the multitrip
nature of purchase behavior.
The variable SPENDPOP, the mean weighted perproduct spending outside of the community of
residence, is the product of spending outside
the home community and the population of
the destination communities, weighted by the
total population of duplicate and nonduplicate
destination communities. The effects of this
interaction variable and its square (SPENDPOP2)
in the final model (Table 11) accentuated the
observation that greater spending generally
occurred in destination communities with
commensurately higher populations. This result
is not surprising, given that larger centers
generally offer a greater number and variety of
outlets for goods and services, particularly for
durable goods.
The effect on the model of the two variables for
the population of the home community, HPOP
and HPOP2 (the square of HPOP), confirmed the
hypothesis that larger home communities would
have lower percent household leakage. This
finding is logical because larger communities can
be expected to offer a greater range of goods and
services than smaller communities. The positive
sign for the HPOP2 coefficient (Table 11) indicates
that the rate of decrease in percent expenditure
leakage increased as the population of the home
community increased. The sum of the populations
of destination communities (including duplicate
communities) was a significant indicator of the
greater purchase opportunities available in larger
centers in earlier models, but it was insignificant
in the final model.
Level of educational attainment has long been
recognized as an indicator of well-being through
its effect on income and other aspects of quality
of life. All of the seven levels of attained education
that were incorporated as dummy variables,
relative to the category “never attended school”
(Table 6), were significant regressors in the final
model (Table 11). The increasing magnitude of
the estimated coefficients from the lowest to the
highest level of education attained suggested
that percent expenditure leakage declined with
greater educational attainment.
Family type (Table 4) was included as an
independent variable in this study because
Statistics Canada’s survey of household
expenditures (Statistics Canada 1998) clearly
showed that household expenditures vary
considerably across family types (other things
being equal). Although not all of the individual
family types were significant, family type was
collectively strongly significant in explaining the
variation in percent expenditure leakage. The
FT10 family type (four adults with any number
of children) was the only family type with a
positive coefficient. Perhaps some members of
large families had more time or resources than
members of smaller families to shop outside of
the home community. All family-type dummy
variables in the model were estimated in relation
to the FT1 family type (a single man or woman
under 45 years of age).
It was expected that percent expenditure
leakage would increase with income, because
higher incomes would permit greater spending
on travel. However, the coefficient for income,
although significant, had an effect contrary to
expectations, its negative sign indicating that
percent expenditure leakage decreased with
increasing income (Table 11). It thus appears
that the opportunity cost of traveling was quite
high. Respondents would rather spend their
time doing something other than traveling
to destination communities to satisfy their
consumer needs, assuming that prices for goods
and services were higher in the home community
than in destination communities, most of which
lie outside of the model forest region. Yanagida et
al. (1991) also found that income was negatively
correlated with leakage (i.e., higher income
contributed to lower expenditure leakage).
Interestingly, no occupational groups were
significant in explaining percent expenditure
leakage, which suggested that occupation had
little or no influence on expenditure leakage,
18
NOR-X-418
average population of a community in the Est
du Lac Témiscouata cluster was 1 039 people,
whereas it was 399 people in the Lac-Metis
cluster. The use of the cluster dummy variables
resulted in a much better model than would have
been the case had the population of destination
communities been used as a regressor.
except through income. It was hypothesized that
some occupations might have been associated
with higher leakage than others because they
afforded opportunities to make purchases for the
household while working away from the home
community. An additional hypothesis that the
number of unemployed adults in a family would
be a significant regressor was based on the
assumption that unemployed family members
would have sufficient time on their hands to
shop outside the home community for bargains
on consumer goods and services. However, this
hypothesis was rejected on the basis of the
regression results. This phenomenon is discussed
further in the “Results and Discussion” portion of
the section on dependence indices, below.
As noted earlier, percent expenditure leakage
was higher for durable goods than for
nondurable goods and services. Initial models
for expenditure leakage for all products indicated
that the percent expenditure leakage for durable
goods was a significant driver of overall percent
expenditure leakage. A model with percent
expenditure leakage for durable goods relative
to total expenditure on all durable goods as the
dependent variable was therefore investigated
to determine any differences between this model
and the overall model (Tables 12 and 13).
Other regressors shown in Table 11 that were
significant in explaining percent expenditure
leakage were dummy variables indicating which
of three clusters of communities associated with
the model forest areas was the respondent’s place
of residence (Fig. 1). The three clusters were
the communities of Auclair, Biencourt, Dégelis,
Lac-des-Aigles, Notre-Dame-du-Lac, Saint-Guy,
Saint-Juste-du-Lac, and Saint-Médard for the Est
du Lac Témiscouata model forest area; EspritSaint, La Trinité-des-Monts, Saint-Eugènede-Ladrière, Saint-Narcisse-de-Rimouski, and
Saint-Valérien for the Nicolas-Riou model forest
area; and La Rédemption, Saint-Charles-Garnier,
Saint-Cléophas, Sainte-Irène and Saint-Zénondu-Lac-Humqui for the Lac-Metis model forest
area. Residents of a community in the Est du Lac
Témiscouata model forest area were predicted to
have household expenditure leakage 9% lower
than that of residents of the Lac-Metis model
forest area, whereas residents of the Nicolas
Riou model forest area were predicted to have
expenditure leakage 5% higher than that of
residents of the Lac-Metis model forest area. The
actual expenditure leakage from each cluster
was 56% from the Est du Lac Témiscouata
cluster of communities, 93% from the Nicolas
Riou cluster, and 92% from the Lac-Metis cluster.
The location of each of the clusters with respect
to major urban communities and the average
population of the communities making up each
cluster may each play a role in this outcome.
Communities in the Nicolas Riou cluster are
relatively close to Rimouski, which would make
trips for consumer purchases less costly than
trips from communities in the other clusters. The
The results of this model were generally similar
to those of the all-products or overall model,
but there were differences for a few of the
variables. Two dummy variables for occupational
groups, MINING (for mining) and TRAP (for
transportation), were significant in the model
for durable goods (Table 13) but not in the
overall model. The positive signs of the MINING
and TRAP coefficients may reflect the greater
opportunities that people employed in these
sectors have to make purchases for the home
while they are working away from home.
The populations of destination communities
were not significant in any of the models for
durable goods that preceded the final model,
but this variable was significant in some of the
earlier versions of the all-products model. This
pattern may have occurred because a greater
proportion of durable goods were purchased in
communities outside of the BSLMF (73.3% of
expenditures on durable goods but only 60.8%
of expenditures on nondurable goods and
services were made outside the model forest).
Most of the communities outside the BSLMF may
have population sizes above the threshold at
which a community has an adequate selection
of durable goods available for purchase. The
average population of surveyed and destination
communities within the BSLMF was 1 061 people,
whereas the average population of destination
communities outside of the model forest
(other than Montréal) was 16 100 people. This
19
NOR-X-418
were individually and collectively insignificant in
the model for durable goods, whereas they were
individually and collectively significant in the
all-products model. Family type and education
clearly had a greater influence on the purchase
of nondurable goods and services than on the
purchase of durable goods, which may reflect a
general lack of selection or availability of durable
goods in the home community. Residents may
simply have no choice but to purchase most
durable goods outside of the home community, as
indicated by the greater leakage of expenditures
for durable goods relative to nondurable goods
and services (Table 7).
substantial difference in average populations
may also explain, to some extent, why the
percent expenditure leakage for all products to
BSLMF communities was only 7.7%, whereas
the percent leakage to communities outside the
model forest was 65.1% (Table 7). The variables
for both the Est du Lac Témiscouata and Nicolas
Riou clusters were significant in the model for
durable goods, as well as in the all-products
model; however, the variable for the Est du Lac
Témiscouata cluster had a positive sign in the
model for durable goods but a negative sign in
the all-products model. It is also noteworthy that
family types and levels of education achieved
Table 12.
Descriptions of basic and derived variables used in the model of percent household expenditure leakage
for durable goodsa
Variable name
and units
DISTKM (km)
DISTKM2
AVDIST (km)
AVDIST2
SPENDPOP ($)
SPENDPOP2
AG
FOREST
CONS
FINPROF
OILGAS
GOVT
MINING
SERV
TRAP
TRAF
Variable description
Mean
Standard
deviation
Sum of all distances/1000b
Square of DISTKM/10
Average of expenditure-weighted distances/100b
Square of AVDIST/10
Average of nonlocal spending, weighted by population
of duplicate and nonduplicate destination
communities/1000f
Square of SPENDPOP
Dummy variable for agriculture sector
Dummy variable for forestry sector
Dummy variable for construction sector
Dummy variable for financial and professional services
sector
Dummy variable for oil and gas (energy) sector
Dummy variable for government sector
Dummy variable for mining sector
Dummy variable for service sector
Dummy variable for transportation sector
Dummy variable for transfers sector
0.348
2.60c
2.94d
1.72e
0.380
0.36
0.67
0.32
0.944
0.390
1.045
NAg
NA
NA
NA
0.887
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
aVariables
that have the same names and the same means and standard deviations as the variables shown in Table 10 are
not presented in this table
bDistances include duplicate distances.
cMedian value; mean was 0.11.
dMedian value; mean was 0.66.
eMedian value; mean was 0.088.
fCommunities include duplicate communities.
gNA = not applicable.
20
NOR-X-418
Table 13.
Results of ordinary least squares regression model of percent household
expenditure leakage for durable goodsa
Variable
nameb
Estimated
coefficient
DISTKM
DISTKM2
AVDIST
AVDIST2
HPOP
HPOP2
INCOME
SPENDPOP
SPENDPOP2
AG
FOREST
CONS
FINPROF
OILGAS
GOVT
MINING
SERV
TRAP
TRAF
EDL
NR
CONSTANT
71.909
–137.250
–28.000
23.502
–3.176
6.841
–1.340
94.920
–30.467
2.445
0.732
6.220
3.747
–1.696
–2.467
8.727
0.719
6.988
2.827
3.971
6.781
41.615
Standard
error
7.336
30.740
4.924
11.134
0.374
0.967
0.451
5.494
2.358
3.132
2.482
3.874
3.133
10.290
2.904
4.118
3.218
3.304
2.475
2.173
2.013
4.525
t-ratio
(df = 477)
p-value
9.8030
–4.465
–5.686
2.072
–8.480
7.077
–2.970
17.280
–12.920
0.781
0.295
1.607
1.196
–0.165
–0.849
2.119
0.223
2.115
1.142
1.828
3.369
9.197
<0.001
<0.001
<0.001
0.039
<0.001
<0.001
0.003
<0.001
<0.001
0.435
0.768
0.109
0.232
0.869
0.396
0.035
0.823
0.035
0.254
0.068
0.001
<0.001
R2 (adjusted) = 0.72. Log-likelihood = –2 018.36.
a
bMost
variables described in Table 12.
Dependence Indices
Household expenditure analysis provides detailed
information about spending leakages from a
local or regional economy. It has the potential to
account for the effects of increased or decreased
investment in particular sectors of the economy
(Davis and Hutton 1992). As the models of
percent leakage expenditure described above
have shown, a portion of the income earned
within a regional economy will leak from that
economy in the form of expenditures on goods
and services, with the magnitude of the leakage
being affected by a number of characteristics,
including income and distances to surrounding
communities.
Methods
A community’s dependence on a particular
sector is commonly measured by allocation of
employment or income. However, in this study,
it has been argued that the location and amount
of consumer household expenditures represent
other measures of a community’s dependence
on a particular sector. Analysis of household
expenditures has been ignored in the community
dependence literature because it requires
detailed information about the geographic
location of expenditures and about the amount
of expenditure occurring within and outside
the region of interest. Such analysis is both
time-consuming and expensive; however, data
from the Canadian Forest Service household
expenditure survey allowed this type of analysis
for the BSLMF region.
The method used in this study to calculate the
dependence indices based on measurements
of employment, income, and household
expenditures is founded on the work of Fletcher
et al. (1991) and Korber et al. (1998). More
specifically, the location quotient method was
21
NOR-X-418
sector employment that is estimated to be basic
is calculated as follows:
employed to estimate the economic base for
various sectors of the economy. The current
study used the method of Korber et al. (1998)
for calculating location quotients. The location
quotient (LQ) of community j for industry i was
calculated as follows:
Fletcher et al. (1991) and Korber et al. (1998)
used equations 5 through 7 to estimate a forest
dependence index using census data. The most
recent economic base data relevant to the BSLMF
can be obtained from the analysis of major
sectors of employment by Korber et al. (1998).
(4)
where E is employment, T is all sectors, and P is
the entire province.
Table 14 provides employment and calculated
base employment by economic sector for the
communities in the BSLMF region. The source of
the raw employment data was the 1996 Canada
Census. The values for provincial output, import,
and export variables that were used to adjust
provincial employment benchmarks (equation 4)
were from Statistics Canada’s 1996 input–output
tables. For example, a total of 1 445 people were
employed in the service industry, of whom only
173 accounted for base employment. Therefore,
1 272 of the service jobs served local needs. In
contrast, there were 1 195 jobs in the forestry
sector, and nearly all of them (1 114) were
counted as base employment, which means that
the forestry sector produced outputs primarily
for the export market.
Equation 4 assumes no net exports or imports
and no inventories. If the province is a net
exporter, location quotients overestimate the
level of employment necessary to provide
for local consumption at the community level
and consequently underestimates the level of
community base employment. Conversely, if the
province is a net importer, the level of community
base employment will be overestimated
(Schwartz 1982; Korber et al. 1998). Therefore,
provincial benchmarks used in the calculation
of location quotients had to be adjusted to
reflect only the output required to meet local or
regional consumption. The adjusted benchmark
employment for use in the equation for a location
quotient is as follows:
From these calculations, an employment-based
sector dependence index can be derived, where
is base employment in sector i for community
j and
is the total base employment in
community j. The employment-based sector
dependence index (ESDI) was calculated as
follows:
(5)
where
is the total provincial output
from industry i,
is the provincial export from
industry i,
is the provincial import from
industry i,
is the provincial employment in
industry i, and * denotes variables of special or
changed significance. The equation for a location
quotient can be rewritten, with the modifications
presented above, as follows:
(6)
(8)
Income-based
sector
dependence
indices
were estimated from average, and not total,
household income per sector to facilitate their
comparison with expenditure-based sector
dependence indices, which are determined from
population samples. The income-based sector
dependence index for each community and
sector was determined by dividing the product of
average base income and base employment for
Base employment is considered to be employment
in a given sector above the provincial average.
The provincial average (that is, the average
across all CSDs in the province) is assumed to
be the sector employment required to serve
local needs in the community. The proportion of
(7)
22
NOR-X-418
23
NOR-X-418
60
0
10
35
0
0
0
0
0
55
65
30
10
20
0
20
35
0
55
25
0
250
0
20
20
560
20
25
15
0
10
20
60
20
140
0
20
20
65
1 195
90
40
80
30
50
0
25
50
30
105
35
35
90
285
75
55
55
0
215
20
20
0
10
25
0
0
0
0
70
0
10
15
35
0
10
0
0
10
0
0
0
0
0
0
0
0
0
0
0
0
0
10
0
0
0
35
915
50
60
30
0
0
0
15
20
0
295
20
45
25
250
15
30
25
0
55
10
20
0
0
0
0
0
0
0
0
0
0
0
0
0
10
15
25
1 445
115
75
20
20
35
0
30
40
0
280
20
90
70
455
55
85
30
10
320
40
45
10
0
0
0
0
0
0
45
0
10
15
80
20
45
0
160
3 675
235
150
270
120
135
0
70
150
115
505
55
115
135
885
120
215
240
0
170
20
10
10
0
20
0
0
0
0
45
0
0
0
55
0
10
0
315
8 810
635
500
455
180
320
0
160
280
175
1 470
150
385
395
2 210
285
490
405
Professional
and
Oil and
financial
gas
Agriculture Construction Forestry services (energy) Government Mining Service Transportation Transfers Other Total
Raw and basic employment, by sector and community, in the Bas-Saint-Laurent Model Forest
Raw employment
Auclair
Biencourt
Dégelis
Esprit-Saint
Lac-des-Aigles
La Rédemption
La Trinité-desMonts
Notre-Damedu-Lac
Saint-CharlesGarnier
SaintCléophas
Saint-Eugène
-deLadrière
Saint-Guy
Sainte-Irène
Saint-Justedu-Lac
Saint-Médard
Saint-Narcisse
-deRimouski
Saint-Valérien
Saint-Zénon
-du-LacHumqui
Total
Community
Table 14.
24
NOR-X-418
0
0
0
39
28
0
45
17
0
105
25
0
0
15
0
18
15
363
1
6
13
22
0
0
18
25
8
12
0
0
0
7
48
14
73
0
5
14
62
1 114
84
38
76
29
48
0
24
47
28
88
34
32
87
262
72
51
52
0
1
0
0
0
0
1
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
106
0
2
0
0
0
0
8
0
0
84
4
8
0
0
0
0
0
0
48
8
18
0
0
0
0
0
0
0
0
0
0
0
0
0
8
14
0
173
5
9
0
6
3
0
15
3
0
8
0
23
12
77
10
2
0
0
71
14
24
0
0
0
0
0
0
0
0
0
0
0
0
8
25
0
58
840
29
0
121
62
33
0
18
61
60
29
6
0
7
166
27
56
107
0
9
0
0
0
0
0
0
0
0
0
0
0
0
0
3
0
6
0
135
2 830
185
147
225
97
125
0
83
118
113
244
62
119
132
581
117
153
194
Professional
and
Oil and
financial
gas
Agriculture Construction Forestry services (energy) Government Mining Service Transportation Transfers Other Total
Concluded
Base
employment
Auclair
Biencourt
Dégelis
Esprit-Saint
Lac-des-Aigles
La Rédemption
La Trinité-desMonts
Notre-Damedu-Lac
Saint-CharlesGarnier
SaintCléophas
Saint-Eugène
-deLadrière
Saint-Guy
Sainte-Irène
Saint-Justedu-Lac
Saint-Médard
Saint-Narcisse
-deRimouski
Saint-Valérien
Saint-Zénon
-du-LacHumqui
Total
Community
Table 14.
dependence index. Expenditure-based measures
alter the employment- or income-based measure
of community dependence and may provide a
more accurate picture of sectoral contributions
to the local economies of the BSLMF with respect
to expenditures.
sector i in community j by the product of average
base income and base employment for all base
workers. The income-based sector dependence
index (ISDI) by community was calculated as
follows:
(9)
Results and Discussion
The sector dependence indices for the
communities within the BSLMF are shown in
Table 15 (the community of Saint-Cléophas is
not included because no local expenditures were
made by survey respondents). For example, the
base employment sector dependence index of
0.105 for agriculture in Biencourt (column 2)
means that 10.5% of Biencourt’s base
employment was in the agriculture sector. The
income-based sector dependence index of 0.012
(column 3) means that 1.2% of the income
received by employees across all sectors making
up Biencourt’s economic base came from the
agriculture sector. Because 1.20% is less than
10.5%, it can be concluded that income from
the agriculture sector was less than the average
income in Biencourt. The community of Biencourt
therefore seems to be less dependent on the
agriculture sector than the base employment
index would suggest. The providers of consumer
goods and services within the community must
therefore rely on other economic sectors if they
are to generate an income around the community
average, assuming that the average income
would be the minimum required to persuade
providers of consumer goods and services to
remain or invest in the community. However, the
expenditure-based sector dependence index of
0.142 (last column of Table 15) indicates that
14.2% of total local expenditures on household
goods and services made by employees of
Biencourt’s economic base industries came from
the agriculture sector. Because 14.2% is greater
than both 10.5% and 1.20%, it can be concluded
that an above-average proportion of agricultural
income was spent within Biencourt. This also
means that Biencourt was more dependent on
the agriculture sector than the employmentand income-based sector dependence indices
indicate, because of low expenditure leakage
from the community for purchases made by
residents working in the agriculture sector. The
agriculture sector displayed this characteristic
for more communities than any other sector
The approach used for the employment- and
income-based sector dependence indices can
also be used to develop a sector dependence
index based on local household expenditure data
(LESDI), calculated as follows:
(10)
where
is the average local annual expenditure
per household for sector i in community j.
Equation 10 indirectly accounts for expenditure
leakage by using expenditures made in the
community of residence (local expenditure)
rather than total household expenditures.
All of the sector dependence indices are
interpreted in the same manner: the level of
dependence decreases as the index approaches
0 and increases as it approaches 1. If a sector
has an index value of 1, the community or region
has 100% dependence on that particular sector.
Expenditure-based indices will be greater than
employment-based indices for sectors in which
employees spend an above-average proportion
of their income within their own community.
Conversely, local expenditure-based indices
will be lower than employment-based indices
for sectors in which employees spend a belowaverage proportion of their income within their
own community. Sectors with above-average
income will have higher values for the incomebased sector dependence indices, and possibly
higher values for the local expenditure-based
sector dependence indices, than those based
on employment data because the income and
expenditure indices are weighted by incomes
and expenditures, respectively. The incomeand expenditure-based sector dependence
indices provide ordinal rankings of income
and expenditures with respect to averages
established by the employment-based sector
25
NOR-X-418
may have more time to shop than people in other
sectors, may be an important factor influencing
high expenditure leakage in this sector. The high
leakage for the agriculture and forestry sectors
is more difficult to explain. Perhaps household
purchases were frequently combined with travel
to work in private woodlots, a predominant
form of forestland tenure in the model forest.
This proclivity to purchase household goods and
services outside of the home community may be
attributable to finer subsets of product classes
than were examined in this work, as exemplified
by previous research. For example, Pinkerton et
al. (1995) observed that households with jobholders working outside of the local area were
more likely to shop outside of the community
for certain goods and services; in addition,
Cowell and Green (1994) noted that purchases
of subsets of the complete range of products
and services were differentially influenced by the
same socioeconomic variables such as income
and education level. Cowell and Green (1994)
also investigated the importance of place, in
terms of community attachment, in explaining
the location of household spending.
(Table 16). If the values for the employment- and
income-based sector dependence indices were
reversed, it could be concluded that income in
the agriculture sector was greater than average
income in Biencourt; however, because the
expenditure-based sector dependence index is
still greater than either of these indices, it would
still be concluded that a below-average proportion
of income was spent outside of Biencourt. The
government sector in Notre-Dame-du-Lac is an
example of this pattern (Table 15).
The employment-based sector dependence
index for forestry in Saint-Valérien exhibited a
different pattern. Although forestry accounted for
25.5% of base employment in this community’s
economic base, the higher value for the incomebased index (0.275 or 27.5%) suggests that
income in the forestry sector was higher than
average income. However, the expenditure index
(0.130 or 13.0%) was lower than both of the
other indices, which suggests that Saint‑Valérien
was less dependent on the forestry sector
than indicated by the employment- and
income-based methods, because of significant
expenditure leakage from the community
through expenditures by employees in the
forestry sector. In other words, a below-average
proportion of forestry income was spent within
Saint‑Valérien. This type of ordinal ranking, with
the local expenditure-based index the lowest of
the three sector dependence indices, accounted
for 35% of all combinations (Table 16). As such,
a high proportion of communities in the BSLMF
were less dependent on their various economic
sectors than is indicated by the employment and
income methods because of high expenditure
leakage. However, the agriculture, forestry, and
transfers sectors, the latter of which consists of
Canada Pension Plan benefits, social assistance
payments, and investment income, contributed
heavily to this observation. These sectors had
the highest concentration of expenditure-based
sector dependence index values that were
lower than those for the employment-based
and income-based sector dependence indices
(Table 16). The inclusion in the transfers sector
of many retired and semiretired individuals, who
The expenditure index for the construction
sector in Auclair (Table 15) provides another
interpretation of the sector dependence indices.
For that sector, the expenditure-based index
(0.061 or 6.1%) fell between the corresponding
employment- and income-based indices (0.064
[6.4%] and 0.033 [3.3%], respectively). An
income index lower than the corresponding
employment index suggests that a belowaverage proportion of construction income was
spent within Auclair; however, the expenditurebased index indicates that Auclair was less
dependent on the construction sector than
the employment-based index suggests but
more dependent than the income-based index
indicates. This can be interpreted to mean that
the proportion of construction income spent
locally was about average with respect to
other economic sectors in the community. The
medium ranking of the expenditure-based sector
dependence index accounted for a third (33%)
of the ordinal patterns in the BSLMF (Table 16).
26
NOR-X-418
Table 15.
Sector dependence indices for communities in the Bas-Saint-Laurent
Model Forest, based on three methods of calculating dependence
Sector; sector dependence indexa
Community and sector
Auclair
Agriculture
Forestry
Construction
Government
Service industry
Biencourt
Agriculture
Forestry
Construction
Service industry
Transfers
Dégelis
Agriculture
Forestry
Service industry
Transfers
Other
Esprit-Saint
Forestry
Service industry
Transportation
Transfers
La Rédemption
Agriculture
Forestry
Construction
Mining
Transfers
La Trinité-des-Monts
Agriculture
Forestry
Government
Transfers
Lac-des-Aigles
Agriculture
Forestry
Mining
Service industry
Transportation
Transfers
Other
Employment
Income
Expenditure
0.408
0.269
0.064
0.071
0.189
0.697
0.285
0.033
0.000
0.024
0.445
0.248
0.061
0.067
0.179
0.105
0.658
0.093
0.093
0.051
0.012
0.945
0.019
0.004
0.020
0.142
0.688
0.060
0.073
0.037
0.126
0.451
0.133
0.285
0.005
0.041
0.569
0.028
0.362
0.000
0.131
0.460
0.131
0.273
0.005
0.611
0.087
0.070
0.232
0.731
0.022
0.006
0.241
0.758
0.065
0.052
0.126
0.071
0.268
0.035
0.072
0.554
0.007
0.073
0.003
0.020
0.897
0.103
0.471
0.008
0.158
0.260
0.287
0.554
0.057
0.102
0.000
0.801
0.040
0.160
0.211
0.700
0.031
0.059
0.034
0.332
0.053
0.016
0.161
0.363
0.041
0.006
0.375
0.019
0.000
0.096
0.502
0.002
0.100
0.407
0.147
0.013
0.217
0.084
0.031
27
NOR-X-418
Table 15.Continued
Sector; Sector dependence indexa
Community and sector
Employment
Income
Expenditure
0.091
0.360
0.052
0.344
0.032
0.120
0.060
0.416
0.011
0.424
0.013
0.077
0.110
0.258
0.023
0.483
0.041
0.093
0.220
0.251
0.529
0.116
0.137
0.747
0.225
0.261
0.514
0.122
0.390
0.200
0.005
0.020
0.435
0.198
0.000
0.017
0.620
0.214
0.005
0.020
0.263
0.000
0.347
0.093
0.051
0.124
0.338
0.538
0.010
0.202
0.788
0.084
0.771
0.145
Forestry
0.302
Service industry
0.059
Transfers
0.639
Saint-Narcisse-de-Rimouski
Agriculture
0.244
Forestry
0.454
Mining
0.045
Service industry
0.026
Transportation
0.075
Transfers
0.156
Saint-Valérien
Agriculture
0.117
Forestry
0.255
Construction
0.264
Government
0.011
Mining
0.123
Service industry
0.063
Transportation
0.166
0.085
0.006
0.908
0.224
0.004
0.772
0.147
0.582
0.005
0.003
0.032
0.230
0.059
0.436
0.310
0.070
0.069
0.056
0.047
0.275
0.479
0.007
0.078
0.000
0.114
0.046
0.130
0.212
0.013
0.099
0.054
0.445
Notre-Dame-du-Lac
Agriculture
Forestry
Construction
Government
Service industry
Transfers
Saint-Charles-Garnier
Agriculture
Forestry
Transfers
Saint-Eugène-de-Ladrière
Agriculture
Forestry
Construction
Professional or
financial services
Service industry
Transfers
Saint-Juste-du-Lac
Agriculture
Forestry
Transfers
Saint-Médard
28
NOR-X-418
Table 15.Concluded
Sector; Sector dependence indexa
Community and sector
Employment
Saint-Zénon-du-Lac-Humqui
Agriculture
0.112
Forestry
0.462
Transfers
0.427
Sainte-Irène
Agriculture
0.213
Forestry
0.289
Government
0.099
Service industry
0.179
Transfers
0.220
Bas-Saint-Laurent Model Forest
Agriculture
0.128
Forestry
0.394
Construction
0.037
Government
0.037
Mining
0.017
Service industry
0.061
Transportation
0.025
Transfers
0.296
Other
0.003
Income
Expenditure
0.017
0.310
0.674
0.024
0.673
0.303
0.183
0.394
0.016
0.115
0.292
0.163
0.387
0.048
0.129
0.273
0.100
0.303
0.049
0.045
0.026
0.037
0.025
0.414
0.002
0.134
0.393
0.036
0.055
0.011
0.071
0.032
0.266
0.004
Note: Columns may not sum to exactly 1.00 because of rounding.
aThere were a total of 11 sectors: agriculture, construction, forestry, professional
or financial services, oil and gas (energy), government, mining, service industry,
transportation, transfers and other. Sectors with a value of zero for all three dependence
measures for a particular community (or the model forest as a whole) are not presented
in this table. A zero value for the income-based sector dependence index means that no
base income data were available in that sector within the community.
29
NOR-X-418
Table 16.
Summary of ordinal ranking of sector dependence indices (SDIs) by economic sectora
Ordinal ranking of the employment, income and
expenditure SDIsb; no. of communities
Economic sector
Agriculture
Forestry
Construction
Professional and financial
services
Government
Mining
Service industry
Transportation
Transfers
Other
No. of combinations of
sector and ranking
MLH
LMH
HLM
LHM
HML
6
4
1
2
1
4
1
4
4
3
MHL
LHc
HLc
1
4
1
1
1d
1
3
2
2
19
1
1
1
4
1
2
18
8
1
1e
6
78
2
10
6
15
2
14
16
7
1
5
4
11
4
14
2
1
1
5
2
2
1
Total no. of
communities
aThe
oil and gas sector is not included in this table because its economic contribution was very small and the SDIs = 0.
rankings for each SDI indicated by letter combinations, where H = high, M = medium, L = low.
ranking of employment and expenditure SDIs, because no base income data were available for the applicable
communities or the employment and income SDIs had the same value.
dAll indexes had the same value; arbitrarily assigned to the LH category.
eAll indexes had the same value; arbitrarily assigned to the HL category.
bOrdinal
cOrdinal
30
NOR-X-418
SUMMARY AND CONCLUSIONS
sector accounted for the sign and significance
of the income coefficient in the broader models.
The coefficient for income remained significant
and its sign did not change from the broader
models shown in Tables 10 and 12.
Community
resource
dependence
has
traditionally been studied in terms of the degree
to which a community depends on the natural
resource sector to provide employment and
income for its residents. The present study has
examined community dependence more broadly
by including all of the sectors that make up a
community’s economy. In particular, this study
has examined the effect of the nonbase sector
on traditional measures of dependence through
the addition of household expenditure leakage
to the economic base model. As mentioned
earlier, Robertson (2003) noted that the effect
of leakage may not be adequately accounted for
in the practical implementation of the economic
base hypothesis. This study accounted for the
effect of leakage to determine its effect on
traditional measures of community dependence
and to place community dependence in a broader
context.
Other household and community characteristics
that were significant drivers of income leakage
were level of attained education, family type,
home
community
population,
populationweighted spending, and the travel distances
to destination communities. The relationship
between
population-weighted
spending,
and expenditure leakage and that between
expenditure
leakage
and
income
were
particularly interesting. The latter suggested
that the incentive to seek the lower prices or
wider selection, or a combination thereof, offered
in destination communities became stronger as
income fell, whereas the former suggested that
a higher proportion of income was spent on
consumer goods and services as income fell. This
may be because, as income declines, less aftertax income is diverted to savings and expenses,
which by definition are in the home community.
Forest dependence, and resource dependence
more generally, can be measured in a number of
ways, but the extent of employment in the sector
relative to employment in all sectors has been
used in many studies (e.g., Stedman et al. 2007).
There are 918 forest-dependent communities in
Canada, with dependence defined as at least
10% of employment in forestry (Stedman et al.
2005). In the current study, forestry was the
only economic sector that played a role in the
economy of all of the communities surveyed
in the BSLMF household expenditure survey.
This characteristic and the high values of the
employment-based sector dependence indices
for the forestry sector (Table 15) suggest that the
CSDs in the survey are communities that depend
heavily on forestry. However, the forestry sector
and the other economic sectors were not by
themselves significant in explaining household
expenditure leakage in the all-products model;
rather, they seemed to influence leakage largely
through income. Because the transfers sector
had the lowest average income of all sectors and
displayed a strong degree of low dependence
(Table 16), two additional models, an all-products
model and a durables-only model, with transfers
removed from their respective data sets, were
investigated to determine whether the transfers
Three types of sector dependence indices were
determined for each of the base economic
sectors for 16 communities in the BSLMF. In 55%
of the combinations of sector and community,
the income-based sector dependence index
suggested that the community in question was
less dependent on the applicable economic sector
than the employment-based index would have
suggested. However, when the employmentand income-based methods were compared with
the index based on household expenditures,
there were notable differences. About 35%
of the combinations of sector and community
(Table 16) had low dependence on their
respective communities when the expenditurebased sector-dependence index was factored
into the analysis. The transfers sector figured
prominently in the calculation of this percentage,
which dropped to 22% when transfers were not
included. Also, and somewhat paradoxically,
there was a preponderance of communities
with low dependence on the agriculture and
forestry sectors, with almost a third (30%) of
31
NOR-X-418
index shown in Table 16. The summary for
the model forest (Table 15) also highlights the
significance of the forestry and transfers sectors,
which collectively accounted for about 72% of
the base income in the region.
the communities displaying this characteristic
(Table 16). The reasons for this finding are
unclear, but it may be a function of the unusual
form of forestland tenure in the BSLMF, which may
promote purchases for the home while working
away from home on private woodlots. Use of
the local expenditure-based sector dependence
indices to gauge dependence in its broader
sense suggested that two-thirds (67%) of the
combinations of income- and employment-based
sector dependence indices were biased: they
did not reflect the degree of dependence that
communities actually had on all of the sectors
making up their economies. This figure dropped
to just over half (54%) when transfers were
omitted and to 27% when agriculture, forestry,
and transfers were omitted. The household
expenditure surveys were conducted about 10
years ago, and community characteristics and
circumstances have changed in the interim.
Nonetheless, this analysis has illustrated the
ability of expenditure leakage data to reveal the
potential biases inherent in sector dependence
indices that are based on employment- or
income-based location quotients.
Percent expenditure leakage can also serve as an
indicator of the impact that an economic shock
is likely to have on a community as a whole. For
example, a community with high expenditure
leakage, such as Saint-Valérien, with seven
economic sectors, may not be significantly
affected by a small decrease in income from
forestry, such as the 0.778% decrease that Lantz
and Yigezu (2003) estimated would occur in
response to a 1% decline in the price of lumber
in a New Brunswick community. Investigation of
the effect of leakage on the impact of positive or
negative economic shocks is an area for further
research, as is an even more fundamental
investigation of leakage changes with respect
to economic shocks. Concurrent developments
such as a decrease in population through outmigration could amplify any negative effects
on a community that a drop in income might
engender.
The degree of bias was even higher when the
model forest as a whole was examined. For the
entire model forest, eight of the nine sectors (see
last section of Table 15) had a “low or high” ranking
for the expenditure-based sector dependence
index, which suggests that these sectors did
not reflect the degree of dependence suggested
by their employment- and income-based sector
dependence indices when examined in relation to
the expenditure-based sector dependence index.
The forestry sector was the only sector that had
a “medium” ranking for the expenditure-based
sector dependence index, which suggests that
the proportion of sector income spent locally was
about average with respect to the other economic
sectors in the model forest. This characteristic
may be a reflection of the relatively high number
of “medium” ordinal rankings of the forestry
sector’s expenditure-based sector dependence
There are, of course, many other model forests
that could be investigated to shed more light on
the factors influencing household expenditure
leakage. Further studies may reveal extensive
differences between regions, such as those noted
by Williamson et al. (1999) between western
and eastern forest-dependent communities.
Further research may lend additional weight to
the finding in this work that employment-based
and income-based measures of dependence
are prone to misinterpretation because of the
common assumption that income earned in
a community is spent within that community
(Jagger et al. 1998). Perhaps, then, the degree
of household expenditure leakage is one of many
valid indicators of community health, well-being,
and, ultimately, sustainability.
32
NOR-X-418
ACKNOWLEDGMENTS
manuscript: Nancy Gélinas, Laval University;
Sylvain Masse, Canadian Forest Service;
Richard Stedman, Cornell University; and Jim
Unterschultz, University of Alberta. Funding for
this study was provided by the Canadian Model
Forest Network.
The authors express their gratitude to residents
of the Bas-Saint-Laurent Model Forest for their
participation in the household expenditure
survey and to Adam Wellstead, Canadian Forest
Service, for his able conduct of the survey
and initial reporting of the survey results. The
authors also thank the following reviewers of the
LITERATURE CITED
Kaufman, H.F.; Kaufman, L.C. 1946. Toward the
stabilization and enrichment of a forest community:
the Montana study. Univ. Montana, Missoula, MT.
US For. Serv., Reg. One.
Andrews, R.B. 1953. Mechanics of the urban economic
base: historical development of the base concept.
Land Econ. 29(2):161–167.
Beckley, T.; Parkins, J.; Stedman, R. 2002. Indicators
of forest-dependent community sustainability: the
evolution of research. For. Chron. 78(5):626–636.
Korber, D.; Beckley, T.; Luckert, M.; White, W. 1998.
Cultural, geographical, and sectoral refinements to
measures of forest industry dependence. Can. J.
For. Res. 28:1380–1387.
Cowell, D.K.; Green, G.P. 1994. Community attachment
and spending location: the importance of place in
household consumption. Soc. Sci. Q. 75(5):637–
655.
Lantz, V.A.; Yigezu, Y.A. 2003. An economic impact
analysis of market and policy changes in a New
Brunswick Fundy Model Forest community. For.
Chron. 79(5):957–966.
Davis, H.C.; Hutton, T.A. 1992. Structural change in the
British Columbia economy: regional diversification
and metropolitan transition. British Columbia
Round Table on the Environment and the Economy.
Gov. B. C., Victoria, BC.
Leake, N.; Adamowicz, W.A.; Boxall, P.C. 2006.
An econometric analysis of the effect of forest
dependence on the economic well-being of
Canadian communities. For. Sci. 52(5):595–604.
(DREE) Department of Regional Economic Expansion.
1979. Single sector communities. Can. Dep. Reg.
Econ. Expans., Ottawa, ON. Occas. Pap. 51 p.
(NRCan) Natural Resources Canada. 2009. The state
of Canada’s forests. Nat. Resour. Can., Can. For.
Serv., Ottawa, ON. 55 p.
Fletcher, S.; White, W.; Phillips, W.; Constantino, L.
1991. An economic analysis of Canadian prairie
provinces forest dependent communities. Dep.
Rural Econ., Fac. Agric. For., Univ. Alberta,
Edmonton, AB. Proj. Rep. 91-05.
Olfert, M.R.; Stadler, J. 1994. Community level
multipliers for rural development initiatives.
Growth Chang. 25(4):467–486.
Overdevest, C.; Green, G.P. 1995. Forest dependent
community well-being: a segmented market
approach. Soc. Nat. Resour. 8:111–131.
Horne, G.; Penner, C. 1992. British Columbia
community employment dependencies final report.
Plan. Stat. Div., Minist. Financ. Corp. Relat., Gov.
B. C., Victoria, BC.
Parkins, J.R.; Stedman, R.C.; Beckley, T.M. 2003. Forest
sector dependence and community well-being: a
structural equation model for New Brunswick and
British Columbia. Rural Sociol. 68(4):554–572.
Horne, G.; Robson, L. 1993. British Columbia
community
economic
dependencies.
British
Columbia Round Table on the Environment and the
Economy. Gov. B. C., Victoria, BC.
Persky, J.; Ranney, D.; Wiewel, W. 1993. Import
substitution and local economic development.
Econ. Dev. Q. 7(1):18–29.
Innis, H.A. 1950. Empire and communications. Toronto
Univ. Press, Toronto, ON. 287 p.
Pharand, N.L. 1988. Forest sector dependent
communities in Canada: a demographic profile.
Can. For. Serv., Labour Market Dev. Branch,
Ottawa, ON. DPC-X-23. 61 p.
Jagger, P.; Wellstead, W.; White, W. 1998. An
expenditure-based
analysis
of
community
dependence: a case study of the Foothills Model
Forest. Foothills Model Forest, Hinton, AB. 38 p.
Also available at http://foothillsresearchinstitute.
c a / C o n t e n t _ F i l e s / F i l e s / SS / SS _ r e p o r t 4 . p d f.
Accessed 17 June 2009.
Pinkerton, J.R.; Hassinger, E.W.; O’Brien, D.J. 1995.
Inshopping by residents of small communities.
Rural Sociol. 60(3):467–480.
33
NOR-X-418
Stedman, R.; White, W.; Patriquin, M.; Watson,
D. 2007. Measuring community forest-sector
dependence: does method matter? Soc. Nat. Res.
20:629–646.
Power, T.M. 1996. Lost landscapes and failed
economies: the search for value of place. Island
Press, Washington, DC. 304 p.
Randall, J.E.; Ironside, R.G. 1996. Communities on
the edge: an economic geography of resourcedependent communities. Can. Geogr. 40(1):17–35.
White,W.; Netzel, B.; Carr, S.; Fraser, G.A. 1986.
Forest sector dependence in rural British Columbia,
1971–81. Can. For. Serv., Pac. For. Cent., Victoria,
BC. BC-X-278. 32 p.
Robertson, G.C. 2003. A test of the economic base
hypothesis in the small forest communities of
southeast Alaska. US Dep. Agric., For. Serv., Pac.
Northwest Res. Stn., Portland, OR. PNW-GTR-592.
Williams, C.C. 1996. Understanding the role of
consumer services in local economic development:
some evidence from the Fens. Environ. Plan. A
28:555–571.
Ross, M.L. 1999. The political economy of the resource
curse. World Polit. 51(2):297–322.
Williamson, T.; Samson, R.; Korber, D. 1999. Economic
performance of forest-reliant census subdivisions
between 1981 and 1991. For. Chron. 75(1):93–
109.
Statistics Canada. 1998. Family expenditure in Canada
1996. Stat. Can., Ottawa, ON. Cat. No. 62-555XPB. 132 p.
Schwartz, H. 1982. A guide to regional multiplier
estimation. Can. Dep. Reg. Econ. Expans. Proj.
Assess. Eval. Branch, Hull, QC.
Yanagida, J.F.; Johnson, B.B.; Young, J.; Lundeen, M.
1991. An analysis of economic and noneconomic
factors affecting retail sales leakages. Rev. Reg.
Stud. 21(1):53–64.
Stedman, R.C.; Parkins, J.R.; Beckley,T.M. 2004.
Resource dependence and community well-being
in rural Canada. Rural Sociol. 69(2):213–234.
Stedman, R.C.; Parkins, J.R.; Beckley, T.M. 2005.
Forest dependence and community well-being in
rural Canada: variation by forest sector and region.
Can. J. For. Res. 35:215–220.
34
NOR-X-418
APPENDIX 1
Telephone Questionnaire Script –
Bas- St.-Laurent Model Forest
36
NOR-X-418
Hello. May I please speak to the person responsible for making your household’s purchasing
decisions?
If this person is not available:
When would be a good time to call this person back?
RECORD this time on the CATI terminal for a callback
Once connected with person:
Good morning/afternoon. I understand that you are responsible for making purchasing
decisions for your household. Is this correct?
Yes
No
Proceed
Ask to speak with correct person
This is [NAME] of Advantage Field Research in Edmonton, Alberta. We are doing a marketing
research study on behalf of the Canadian Model Forest Network to find out about household
spending patterns in the Bas-St.-Laurent area. Your number was one of about two thousand
households in the area which have been randomly chosen for an interview. All responses will remain
confidential. Could you take about 10 minutes to answer a few questions for me?
Yes
No
Proceed
I thank you for your time. Goodbye.
1. Where do you live? [Do not read list]
1.
2.
3.
4.
5.
6.
7.
8.
Notre-Dame-du-Lac
St.-Juste-du-Lac
Lots–Renverses
Auclair
Lejeune
Biencourt
Lac-des-Aigles
St.-Médard
9. St.-Eugène-de-Ladrière
10. St.-Valérien
11. St.-Charles-Garnier
12. Ste.-Angèle-de-Mérici
13. Ste. Jeanne-d’Arc
14. Ste.-Irène
15. Lac Humqui
2. For the next question, I will read a list of different types of products and services. For each, we
would like you to tell us in which town or city you USUALLY buy each type of product or service
for your household. Answer the following questions as a representative of your household.
2a. To begin with, in which town or city do you usually buy the food you purchase from a
grocery store? Do you usually buy it in [Read list except Never Purchase]
1. Notre-Dame-du-Lac
2. St.-Juste-du-Lac
3. Lots-Renverses
4. Auclair
5. Lejeune
6. Biencourt
7. Lac-des-Aigles
8. St.-Médard
9 St.-Eugène-de-Ladrière
10. St.-Valérien
11.
12.
13.
14.
15.
16.
17.
18.
19.
20.
37
St.-Charles-Garnier
Ste.-Angèle-de-Mérici
Ste.-Jeanne d’Arc
Ste.-Irène
Lac-Humqui
Rivière-du-Loup
Rimouski
Edmunston
OTHER (SPECIFY) ______________
NEVER PURCHASE [Do not read]
NOR-X-418
2b. Next, where do you usually buy the meals you purchase from restaurants? [Read list
except Never Purchase]
1. Notre-Dame-du-Lac
2. St.-Juste-du-Lac
3. Lots-Renverses
4. Auclair
5. Lejeune
6. Biencourt
7. Lac-des-Aigles
8. St.-Médard
9. St.-Eugène-de-Ladrière
10. St.-Valérien
11. St.-Charles-Garnier
12. Ste.-Angèle-de-Mérici
13. Ste.-Jeanne d’Arc
14. Ste.-Irène
15. Lac-Humqui
16. Rivière-du-Loup
17. Rimouski
18. Edmunston
19.OTHER (SPECIFY) _____________________
20. NEVER PURCHASE [Do not read]
2c. Where do you usually buy household supplies such as child care products, pet supplies,
and/or household cleaning supplies? [Read list if necessary]
1. Notre-Dame-du-Lac
2. St.-Juste-du-Lac
3. Lots-Renverses
4. Auclair
5. Lejeune
6. Biencourt
7. Lac-des-Aigles
8. St.-Médard
9. St.-Eugène-de-Ladrière
10. St.-Valérien
11. St.-Charles-Garnier
12.
13.
14.
15.
16.
17.
18.
19.
20.
21.
Ste.-Angèle-de-Mérici
Ste.-Jeanne d’Arc
Ste.-Irène
Lac-Humqui
Rivière-du-Loup
Rimouski
Edmunston
FROM A CATALOQUE OR THE INTERNET
OTHER (SPECIFY) ____________________
NEVER PURCHASE [Do not read]
2d. Where do you usually buy clothing? [Read list if necessary]
1. Notre-Dame-du-Lac
2. St.-Juste-du-Lac
3. Lots-Renverses
4. Auclair
5. Lejeune
6. Biencourt
7. Lac-des-Aigles
8. St.-Médard
9. St.-Eugène-de-Ladrière
10. St.-Valérien
11. St.-Charles-Garnier
12.
13.
14.
15.
16.
17.
18.
19.
20.
21.
Ste.-Angèle-de-Mérici
Ste.-Jeanne d’Arc
Ste.-Irène
Lac-Humqui
Rivière-du-Loup
Rimouski
Edmunston
FROM A CATALOQUE OR THE INTERNET
OTHER (SPECIFY) ______________
NEVER PURCHASE [Do not read]
2e. Where do you usually buy fuel for your vehicles? [Read list if necessary]
1. Notre-Dame-du-Lac
2. St.-Juste-du-Lac
3. Lots-Renverses
4. Auclair
5. Lejeune
6. Biencourt
7. Lac-des-Aigles
8. St.-Médard
9. St.-Eugène-de-Ladrière
10. St.-Valérien
11.
12.
13.
14.
15.
16.
17.
18.
19.
20.
38
St.-Charles-Garnier
Ste.-Angèle-de-Mérici
Ste.-Jeanne d’Arc
Ste.-Irène
Lac-Humqui
Rivière-du-Loup
Rimouski
Edmunston
OTHER (SPECIFY) ________________
NEVER PURCHASE [Do not read]
NOR-X-418
2f.
Where do you usually go for dental and optical services? [Read list if necessary]
1. Notre-Dame-du-Lac
2. St.-Juste-du-Lac
3. Lots-Renverses
4. Auclair
5. Lejeune
6. Biencourt
7. Lac-des-Aigles
8. St.-Médard
9. St.-Eugène-de-Ladrière
10. St.-Valérien
11.
12.
13.
14.
15.
16.
17.
18.
19.
20.
St.-Charles-Garnier
Ste.-Angèle-de-Mérici
Ste.-Jeanne d’Arc
Ste.-Irène
Lac-Humqui
Rivière-du-Loup
Rimouski
Edmunston
OTHER (SPECIFY) _______________
NEVER PURCHASE [Do not read]
2g. Where do you usually buy medication purchased at a pharmacy? [Read list if
necessary]
1. Notre-Dame-du-Lac
2. St.-Juste-du-Lac
3. Lots-Renverses
4. Auclair
5. Lejeune
6. Biencourt
7. Lac-des-Aigles
8. St.-Médard
9. St.-Eugène-de-Ladrière
10. St.-Valérien
11.
12.
13.
14.
15.
16.
17.
18.
19.
20.
St.-Charles-Garnier
Ste.-Angèle-de-Mérici
Ste.-Jeanne d’Arc
Ste.-Irène
Lac-Humqui
Rivière-du-Loup
Rimouski
Edmunston
OTHER (SPECIFY) ____________________
NEVER PURCHASE [Do not read]
2h. Where do you usually go for spectator events and entertainment such as concerts, live
sporting events, and movies? [Read list if necessary]
1. Notre-Dame-du-Lac
2. St.-Juste-du-Lac
3. Lots-Renverses
4. Auclair
5. Lejeune
6. Biencourt
7. Lac-des-Aigles
8. St.-Médard
9. St.-Eugène-de-Ladrière
10. St.-Valérien
2i.
11.
12.
13.
14.
15.
16.
17.
18.
19.
20.
Where do you usually buy computer equipment and software including computer games?
[Read list if necessary]
1. Notre-Dame-du-Lac
2. St.-Juste-du-Lac
3. Lots-Renverses
4. Auclair
5. Lejeune
6. Biencourt
7. Lac-des-Aigles
8. St.-Médard
9. St.-Eugène-de-Ladrière
10. St.-Valérien
11. St.-Charles-Garnier
St.-Charles-Garnier
Ste.-Angèle-de-Mérici
Ste.-Jeanne d’Arc
Ste.-Irène
Lac-Humqui
Rivière-du-Loup
Rimouski
Edmunston
OTHER (SPECIFY) ____________________
NEVER PURCHASE [Do not read]
12.
13.
14.
15.
16.
17.
18.
18.
19.
20.
39
Ste.-Angèle-de-Mérici
Ste.-Jeanne d’Arc
Ste.-Irène
Lac-Humqui
Rivière-du-Loup
Rimouski
Edmunston
FROM A CATALOQUE OR THE INTERNET
OTHER (SPECIFY) ____________________
NEVER PURCHASE [Do not read]
NOR-X-418
2j.
Where do you usually buy tobacco and alcohol products? [Read list if necessary]
1. Notre-Dame-du-Lac
2. St.-Juste-du-Lac
3. Lots-Renverses
4. Auclair
5. Lejeune
6. Biencourt
7. Lac-des-Aigles
8. St.-Médard
9. St.-Eugène-de-Ladrière
10. St.-Valérien
11.
12.
13.
14.
15.
16.
17.
18.
19.
20.
St.-Charles-Garnier
Ste.-Angèle-de-Mérici
Ste.-Jeanne d’Arc
Ste.-Irène
Lac-Humqui
Rivière-du-Loup
Rimouski
Edmunston
OTHER (SPECIFY) ____________________
NEVER PURCHASE [Do not read]
2k. Where do you usually buy reading material and other printed matter? [Read list if
necessary]
1. Notre-Dame-du-Lac
2. St.-Juste-du-Lac
3. Lots-Renverses
4. Auclair
5. Lejeune
6. Biencourt
7. Lac-des-Aigles
8. St.-Médard
9. St.-Eugène-de-Ladrière
10. St.-Valérien
11. St.-Charles-Garnier
2l.
12.
13.
14.
15.
16.
17.
18.
19.
20.
21.
Ste.-Angèle-de-Mérici
Ste.-Jeanne d’Arc
Ste.-Irène
Lac-Humqui
Rivière-du-Loup
Rimouski
Edmunston
FROM A CATALOQUE OR THE INTERNET
OTHER (SPECIFY) ____________________
NEVER PURCHASE [Do not read]
Where do you usually buy other small items such as small gifts, toys, games or
accessories? [Read list if necessary]
1. Notre-Dame-du-Lac
2. St.-Juste-du-Lac
3. Lots-Renverses
4. Auclair
5. Lejeune
6. Biencourt
7. Lac-des-Aigles
8. St.-Médard
9. St.-Eugène-de-Ladrière
10. St.-Valérien
11. St.-Charles-Garnier
12.
13.
14.
15.
16.
17.
18.
19.
20.
21.
Ste.-Angèle-de-Mérici
Ste.-Jeanne d’Arc
Ste.-Irène
Lac-Humqui
Rivière-du-Loup
Rimouski
Edmunston
FROM A CATALOQUE OR THE INTERNET
OTHER (SPECIFY) ____________________
NEVER PURCHASE [Do not read]
3. What is the most important reason why you purchase your day to day goods in these
communities? Is it because of [Read list, Single response]
1.
2.
3.
4.
5.
Lower Price
More Convenient
Wide selection of products
To support local business or
Other Record response ______________
4. Now we would like to continue with some larger items. In which town or city do you, or would
you, buy the following items?
40
NOR-X-418
4a. Furniture, appliances [Read list except Never Purchase]
1. Notre-Dame-du-Lac
2. St.-Juste-du-Lac
3. Lots-Renverses
4. Auclair
5. Lejeune
6. Biencourt
7. Lac-des-Aigles
8. St.-Médard9.
9. St.-Eugène-de-Ladrière
10. St.-Valérien
11. St.-Charles-Garnier
12.
13.
14.
15.
16.
17.
18.
19.
20.
21.
Ste.-Angèle-de-Mérici
Ste.-Jeanne d’Arc
Ste.-Irène
Lac-Humqui
Rivière-du-Loup
Rimouski
Edmunston
FROM A CATALOQUE OR THE INTERNET
OTHER (SPECIFY) ____________________
NEVER PURCHASE [Do not read]
4b. Home entertainment equipment [Read list if necessary]
1. Notre-Dame-du-Lac
2. St.-Juste-du-Lac
3. Lots-Renverses
4. Auclair
5. Lejeune
6. Biencourt
7. Lac-des-Aigles
8. St.-Médard
9. St.-Eugène-de-Ladrière
10. St.-Valérien
11. St.-Charles-Garnier
12.
13.
14.
15.
16.
17.
18.
19.
20.
21.
Ste.-Angèle-de-Mérici
Ste.-Jeanne d’Arc
Ste.-Irène
Lac-Humqui
Rivière-du-Loup
Rimouski
Edmunston
FROM A CATALOQUE OR THE INTERNET
OTHER (SPECIFY) ____________________
NEVER PURCHASE [Do not read]
4c. Sporting and recreational equipment such as camping equipment, skiing and golf
equipment, bicycles, or fishing equipment [Read list if necessary]
1. Notre-Dame-du-Lac
2. St.-Juste-du-Lac
3. Lots-Renverses
4. Auclair
5. Lejeune
6. Biencourt
7. Lac-des-Aigles
8. St.-Médard
9. St.-Eugène-de-Ladrière
10. St.-Valérien
11. St.-Charles-Garnier
12.
13.
14.
15.
16.
17.
18.
19.
20.
21.
Ste.-Angèle-de-Mérici
Ste.-Jeanne d’Arc.
Ste.-Irène
Lac-Humqui
Rivière-du-Loup
Rimouski
Edmunston
FROM A CATALOQUE OR THE INTERNET
OTHER (SPECIFY) ____________________
NEVER PURCHASE [Do not read]
4d. Recreational vehicles such as ATV’s, boats, snowmobiles, and dirt bikes [Read list if
necessary]
1. Notre-Dame-du-Lac
2. St.-Juste-du-Lac
3. Lots-Renverses
4. Auclair
5. Lejeune
6. Biencourt
7. Lac-des-Aigles
8. St.-Médard
9. St.-Eugène-de-Ladrière
10. St.-Valérien
11.
12.
13.
14.
15.
16.
17.
18.
19.
20.
41
St.-Charles-Garnier
Ste.-Angèle-de-Mérici
Ste.-Jeanne d’Arc
Ste.-Irène
Lac-Humqui
Rivière-du-Loup
Rimouski
Edmunston
OTHER (SPECIFY) ____________________
NEVER PURCHASE [Do not read]
NOR-X-418
4e. New cars and trucks [Read list if necessary]
1. Notre-Dame-du-Lac
2. St.-Juste-du-Lac
3. Lots-Renverses
4. Auclair
5. Lejeune
6. Biencourt
7. Lac-des-Aigles
8. St.-Médard
9. St.-Eugène-de-Ladrière
10. St.-Valérien
4f.
11.
12.
13.
14.
15.
16.
17.
18.
19.
20.
St.-Charles-Garnier
Ste.-Angèle-de-Mérici
Ste.-Jeanne d’Arc
Ste.-Irène
Lac-Humqui
Rivière-du-Loup
Rimouski
Edmunston
OTHER (SPECIFY) ____________________
NEVER PURCHASE [Do not read]
Used cars and trucks [Read list if necessary]
1. Notre-Dame-du-Lac
2. St.-Juste-du-Lac
3. Lots-Renverses
4. Auclair
5. Lejeune
6. Biencourt
7. Lac-des-Aigles
8. St.-Médard
9. St.-Eugène-de-Ladrière
10. St.-Valérien
11.
12.
13.
14.
15.
16.
17.
18.
19.
20.
St.-Charles-Garnier
Ste.-Angèle-de-Mérici
Ste.-Jeanne d’Arc
Ste.-Irène
Lac-Humqui
Rivière-du-Loup
Rimouski
Edmunston
OTHER (SPECIFY) ____________________
NEVER PURCHASE [Do not read]
4g. Motor homes and travel trailers [Read list if necessary]
1. Notre-Dame-du-Lac
2. St.-Juste-du-Lac
3. Lots-Renverses
4. Auclair
5. Lejeune
6. Biencourt
7. Lac-des-Aigles
8. St.-Médard
9. St.-Eugène-de-Ladrière
10. St.-Valérien
11.
12.
13.
14.
15.
16.
17.
18.
19.
20.
St.-Charles-Garnier
Ste.-Angèle-de-Mérici
Ste.-Jeanne d’Arc
Ste.-Irène
Lac-Humqui
Rivière-du-Loup
Rimouski
Edmunston
OTHER (SPECIFY) ____________________
NEVER PURCHASE [Do not read]
5. What is the most important reason why you purchase your major goods in these communities?
Is it because of [Read list, Single response]
1.
2.
3.
4.
5.
Lower Price
More Convenient
Wide selection of products
To support local business or
Other Record response ______________
6. When you go on vacation, where do you usually go? [Do not read list. Probe if you are
unsure which category the response falls into. Single response]
1.
2.
3.
4.
Camping at the beach in the Bas-St.-Laurent area
Other Places Within Quebec
Outside Quebec
Don’t Go On Vacation
42
NOR-X-418
Finally, I would like to ask you some questions about yourself and your household to help classify
individuals for this study.
7. What is the highest level of education that you have completed? Please stop me when I have
read your category. [Read List except refused]
1.
2.
3.
4.
5.
Never attended school
Grade school (Grade 1 to 9)
Some high school
High school graduate
Technical school
6.
7.
8.
9.
Some college or university
Undergraduate university degree
Graduate degree
Refused [Do not read this response]
8. Which of the following categories includes your age? Again, please stop me once I have read
your category. [Read list except refused]
1.
2.
3.
4.
18
26
36
46
to
to
to
to
25
35
45
55
years
years
years
years
5. 56 to 65 years
6. Over 66
7. Refused [Do not read this response]
9. How many of the OTHER people that live in your household fall into each of the following age
categories? [Read list except refused]
_____
_____
_____
_____
Under 18 years
18 to 25 years
26 to 35 years
36 to 45 years
_____
_____
_____
_____
46 to 55 years
56 to 65 years
Over 66
Refused [Do not read this response]
10.Household members may be employed outside the home, self-employed, homemakers, retired,
students, or others who are not working. Remembering to include yourself, how many of the
people over 18 in your household fall into each of the following groups? [Read list]
_____
_____
_____
_____
Employed outside the home
Self-employed
Homemakers
Retired
_____ Students
_____ Others not working
_____ Refused [Do not read this response]
11.Please identify which of the following industries or other sources of income currently contribute
to your household’s income. [Read complete list except refused. Multiple responses]
1. Agriculture
2. Construction
3. Forestry
4. Financial Sector
5. Oil and Gas
6. Government Worker
7. Professional
8. Mining
9. Service Industry
10. Transportation
11. Canada Pension Plan and/or Private Pension Income
12. Investment Income
13. Social Assistance Payments and/or Unemployment Insurance
14. Other industry [Record response] ______________
15. Refused [Do not read]
43
NOR-X-418
12.Of the above sources of income, which source provides most of your household’s income?
[Read list of their responses if necessary. Single response only.]
1. Agriculture
2. Construction
3. Forestry
4. Financial Sector
5. Oil and Gas
6. Government Worker
7. Professional
8. Mining
9. Service Industry
10. Transportation
11. Canada Pension Plan and/or Private Pension Income
12. Investment Income
13. Social Assistance Payments and/or Unemployment Insurance
14. Other industry [Record response] ______________
15. Refused [Do not read]
13.Which of the following categories best describes your household’s total earnings, before taxes
for 1998? [Read list except refused]
1. $15,000 or less
2. $15,001 to $20,000
3. $20,001 to $30,000
4. $30,001 to $40,000
5. $40,001 to $50,000
6. $50,001 to $60,000
7. $60,001 to $70,000
8. $70,001 to $80,000
9. $80,001 to $90,000
10. $90,001 to $100,000
11. $100,001 to $110,000
12. $110,001 to $120,000
13. More than $120,000
14. Refused [Do not read this category]
Those are all of the questions I needed to ask you. Thank-you for taking the time to help the
Canadian Model Forest Network with this study. Goodbye.
End of Survey
Record Gender
1. Male
2. Female
44
NOR-X-418
APPENDIX 2
An example of a calculation of household
expenditure for a product class using the
modifier calculation
expenditure for clothing was $1 892 leaving a
difference of $125 or about 6% less than the
overall Quebec average. The average expenditure
for clothing by all Quebec households with
incomes of $50 000 to $59 999 was $2 492 or
a difference of $475 or 23% more. Finally, we
compared the average family expenditure for
clothing with the average clothing expenditure
for this family type in Quebec. A Quebec
household with two adults and two children
spends an annual average of $2 795 on clothing,
$778 (39%) more than the Quebec average. As
we are weighting these modifications equally, we
sum the three modifications and divide by three
to obtain the overall modifier for this case. This
is shown below.
We have modified the data in the Family
Expenditure in Canada 1996 survey to align it
as closely as possible with actual expenditures
in the BSLMF region. This was done by taking
the expenditure level reported in the survey
for a given product class and modifying it to
concurrently reflect the rural nature of BSLMF
communities and the income and the family
structure of the survey respondents. Each of
these three components was given equal weight
in calculating the modifier. An example for
clothing expenses is provided below.
From the Family Expenditure in Canada 1996
survey we know that the average Quebec family
spends $2 017 on clothing. For a household
included in the BSLMF expenditure survey with
an income of $50 000 to $59 999, composed of
two adults and two children, we would make the
following modifications. First, we determined the
difference between expenditures of rural nonfarm inhabitants (as residents of the BSLMF are
classed) and the average Quebec expenditure for
clothing as reported in the Family Expenditure in
Canada 1996 survey. The average Quebec rural
Rural non-farm (-0.062)+ Income (+0.235)+
Family type (0.386), divided by 3 = +0.186. The
modifier used for this particular product class for
the above family type was +0.187 or an increase
of 18.7% over the Quebec average expenditure
of $2 017. In this example the family would have
spent $2 392 on clothing. This was repeated for
all product classes for all survey respondents.
46
NOR-X-418
Causal factors behind household expenditure leakage
and its effect on community resource dependence in Quebec
D. H. Kuhnke and W. A. White
To order publications on-line, visit the Canadian Forest Service Bookstore at:
Information Report NOR-X-418
bookstore.c fs.nrcan.gc.ca
NORTHERN FORESTRY CENTRE
CANADIAN FOREST SERVICE
EDMONTON, ALBERTA
QUEBEC
N
Bas-Saint-Laurent
Model Forest
E
W
S
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

advertisement