INFORMATION TO USERS

INFORMATION TO  USERS

INFORMATION TO USERS

This was produced from a copy of a document sent to us for microfilming. While the most advanced technological means to photograph and reproduce this document have been used, the quality is heavily dependent upon the quality of the material submitted.

The following explanation of techniques is provided to help you understand markings or notations which may appear on this reproduction.

1. The sign or "target" for pages apparently lacking from the document photographed is "Missing Page(s)". If it was possible to obtain the missing page(s) or section, they are spliced into the film along with adjacent pages.

This may have necessitated cutting through an image and duplicating adjacent pages to assure you of complete continuity.

2. When an image on the film is obliterated with a round black mark it is an indication that the film inspector noticed either blurred copy because of movement during exposure, or duplicate copy. Unless we meant to delete, copyrighted materials that should not have been filmed, you will find a good image of the page in the adjacent frame.

3. When a map, drawing or chart, etc., is part of the material being photo­ graphed the photographer has followed a definite method in "sectioning" the material. It is customary to begin filming at the upper left hand corner of a large sheet and to continue from left to right in equal sections with small overlaps. If necessary, sectioning is continued again—beginning below the first row and continuing on until complete.

4. For any illustrations that cannot be reproduced satisfactorily by xerography, photographic prints can be purchased at additional cost and tipped into your xerographic copy. Requests can be made to our

Dissertations Customer Services Department.

5. Some pages in any document may have indistinct print. In all cases we have filmed the best available copy.

University

Microfilms

International

300 N. ZEEB ROAD, ANN ARBOR, Ml 48106

18 BEDFORD ROW, LONDON WC1R 4EJ, ENGLAND

8100302

CAO, THAN VAN

MIXED LAND USES, EXTERNALITIES, AND RESIDENTIAL PROPERTY

VALUES: AN EMPIRICAL ANALYSIS OF THE MUNICIPAL ZONING

ORDINANCE OF TUCSON, ARIZONA

PH.D. 1980

The University of Arizona

University

Microfilms

International

300 N. Zeeb Road, Ann Arbor, MI 48106

PLEASE NOTE:

In all cases this material has been filmed in the best possible way from the available copy. Problems encountered with this document have been identified here with a check mark .

1. Glossy photographs 's/

2. Colored illustrations

3. Photographs with dark background

'4. Illustrations are poor copy

5. °rint shows through as there is text on both sides of page

6. Indistinct, broken or small print on several pages

7. Tightly bound copy with print lost in spine __

8. Computer printout pages .with indistinct print _________

9. Page(s) lacking when material received, and not available from school or author

10. Page(s) follows seem to be missing in numbering only as text

11. Poor carbon copy

12. Not original copy, several pages with blurred type

13. Appendix pages are poor copy

14. Original copy with light type

15. Curling and wrinkled pages

16. Other

Jniversity

Micrdrilms

Tternarional

D O N

255= RD.. ANN ARBOR Ml 48106 '313) 761-4700

MIXED LAND USES, EXTERNALITIES, AND

RESIDENTIAL PROPERTY VALUES: AN

EMPIRICAL ANALYSIS OF THE MUNICIPAL

ZONING ORDINANCE OF TUCSON, ARIZONA by

Than Van Cao

A Dissertation Submitted to the Faculty of the

DEPARTMENT OF ECONOMICS

In Partial Fulfillment of the Requirements

For the Degree of

DOCTOR OF PHILOSOPHY

In the Graduate College

THE UNIVERSITY OF ARIZONA

THE UNIVERSITY OF ARIZONA

GRADUATE COLLEGE

As members of the Final Examination Committee, we certify that we have read the dissertation prepared by

Than Van Cao

entitled

MIXED LAND USES, EXTERNALITIES, AND RESIDENTIAL PROPERTY VALUES:

AN EMPIRICAL ANALYSIS OF THE MUNICIPAL ZONING ORDINANCE OF

TUCSON, ARIZONA

and recommend that it be accepted as fulfilling the dissertation requirement for the Degree of

Doctor of Philosophy .

Date

Date

V

'

<2 o ~ % Z!

Date

^

/

o

Date

Date

Final approval and acceptance of this dissertation is contingent upon the candidate's submission of the final copy of the dissertation to the Graduate

College.

I hereby certify that I have read this dissertation prepared under my direction and recommend that it be accepted as fulfilling the dissertation requirement.

Dissertation Director Date

STATEMENT BY AUTHOR

This dissertation has been submitted in partial fulfillment of requirements for an advanced degree at The University of Arizona and is deposited in the University Library to be made available to bor­ rowers under rules of the Library.

Brief quotations from this dissertation are allowable without special permission, provided that accurate acknowledgment of source is made. Requests for permission for extended quotation from or re­ production of this manuscript in whole or in part may be granted by the head of the major department or the Dean of the Graduate College when in his judgment the proposed use of the material is in the in­ terests of scholarship. In all other instances, however, permission must be obtained from the author.

-

<

_

SIGNED:

ACKNOWLEDGMENTS

I would like to thank my dissertation committee, Professors

Harry W. Ayer, David L. Barkley, Dennis C. Cory, Jimmye S. Hillman,

David E. Pingry, and Lester D. Taylor for their comments on numerous drafts of this study. Errors of analysis and exposition that remain are, of course, my responsibility.

Financial assistance came from the Department of Agricultural

Economics and the College of Agriculture of The University of Arizona.

I would like to take this opportunity to express my appreciation for their continued support throughout my doctoral program.

I wish to thank my family for the support and understanding that enabled me to complete this work. Finally, I thank Raye Renfroe for typing the final copy of the manuscript.

iii

TABLE OF CONTENTS

LIST OF TABLES

LIST OF ILLUSTRATIONS

1. REVIEW OF THE LITERATURE ON ZONING AND NEIGHBORHOOD

EXTERNALITIES

Introductory Remarks

Review of Economic Literature

Review of Econometric Literature

Summary and Conclusions

Objectives of the Study

2. EXTERNALITIES AND RESIDENTIAL PROPERTY VALUES:

A THEORETICAL FRAMEWORK

Introduction

Neighborhood and Land Use Externalities

Nonsingle-Family Land Use Externalities

A Model of Consumer Bidding Behavior in an Urban

Property Market

Specification of the Model

Page vi vii

3. THE DATA BASE

Introduction

Zoning Analysis of Land Use

Units of Observations

The Variables

Data Sources and Data Characteristics

55

55

57

58

60

73

4. ESTIMATION RESULTS 77

Introduction

Zoning Model

77

80

Land Use Model 83

Further Discussions on Both Zoning and Land Use Models. . 100

1

1

6

10

22

24

26

26

27

30

39

51

iv

V

TABLE OF CONTENTS—Continued

5. MIXED LAND USE AND POLICY IMPLICATIONS

Introduction

Mixed Land Use

Costs of Land Use Separation

Benefits of Mixed Land Use

Policy Implications

Future Research Issues

APPENDIX A: LAND USE IN ACRES BY ZONING—CITY OF TUCSON. . . 118

APPENDIX B: CITY OF TUCSON ZONING DISTRICT NARRATIVE

SUMMARIES 120

APPENDIX C: DEPENDENT AND INDEPENDENT VARIABLES

Description, Notation and Sources of Data

BIBLIOGRAPHY

122

122

127

Page

106

106

107

107

110

113

115

LIST OF TABLES

Table

3-1 Generalized Land Use, City of Tucson

4-1 Alternative Estimates of a Valuation Equation for Single-

Family Homes in the City of Tucson in 1970; Regression

Coefficients (Absolute T-Ratios in Parentheses)

4-2 Range of Various Land-Use Parameters

4-3 Correlations among Variables

4-4 Zoning and Land Use in the City of Tucson (Nonsinglefamily Uses)

A-l Land Use in Acres by Zoning--City of Tucson

Page

57

73

96

98

104

119

vi

LIST OF ILLUSTRATIONS

Exhibit

1 Land-Use Map, City of Tucson

2 Map of 52 Census Tracts, City of Tucson

Page

125

126

vii

ABSTRACT

This dissertation is primarily concerned with a number of theo­ retical and empirical problems in the economics of land-use control.

Chapter 1 sets the stage for the review of the economic litera­ ture on zoning and neighborhood externalities. This chapter concludes that there are important research issues which need to be studied, viz. the recognition of the multi-nucleated character of the contemporary urban areas, and the need to take account of both the advantages and disadvantages of proximity of single-family homes to nonresidential activities.

Since economic research so far has failed to establish conclu­ sively that neighborhood externalities affect adversely or advantageous­ ly the market value of residential properties, Chapter 2 shall discuss household behavior when confronting ^neighborhood externalities, with special reference to land-use externalities. The discussion suggests, among other things, that the existence of nonsingle-family land uses in a neighborhood does not necessarily tend to depress the price of singlefamily homes.

Chapter 3 provides a data base for the research. Then, in

Chapter 4 the hypothesized relationship between neighborhood externali­ ties and residential property value is tested econometrically using aggregate data for 52 neighborhoods in the City of Tucson. The results of the estimations indicate that the value of a single-family home

viii

ix

depends, among other things, on its physical characteristics, its ac­ cessibility to employment and shopping, and local public services. For the first time, there is statistical evidence that over the ranges studied in this research nonsingle-family land uses exert a positive influence on residential property value. These results suggest that the time has come to redirect future research or policy efforts toward viewing mixed land uses possibly as a beneficial contribution to con­ temporary urban development. That is, zoning ordinances could legiti­ mately move away from a "separate facilities" philosophy to a "mixed land use" philosophy without lowering property values. Issues of ac­ cessibility and restrictions in the availability of energy resources could have much to do with the lessening importance of conventional belief in separatory land planning doctrines.

CHAPTER 1

REVIEW OF THE LITERATURE ON ZONING

AND NEIGHBORHOOD EXTERNALITIES

1. Introductory Remarks

The basic methods for controlling the use of private real prop­ erty in the United States are zoning and subdivision regulations.^"

Zoning simply is the division of land into districts within which per­ missible uses are prescribed and restrictions on building height, layout, and other requirements are defined. The regulations are uni­ form in all districts zoned for the same use. Today zoning statutes are a nearly ubiquitous phenomenon in urban America. Most of American cities with a population of over 5,000 now have some kind of zoning

2

statute. Zoning is not a right of a municipality. Local governments need a specific grant of power from the State to zone. Usually a zoning ordinance has one or more of the following purposes:

(1) to protect the market value of property;

(2) to control population density;

(3) to restrict land-uses and regulate external effects between conflicting uses;

(4) to control the costs of public services.

^Subdivision regulations are concerned with specifying the mini­ mum standards that apply to new residential development (street design, arrangement of lots, drainage, sewage, etc.). Subdivisions are usually part of the municipal zoning code.

^A. Manvel (1968), p. 31.

1

It is not uncommon to find that most zoning ordinances aim at promoting the health, safety, morals, and general welfare of the community.

These provisions are said to be necessary for the restriction of nui­ sances which the use of one piece of land for a particular purpose might

3 incur.

Proponents of municipal zoning regulations argue that there exist externality effects between various types of land uses in the urban property market and the raison d'etre for zoning is market failure due to external effects. Since externalities cause inefficiency in urban land markets, the government frequently intervenes in this situa­ tion to protect property owners against the possible depression of their property values and capital losses.

Before proceeding further, it is necessary to explain briefly the concept of externalities as used in the economic literature. An externality is an effect of an activity on the welfare of people in ways that cannot be compensated adequately by private agreements. More spe­ cifically, there are two main components in the definition of external­ ity: an interdependent utility (or production) function and a lack of compensation. Much has been written on externality and there is still

4 no complete agreement on this subject in the profession. Regardless of the source of external effects, their existence implies that resources

3

Barlowe (1978) indicated that zoning can be used for a broader objective than merely the control of nuisances: modern zoning practice is also a tool for the effectuation of land-resource planning.

4

See Pigou (1920) and generations of welfare economists writing since his time. It is noteworthy to see Coase (1960) for a "small numbers case" of externality.

3

are not allocated so as to satisfy conditions for Pareto-optimality.

That is, it is possible to reallocate resources in such a way that at least one individual is better off with no individuals being worse off.

(A more detailed discussion on externalities will be presented in the subsequent chapters). Advocates of land use planning generally support the idea that some zoning regulations are needed because direct controls on externalities are imperfect and they assume that the government is possessed of knowledge in its establishment of land use patterns.

Critics of land-use planning and zoning disagree on theoretical grounds. Otto Davis has found zoning of dubious value, eliminating external diseconomies in some instances but actually creating them in others and preventing people from developing land as they wish even if no external diseconomies are created.^ He concluded that mutually in­ compatible uses would not locate near each other under market conditions and that zoning could only be useful to keep out an a . ^.ivity that causes harm to adjacent activities, and in turn either benefits from them or is not harmed by them. Furthermore, in a subsequent article, Davis argued that municipal zoning has been entrusted to the political process for its administration. And a rational political decision-maker would only support those zoning actions which will increase the number of votes for him. Davis, therefore, concluded that a democratic political process sometimes may impose regulations which result in "over-zoning" of

^Davis (1960).

^Davis (1963).

4

existing land uses and, hence, impede the transition of the urban prop­ erty market toward more efficient land-use patterns.

Similarly, Siegan contends that governmental control over land use through zoning has been unworkable, inequitable and a serious impedi-

7 ment to the operation of the real estate market. According to him, there are alternatives to zoning as a means of effecting optimal land use. He indicates that in Houston, Texas, where contractual arrange-

0 ments (such as restrictive covenants) rather than zoning govern land use, land use changes in response to market forces are swifter than in zoned cities and spatial configurations and land use patterns are more flexible. Siegan concludes that although the contracts are more flexi­ ble than zoning, they provide the same types of guarantees against unwarranted and unexpected changes in land use.

Another critic of zoning is Ellickson who argues that zoning standards are neither as efficient nor as equitable as available alter-

9 natives. Zoning impairs efficient urban growth and discriminates against migrants, minority groups, lower classes, and landowners with little political influence. The alternative proposed by Ellickson relies primarily on three decentralized devices to internalize the ex­ ternal costs of "unneighborly" land use activities:

1. in areas where much of the land is still undeveloped, a con­

7

Siegan (1970).

8

The contractual arrangements are contracts that prohibit the reVale of land for various injurious uses.

9

Ellickson (1973).

5

land parcels and covenants between their owners—is a good mechanism for handling the problem of externality;

2. in the case of substantial injuries from nuisances, internal­ ization can be achieved through private lawsuits by the injured neighbors;

3. trivial harms caused by noxious land uses cannot be efficiently internalized through nuisance suits; in this case, external costs can best be deterred through fines levied against the damage of nuisances.

While critics of zoning disagree on theoretical grounds, as demonstrated in the above paragraphs, empirical evidence on zoning efficiency in the allocation of land uses is mixed. In recent years, some authors have statistically investigated the economic rationale for zoning by studying the relationships between market values of residen­ tial properties and various physical and neighboring land-use character­ istics of these properties. Crecine et al., Rueter, and Maser et al. have all found that the introduction of a zoning ordinance does not have any significant effect upon the market prices of the properties."''

0

The above results are, however, in contrast with the findings of

Stull who found that households are sensitive to land use environments of the neighborhoods in which their homes are located.

Sections 2 and 3 will review rather carefully the theoretical and empirical literature on land-use control and neighborhood externalities.

"^Crecine, Davis and Jackson (1967), Rueter (1973), and Maser, n stull (1975).

2. Review of Economic Literature

The economic literature on land-use controls and neighborhood externalities is not abundant. There have been a few important writing in the field, however, and this section will review them in some detail

Chapter XI of A. Marshall's Principles of Economics is devoted

12

to the determination of urban land value. However, chapter XI is not directly related to the specific topics of this study. Marshall was essentially interested in nonresidential land uses and the role of ac­ cessibility in the determination of the site value for such uses. Res.i dential land use was peripherally discussed in one paragraph in which

Marshall stated that a portion of the value of a residential site is

13 due to its neighborhood characteristics.

A.C. Pigou later observed in his Economics of Welfare that neighborhood effects are only special cases of a broader class of eco­ nomic phenomena. According to him, a divergence between marginal pri­ vate net product and marginal social net product occurs when a factory is built in a residential quarter of the city and consequently part of

14 the amenities of the neighboring sites is destroyed. Pigou then argued that the activity generating the externality should be held liable for the damage that results from it. He proposed that the com­ pensation (and tax-subsidy) principles should be applied to all situa­ tions where marginal private and social products differed.

1 2

A. Marshall, Principles of Economics, London: MacMillan and

Co., 8th edition, 1930.

"^Ibid., p. 445.

"^A.C. Pigou (1920), pp. 184-185.

7

It is noted that land-use controls and zoning were not discussed by Marshall or Pigou. Marshall was basically interested in the opera­ tion of the real estate market and how land values were determined in this market. Pigou was mainly interested in the possibility of govern­ ment intervention to improve the efficiency of the market economy. He was concerned primarily with "solutions" to the externality problem in general. (The real estate market was not of interest to him.)

Early discussion of land-use controls and zoning was contained in the works of major writers who have come to be known as the Land

Economists (Robert Haig and Richard Ratcliff among others). They usu­ ally observed that neighborhood externalities might produce important inefficiencies and inequities in the urban land markets. Some form of zoning might be necessary to eliminate or mitigate the effects of these externalities. Robert Haig, for example, argued that:

. . . unless social control is exercised, unless zoning is fully and skillfully applied, it is entirely possible for an individual to make for himself a dollar of profit but at the same time cause a loss of many dollars to his neighbors and to the community as a whole. . .

A survey of all the important land economists' works is beyond the scope of this study. It suffices to remark here that these authors never attempted to establish a linkage between their insight and the literature on externalities and welfare economics initiated earlier by

Pigou.

Modern writings on land-use externalities and public policy measures really began in the early 1960s. Otto A. Davis and Andrew B.

^Robert Haig (1926), p. 433. V

8

externalities are present in urban property markets and demonstrated that the existence of neighborhood externalities may produce market

15 equilibria which are not Pareto optimal. They discussed alternative arrangements which may prevent inefficiencies resulting from externali­ ties and also proposed a cost-benefit criterion for urban renewal

16

purposes.

A series of articles followed and the same authors (Davis and

Whinston) published in 1964 their "Economics of Complex Systems: The

Case of Municipal Zoning." In this article, they attempted to determine the conditions under which a market mechanism can be expected to operate efficiently in urban property. The results obtained can be stated brief­ ly: if each household's well-being at a specific location depends on who its neighbors are, then a completely decentralized and unregulated

17 market mechanism cannot be expected to achieve Pareto-optimality. To mitigate the effects of interdependencies in the urban property markets, they observed that zoning constraints may be used to permit the market to perform more efficiently, and they made the following policy

^Davis and Whinston (1961) used a game-theoretic model in their analysis with an example known as "The Prisoner's Dilemma." For an ex­ planation of this type of game theory, see Duncan and Raiffa, Games and

Decisions (1957), pp. 94-102.

16

Notice that although the importance of neighborhood externali­ ties was well established, scholarly work on the economics of zoning was not written until a few years later.

17

Davis and Whinston (1964), p. 443. The authors also stated that "if independence is present, then individual action is sufficient for the market mechanism to produce prices with sufficient information content to lead the system to Pareto-optimality" (Ibid., p. 443).

recommendations: (a) municipal zoning ordinance should encourage sub­ division type developments; (b) special restrictions should be imposed

18

on land usage near the boundary lines of zoning districts.

The article by Davis and Whinston is not a definitive work on the economics of zoning. It is not concerned with land-use control problems. For example, it ignores the critical question of how much land to allocate to each of the homogeneous zoning districts. However,

Davis and Whinston have made a basic contribution by investigating in detail the relationship between zoning regulations and the traditional economic literature on externalities and welfare economics.

Another study of the urban land market which takes into account neighborhood effects is Martin Bailey's "Note on the Economics of Resi-

19 dential Zoning and Urban Renewal." He argued that the public policy on land use control is based on the proposition that "the completely unrestricted use of land (especially urban land) by its owners according to what each deemed to be in his interest would lead to results injurious to all of them as a group. . since "most forms of land use . . . have

2 0 beneficial or harmful effects on neighboring properties." An example is that of nuisance. The particular kind of nuisance with which Bailey is concerned is, in his words, "the nuisance of people themselves when

18

The reason that boundary areas should be regulated is presum­ ably to minimize externality spillovers; this notion is similar to the

"greenbelt" principle (the interested reader can see Clarence S. Stein

(1966), pp. 119-187, for a discussion of "greenbelt").

19

Martin J. Bailey (1959).

20

Ibid., p. 288.

10

they live adjacent to other people whose tastes, habits, and incomes are

21

markedly different from their own." He made the distinction between mutual and unilateral nuisance. (In the case cited above the nuisance is unilateral rather than mutual.) There would be no reasons for inter­ vention if the nuisance were mutual. Bailey's main contribution was the establishment of two alternative criteria for determining whether or not there is a need for zoning restrictions in any particular residential area. These criteria relate to the land values pattern across the neighborhood in question. He maintained that neighborhood externalities will cause reductions in land value in any residential area which is adjacent to an "undesirable" land use; these price reductions will in turn cause locational adjustments along the boundary area separating the two kinds of use. The result: is a suboptimal or inefficient allocation of land between the two activities. Land use zoning would be used as one method of effecting optimal spatial arrangements. Thus, Bailey's article directly attacks the land-use control problem, i.e., the deci­ sion about how much land should be allocated to each economic activity in an urban area.

3. Review of Econometric Literature

The theoretical studies cited in the previous section have as­ sumed that neighborhood externalities are important. Neighborhood ex­ ternalities in the econometric literature discussed in this section are instead viewed as a phenomenon whose existence and importance is a hypothesis to be tested empirically. If the hypothesis is accepted,

21

Ibid., p. 288.

11

then the effects of neighborhood characteristics appear in the prices which single-family homes receive in the urban property market. In reality, there are several factors that are expected to determine the price of a particular house. These factors are, among other things, the physical features of the structure, its accessibility to work place and shopping centers, the property tax/public service package, and its environmental or neighborhood characteristics. To isolate the influence of neighborhood externality effect alone, it is necessary to control for the independent influences of all the other variables. Two general approaches have been employed in the econometric literature: (a) com-parative approach; and (b) multiple regression techniques.

In the comparative approach, the effect of neighborhood on prop­ erty values can be estimated by matching neighborhoods with all but one similar characteristic. For instance, the value of similar single-family homes in two neighborhoods, one near a shopping center and the other contiguous to other detached single-family homes, can be compared. Price differences found in the two neighborhoods would then be attributed to

22 the single variable which has not been held constant. Laurenti ' and

23

Nourse are probably the two most notable studies ever to employ this approach. This approach is quite simple but has one major drawback--it is very difficult to identify two neighborhoods that are similar in all ways except one.

The second method, favored by many researchers, is the use of multiple regression techniques to examine change in property values.

"^Laurenti (1960).

23

Nourse (1963).

12

House price is regressed on a set of explanatory variables to examine why price differs from one property to another. It is assumed, a priori, that these variables affect property value. The results of the analysis test the relative importance, if any, of each variable in explaining the value of the property. A major advantage of this technique is that

"control" properties or neighborhoods are not necessary. One major dis­ advantage is that considerable effort is required to obtain data on all variables included in the regression equation.

The econometric studies which are discussed in this section use

24 this "hedonic" regression technique.

Eugene Brigham's article on the "Determinants of Residential

Land Values" attempted to determine how land values varied along two radial lines (Ray 1 and Ray 2) emanating from the center of Los Ange-

25 les. Variables included in his regression equation were: a. accessibility index (either distance to the Central Business

District (CBD) or an employment potential index)? b. a topography dummy to account for the land height variations around Los Angeles; c. a neighborhood quality variable.

24

The "hedonic" regression technique is discussed in Section 5 of Chapter 2.

25

E. Brigham (1965). Other studies of determinants of residen­ tial land values include Robert Edelstein (1974), and Grether and

Mieszkowski (1974). A number of publications sought to study one partic­ ular aspect of housing value: for studies of air pollution, see Ridker and Henning (1967), Polinsky and Shavell (1975), K. Small (1975), V.

Kerry Smith (1977); for studies of racial discrimination, see R. Muth

(1969), M. Straszheim (1974), Kain and Quigley (1975), C. Daniels (1975),

A. Schnare (1976). They are not reviewed in this study.

13

Unable to obtain direct measures of neighborhood quality, he chose such proxies as average block income, average block building value and extent of dwelling-unit crowding to approximate neighborhood charac­ teristics. The results of his study indicated that distance and job accessibility were important in both rays. Neighborhood quality was also important, though not as significantly as one would have expected

(the t-statistic for this coefficient in his equation 1 -was 1.15). In general, Brigham's findings somewhat support the neighborhood externality hypothesis.

Following Brigham there were several more articles of a similar type. As mentioned in the introductory section of this chapter, the more important studies by Crecine et al., Rueter, Maser et al., and

Stull will be discussed.

The results of the study by Crecine, Davis and Jackson do not support the notion that either external diseconomies or economies abound in the urban property market, but instead suggest that there is a great

26 deal of independence as a characteristic of this market. The Crecine et al. article was written in an attempt to verify or to refute the belief that non-market land use interactions were important features of urban areas. This view was stated earlier by Davis when he noted that:

Despite various statements on the subject it should be abun­ dantly clear at least to economists that the desire for zoning restrictions arises because of the presence of external effects in the urban property market .... An external economy will increase the capital value of affected properties and an ex­ ternal diseconomy will decrease the value.

26

Crecine et al. (1967), p. 93.

27

Davis (1963), p. 375.

14

Crecine et al. found evidence contrary to this belief by studying a hedonic index of single-family house prices in Pittsburgh between 1956 and

28

1963. These authors have included in their regression equation as ex­ planatory variables various zoning type externalities and non-zoning type externalities. The zoning externalities, or externalities which should be related to land uses, were measured as the percentage of the

29 total land area devoted to the particular use in the neighborhood.

The dependent variable was the price per square foot of single-family dwellings. Crecine et al. stated that the sign of an estimated coeffi­ cient should indicate whether the associated independent variable caused an external economy (positive) or diseconomy (negative). Similarly, the magnitude of an estimated coefficient should give an indication of the relative strength of the externality since the independent variables are measured in terms of percentages. Their empirical results showed that the signs of the estimated coefficients varied. This suggested to Cre­ cine et al. that land uses could give rise to either external economies or diseconomies, and these results "cast serious doubt upon this

• 23

They credited Brigham (1965) among others for this technique of hedonic price index. The basic references for the construction of this type of index ar«D Court (1938), Stone (1956), Adelman and Griliches

(1961), and Griliches (1961 and 1967). Note that one of the earliest attempts to apply this technique to house prices was Bailey, Muth and

Nourse (1963). Other recent studies that use hedonic price technique are Kain and Quigley (1970), King and Mieszkowski (1973), Grether and

Mieszkowski (1974), and Peterson (1974).

29 .

Neighborhood in Crecine et al. study was defined to be the census block in which the sale took place. Note that when a sale was on the boundary of such a block, this definition does not include the properties across the street. Crecine et al. estimated a hedonic price index for each zoning category in each of five census tracts.

30

uniformity (of taste) assumption" within the urban property market.

Also, only 16 out of 108 coefficients (in 10 equations) were statisti­ cally significant. In fact, several coefficients had t-statistics below

31

0.01. This suggested to the authors that the results might be

"random." Their hypothesis was that the independent variables under consideration produced neither external economies nor diseconomies upon single-family dwellings. To support this hypothesis they pointed out that if the types of land uses considered gave rise to random effects then a regression equation estimated on the basis of a set of random observations would be expected to yield some significant coefficients.

In fact, one would expect to find 10 percent of the coefficients sta­ tistically significant at the 10 percent level, 20 percent at the 20 percent level and so forth. Crecine et al. presented a table of ex­ pected and actual number of coefficients falling in each significance

32 level. They performed a Chi-square goodness of fit test and found that the resulting Chi-square statistic obtained was 14.4 which, with

33 nine degrees of freedom, was not significant at the 10 percent level.

Their conclusion, therefore, was that the empirical results "do not support the notion that external diseconomies (or external economies)

^Crecine et al. (1967), p. 92.

31

Op. cit., Table A, pp. 97-98.

32

The expected numbers were based on the hypothesis of complete randomness so there were equal numbers in each cell (see Crecine et al.,

Table 5, p. 93).

33

The critical value at the 10 percent level with nine degrees of freedom is 14.7.

16

abound in the urban property market. Instead, these results suggest

34 that there is a great deal of independence in that market."

Rueter repeated the Crecine et al. study and arrived at simijar conclusions. He stated that "the only defensible interpretation of these

35 results is that these markets exhibit considerable independence." The general approach adopted by Rueter and Crecine et al. to study the prob­ lem is almost identical. Market transactions in Pittsburgh were used to

36 construct a hedonic price index. The magnitudes and signs of the co­ efficients of those indexes were compared with those suggested by the zoning ordinances. However, there were some differences in these two studies. Rueter used a larger data base and included more variables concerning the specific houses sold (e.g., lot size, slope of the lot, height of structure, assessed value of the structure). The dependent variable in his equation was the house price. Another change made by

Rueter was to include several new land use variables into the price indexes. These new variables consisted of 26 new zoning type and nonzoning type externalities. Also, in Rueter

1 s study, the observations were from 1958 to 1969 as compared to 1956 to 1963 for Crecine et al.

The concept of neighborhood was redefined by Rueter. It is noted earlier that Crecine et al. had used the percentage of the census block in which the transaction took place that was devoted to each particular land use.

34

Crecine et al. (1967), p. 93.

36

As noted earlier, the hedonic price approach is discussed in

Section 5 of Chapter 2.

17

Thus, the neighborhood effect was considered to be that included in the relevant land use block. The critique of this definition is that when­ ever a transaction took place on the boundary of a block uses that might lie across the street, but yet in a different census block, it would not be included in the neighborhood. Rueter redefined the neighborhood to be an area within 150 feet of the property whose transaction price

37 was observed. This definition was for the zoning type externalities only. The data describing non-zoning type externalities (e.g., popula­ tion density, racial composition, size of the structure, state of dis­ repair of adjacent dwellings, owner-occupation or renter-occupation of the buildings) were not available at a level of disaggregation smaller than the census block. Hence, the non-zoning type externalities were measured in all census blocks within 150 feet of the property which has been sold.

Obviously, by improving the empirical definition of a neighbor­ hood, extending the period of observations, expanding the study to 11 census tracts, including better controls for house specific amenities,

Rueter has corrected many of the shortcomings of the Crecine et al. study. Despite all those changes, Rueter arrived at almost the same conclusions found in the earlier study. His empirical results showed that the magnitudes of the estimated externality coefficients varied both across zoning categories and within zoning classifications across census tracts. Variations also were observed in the signs of these estimated coefficients. Furthermore, most of the coefficients were

37

Rueter also used a 300-foot radius as an alternative defini­ tion of neighborhood.

13

statistically insignificant. Following the lead in Crecine et al. study,

Rueter performed a Chi-square goodness of fit test and found that only five of the 16 resulting statistics were significant at the 10 percent level. He concluded that: the only strongly defensible implication that can be derived from this exercise is that there is much more independence in^g urban property markets than the zoning ordinance anticipates.

Further evidence regarding land use interactions comes from a

39 study by Maser et al. These authors analyzed the effects of zoning in the urbanized area of Rochester, New York (Monroe County) in order to determine whether zoning modifies outcomes in the urban land market, or whether market forces negate the forces of land use controls. The tests again involved studying the coefficients of a hedonic price index.

Maser et al. selected a sample of about 11 percent of all real estate transactions recorded in Monroe County during each of three years:

1950, 1960, and 1971. This index included the assessed value of the structure, the site's zoning category, the site's accessibility to em--

40 ployment and services, and other measures of environmental and house characteristics. Maser et al. found no land price effects attributable to zoning. They concluded that "the externalities which zoning is sup-

41 posed to prevent could not be detected. ..." Zoning, in their view,

3R

Rueter (1973), p. 334.

39

Maser, Riker, Rosett (1977).

40

Maser et al. used "traffic density" to account for access to local subcenters. They also used "population density" and "house size" as proxies for proximity to centers of employment.

41

Maser et al. (1977), p. 128. They did find, however, in an­ other part of their study that airport noise and the presence of nearby bodies of water (lakes) significantly affected land values.

is ineffective. Consequently, they recommended consideration of alterna­ tives to zoning, particularly a greater reliance on the judicial system. contain some evidence supporting the hypothesis that non-market land-use externalities are relatively unimportant. Despite many shortcomings, the effect of these articles would have been very strong if it was not for a countervailing study with significantly contrasting conclusions.

A study that finds important land-use externalities is that of

42

William Stull. Using 1960 census data and other information on land use patterns and tax/assessment practices from 40 suburban cities and towns in the Boston metropolitan area, Stull estimated the effect of nonsingle-family land uses on the value of owner-occupied single-family homes. The procedure again involved studying the coefficients of hedonic price equation. This equation included a number of variables to control for a particular structure's accessibility and physical charac­ teristics, and tax/public service package. He also included a set of environmental variables to account for the effect of various nonsinglefamily land uses. His measure of these variables was the proportion of community land devoted to multiple-family, commercial, industrial, in­ stitutional, or vacant/agricultural uses. The estimation technique

42

Stull (1975). In an earlier study, Kain and Quigley (1970) have attempted to measure the value of housing quality and calculated an index that measured the effect of commercial and industrial uses. They found that this index exerted a negative effect on the value of single family homes. Nevertheless, it is difficult to interpret this result because the index is the product of a factor analysis involving 39 vari­ ables and therefore subject to the influences of many characteristics other than commercial and industrial uses. For this reason, the Kain and Quigley article is not to be examined more thoroughly in this study.

20

involved both ordinary least squares (OLS) and two-stage least squares

43

(2SLS). The results of his OLS and 2SLS estimating procedure revealed that "in the study area households were fairly sensitive to the land use

44 environments of the communities in which they purchased homes."

Stull's conclusions are directly contrary to those obtained by the two Pittsburgh studies (Crecine et al. and Rueter) and the Monroe

County, New York study (Maser et al•). According to Stull, "property markets in the Boston SMSA were characterized by a substantial amount of interdependence in that the value of the typical single-family property

45 depended significantly upon community land use patterns." The coeffi­ cients of the variables representing multiple-family, commercial, vacant, and industrial uses are statistically significant. These coefficients have negative signs, suggesting that these uses exert a negative effect

46 on the value of single-family homes. Specifically, Stull's use of regression techniques led to the conclusion that if 10 percent of a community's land is converted from single-family homes to other uses the

43

Stull used a 2SLS procedure because he suspected a simultaneity between property values and public sector characteristics. His 2SLS estimates were very similar to their OLS counterparts.

44

Stull (1975), p. 551.

45

Op. cit., p. 552 (emphasis added).

46

Stull includes commercial uses and also commercial uses squared variables. His hypothesis is that a small amount of commercial activity is desirable and larger amounts are undesirable because of increased traffic or noise. The coefficient of commercial uses is found to be positive and that of commercial uses squared negative. According to him the maximum house value occurs when the proportion of commercial is about

5 percent, other things equal (Stull, p. 551).

average 1960 value will be lowered by $630. It is noted that Stull de­ fined a home's neighborhood to be its entire town. His conclusion that neighborhood externalities exist was based on his findings according to which the coefficients of two of the five land use variables were sig-

47 nificantly negative at the 0.05 level.

Note that Stull has utilized two-stage least squares to consider the simultaneity between the local fiscal sector and housing value. The lack of a theoretical framework for this procedure, however, makes its use ineffective. Stull recognized the fiscal variable as endogenous, but he eliminated local expenditures from his study. Also noteworthy is his accessibility measure. Stull's sample is taken from Boston sub­ urban towns, yet his only measure of accessibility to work place is the ratio of employment in the individual towns to the number of housing units in them (this ratio actually is "jobs per household"). He did not use any measure of the access of the housing units in one particular town to employment sites in adjacent towns of the sample. This question will be discussed in more detail in Chapter 4.

R.N. Lafferty and H.E. Freeh III extended Stull's work by view-

48 ing each land use as generating externalities at two levels. One has effects at short distances, thus small neighborhoods, while the other has city-wide effects. They repeated Stull's experiment using a two-part definition of neighborhood that allowed for the existence of two types

47

At the 0.10 level, four of the five land use coefficients were significant.

48

Lafferty and Freeh III (1978).

22

of land use externalities. Use of this definition resulted in the same conclusion as Stull, i.e., neighborhood externalities exist. But with this neighborhood definition, the coefficients of Stull's city-wide land use variables changed sign and were positive. The authors asserted that the positive sign of the regression coefficients for the multiple-family uses and for the industrial uses may seem to run counter to expectations.

But, they argued, "the positive signs are not as surprising when it is considered that many of the negative externalities affect local neighbor­ wide positive externalities such as fiscal inputs (not captured by the employment and tax variables) that overcome the negative externali-

49 ties." This question will be taken up and amply commented upon in

Chapters 4 and 5.

4. Summary and Conclusions

(a) Despite the widespread practice of land-use planning and zoning in the United States, it is somewhat surprising that there exist only a few empirical studies designed to test for the existence and magnitude of externalities of neighborhood land uses, and that the re­ sults of these studies are contradictory. The evidence on the issue is mixed.

(b) The use of various definitions of neighborhood has probably resulted in the discrepancy in findings of the previous empirical stud­ ies. With the town-wide externalities being generally positive and

49

Ibid., p. 389.

23

50

fairly highly correlated with the negative local neighborhood measures, as indicated by Lafferty and Freeh III, the insignificant results of the small neighborhood studies of Pittsburgh and New York were possibly caused by bias resulting from omitting the measures of land use at a wider level (census tract, group of tracts, or city-wide level). And

Stull's study, using only city-wide variables, would be biased in the direction of finding a negative influence where the true one was zero or positive.

(c) The empirical evidence for the role of externalities in the determination of property values is far from conclusive. As noted in the above paragraphs, the contradictory results may have been attribu­ table to the authors' definition of a neighborhood. But, more important perhaps, the above studies did not fully explore the theoretically am­ biguous nature of the effects of distance to nonresidential sites on single-family homes. As Edwin Mills has judiciously emphasized, people must travel to CBDs to work and shop, and land values close to the CBD are high because such sites provide cheap ac­ cess to the CBD. The same is true, on a reduced scale, of any nonresidential land use anywhere in a metropolitan area. Every commercial and industrial site generates employment and/or shopping, and proximity to them is valuable. Thus, residential land values may even fall with distance from a nonresidential site; but that does not imply that there are no external dis­ economies from the site, only that they are more than offset by the advantages of proximity. Future studies of externalities from nonresidential activities should take account of both the advantages and disadvantages of proximity to them.^

1

The use of cities as observations required that the local neighborhood effects be aggregated to the city level, hence the correla­ tion between the city-wide measure of land use and the local neighborhood measure.

51

Mills (1979), pp. 525-526.

24

(d) In reviewing the existing literature it is apparent that in both theoretical and empirical models of residential location, the con­ cept of accessibility and its related transport costs were of primary

52 concern throughout the articles discussed in Sections 2 and 3 above.

But it is also noted that, almost without exception, these models have adopted an unrealistic assumption that there exists in a city only one center of economic activity (the Central Business District), where all employment is concentrated. One does not expect these monocentric models

53 to produce accurate representations of modern cities.

5. Objectives of the Study

Given the contradictory results and methodology limitations of previous research, this dissertation attempts to accomplish the follow­ ing objectives:

(1) Most narrowly classified as a study of land use or neighbor­ hood externalities, the primary objective of this research is to identify and estimate the quantitative impact of nonsingle-family land uses on the value of single-family homes;

(2) Other objectives are: (a) to expand the previous methodology

54 by recognizing the multi-nucleated character of modern cities and by

52

In the more recent papers, other considerations have also be­ come important, such as environmental quality, amenities, neighborhood characteristics, etc.

53

The original concept of a monocentric city was developed by

Von Thunen in 1826. For a discussion of monocentric models, see Romanos

(1976) and Wheaton (1979); both discuss Von Thunen's work.

54

This study argues that urban areas contain more than one center of economic activity and a household takes into account the location and magnitude of all these centers.

25

taking into account both the advantages and disadvantages of proximity of residential sites to nonresidential activities; (b) to evaluate the influence of public sector characteristics on property values, in particu­ lar to estimate empirically the impact of property taxes and local public

55 school quality on the value of single-family homes; and (c) finally, to draw policy implications for which this research is relevant.

This is very much in the spirit of what has come to be known as the Tiebout Hypothesis. Indeed, Charles Tiebout (1956) developed a theoretical model involving consumer location in accord with preferences for local public goods and services; he argued that households would choose a residential neighborhood on the basis of its package of public services and taxes; this implies that households will compete for de­ sirable locations and will bid urban property values up, capitalizing locational advantages.

CHAPTER 2

EXTERNALITIES AND RESIDENTIAL PROPERTY

VALUES: A THEORETICAL FRAMEWORK

1. Introduction

The analysis presented in this chapter considers theory useful in analyzing the effects of externalities on the urban property market.

In Chapter 1, the term "externalities" has been briefly defined. it will be examined here in more detail.

It is generally agreed that households choose among prospective residential locations at least partially on the basis of neighborhood.

To understand urban land use control and the relationship between ex­ ternal effects and property values, it is necessary to devote some anal­ ytical attention to the neighborhood externality concept and the theoretical basis for government controls of land use external effects.

This will be taken up in Section 2.

To help clarify the concept of external effects, some important nonsingle-family land use externalities will be discussed in Section 3.

Then, in Section 4, a model of consumer bidding behavior in an urban property market will be developed. The conclusions of this anal­ ysis about consumer bids are found to have significant empirical applicability.

After the discussion on theoretical foundations, Section 5 will present a functional specification of the model, using the hedonic price

26

27

approach to study the influence of externalities on residential property values.

2. Neighborhood and Land Use Externalities

Externalities are typically defined as arising whenever the value of one agent's objective function (a consumer's utility function or a firm's production or profit function) depends on the unintended or inci­ dental byproducts of some activity of others."*' Consider the case of a consumer. His utility function would be written as:

U. (X..,..., X. l 11 In

X ml ran

X ) (2.1) where X^ is the amount of j consumed by person i. This formulation assumes n goods indexed by j and m individuals indexed by i f and illus­ trates the definition of an externality in that individual i's utility is

2 shown to be dependent on other individuals' consumption.

Arrow argued, however, that this definition is inappropriate be-

3 cause externality has been defined too broadly. He has shown that in a barter economy, for example, where aggregate supplies of consumption commodities are fixed, this definition would label all goods as external­ ities. Each individual's utility would directly depend on what his trad­ ing partner is willing to give up. There would be a direct link between utility functions as specified in (2.1), and each good would give rise

^Steven A.Y. Lin (1976), p. 1.

2

In general, this utility function will assign a zero weight to many of the X..'s.

ID

3

K. Arrow (1970).

28

to externalities. If this is the case, the term externality becomes meaningless since it does not define a specific class of phenomenon. A more accurate definition of externalities must go beyond expression (2.1) to include some reference to the market structure.

The relationship between the existence of externalities and

4 market structure has been studied by Arrow. According to him, rela­ tionships of the type specified by (2.1) do not cause inefficiencies if there is a sufficiently rich set of markets. The existence of a competi­ tive market system would solve the problem of having everything classi­ fied as an externality. Specifically, the definition of commodity space has been reinterpreted. It consisted basically of introducing a good

X. which is defined by X. = X. . for all individuals. X. is the ljk 13k lj

13k individual k's perception of individual j's consumption of good i. The utility function now is U, (X n

,, k Ilk

X , ), and only variables proper mnk to the individual will appear in his utility function. With this reinterpretation, Arrow showed that the "equivalence'' between the set of allocations that are Pareto optimal and those that represent competitive equilibria are still valid. This "equivalence" will hold unless there is a market failure.

Externalities thus require a two-part definition: a relationship as specified in (2.1) and the failure of a market. The study of neigh­ borhood and land use externalities must therefore start with a study of the urban land market. If Arrow's conclusions are accepted, there would be no land use externalities in a perfectly competitive land market. If

4

Arrow (1970).

a person A is offended by his neighbor B's land use, he would simply purchase the right to have B reduce the amount of the land use. What

29

would develop in this situation would be a market for property rights.

However, because of their peculiar nature, the goods to be mar­ keted in this situation create many difficulties. A private agreement between A and B to decrease the level of the offending land use cannot prevent other residents in the neighborhood from enjoying the reduced external effects even if they refuse to contribute to the cost of ex­ ternality reduction. In other words, there is a free rider problem which can lead to a sub-optimal supply of the good. Thus, it is claimed, the government must ensure adequate production of public goods by land use controls or other means.

Essentially, there are two considerations that together may justify government intervention to limit land use external effects: the public good nature of the effects (free rider problem) and high trans­ actions costs of private agreements. These are the necessary conditions for government intervention. The sufficient condition requires that the government can estimate accurately the benefits and costs of externality abatement to each resident. If the government does not know the magni­ tude of benefits, there is no apparent mechanism by which it can be more successful than private parties in inducing people to reveal their pref­ erences for public goods.

It was noted earlier that the transaction costs of private nego­ tiations for public goods may be so high that the goods will be under­ produced. But, government intervention also involves transaction costs.

To say that government transaction costs are less than private transaction

30

costs in optimizing public good production is to admit the use of co­ ercion by the government when dealing with unwilling participants.

The theoretical basis for government control of external effects having been examined briefly, the next section shall discuss the nature and magnitude of nonsingle-family land-use external effects.

3. Nonsingle-family Land Use Externalities

As indicated in Chapter 1, evidence provided by previous studies on the magnitude of nonsingle-family land use externalities is mixed.

In other words, the findings of these studies concerning land use control

(and zoning, in particular) are far from being conclusive. To study the effects of zoning in controlling land use externalities, it is necessary to examine first a typical zoning ordinance. In Tucson, for example, a typical set of zoning district classes is the following:

—Residential, single-family

—Residential, multiple-family

—Commercial, local

—Commercial, general

—Industrial, light

—Industrial, heavy.

In addition, there are two other general categories of land use which are controlled in one way or another by most zoning ordinances but for which special districts are not usually set up:

--Public and semi-public

—Vacant land.

\

31

The land uses listed above shall be examined one at a time, in order to determine the probable effect of each on single-family home property values.

Multiple-family land-uses in this study include two-family homes, condominiums, apartment complexes, and multiple residential with limited commercial and professional uses permitted. They are conventionally thought to be undesirable additions to single-family neighborhoods.

Some judicial opinions even gave derogatory and invective comments about

5 apartment houses. If one examines zoning and rezoning disputes in which homeowners try to keep multiple-family structures out of their neighborhood, it is apparent that class, ethnic, and racial considera­ tions play as important a role as the eventual erection of the struc­ tures. Another possible objection to the construction of multiple-family housing units is the burden they will allegedly impose upon the local public sector, especially the school system, by the influx of new resi­ dents.^ If these allegations are true then multiple-family land uses probably exert a negative influence on price due to the neighborhood externalities which they generate.

However, some authors argued that residential areas composed of mixed dwelling types (single-family, multiple-family) are more interest­ ing esthetically, more satisfying socially, and have the advantage of

^Justice Sutherland's opinion in Village of Euclid vs. Amber

Realty Co., 272 U.S. 375, 1926, see Williams (1966).

6

Norman Williams (1966), p. 51, observes that the legal status of such arguments is still in doubt and there are several cases on both sides of the question.

32

offering housing appropriate to the different stages of family needs

7 without the necessity of the family leaving the neighborhood.

The idea of the single-family residential use primacy has been often under attack in the literature, and the views by homeowners that residents in multiple-family housing units are less "socially desirable" were frequently based on questionable information: while residents of these multi-family dwelling units often have lower income and are rent­ ers, none of these facts per se should automatically reduce the commung ity's quality of life.

Also, a more positive view is found in a recent judicial opin-

9 ion. In a Pennsylvania case involving a proposal to build a multiple dwelling unit in a residential district, the Court upheld the decision to approve the proposed development, despite neighboring homeowners' objection. In the Court's judgment "the constant augmenting of our population . . . and the increasing migration of people from countryside to urban or near-urban sites are making multiple dwellings more and more a necessity . . . Multiple-dwelling living indeed became a popular trend over the past two decades and proponents of multiple-family homes, have grown considerably in numbers. They are not unaware of the negative influence created by an eventual increase in population density in their neighborhoods. However, they seem to be motivated by economic incentives

7

Leary (1963), p. 425.

8

, p. 353.

9

.1952, Williams (1966).

"^Ibid., cited in N. Williams (1966), p. 65.

to accept the existence of multiple-family homes in their area. When­ ever apartment complexes are permitted to be built in a district, land in the neighborhood would tend to increase in value. And land price is an important component of real property values. The probability of rezoning and speculation about future appreciation of property values in this context of land use admixtures would exert a positive influence on single-family home price. Furthermore, in two recent empirical stud­ ies, several authors have found that the effects of apartment develop­ ments are fairly localized, within only 80 to 250 feet; as the distance between a household and the apartments increases, the effects seem to be

12

dissipated rapidly. Therefore, rational behavior dictates that home­ owners weigh the negative effects against the positive ones. The net effect would be indeterminate. It all depends on the relative strength of each of the two opposing forces. There is no a priori conclusive answer to this problem. Essentially it is an empirical question.

Commercial land uses as they are defined in this study consist of local business and retail business with wholesale and entertainment enterprises permitted. Most zoning codes handle this diversity of uses by dividing commercial activities into "local business" and "general

11

Note that the multiple-family land-use category includes mul­ tiple residential with compatible commercial and professional uses per­ mitted. A survey of the per acre values of residential and commercial/ industrial land is presented in Irving Hoch (1969), Table 5, p. 109.

The weighted average value in 19 large cities in the U.S. is $201,400 per residential acre and $277,700 per nonresidential acre (vacant land excepted).

12

Tideman (1969), Grether and Mieszkowski (1974).

34

business" subclassifications. The corner grocery stores go in the former; automobile dealerships go in the latter.

It seems likely that most neighborhoods would support some local commercial enterprises. Substituting some amount of local commercial activity for residential homes may raise property values in the neighbor­ hood rather than lower thern. However, the increase in the scale and intensity of commercial activity beyond some acceptable level would be ejected to depress property values. In general, commercial land uses offer both "desirable" characteristics (accessibility, proximity) and

"undesirable" characteristics (noise, traffic congestion). Their net influence on residential property values is a question which cannot be settled theoretically. Empirical evidence must be brought to bear on this question.

Some recent literature suggested that mixed land uses (e.g., commerce and recreation) should be viewed as a beneficial contribution to contemporary urban development. D. Procos, for example, asserted that commercial activities, far from exerting a drag on recreational and cultural innovation, have enhanced recreational activities. He indicated that complete "social disengagement" from commercial land uses is not advisable, and there is a stronger case to be made for diluting commer­ cial activities in other land uses than for isolating them in enclaves."^

13

D. Procos (1976), pp. 48-49, cited the Canadian example of the

Leaf Rapids, Manitoba, Town Center, which is an agglomeration of public offices, health facilities, shopping center, athletic facilities, school, cultural center, library, hotel.

35

Proximity of single-family homes to such complexes offering a mix of commerce and leisure may well exert an advantageous effect on residential property values.

Industrial activities, according to the conventional belief, belong to the class of "undesirable" land uses. This is generally true in the case of heavy industry. Included in this category are: factor­ ies, quarries, etc. Most zoning ordinances distinguish between "light'' and "heavy" industrial activities on the basis of various nuisance cri­ teria. Obviously light and clean industries affect the residential property values less than do heavy industries, since they are respon­ sible for a smaller degree of nuisance in terms of noise, air and/or water pollution, etc. In other words, the strength of their effects is different. The two uses could conceivably influence housing prices in different directions. It is necessary to observe that industries are not only an important (or even unique) source of local employment but also a substantial generator of revenue for local governments. One can very well argue that it is not unreasonable to assume that there is a positive relationship between manufacturing activity and property values which operates through local taxes and local employment. Other positive aspects are that many modern industrial enterprises are beautifully de­ signed and occupy a relatively large area with nice landscaping. Thus they provide both open space and visual amenity values. Households need to take into consideration in their housing bids both the "desirable" and "undesirable" aspects of industrial activities. Again, just as in the cases of multiple-family and commercial land-uses, the sign of the

14

industrial land coefficient is a priori indeterminate. The signifi­

36

cance of industrial land use as a determinant of residential property values is an empirical question.

Public and semi-public land uses include churches, schools, universities, museums, hospitals, parks, public buildings. Perhaps none

15 of them generate important nuisance. They are generally attractive in appearance. In many cases they provide open space and recreation facil­ ities (e.g., public parks) to local residents. Uses in these classes should have a comparatively favorable effect on property values. The problem of concern here is whether public and semi-public land uses raise property values more than single-family uses do. If so, then the net effect on price will be positive. If not, it will be negative.

There is no a priori answer to this question. Some of these public and semi-public land uses could have a positive net influence on price. A university may have this effect because of the cultural

14

According to J. Jacobs (1961), mixing of land-uses in cities is not harmful. In her words, "A second category of uses is convention­ ally considered, by planners and zoners, to be harmful, especially if these uses are mingled into residential areas. This category includes bars, theaters, clinics, businesses and manufacturing. It is a category which is not harmful; the arguments that these uses are to be tightly controlled derive from their effects in suburbs and in dull, inherently dangerous gray areas, not from their effects in lively city districts. . .

In lively city districts . . . they are positively necessary, either for their direct contributions to safety, public contact and cross-use, or because they help support other diversity which has these direct ef­ fects." (pp. 231-232) on Sundays; students coming to and leaving schools, etc.). Proponents of mixed land-uses (residential-commercial, or residential-industrial) argue that shopping centers do not generate more traffic congestion than churches and schools, cf. Jane Jacobs (1961), D. Procos (1976).

37

facilities which it provides to the community. Churches and schools probably have similar influence, since schools make the neighborhood a more desirable place to live in the eyes of potential home buyers.

Hospitals, however, are not as highly regarded as the previous categories. They may very well have a negative net effect on property values because of various nuisances which they may create (ambulance siren, traffic).

Vacant land is a kind of residual category (open space and other undeveloped lands). In general, they are not distinctly undesirable and most do not provide important local services. Consequently, the net effect of vacant land on property values is theoretically indeterminate.

In this study, two distinct sets of land-use data are used: the

17 actual land-use variables and the "zoning" variables. Land-use vari­ ables describe the actual situation of a neighborhood environment while

Nationally, a city the size of Tucson contains on average about

3,550 vacant acres; Tucson contains over 15,400 vacant acres or nearly

4.35 times the national median. (See Update, an information series of the Tucson Planning Department (n.d.), p. 4). In one-half of all ob­ servation units (26 census tracts), the proportion of vacant land varies between 26 percent and 82 percent of the total acreage. For this reason, vacant land will not be included in the regression equation.

17

An actual land-use variable describes one of the following situations: residential single-family use, multiple-family use, commer­ cial use, industrial use, public and semi-public use. A zoning vari­ able describes the land-use situation as it is appeared on the city zoning map. This variable can be distinguished from a land-use variable by a simple example. Take the case of a R1 zoning district (which is a residential single-family zone) in Census Tract No. 4. It is found in this Rl district not only the acreage devoted to single-family uses

(24.31 acres) but also 2.44 acres of multiple-family uses, 19.00 acres of commercial uses and 2.50 acres of public and semi-public uses. These nonsingle-family uses were allowed to exist by virtue of the non­ conforming use provisions.

38

the zoning variables represent what the situation should be in the eyes of the city planners—situation which exists only on the zoning map and does not reflect the actual land-use conditions in the neighborhood.

Presumably the interaction between household behavior and the neighbor­ hood environment is much more direct in the former case (actual land-use) than in the latter (zoning). In other words, a prospective home pur­ chaser may very well "perceive" the actual land-use situation in a neigh­ borhood while he may completely ignore its zoning condition. And what constitutes "quality of a neighborhood" is exactly a function of the human process called "perception." Consequently, it may be possible that the regression equation using zoning variables would produce insignifi­ cant zoning coefficients due to the phenomenon of non-perception by residents of land use dissimilarities in neighborhood environment. Dif­ ferences in land uses which are actually perceived should be capitalized in residential property values. It is expected, then, that the coeffi­ cients of various land-use variables will be statistically significant.

In summary, the discussion in this section has attempted to il­ luminate the concept of externalities, especially land-use externalities.

The clarification of this concept will help to understand section 4 in which a model of residential bidding behavior is formalized—a model which takes account of both various attributes of the chosen residence and the land-use characteristics of a neighborhood in which the dwelling is located.

39

4. A Model of Consumer Bidding Behavior in an Urban Property Market

An urban resident's life consists of a series of interconnected events or activities: work, recreation, family activities, religious and organizational participation, and social activities. This series of activities is manifested by many episodes of going to work, making a trip to the stores, visiting friends, etc. Intuitively it is not diffi­ cult for an urban resident to see that some notion of accessibility will emerge relative to the dual "desirable-undesirable" concept of land use that he experiences in pursuing these activities. His housing choice can be expressed as a function of some mix of accessibility, amenities, neighborhood qualities, and some other factors involved in his residen­ tial decision. Applying consumer theory to the phenomena under discus­ sion, it may be stated that an individual makes his residential choice by searching for an optimal combination of satisfactions of all those needs essential to his sense of well-being, taking into consideration his series of interconnected activities and his income constraints.

To formalize the residential choice problem, consider a consumer whose well-being depends not only on the goods and services which he himself purchases but also on the way in which land is used in the neighborhood of his residence. The classic theory of consumer choice can be extended usefully to permit this situation to be handled analyti­ cally. It would seem appropriate to include some neighborhood land use variables as arguments in the consumer's utility function, assuming that land use variable is an adequate proxy for neighborhood quality.

Households do not generally object to the presence of nonresi­

40

dential structures in their neighborhoods. They are really concerned with the characteristics of the typical structure. For example if the typical building in the neighborhood is ugly and noisy, then the resi­ dents there would be very displeased. On the contrary, if a well de­ signed and quiet nonresidential site is located in the neighborhood, households in the area might find it desirable or even prefer it to a single-family home since it can generate employment and/or shopping and recreation. In other words, it is the characteristics (e.g., noise, pollution, traffic) of the use rather than the nature of the economic

18

activity being conducted on the land which matters. Land use vari­ ables seem to be appropriate proxies for a set of neighborhood quality arguments to be included in the consumer's utility function.

It is noted that many zoning ordinances are based on the prin­ ciple that "incompatible" land uses should be spatially segregated to avoid their harmful effects on neighborhood amenities. Mishan would

19 call this the principle of "separate facilities." The separate facil­ ities solution to neighborhood amenity problems is the analogue to the smoking and no-smoking car solution of railroad passenger trains. Mishan asked: why not provide separate facilities for everybody in a society in virtually all aspects of the environment? An area in each community could be set aside for people who wish to live in a pollution-free or

18

See K. Lancaster (1966) for a more detailed discussion on

"goods as a bundle of characteristics."

19

E.J. Mishan (1977).

41

noise-free environment. Of course, Mishan's proposal is a controversial issue. It raises many objections and Chapter 5 will probe deeper.

Some recent zoning ordinances have a comparatively less restric­ tive approach on these matters. With modern technology, most industrial processes can be made nuisance-free; it is simply a matter of relative

20 . .

costs. Many cities have turned toward the use of performance standard regulations. Commercial and industrial activities are given a v/ide range of site choices if they are conducted in conformance with speci­ fied performance standards, phrased in terms of the permitted amount of nuisance which may be emitted in each zoning district (i.e., noise, smoke, glare or heat, odors, vibration, dust, fumes, etc.). A study on neighborhood externalities might find "performance" data more useful than land use data. But unfortunately performance data are not avail­ able. Thus, in this research land use data will be employed, as dis­ cussed earlier.

These preliminary remarks having been made, the discussion will proceed by considering a household seeking to purchase a single-family home in an urban area. Since accessibility to job is a key factor in much of the studies on residential site choice, it is assumed here that

21

the household's head is employed. In buying a single-family home, the household not only wants proximity to employment but also a wide variety of housing attributes, such as the structural characteristics of the

2 0

N. Williams (1966), p. 218.

21

Should he be unemployed, work accessibility would be a mean­ ingless concept for him. Furthermore, mortgage financing requires him to provide evidence of regular employment.

42

unit, the local public services, the neighborhood quality, inter alia.

These notions can be formalized by writing the household's utility func­ tion in the following way:

U = U (H, X)

( 2 . 2 )

where H = a vector of housing consumption characteristics,

X = a vector of non-housing commodities.

The H vector consists of several subvectors and can be written as follows:

H = (L, S, A, E)

(2.3) where L is surrounding land use mix characteristics associated with housing bundle (nonsingle-family land use),

S is structural or physical characteristics associated with housing bundle,

A is accessibility characteristics associated with housing bundle,

E is public sector characteristics associated with housing bundle.

Each land use included in the L subvector is assumed to have a dual function:

—desirable function, since it provides accessibility to employment and other benefits;

--undesirable function, because of the amount of noise, pollution, traffic, etc. which it generates.

In mathematical terms, the subvector L can be written as follows:

L = (L

11

L

12

L

21

L

2 2

(2.4)

where L., = desirable characteristic of land use i il

L._ = undesirable characteristic of land use i i2 for i = 1, ..., N.

It is assumed that L., would raise the household's utility, while L. il

1 i2 would lower it, or:

43

au/3L i;L

> 0

(2.5)

8U/8

Li2

< 0

( 2 . 6 )

Using the standard nomenclature, those land uses which raise utility generate positive externalities while those which lower it generate negative externalities. Under these definitions, each land use generates both positive and negative externalities. That is, L., = L.,(L.) il il l and L.„ = L (L.) and 3L._/8L. > 0 and 3L. /3L. > 0. The net effect of i2 l il l i2 l each land use depends on which externality has a relatively stronger influence. (This study is concerned with technological and not pecuniary externalities.) Then, by substituting (2.3) into ('2.2), the utility function becomes:

U = U (L, S, A, E, X)

(2.7)

It is convenient to enumerate the various sets of arguments in the util­ ity function (2.7) as follows:

1. L (land uses)

2. S (structural characteristics)

3. A (accessibility)

44

4. E (public services)

5. X (non-housing commodities)

The partial derivative of U with respect to a particular argument will be noted u^. The i refers to a set of arguments and j to a specific

2 element of that set. Thus, u^ is the partial derivative of the utility function with respect to a small change in the argument corresponding to the second enumerated set, i.e., the structural characteristics of the

22

housing unit.

It is assumed that the signs of the u^ are as follows: u^ ^ 0 u^ > 0

(2.8)

(2.9)

u^ > 0 (2.10)

u

4

> 0

1 u

5

> 0 l

(2.11)

(2.12)

The sign of the total differential of U with respect to , as noted earlier, is indeterminate a priori because the positive external effects may outweigh the negative external effects, or vice versa. The signs of

2 3 4 5 u., u., u., and u. are positive because it is assumed that all the relei l l l vant variables are defined so that increases in them increase the house­ hold's well-being.

2 2

The conventional notation 3U/3 is employed when referring to the partial derivative of utility with respect to a variable.

45

The household's choice is subject to its budget constraint which can be written as follows:

Y = (2.13) where Y = annual income,

P^ = price of non-housing commodity i,

X^ = quantity of non-housing commodity i,

R = annual housing payment,

T = annual property tax payment, g = the transportation cost function,

A = accessibility to work, shopping, etc., as before.

It should be pointed out that A is defined here as accessibility to em­ ployment, shopping, recreation, etc. outside of census tract which is the unit of observation in this study. Accessibility inside census tract is captured by Thus, the household's annual income is ex­ hausted by payments for non-housing commodities, for housing, for prop­ erty tax, and for commuting costs (assuming that there are no savings and no other taxes than property tax).

Let V represent the market value of the housing unit and assume further that:

R = a^v

T = a

2

V where is a credit parameter and is the property tax rate. Substi­ tution of the above into (2.13) yields:

Y = 2P i

X i

+ (a + a

2

)V + g(A, L, X)

46

(2.14)

Obviously, the more accessible the site, the smaller is the com­ muting cost. The sign of 3g/3A is therefore negative:

3g/3A < 0 (2.15)

Also, the more commodities a household consumes, the more trips it makes, the higher is the transportation cost, or:

8g/3X > 0 (2.16)

Since L describes the amount of land in the neighborhood having "de­ sirable" characteristics, it is reasonable to assume that:

9g/3L il

< 0

3g/3L i2

1

(2.17)

In words, this means that if the land in the neighborhood is more de­ sirable and convenient (e.g., more varieties of commodities in a shop­ ping center), the household would make fewer trips to the stores, hence the lower transportation cost. By a contrario reasoning it can be assumed that:

(2.18)

Now, suppose that a single-family home is up for sale and the question is the following: what is the maximum value a consumer will bid for that particular house? Alonso argued that unless one specifies in ad­ vance a given level of utility of the household, the answer is

47

23 — indeterminate. An arbitrary utility level U = U is selected in this

24 analysis. Equation (2.14) can be rearranged as follows:

V = {l/(a + a

2

)}{Y - (2.19)

To determine the maximum bid for a piece of property is to maximize

(2.19) subject to the constraint:

(2.20) U = U (L, S, A, E, X)

The above problem can be expressed in a Lagrangian form:

(2.21) where A is a Lagrangian multiplier. The piece of property in question being fixed, the following variables are parametric to the bidder's decision: A, L, S and E. When the property is varied, they vary.

The remaining parameters are not associated with the property: Y, a^,

P, U. The decision variables under the control of the bidder are the

X^'s. Differentiating (2.21) with respect to X^ and A and setting these partial derivatives equal to zero yield the following first-order conditions:

(2.22) 9xy3x. = {l/(a n

+ a_)}{-P. - (9g/9X.)} + Au

5

= 0 l 1 2 l l l

9£/9A = U - U(-) = 0 (2.23)

23

W. Alonso (1964).

24

The following analysis draws heavily from Alonso, op. cit.,

Chapter 4.

48

Equation (2.22) can be rewritten to obtain the value for A:

}/{

2

) u^} (2.24)

The numerator is positive by (2.16) and because P^ is positive. The denominator is also positive by (2.12) and because (a^ + a^) is posi­ tive. Therefore, A is positive.

A > 0 (2.25)

Repeating the calculation for and dividing the result into (2.24) yields: i

, + (9g/3X

5

/u

5 l :

j l j

(2.26)

This is the usual utility maximization condition from consumer theory which requires that the marginal rate of substitution between any two commodities be equal to their price ratio. Note that the term on the left includes both price (P^ and P ) and the transportation costs

Og/8X^ and 3g/3X^.) required to purchase the last unit of the good.

If the maximum bid is denoted by B, then for every consumer there will be a B associated with each property on which he bids. B would be obtained by substituting the equilibrium values of the x

^' s into (2.19). Recall that the following parameters are associated with the property: a^, A, L, S and E. If one or more of these parameters changed, B would change. The relationship between B and these param-

25 eters can be written as follows:

25

Those parameters not associated with the property are not in­ cluded in (2.27).

B = B(a

2

, A, L, S, E, U)

49

(2.27)

The signs of the partial derivatives of B can be obtained by using the

Envelope Theorem which, in this context, states that:

(2.28) where is the Lagrangian function defined in (2.21) and 6 is any vari-

26 able parametric to the bidding process.

Application of (2.28) yields:

9B/8U = 9</3U = -A < 0 (2.29)

The above result indicates that an increase in the given level of util­ ity of the consumer causes B to fall, other parameters held constant.

A similar conclusion is found in Alonso's work. He stated that "Lower bid price curves imply greater satisfaction. This is common sense: the

27 lower curves signify lower land prices."

Other interesting results are also obtained by applying (2.28):

(2.30) 3B/3A. = = 7 (- T

2

-) + Xu

3

> 0 i Cj + a

2

SR. using (2.10), (2.15), and (2.25).

^ = Au

2

> 0

9S. 9S. i l l using (2.9) and (2.25)

26

See, for example, E. Silberberg (1978), pp. 168-171.

27

, p. 69.

(2.31)

50

( 2 . 3 2 )

If-

= IJT" = - ( - — 7 — )

2

( Y - ZP.X. -

g(')

2 2 12 11 using (2.14).

3B _ dE

±

8E i

. 4

U i

> using (2.11) and (2.25).

(2.33)

0L,

1 oL,

1

(7—^-—) I?—)

0i_ + a

1 2

9L. 9L._ l l i 2

+

X

(|?—

+

If—)

0

(

2

34)

3L. 3L._ < l l 1 2 using (2.5), (2.6), (2.17), (2.18) and (2.25).

In words, (2.30) shows that the more accessible the site to employment, shopping and recreation, the higher will be the bid. The positive sign of (2.31) means that the maximum bid a household will make on a particu­ lar piece of property varies directly with the structural qualities of the property itself. The property tax rate varies inversely with the maximum bid, since (2.32) is negative. The quality of local public services has a positive relationship with the household's maximum bid, because (2.33) is positive. Finally, (2.34) indicates that land uses a priori have an indeterminate effect on the maximum bid, because of the two simultaneous and opposing forces of positive and negative externali­ ties, due to the "desirable" and "undesirable" characteristics of each

2 8 land use in the neighborhood, already discussed in Section 3.

28

The theoretical model discussed in this Section can easily be extended to include additional explanatory variables. For example, if one also wants to study the effects of racial composition on property values, one has to repeat the above analytical procedure by doing the following steps:

—include in the utility function U an argument denoted M;

(continued)

It is assumed that the results obtained in the above analysis

51

for the bidding behavior of a single consumer can be used to predict the pattern of final valuations on an aggregative basis. Obviously, when all households have the same utility function and the same level of in­ come, their valuations would be identical. In a real-world situation, taste and income would differ among consumers. If these differences are too great, they would atrophy the presumed relationship between the value of a piece of property and its various housing characteristics.

Therefore, application of (2.30) through (2.34) without sufficient modi­ fications or necessary assumptions will be an unfruitful exercise.

5. Specification of the Model

Using the notation of the previous section, a general relation­ ship between the property value and various characteristics associated with the housing bundle can be formulated as follows:

MV = b + Zb n

.S. + Eb„.A, + Eb_.E. + Zb„a„ + Eb r

.M.

0 li l 2i l 3i l 4 2 5i l

+ b,.L. + e

6i i

(2.35)

28(continued)

M ^

-assume u^

<

0, because the presence of this type of neighbors

(i.e., ethnic minorities) and their characteristics may decrease, increase, or leave neutral one's level of utility;

-argument M also appears in the Lagrangian function, via the utility function;

-finally, M is included in the maximum bid function B and the par­ tial derivative of B with respect to M. is 9B/9M. = 3^/9M. =

M >

1 1 1

Au^

<

0, which means that the effect of the presence of ethnic minorities in a neighborhood on property values is a priori in­ determinate.

52

where MV = market value of the property (assuming MV is fairly close approximation to B),

S, = a structural characteristics variable,

1

= an accessibility variable,

= a local public service variable,

(*2 = the property tax rate,

= a variable representing "social" externalities (e.g., ethnic minorities in neighborhood),

= a neighborhood land use variable, b^ = a constant term, b^j = a regression coefficient, e = a stochastic term.

Note, first of all, the similarity between (2.35) and the "hedonic" regressions used by Griliches and others to study the automobile

29 market. Secondly, as discussed in the previous sections, housing is not a homogeneous good but consists of a bundle of many separate items

(e.g., number of rooms, plumbing facilities, accessibility to work, etc.). If the value of each item contributes to the total price of the housing "bundle," one may wish to know what this contribution is. This technique is known as "hedonic price analysis," which has become a fre­ quently used procedure in many urban property value studies. The im­ plicit assumptions of this approach are that the consumers (viz. home buyers) perceive housing as a bundle and that the different items of the

29

Griliches (1961), p. 175.

53

housing bundle are bought and sold together, although in many cases they can be sold separately.

In this study, the above technique is adopted. The property's market price is to be regressed on a vector of neighborhood and property characteristics. Through regression analysis, one shall obtain what is called a hedonic price equation whose coefficients can be interpreted as the implicit market prices of the various characteristics at a particu-

30 lar point in time. Multiple regression analyses have, in fact, been used by several previous empirical studies on the determinants of rela-

31 tive house prices. The authors of these empirical works based their conclusions on the magnitude and signs of the regression coefficients.

This study shall follow their lead: a negative sign indicates that ex­ ternal diseconomies are anticipated and a positive sign denotes external economies. The coefficients of the explanatory variables may have ex­ pected signs, but to have an influence on the property values they must also be significantly different from zero at some significance level.

The t-ratios shall be used to perform this statistical test.

It is noted that the estimated coefficients of the hedonic price regression equation are usually interpreted as implicit prices of the

32 urban property's characteristics at a given point in time. The hedonic

30

For more details on hedonic prices, see Adelman and Griliches

(1961) and Bailey et al. (1963) among others. Essentially, the hedonic price method assumes that the market value of a house is a function of the "characteristics" of the house.

31

For a brief summary and examination of important empirical works, see Ball (1973).

32

These implicit prices are estimated in the form of the regres­ sion coefficients, e.g. 8MV/3S^ t

= hedonic price of the i^ of set of structural characteristics S, at time t.

prices obtained in this manner are not necessarily long-run equilibrium prices. They are rather a set of short-run market prices attached to various housing characteristics. The particular hedonic prices which are the result of a single urban property value study cannot be consid­ ered as invariant across time and space. Griliches"

5

"^ noted that there is no reason to expect that the relationship between the overall price of the housing bundle and the level or quantity of various characteris­ tics will remain constant. He asserted that both the relative and the

34 absolute prices of the various components may change. Stull also warned against expansive generalizations of the hedonic price equation's results. The present study also suffers these limitations.

On this note of caution, the discussion about theoretical frame­ work and functional specification for a model of externalities and urban residential property is concluded. Chapters 3 and 4 which follow shall provide a data base for this research and present many interesting empirical results.

33

Griliches (1971), p. 4.

34

Stull (1975), p. 552.

CHAPTER 3

THE DATA BASE

1. Introduction

The area under study is the City of Tucson, Arizona. It occu­ pies an area of 54,177 acres (or about 80 square miles) of which 10,157 acres are devoted to streets. The net area, exclusive of streets, is about 44,020 acres.

In 1970, the population of Tucson was estimated at 262,933.

There were 89,264 year-round housing units. Tucson had a relatively low vacancy rate since 84,226 (or 94.4 percent) of these housing units were occupied. Persons of Spanish language and the Negro population occupied respectively 14,911 (or 17.7 percent) and 2,695 (or 3.2 percent) of the city's housing units.

A graphic picture of the extent and arrangement of the type of land uses within the study area is portrayed in the accompanying land use map (see Exhibit 1). The map depicts a broad and simplified por­ trait of the patterns of land use on a city-wide basis. The four general land use categories shown on the map are: residential, commercial, in­ dustrial, and public (government) lands.

Various levels of governments have large holdings of land (4.509 acres) within the study area. These lands range from public reserves to city-owned well sites and neighborhood parks.

"*"City of Tucson, Arizona, Planning Division (1973), p. 6-5. (All figures in this section are taken from the Interim Concept Plan.)

55

56

Of the approximately 44,020 acres of land (net area) within the

City of Tucson, 16,353 acres, or 37.2 percent of the land, is classified vacant developable. Vacant land is defined as parcels of land which are presently undeveloped. It excludes areas subject to flooding, drainageways, slopes of greater than 12 percent, and major rights-of-way.

Detailed land use survey in Tucson was accomplished in 1958-59 for the purpose of the General Land Use Plan and a summary update was completed in 1968. The results of the survey and update provided general information on the present (1970) status of land use in the City of

Tucson (see Table 3-1).

The general land use pattern is one of low density single-family residential development extending from the downtown area to Harrison Road and north to the Rillito River. The average population density is 14.4 persons per developed residential acre. The area averages 4.9 dwelling units per developed residential acre. But within the central portion of the city, population and housing densities increase. In the University area, for example, population densities increase to 31.6 persons per residential acre and housing densities increase to 10.8 dwelling units per developed residential acre.

Single-family residential uses within the City of Tucson take up the largest portion of land area (14,442 acres). Approximately 8 per­ cent, or 3,436 acres, is presently classified as multiple family resi­ dential. Commercial uses make up 6 percent, or 2,721 acres, of land within the City of Tucson. About 4.8 percent, or 2,095 acres, is in industrial use.

57

Table 3-1. Generalized Land Use, City of Tucson.

Land Use Category

Single-family

Multi-family

Commercial

Industrial

Public, Semi-public

Washes, medians

Developed area

Vacant land

Net area

Streets

3

Gross area

Area

(acres)

14,441.61

3,435.61

2,721.49

2,094.98

4,508.59

464.05

27,666.33

16,353.43

44,019.76

10,157.26

54,177.02

Percent­ ages

32.81

7.80

6.18

4.76

10.24

1.05

62.85

37.15

100.00

18.75

100.00

S

Residual area, includes streets, alleys and similar areas.

Source: Interim Concept Plan, Department of Community Development, City of Tucson, 1973, p. 6-5.

Because of various special provisions of the zoning code, such as those dealing with non-conforming uses, actual land-use figures are different from zoning figures. It is thus necessary to compare actual land use with a zoning analysis of land use.

2

2. Zoning Analysis of Land Use

Of the 44,020 acres within Tucson city boundaries, 35,166 or

80 percent, are designated for residential uses. Presently undeveloped are 12,691 acres in residential zones. Of the 12,691 acres, 5,280 are zoned for single-family residences under the R-1 zoning category. (Refer to Appendix B for a summary of zoning categories.)

See Appendix A for details concerning the total acreage of various land use classifications within each zoning category.

58

Multiple family zoning categories make up 11,264 acres, or 25.6 percent of the city area. The R--2 (low density residential) zoning cate­ gory which contains 8,510 acres of zoned area within the city, has 6,400 acres developed. Of this total, close to one-half, or 3,177 acres is currently in singld family usage, indicating that even though it is zoned for a higher density, single family uses predominate the R-2 category.

The commercial zoning categories make up 10.4 percent, or 4,593 acres within the city. Approximately 4,000 acres are found in the B-1

(local business) and B-2A (general and intensive business) categories.

Almost half of the B-1 zoning classification is presently undeveloped as

743 acres of the 1,675 acres are classified vacant. The majority of the developed commercial uses are zoned B-2A as 1,082 acres, or 53 percent, of all developed commercial land is found in this zoning category.

Industrial zoning within the city accounts for 4,261 acres of which 2,209 acres, or 52 percent, are presently developed.

The 1-1 (light industry) zoning category makes up 3,891 acres of the total 4,261 acres zoned for industrial use. Of this 2,049 acres or

52.7 percent are presently developed, leaving 1,841 acres vacant. The remainder, or 370 acres, is zoned 1-2 (heavy industry) and P-I (Park

Industrial).

3. Units of Observations

The sample area under study consists of observations on 52 census

3 tracts in the City of Tucson, Arizona. For the purpose of this research

3

See Exhibit 2.

59

a "neighborhood" is defined as a census tract. This definition is adopted because "tracts were generally designed to be relatively uniform with respect to population characteristics, economic status, and living

4 conditions." It is hypothesized that there is a relationship between property values and various characteristics of the dwelling units as well as neighboring land-uses, since it is not unreasonable to suppose that the purchase of a house involves more than simply the acquisition of amenities contained in the building. Housing is more accurately de­ fined to include external or neighboring attributes of a dwelling and it is therefore worth considering how these neighborhood factors influence relative housing prices.

The concept of neighborhood has been variously defined in the economic literature. For example, Stull's definition of what might con­ stitute a neighboring land use was different from that found in previous studies. In effect, he used the proportion of a whole community's land devoted to various nonsingle-family home uses as the land use environ­ ment variable, i.e., he defined the neighborhood of a home to be the entire community. Rueter on the other hand, defined neighborhood as either a 150-foot radius or a 300-foot radius from the home. In other words, in Rueter's model, there was only a very small local neighborhood.

The above definitions of the relevant neighborhood seem to be too simple.

Land use can generate externalities at several levels. It could have effects at relatively short distances. It could also have city-wide effects. In the former case, Stull's broad definition was inappropriate,

4

U.S. Department of Commerce, Bureau of the Census (1972),

Appendix A, p. 1.

60

while in the latter case Rueter's narrow definition of neighborhood did not accurately reflect the actual situation. It follows that both Stull and Rueter studies could have yielded spurious results by excluding the effects of the relevant neighborhood.

In the present study neighborhood is defined neither as a 150foot radius (only about two houses away) nor as the entire city (about

80 square miles in the case of Tucson). Neighborhood is seldom easily defined. However, the general consensus is that it should be homogeneous or uniform in some respect. For the purpose of this research, a census tract is probably the best candidate for an acceptable and appropriate definition of neighborhood. It is therefore chosen as the unit of observation.

4. The Variables

The bundle of housing characteristics affecting property values can be exhaustively classified into four categories:

—physical characteristics,

—accessibility characteristics,

—public sector characteristics,

—neighborhood or environmental characteristics.

As does much of the empirical literature concerning housing markets, this study uses multiple regression techniques as the principal tool of analysis. A description of the variables used in the estimation equa-

5 tion follows.

^The discussion of each variable is generally brief. A detailed description of the variables used and their sources is presented in

Appendix C.

61

(a) Dependent Variable. The dependent variable used is the median value of owner-occupied single-family homes in 52 census tracts in Tucson. Data for this variable is taken directly from the 1970 U.S.

Census of Population and Housing. Single-family homes on very large lots (10 acres or more) are excluded since their median value is un­ available and also because this category (ranch house/estate) makes up less than 4 percent of total single-family acreage.

There is a practical reason to use the median value of singlefamily homes as the dependent variable. As mentioned earlier, the

Interim Concept Plan indicated that within the city, single-family uses occupied the largest amount of residential land. Expressed as a per­ centage of the total residential land use, single-family use represented

81 percent, and multiple-family 19 percent. In other words, singlefamily homes made up the majority of land acreage within the city. This fact justified the emphasis placed on the importance of single-family dwelling units. This variable is regressed on various independent vari­ ables describing the characteristics of the single-family homes in the census tracts to determine which of these characteristics are of sig­ nificance to home buyers.

(b) Independent Variables. The discussion of these variables is organized around the four categories of housing characteristics described in the first paragraph of this section.

(1) Physical characteristics. It is generally agreed that market value is higher for larger homes, ceteris paribus. But data on the size of the houses (e.g., square feet of living area) are not available. The median number of rooms is used to capture this size effect. Additional

62

information about the structures includes whether or not there is a basement. The houses must also have a significant remaining economic life expectancy to exert an influence on their price. The variable used to account for this effect is the age of the dwelling unit. In addi­ tion, the value of the houses depends on the existence (or nonexistence) of certain basic facilities. Thus an attempt is made to control for these effects by including a variable denoted as lack of plumbing facil­ for these variables are taken from the 1970 Housing Census.

(2) Accessibility characteristics. This analysis deals with the area of the City of Tucson proper. (Suburban and rural areas, and

Indian Reservations in the Tucson SMSA are excluded.) Within the city, there is no intrinsic economic value in distance or closeness to the

Central Business District (CBD). The CBD is important mainly as a sur­ rogate for such attributes as:

—convenience to major shopping centers,

—convenience to work areas, etc.

In Tucson, the CBD does not offer these conveniences to the pop­ ulation on a more concentrated basis than do other areas within the city.

Tucson has an abundance of shopping malls and centers outside of the CBD

(35 major shopping centers in 1970). Also, local industry is not con­ centrated in the CBD (there were about 69 manufacturing companies scat­ tered in various areas of the city).

Several previous studies have assumed that a significant amount of economic activities occurred in the CBD. Therefore, the road distance from the geographical center of a neighborhood to the CBD constituted an

63

appropriate measure of accessibility in each neighborhood to sources of employment. The closer a neighborhood is to the CBD the higher its real property values ought to be. This measure of accessibility, as mentioned in the previous paragraph, is deficient in the case of Tucson where its

CBD employs only 5.7 percent of the labor force.^ Therefore it was necessary to devise a new accessibility measure in this study. If one can assume that the city is divided into N census tracts, and let d.. be the distance from the geographical center of tract i to the center

9

of tract j, and e^ be the total employment in tract i and E be the total employment in the city, then one measure of accessibility to employment for tract i is given by

A. =

1

N e.

Z

d.. • (-

1

) j=l

13 E

N i = 1, ..., N and where E = E e. i=l

1

(3.1)

It is assumed that, ceteris paribus, an increase in A^ results in a decrease in V^, where is the value of the real property in tract i.

As mentioned in Chapter 1, the monocentric assumption of previ­ ous models was not an accurate description of modern cities. This study achieves greater realism by generalizing the monocentric assumption to recognize that the urban area has numerous centers of economic activity to which access has value. A.T. King also acknowledged the importance

6

Seventy-five percent of the labor force work in the remainder of Tucson and 19.3 percent of them work outside the city; for details see Table P-2 of 1970 Census of Population and Housing.

64

of a measure of accessibility such as equation (3.1). But he admitted that, for empirical work, this measure is very tedious to calculate when many places of employment are distinguished. King used instead a simple version which recognizes only two employment centers in his study of the

New Haven SMSA, assuming that these two centers are a reasonable approx-

7 imation of the more complete measure of employment potential. The present study, on the contrary, recognizes 52 employment centers in 52 census tracts. This recognition involves a more complex calculation, but it certainly is a more appropriate measure of accessibility.

A second employment accessibility variable is used in this study.

It has been employed by Stull to calculate for each unit of observation the ratio of: total employment total number of owner-occupied single-family homes

The purpose of this variable is to account for the influence of proxim­ ity to local employment. The sign of this variable's coefficient is expected to be positive. For example, 100 jobs in a particular census tract where there are many single-family homes would have a different effect on the median value of the property than would the same 100 jobs with very few homes.

There are other types of accessibility, for example accessibility to local shopping centers. One can attempt to take the influence of this

7

This measure of accessibility is known in King's study as

GRAVITY; he found a complete insignificance of this variable. In fact, he used other measures, such as logarithm of distance to CBD, inverse of distance, logarithm of gravity; see King (1973), pp. 81-83.

65

type of accessibility into account by including a "commercial land-use variable" in the regression equation. The land-use variables will be discussed in the section on Environmental Characteristics.

(3) Public sector characteristics. It is noted in Chapter 1 that, in a Tiebout world, consumers (home buyers) "shop" among different neighborhoods offering varying packages of tax and local public services and select as a residence those neighborhoods which offer a program of tax-public services combination best suited to their preferences. If in fact households make their residential choice on this basis then property tax and local public output are expected to exert some influences on urban property values.

To account for these effects one can use two explanatory vari­ ables: property tax rates (or an index representing excess tax burden) and a local public school quality index represented by a proxy vari­ able—the Reading Achievement Tests. It can be assumed that property owners (especially those having children) make residential choices at least partly on the basis of this education quality criterion. This research departs from previous studies by not using the total school expenditure per pupil to measure school quality. Indeed, school expend­ iture per pupil may not reflect education quality at all. Two schools

(with same enrollment) can have two different expenditure outlays, if there is for example a higher proportion of older teachers (i.e., higher salaried teachers) in one school. The other with a larger proportion of younger teachers (i.e., lower salaried teachers) can have a smaller expenditure budget. Nevertheless, one cannot correctly infer, by look­ ing at the expenditure figure alone, that the school with a higher

66

proportion of young teachers must necessarily produce inferior education quality. For this reason the scores of the Reading Achievement Tests are used as a proxy to account for local public school quality.

On the tax side one can expect that tax burdens also influence the market value of the property. Within the city of Tucson, property tax rates are greater in some areas than in others. Variations in taxes can affect the marketability of dwellings. Properties that are other­ wise comparable to those not subject to special and heavier assessments can be expected to bring a lower sale price. To account for the effect of this burden, one can use a variable denoted as "Excess Tax Burden."

In Tucson, property tax rates differ primarily because of the differing levels of two main tax components: school district tax (there are four school districts in the city of Tucson) and improvement district tax

(e.g., street lighting). Excess Tax Burden is an index defined as being equal to 1 +

T. - T

T

L where T. is the tax rate in tract i and T is the

1

L lowest tax rate. It is assumed that buyers would compare the tax burden in various neighborhoods. The "Excess Tax Burden" variable is adopted in this study to measure the household's tax burden in one neighborhood relative to that of another neighborhood.

The possibility of bias from the simultaneous relationship be­ tween tax rates and house values was not considered serious for this sample because of the relatively low amount of variation in the tax variable.^

8

For the tax rate variable the coefficient of variation measured: a(x) 1.43 . . . ,,

67

Furthermore, as was the case in a previous study, the argument that house values and taxes should be regarded as simultaneously deter-

9 mined appears to be unimportant with respect to the present study.

This follows from an institutional characteristic of an assessment prac­ tice in Tucson which tends to reduce the simultaneity between taxes and house values. Specifically, general reassessment of all houses in rela­ tion to market value occurred infrequently, thus holding the tax base constant regardless of changes in taxes which might be capitalized into

10

market value. If houses were routinely reassessed after each sale, the simultaneity would be a more serious problem.

(4) Environmental characteristics. It is commonly suggested that environmental factors would be an important i'jpncern for property owners.

However, no direct measure of this attribute is available. The selec­ tion of the independent variables to be used to control for environmental influences is mainly dictated by the availability of data. It is hy­ pothesized that there exists a relationship between the value of real property and

(a) various demographic, social, and economic features that charac­ terize the environment in which the property is located, and

8(continued)

For further explanation of the coefficient of varia­ tion, see Emanuel Parzen, Modern Probability Theory and its Applications,

New York, John Wiley and Sons, 1960, p. 379.

9

.

10

Before the advent of electronic data processing technique in

1972, less than one-third of residential houses in Tucson were re­ assessed each year; there was no routine reassessment after each resi­ dential sale.

68

(b) a set of environmental (or land-use) characteristics influenced by the government.

For the purposes of this study, the environmental characteris­ tics described in (a) and (b) above can be labeled as non-zoning vari­ ables and land-use or zoning variables, respectively.

Non-zoning variables consist of such socio-economic characteris­ tics of the environment as: racial composition, employment (or unem­ ployment) status, proportion of poor families, percentages of housing units which are crowded, noise, crime, etc. All together there are seven such non-zoning variables. Many of them are self-explanatory.

Because of their relative importance, only the following variables are considered and briefly discussed: racial composition, noise, and crime.

It is common practice to examine the ethnic composition of a neighborhood. This study shall attempt to determine whether or not hous­ ing prices are influenced by this racial element. Included in the re­ gression equation is an explanatory variable denoted as percent Negro in neighborhood's population. Several previous studies have explored the impact of race on housing values with somewhat mixed results. Lapham

(1971) found no evidence of housing price differences in black versus white housing areas in Dallas. Bailey (1966) similarly found no such evidence in his study of housing in Chicago. Kain and Quigley (1970) uncovered possible race discrimination effects, as did King and Mieszkowski (1973).

The black population of Tucson is not, however, the main com­ ponent of the minority population. In fact, the largest ethnic minority group consists of persons of Spanish language and Spanish surname. They

account for 23.9 percent of the total population. This group is dis­ tributed mainly on the south and west sides of Tucson, between Tucson

69

International Airport and Mission Road. Their major concentration ex­ tends upward from the San Xavier Indian Reservation to the vicinity of

Grant Road. Ethnic minority density declines toward the north and east limits of the city. To account for the influence of this racial factor on housing prices, included in the regression equation is a second ex­ planatory variable known as percent of persons of Spanish language or

Spanish surname in neighborhood.

These two variables can be conveniently combined to constitute the "ethnic minority" variable, or for short, the "minority" variable.

There is conceivably a strong correlation between this variable and the socio-economic composition of the neighborhood (e.g., percentage of families that are poor; percentage of labor force unemployed; or per­ centage of housing units which are crowded). Consequently, the "minor­ ity" variable alone is used as a proxy for the entire set of socio­ economic characteristics, in order to avoid the problem of multicollinearity.

Tucson traffic makes a substantial contribution to noise pollu­ tion. The College of Engineering of the University of Arizona has con­ ducted a study on daytime noise environment in Tucson and concluded that most outdoor noise in the city is produced by automobiles, trucks, air­ craft, and railroad trains and that industry is not a major noise source.

A decibel is used as a unit of measurement for expressing the relative intensity of sounds on a scale from zero (least perceptible) to the

70

sound pain level. The results of traffic noise measurements in deci­ bel values are available for most major intersections of Tucson. Both an isodecibel contour map of automobile traffic noise (ranging from 73 to 92 decibels) and a map of city's areas in which aircraft noise ex­ ceeds 75 decibels are provided.

To study the possible effects of noise pollution on the real property values, an explanatory variable representing the automobile traffic noise level for each census tract and a. dummy variable describing the degree of aircraft noise pollution are included in the

12

regression equation.

It is obvious that crime imposes substantial costs upon individ­ ual members of society, as well as upon society as a whole. There is another less obvious but real cost to society—the decline of neighbor­ hoods due to fear of crime. If crime and the fear of crime result in less demand for housing in those areas (the fear of victimization would cause people to move out of high crime areas) then homeowners suffer a loss of wealth because of a decline in their property values relative to other areas of the city. But as yet, no empirical research has been done toward estimating the impact of crime on housing values. Bogue

"'""'"The standard measure for loudness is decibels on the A scale

(dbA). This "A scale" is the legally accepted sound level for environ­ mental impact studies (see Colin G. Gordon, et al., Highway Noise,

NCHRPR 117, National Research Council, V7ashington, D.C., 1971).

12

The dummy variable is assigned the value of 1 for area in which aircraft noise exceeds 75 decibels and the value of zero if this noise level is below 75 decibels. Gautrin (1975) found that airport noise reduces the desirability of neighborhoods and dampens housing values.

71

alludes to this issue by saying that ". . . the impact of crime and de­ linquency in lowering land values and hastening the exodus of people from neighborhoods . . . needs to be appreciated and given a place in

• i,13 economic analysis.

In this study, it is hypothesized that there is a relationship between crime rates and urban property values. It is not unreasonable to believe that people want to live in areas of the city which are safe.

The regression equation includes two explanatory variables representing crime rate against property (rate per 100 residential single family

14 homes) and crime rate against persons (rate per 1,000 population).

Land-use or zoning variables. Estimation of real property values cannot adequately be made without data concerning government restrictions on urban land-use. These include zoning, subdivisions and other regula­ tions. The hypothesis of this study is that the residential property value in a given neighborhood is influenced by the externalities emanat­ ing from neighboring land uses.

The city of Tucson zoning ordinance has created 24 zoning dis­ tricts (or zones). In general, the city is divided into zones as follows: residential, commercial, industrial land uses. Residential districts are divided into single-family and multiple-family

13

Bogue, D.G. "Discussion" in Perloff, H.S. and Wingo, L. Jr.,

Issues in Urban Economics, Baltimore, The Johns Hopkins Press, 1968.

14

Crimes against property consist of burglary, larceny, and auto theft. Crimes against persons are criminal homicide, forcible rape, robbery, and assault.

districts. Commercial zones are divided into local business districts and general (and intensive) business districts. Similarly, industrial zones are divided into heavy and light industry districts. A summary of zoning classifications and descriptions is presented in Appendix B.

It is axiomatic that in single-family district only one house per lot occupied by a single family is permitted. Thus apartments, for example, are excluded from this zone.

In each zoning district, the ordinance limits the height, bulk, and uses of buildings and other structures, the density of population, the use to which land may be put, and other matters. However, the city zoning ordinance has also made some provisions for the continuation of non-conforming uses, since to invalidate an existing store in a residen­ tial area would be an unconstitutional deprivation of property.

The land use and zoning data used by this study are obtained from the Department of Planning, City of Tucson. This Department com­ piled in 1973 a report entitled "Land Use, Zoning, and Census Data by

Census Tract Number." In its original form these land use data are broken down into 26 different categories. In this study they are ap­ propriately aggregated to yield a set of five categories, corresponding to those listed in the Tucson Zoning Code. Four land-use variables are

15

Rl Residence Zone or single-family residential district, as its name implies, is primarily for the use of one-family dwellings.

Schools, churches and public buildings are permitted in this district.

Multiple-family dwellings belong to R2, R3, R4 Residence Zones or Dis­ tricts. In these districts, multi-family residences, apartment houses, as well as Rl uses are permitted. (See Tucson Zoning Code 1967, Arti­ cle 1, Divisions 7-12).

16

then included in the regression equation. They are proportions of

73

land in the neighborhood devoted to:

—multi-family uses,

--commercial uses,

—industrial uses, and

—public and semi-public (or institutional) uses.

It should be mentioned again that this study makes a distinction between "zoning" and "land use" variables. Therefore, three zoning variables are also employed. They are proportions of land in the neigh­ borhood zoned multiple-family, commercial, and industrial. (Note that the zoning ordinance does not set up special districts for public land uses.)

5. Data Sources and Data Characteristics

The data for this study have been obtained from several separate sources:

1. detailed information on housing and socio-economic characteris­ tics of the neighborhoods was obtained from the U.S. Bureau of

Census, 1970 Census of Population and Housing;

2. property tax rates and data on local public service were obtained from various levels of government in Pima County and the State of Arizona;

"^It should be noted that: (a) the variable representing "pro­ portions of land devoted to single-family uses" is not included in the regression equation in order to avoid the singularity of the X'X matrix; this technique is also used by Stull (1975); (b) land use data used in this study are net of streets, washes, medians.

74

3. land use data were obtained from the Department of Planning,

City of Tucson;

4. crime statistics were provided by the Department of Police,

City of Tucson;

5. data for the noise pollution variables were taken from a Tech­ nical Report published by the University of Arizona.

17

With a few exceptions, all data gathered were for the year 1970.

It is noted that all Census data used in this study (except AGE and BASEMENT) were based on a sample of 100 percent of the population.

Data concerning AGE and BASEMENT were based on sampling rates of 20 or

15 percent.

The reliability of Census data depends, among other things, on the accuracy of the Census count. It is believed that the enumeration has been carried out carefully, and great efforts have been made to reach every member of the population. The 1970 Census was carried out to the accompaniment of a considerable amount of publicity and discussion largely concerned with the confidentiality of returns. The immediate cause of concern appeared to have arisen from the insertion of questions concerning ethnic origin. This easily aroused fears of additional har­ assment among the immigrant population. Some omissions might have pos­ sibly occurred. But in overall terms, the omissions were probably negligible. Census data can be considered as accurate and reliable.

17

Land use data were for 1971, but City Planning officials as­ serted that there was little change between 1970 and 1971 land use patterns. Data on crimes and noise were for 1972 and 1973, respec­ tively.

An alternative source of data would be the Multiple Listing

Service (MLS). As a source of information about houses and housing transactions, the files of the MLS are generally very valuable. When

75

a house is offered for sale, the real estate brokers who are members of the MLS agree that a description of the house in question will be sent to the central MLS office if they cannot sell it themselves in a short time. When the central MLS office is notified, it prepares a descrip­ tion of the house's features in a standard format and distributes this to all member brokers, and only member brokers. Several attempts made by the author to obtain the MLS files have been in vain. Moreover, the

Tucson MLS office maintained that MLS records are not to be kept in files for many years; MLS records for 1970 were non-existent.

It is generally agreed that the MLS records provide many details about the houses passing through the housing market. However, for these data, as for any other, it is important to consider the possible inaccu­ racies and biases. Houses sold through the MLS are not representative of all houses sold, since MLS sales account for only a fraction of total housing transactions. A test of how representative MLS sales are of the total market is not possible because no statistics for non-MLS transac­ tions are available. Another caveat is that the MLS sample may include too many of the high-priced houses. All this suggests that inaccuracies and biases of the MLS data would impair the generality and usefulness of the conclusions which can be drawn.

Compared with the information provided by the MLS records, Census data used in this study can be considered as possessing a relatively higher degree of completeness with respect to both population and housing

76

units characteristics. The same can be said concerning the land-use data (which are unavailable in MLS records) obtained from the detailed land use survey in Tucson.

Utilizing these data, a collection of 19 variables is included in the regression equation, variables which, in one way or another, are hypothesized to exert influences on housing prices. There are possibly other influences, many of which data limitations preclude their in­ corporation into the analysis, but the explanatory variables of this study are comprehensive enough to deal adequately with the problem of land use and other externalities in the urban property market.

Chapter 4 will present the results of regressing the median value of owner-occupied single-family homes against the complete set of explanatory variables just described.

CHAPTER 4

ESTIMATION RESULTS

1. Introduction

In this chapter, the results of the empirical analysis will be presented.

As mentioned earlier, the data base of this study consisted of observations on all the variables described in Chapter 3 for 52 neigh­ borhoods in the City of Tucson in 1970. The procedure was to regress the median value of single-family homes for these neighborhoods on vari­ ous combinations of the independent variables discussed in the previous chapter."'' The estimation procedure used was ordinary least squares.

Table 4-1 presents the important results of these estimations. This chapter will deal with two models separately. The first model concerns the analysis with samples based upon zoning data. The second model utilizes instead actual land use data in its analysis of neighborhood externality effects. The results will be discussed sequentially, start­ ing with the first (or zoning) model.

"'"The median value of owner-occupied dwelling units in singleunit structures used in this study is based on the homeowner's estimate of how much the property would sell for if it were for sale. This pro­ cedure would produce errors in the valuation of individual properties.

However, there is evidence that these errors cancel out when the indi­ vidual prices are aggregated into average values. (cf. L. Kish and J.

Lansing, "Response Errors in Estimating the Value of Homes, Am. Stat.

Assoc. Journal, September, 1954). These results also apply to the median values, presumably.

77

Table 4-1. Alternative Estimates of a Valuation Equation for Single-

Family Homes in the City of Tucson in 1970; Regression

Coefficients (Absolute T-Ratios in Parentheses)

78

Equation

Number

3

1 2 3

4 5 6

Rooms

Age

2588.3

(4.20) b

2633.4

(4.04) b

3296.2

(6.52) b

3647.4

(6.16) b

3639.2

(5.98) b

3985.7

(6.41) b

-2564.8

(1.61) c

-2575.6

(1.57) c

- 1957.2

(1.37) c

- 2099.9

(1.52) c

- 2110.5

(1.50) c

- 3267.4

(2.07) d

Plumbing -4504.5

Basement

Proximity

(0.31)

5967.9

(0.51)

513.9

(3.03) b

-4501.9

(0.30)

4876.0

(0.38)

524.8

(2.92) b

- 8259.1

(0.63)

-13794.0

(0.98)

13184.0

(1.25)

253.8

(1.53) c

12556.0

(1.16)

133.7

(0.70)

-13855.0

(0.97)

12321.0

(1.09)

-14693.0

(1.11)

10892.0

(1.04)

134.7

(0.70)

Schools 4402.7

(2.30) d

4236.3

(2.03) d

5548.4

(3.21) b

5249.2

(3.10) b

5186.3

(2.77)b

4475.8

(2.66) b

Excestax -5868.5

(1.89) d

-5468.8

(1.48) c

- 7283.8

(2.61) b

- 6441.6

(2.33) d

- 6380.6

(2.20) d

- 5182.0

(1.88) d

Minority -4187.6

(2.83) b

-4280.6

(2.46) b

- 4376.6

(3.42) b

- 4079.5

(3.24) b

- 4140.5

(2.82) b

- 4412.1

(3.52)b

NSF1 - 937.0

(0.67)

MF

Commercial

Industrial

- 660.7

(0.34)

-1174.6

(0.48)

-1382.0

(0.58)

NSF2

6266.7

(3.41) b

Mulfam

Commerce

Industry

9436.4

(2.63) b

9901.5

(2.56) b

4429.9

(2.29) d

9305.9

(2.35) d

9516.9

(2.70) b

10356.0

(1.55) c

10322.0

(3.17) b

4386.2

(2.17) d

4734.1

(2.59) b

Public

Commercesq

(3.69) b

9322.3

(3.63) b

0.98

(0.08)

9219.5

(4.03) b

79

Table 4-1 (Continued).

Equation

Number

3

1 2 3 4 5 6

Distance - 26.8

(1.47) c

Constant

Term

R

2

60.3

(0.01)

0.83

Degree of

Freedom

(for t-tests) 42

- 73.7

(0.01)

0.83

40

- 8580.5 -11148.6

(1.61)

42 c

0.86

(2.01) d

0.88

39

-10972.8 - 8552.4

(1.83)

0.88

38 d

(1.49)

39 c

0.89 d

The dependent variable is the median value of owner-occupied dwelling units in single unit structures.

Significant at the 0.01 level (one-tailed test).

Q

Significant at the 0.10 level (one-tailed test).

Significant at the 0.05 level (one-tailed test).

Equation 1 includes a conventional set of physical characteristics, ac­ cessibility characteristics, local public sector characteristics, and an additional variable describing the proportion of neighborhood land zoned non-single family home uses (denoted NSF1). Equation 2 includes all of the variables in equation 1, with the exception that NSF1 is decomposed into its three distinct components. This is done by subdividing the proportion of land zoned nonsingle-family into three general zoning categories: proportion of land zoned multiple-family (denoted MF); pro­ portion of land zoned commercial (denoted (COMMCIAL); and proportion of land zoned industrial (denoted INDTRIAL). Each is then treated as a separate variable for estimation purposes. The results of these esti­ mations are given in Table 4-1.

2

Note first that the R for these estimations (0.83) is fairly

2 high for a cross-section study.

Turning to the estimates themselves, observe that the variables have coefficients whose signs are as predicted in Chapter 2. In both equations, the variables ROOMS, AGE, PROXIMITY, SCHOOLS, EXCESTAX, and

MINORITY have coefficients which are significant (using a one-tailed test).^

2. Zoning Model

Consider the first two equations: equation 1 and equation 2.

80

2 2

In Crecine et al. (1967), R for 10 equations varies from 0.24 to 0.79 (6 equations have R^ < 0.51); in Rueter (1973), R^ for 12 equa­ tions varies from 0.56 to 0.80; in Kain and Quigley (1970), R^ is 0.73.

3

One-tailed test was used because it is assumed that most of the coefficient signs in Table 4-1 were known a priori. Had two-tailed test been applied, the interpretation of the results shown in Table 4-1 would have been unaffected.

81

Among the variables describing the physical characteristics of the house, ROOMS and AGE have coefficients whose sign is right. The co­ efficient of ROOMS is significant at the 0.01 level and that of AGE at the 0.10 level. The variables PLUMBING and BASEMENT also have coeffi­ cients with right signs, but they are not significant. This is not a surprise since the proportion of homes having a basement is three percent

5 and the proportion of homes lacking plumbing facilities is 1 percent.

The local employment accessibility effect in both equations (as represented by PROXIMITY) is rather significant. This result suggests that the local employment accessibility is an important determinant of property value.

The public sector variables SCHOOLS and EXCESTAX have coeffi­ cients whose signs are as expected, i.e., school quality has a positive coefficient and excess tax burden has a negative one, as predicted.

Both have significant coefficients. The message v/hich comes through all of this is clear: the present study has found evidence that local prop­ erty tax rate and local public school quality are significant factors in the residential site choice. These public sector variables have exerted an independent influence on the market price of the neighborhood's single-family homes, an effect which this study has been able to iso­ late. There is empirical support for the tax capitalization hypothesis.

These results are similar to those obtained by Wallace Oates.^ He also came out with coefficient estimates for the property tax rate and

5

1970 Census of Housing, Tables HI and H2.

G

W. Oates (1969), p. 964.

82

educational expenditures per pupil (proxy for school quality) which were of the right sign and highly significant.

The variable MINORITY in both equation 1 and 2 has coefficient with negative sign and significant t-ratios. It is evident that in the present study housing prices are influenced by the ethnic composition of

7 a neighborhood.

The statistical test in this study could not demonstrate a causal relationship between CRIME (or NOISE) and urban property values.

In various trial regression equations, the coefficient of NOISE is found to be negative and that of CRIME positive. Both variables are, however, completely insignificant. They are thus dropped from the final regres­ sion equations shown in Table 4-1. This does not imply that CRIME and

NOISE are not expected to affect property values, but simply indicates that, due perhaps to a lack of an adequate data base for these variables, this study fails to support the hypothesized relationship.

One very striking and interesting result is that none of the zoning variables in equation 1 and equation 2 has a significant coeffi­ cient. These results, of course, lend some support to the contention in

Chapter 2 that neighborhood effects not perceived by the residents are ignored in their property valuation process. However, the low t-ratio of the zoning variables does not imply that there is no relationship between land-use externalities and property value. The above results simply indicate that, because of the non-perception phenomena involved

7

For an excellent review of what is known about the effects of racial prejudice and racial discrimination on the urban housing market, see J. Yinger (1979), pp. 430-468.

83

in the valuation process by households, this study has failed to support the hypothesis of zoning external effects on residential property value.

It is hoped that the estimations will reveal more significant results when the actual land use data are used.

In both equations 1 and 2, the coefficients of various zoning variables have a negative sign. As shown above, the insignificance of these coefficients having not supported the hypothesis of externalities between zoning and property value, this negative relationship between them may simply reflect the zoning board's behavior and imply that zoning administrators tend to allocate nonsingle-family land-use activities to census tracts with low residential property values to avoid political pressure.

3. Land Use Model

Various land-use variables have been included in Equations 3 through 6 (in lieu of zoning variables). Other variables remain un­ changed.

2

Again, note that the R for these estimations are fairly high

(0.86 to 0.89).

All the variables (except PLUMBING, BASEMENT, and PROXIMITY)

0 have coefficients which are significant. The signs of the coefficients are as predicted.

Consider equations 3, 4 and 5. As was the case in equations 1 and 2, the variables ROOMS and AGE have coefficients whose sign is right.

8 equation 3, the coefficient of PROXIMITY is significant at the 0.10 level.

84

Note that the coefficient of ROOMS has a .high t-ratio. It is signifi­ cant at the 0.01 level. The coefficient of AGE has a negative sign and is significant at the 0.10 level. These results suggest, as argued in

Chapter 2, that the value is higher for larger homes and lower for older homes, ceteris paribus. The variables PLUMBING and BASEMEHT do not perform very well. Their coefficients have the correct sign; however, they are not statistically significant. The reason for this is the same as that presented in the section dealing with equations 1 and 2.

It is noted that the PROXIMITY coefficient is not significant

(except in equation 3). One possible explanation is that households really evaluate neighboring commercial and industrial land uses in about the same way as they evaluate PROXIMITY, so that in the true model the

PROXIMITY coefficient is zero. What is worth noting is that this local accessibility effect (i.e., PROXIMITY) has probably been controlled for through the commercial and industrial land-use variables. The discus­ sion of this matter is deferred until a little later, when various land use variables are examined in more detail.

Consider now the public sector variables. The coefficient of the variable representing local public school quality (denoted SCHOOLS) has a positive sign, and that of the tax variable has a negative sign, as expected. The coefficients are significant at the 0.01 and 0.05

9 level, respectively. (See equations 3 through 6.) This provides addi­ tional empirical evidence that property taxes and local public services are capitalized into location values.

9

The coefficient of the tax variable is significant even at the

0.01 level in equation 3.

85

The variable MINORITY in all equations (3 through 6) has coeffi­ cient with negative sign and significant t-ratios. Indeed, this coef­ ficient is significant at the 0.01 level. It is found that the variable

MINORITY is highly correlated with some other independent variables de­ scribing various socio-economic characteristics of the neighborhood, such as the "percent of housing units which are crowded" (R = 0.33), or the "percent of families with income below poverty level" (R = 0.75).

As mentioned earlier, in order to avoid the possible problem of multicollinearity, only the variable MINORITY is included in the regression equation. It serves as a proxy variable for an array of social and eco­ nomic conditions of the neighborhood. In all the alternative estima­ tions of this study, there is ample evidence that housing prices are influenced by the ethnic composition.

As mentioned before, the variables CRIME and NOISE have com­ pletely insignificant coefficients. These two variables do not appear in the final equations 1 through 6. There are not adequate data on

CRIME and NOISE to conduct a more meaningful research on the presumed effects of these independent variables on property values.

The land use variables are now considered. Note that all the land use variables in equations 3 and 4 have coefficients whose sign is positive and whose t-ratios are relatively high.

These land use variables, it is recalled, are the proportion of neighborhood land devoted to land uses other than single-family homes.

In Chapter 2, it is argued that the effects of nonresidential sites on residential property values are rather ambiguous. Residential property values may even rise with proximity to a commercial or industrial site.

86

The present study of externalities from nonresidential land uses has taken account of both the advantages and disadvantages of proximity to them. It has not been possible theoretically to predict a priori the sign of the nonresidential variables' coefficients. Only empirical tests can determine the direction and the magnitude of each of these nonresi­ dential land use effects. It turns out that these neighborhood external effects are positive. This does not imply that there are no external diseconomies from the nonresidential site (e.g., noise, traffic, etc.), only that they are more than offset by the advantages of proximity. What equation 4 says is that converting 10 percent of a neighborhood's land from single-family homes to another land-use (say, commercial activity) will raise median home values in that neighborhood by around $990, ceteris paribus.^ The most important message of this equation is that neighborhood effects are significant and that there is net beneficial impact of nonresidential activities on home values.^ This evidence would lend some credence to arguments favoring mixed land use. It seems clear that the cliche belief according to which homeowners prefer neigh­ borhoods whose land is exclusively occupied by single-family homes has no firm empirical foundation in the context of land use in Tucson.

These results are directly contrary to the conclusions of Crecine et al., Rueter, and Stull (among others). Crecine et al. argued

"^This statement does not stand without qualifications. Cf. infra, p. 94 and Table 4-2, for restrictions in meaning and application regarding each land-use category.

11

The public/semi-public (or institutional) land-use coefficient is also positive, suggesting that over some range it would be profitable for homeowners to promote institutional locations in their neighborhood, ceteris paribus.

that there is a strong possibility that "the urban property market is

87

not characterized by great interdependence and that externalities do not

12

abound in that market." Rueter repeated this study and found that the external effects expected by the municipal land-use control did not actu­ ally materialize, and he concluded that "there is much independence in

13 urban property markets than the zoning ordinance anticipates." Stull, on the contrary, found that zoning does affect property values in the

Boston metropolitan area. His conclusion is that "the data seem to re­ veal that in the study area households were fairly sensitive to the land multiple-family, commercial, industrial or vacant land sold at a dis-

14 count, other things equal." In other words, Stull not only identified the existence of nonresidential land use externalities but also found that these externalities unambiguously caused negative effects on single-family home values.

The results given in Table 4-1 of this study did not corroborate the above findings. In fact, whether one considers equation 3 which contains a single nonresidential land use variable (denoted NSF2) or equations 4 through 6 where this variable is decomposed into its various components (multiple-family, commercial, industrial, public), the ex­ ternal effects of nonsingle-family land uses—after both the advantages and disadvantages of proximity have been taken into account—are not

12

Crecine et al. (1967), pp. 93-94.

14

Stull (1975), p. 551.

88

necessarily negative. In the study area of Tucson, propinquity to non­ residential sites is valuable and residential property values even in­ crease with proximity to them.

Equation 4 in Table 4-1 illustrates these results a little more explicitly. First, observe that the t-ratios of four land use coeffi­ cients in this equation vary. The multiple family, commercial and public/semi-public land-use coefficients are significant at the 0.01 percent level, and the industrial coefficient is significant at the

0.05 level.

Earlier it was suggested that the sign of the multiple family land use coefficient could be positive, negative or zero. Homeowners in some neighborhoods may have strong antipathy toward multiple-family structures (due for example to social and economic characteristics of their occupants). To the extent that this is true, the multiple-family parameter in equation 4 can be negative. However, if the multiple-family buildings were well designed and homeowners were more concerned with the appearance of the multiple residential units rather than with the social and economic background of their occupants, then the same parameter can very well be positive. The desire of homeowners to be mingled with a variety of visual or structural appearances in the neighborhood is an­ other explanation. Diversity is indeed a natural phenomenon to cities of the size of Tucson. Single-family residential monotony could have as by-products such things as accessibility inconvenience, lack of public street life, and even danger or the fear of the streets after dark.

J. Jacobs went further to assert that "some people fear to be alone in

89

their houses by day. ... We can see very well how fatal is its (resi­ dential) monotony.""'"

5

From a social viewpoint, Schafer, in his study of multifamily housing market, stated that "since the number of working wives is in­ creasing, households may spend less time in their homes, a trend which

16

should favor multifamily housing."

It is also found that many of the land uses controlled by zoning do not appear to have detrimental effects on residential property. In particular, some recent findings indicate that apartments and minor com­ mercial centers appeared to be relatively harmless; even the highway did

. . . . 1 7 not have significant negative effects on adjoining properties.

Some authors pointed out that the negative attitude toward apart­ ment houses is due to lack of information and families in residential areas were more opposed to multifamily housing than families living in areas of mixed land uses.

18

In Chapter 2, both positive and negative externalities from multiple-family residential sites have been discussed. The above addi­ tional comments (and those in Chapter 2) are some of the possible ex­ planations for the positive sign of the multiple-family coefficient.

What equation 4 says is that converting 10 percent of a neighborhood's land from single-family homes to multiple-family uses would raise home values in that neighborhood by around $940, ceteris paribus.

15

, p. 144 (comments in parentheses added).

16

Schafer (1974), p. 122.

17

Grether and Mieszkowski (1977).

18

Grossman (1966), pp. 3-6.

90

The commercial land use variable is now discussed. It is argued in Chapter 2 that the sign of this coefficient is a priori indeterminate.

It turns out in equation 4 that the commercial coefficient has a posi­ tive sign and is significant at the 0.01 level. Again this result does not imply that there are no neighborhood external diseconomies from the commercial site, only that they are more than offset by the advantages of proximity, shopping convenience, recreation facilities, and other positive external effects. However, as argued in Chapter 2, there could be a possibility that the relationship between commercial activity and residential property value was nonlinear. The estimation results of equation 5, in which a squared commercial term has been included (in addition to a linear term), show that the linear term has a positive coefficient which is significant at the 0.10 level, while the coefficient of the nonlinear term is negative. From the "coefficient's sign" stand­ point, these results indicate that the functional relationship being estimated has the shape of an inverted U. They can be used to defend the contention that a "right" amount of commercial activity may very well be preferred to additional single-family homes in a neighborhood.

Beyond a certain "optimal" point, large amount of commercial activity becomes undesirable, and this is the reason for the negative coefficient of the squared term. But this squared term's coefficient is clearly insignificant. One possible explanation for this is that the luw t-statistic is simply the consequence of stochastic influences peculiar to the specific set of data being used. Observe that there are 13 distinct regressors in equation 5, five of them dealing with land use.

Even if all were in fact variables in the true model, there is a

91

probability that one or two will show up with low t-ratios due to "noise" in the valuation process. The order of magnitude of the squared term's coefficient is not consistent with that of other variables. One does not expect that the various land uses will have exactly the same effects.

Nevertheless, the critical thing is that the magnitude of the squared term's coefficient is disproportionately smaller than that of the others.

Obviously, this estimation has failed to settle the question of nonlinear relationship between commercial activity and residential property values.

Industrial land use is accorded some special treatment in this study. Indeed, industrial activity is an essential ingredient in the development of Tucson's economic well-being. According to the City of

Tucson Planning Department "the industrial sector can help attain many other community advantages and objectives. Not only do they serve as a source of primary employment, but can act as a strengthening force in

19 our local economy and as major contributors to our local tax base."

Despite the importance of industrial activity, past industrial planning and zoning have been carried on, in the words of the Tucson Planning

Department, "under an approach filled with misconceptions and generali­ se) zations.

For instance, the Tucson's General Land Use Plan (GLUP), adopted in 1959, assumed that industrial uses are rarely, if ever, compatible with residential or commercial developments. For this reason, industrial

19

City of Tucson Planning Department, Industrial Patterns and

Trends, Executive Summary, March 1976, hereinafter cited as Tucson

Industrial Patterns (1976).

20

Op. cit., p. 1.

92

uses needed to be segregated to protect the community from possible

"harmful" effects. Recent efforts of industrial planning and zoning in

Tucson have attempted to refine on earlier approach. Specifically, these attempts are:

(a) to distinguish what types of industrial activities can or cannot be integrated into primarily non-industrial areas, and

(b) to have industry actively participate in attaining predetermined community goals.

In other words, industrial enterprises should not only provide the com­ munity with increased local employment or payroll income but they should also, after all factors (including environmental, ecological, etc.) are considered, be an asset to the community.

In order to offer some counter-arguments to the planning doc­ trine of industrial land-use segregation, it is necessary to understand the term "industrial use." Industrial development usually is equated with heavy manufacturing plants and warehouse facilities. In actuality, industry encompasses a broader array of activities. Industrial develop­ ment involves all those operations concerned with the production, stor­ age, and distribution of goods and services. Because of Tucson's loca­ tion and natural resource conditions (for example: a comparatively limited supply of water), companies which are involved in heavier pro­ cesses of manufacturing are nonexistent. Rather, Tucson's industrial profile is one of "light" and "clean" industrial operations: electronics assembly plants, regional warehousing and distribution centers, research and development laboratories, technical offices, and communication out­ lets. Local community attitude concerning industrial development in

Tucson is also an important factor. Surveys conducted in 1973 suggest that nearly all the community of Tucson favors the existence of select clean industry.

2

"*"

In the City of Tucson, as mentioned earlier, about 4.8 percent, or 2,095 acres, of total land is in industrial use. As far as zoning statutes are concerned, light manufacturing, research and development, and other light industrial operations are permitted within the commer­ cial classifications. In addition, the city code permits for research and development activities in its combination residential-office zone.

The Zoning Code of Tucson makes a distinction between light (industrial)

2 2

activity and heavy (industrial) activity. Within the city limits, light industrial activities predominate (95% of total industrial areas).

Land zoned for industry has increased to a significant degree over the years. In 1958 there were 1,755 acres in Tucson. By 1970 the figure had risen to 2,095 acres (see Table 3-1) and by 1975 industrial land

24 area had increased to 4,100 acres. However, due to its particular nature, industrial activities in Tucson were responsible in 1978 for only about 5 percent of total air pollution in Tucson. In fact, of the approximately 855 tons of total air pollution emitted in Tucson each

21

Tucson Industrial Patterns (1976), p. 11. Only one-third of the residents suggest to locate industry in new sections of the com­

2 2

, Article 1, Division 22 and

Article 1, Division 23.

23

Tucson Industrial Patterns (1976), p. 13.

24

Op. cit., p. 13.

25

day, 75 percent was related to vehicular traffic. It is clear that

94

the automobile, and not industry, is responsible for the majority, by weight, of the air pollution in the city. Furthermore, in a University of Arizona noise environment report "it was discovered that most outdoor noise in Tucson .is produced by automobiles, trucks, aircrafts, and rail-

2 road trains. Industry is not a major noise source at this time (1973)."

6

The above facts and observations on industrial activity may help explain the positive sign of industrial use's coefficient in equations 4 and 5.

Note that this coefficient is significant at the 0.05 level. Again, it should be pointed out that this does not in any way imply that there are no external diseconomies from the industrial site, only that they are more than offset by the advantages associated with industrial activities

(employment, proximity, etc.).

It is necessary to emphasize that the statement about the posi­ tive relationship between nonsingle-family land uses and urban property values does not stand without qualifications. The tests for individual regression coefficients presented in Table 4-1 show that an increase of nonsingle-family land uses tends to raise value of homes (significant at least at the 0.05 level). But obviously the above result is only appli­ cable over the range of this study's observations. For all four cate­ gories of land use under discussion, these observations are bunched

25

Pima County Air Pollution Emissions Inventory, AQ-129, April

1980, p. 2. The remainder 20% is attributed to other activities such as: pesticide usage, waste burning, wild fires, agricultural and com­ mercial operations, etc.

2 6

Vesta B. Conley and M.R. Bottaccini, Daytime Noise Environment in Tucson, Arizona, EES Series Report No. 40, University of Arizona,

1973, p. 1.

between zero and 10, 12, 20, or 21 percent (see Table 4-2). Therefore, the conclusions that an increase of nonsingle-family land uses tends to raise residential property values is practically appropriate over l: lov" ranges of these land uses, and less applicable to high ranges. There is a need in future studies to analyze the relationship between land uses and property values at high ranges of various nonsingle-family land uses

It is apparent that any statistical inference drawn from the findings of this study should take into account these range limits. For example, the range of multiple-family land use, as shown in Table 4-2, is between zero and 20 percent. The results presented in Table 4-1, together with the range qualifications, can now be understood in the following way: property value tends to vary directly with multiplefamily land use when this land use is within the range of zero and 20 percent of the neighborhood's total area. In Tucson, the actual propor­ tion of multiple-family land use varies from 1.7 percent to 47.4 percent

However, there are not enough observations (only 12 observations, or

23 percent of total observations) to infer about the effect of multiplefamily land use on property value when this proportion is beyond the range of 20 percent.

Similarly, homes value tends to vary directly with commercial, industrial, and public land uses, when their range is respectively be­ tween zero and 12 percent, zero and 10 percent, and zero and 21 percent.

In this study's sample, the actual proportion of commercial, industrial, and public land use varies from zero to 69.5 percent, 95.9 percent, and

57.7 percent, respectively. However, beyond the specified range of

12 percent (for commercial), 10 percent (for industrial), and 21 percent

96

Table 4-2. Range of Various Land-Use Parameters.

Land-use category

Multiple-family

Commerce

Industry

Public

Note: MF means Multiple-family.

Range in which at least

75% of the total observations occur

0 < % MF

<_ 20%

0 < % COMMERCE <_ 12%

0 < % PUBLIC

<_ 21%

(for public land use), the number of observations is so small that it is not appropriate to make any significant inference about the relationship between property value and these land-use categories.

The presentation of this study's empirical results is concluded with equation 6. It was previously noted that the coefficients of the

"squared commercial term" and "proximity" are not statistically signifi­ cant, although they have the correct sign. In equation 6 PROXIMITY is replaced by another accessibility variable denoted DISTANCE. The squared commercial term is dropped from the regression because of its poor per-

2 formance. As a result of this technical refinement, the R for estima­ tions of equation 6 slightly improved. All the coefficients have the right sign as in previous equations. Again, except for PLUMBING and

BASEMENT, all variables have significant coefficients. Note that the

27 coefficient of DISTANCE has a negative sign. In terms of t-ratios, this variable performs better than PROXIMITY, whose coefficient has a

27

This index of accessibility is negative, as expected, since decreased access implies lower prices for the property (a result pre­ dicted in the Alonso model of locational choice).

97

low t-ratio of 0.7 in both equations 4 and 5. After all PROXIMITY, as devised by W. Stull and employed in equations 1 through 5 of this study, is simply the ratio of local employment to the number of single-family homes in a neighborhood. It is a proxy to measure the local accessibil­ ity characteristics. DISTANCE, on the contrary, is a measure of general accessibility, since it takes into account both: (a) the separation in space between one neighborhood and all others; and (b) a weighing factor representing the total employment situation of the entire city (see Ap­ pendix C). It turns out that DISTANCE is, in fact, a significant vari­ able in the model. Table 4-3 shows the simple correlations for the

2 8 variables included in equation 6.

Nov; that the presentation of empirical results is completed, it is worth noting that almost without exception these results are in con­ formance with those derived mathematically in Chapter 2, especially those given by relationships (2.30) through (2.34).

Before proceeding to further discussion, it is however necessary to recapitulate the major findings:

1. The general accessibility should be measured using "DISTANCE"; local accessibility measured by "PROXIMITY" is largely captured by land use variables. "DISTANCE" is appropriate for comparing relatively small neighborhoods like census tracts; "PROXIMITY" is relevant for large areas like suburbs.

28

These generally moderate correlations suggest that the present study is not stymied by problems of multicollinearity.

Table 4-3. Correlations among Variables

Variables ROOMS AGE PLUMBING BASEMENT DISTANCE SCHOOLS EXCESTAX MINORITY MULFAM COMMERCE INDUSTRY PUBLIC

ROOMS

AGE

1.00

-0.42 1.00

PLUMBING -0.15 0.07 1.00

BASEMENT -0.16 0.34 -0.14

DISTANCE 0.57 -0.68

-0.01

SCHOOLS 0.68 -0.29 -0.40

EXCESTAX 0.11 -0.16 -0.01

MINORITY -0.33 0.30 0.43

MULFAM

-0.49 0.15 -0.09

COMMERCE -0.49

0.21 -0.09

INDUSTRY -0.09 -0.11

0.06

PUBLIC -0.06

0.05 0.40

1.00

-0.38

-0.17

0.01

0.01

0.07

0.02

-0.12

0.14

1.00

0.36

0.27

-0.26

-0.19

-0.39

0.13

-0.20

1.00

0.10

1.00

-0,67 -0.03

-0.15

-0.40 -0.14

-0.24

0.07

0.17

-0.27 -0.02

1.00

0.04

0.04

0.20

0.21

1.00

0.02

-0.21

-0.11

1.00

-0.07

-0.01

1.00

-0.19 llOO

See Appendix C for definitions of the variables.

2. The tax capitalization hypothesis is given additional support with new evidence in this study that tax levels had a signifi­ cant negative impact on property values.

99

3. School quality is an important determinant of home values.

(Parents' income and education can be significant predictors of student outcomes. To the extent that this is true, income can be correlated with property value, via effect of schoo.1 quality.)

4. Zoning variables are inferior to actual land-use variables; homeowners are more concerned with existing land-use mix than with probable future land-use configurations.

5. Over the ranges studied in this research, increase of nonsinglefamily land uses in a census tract tends to raise value of homes. That is, zoning ordinances could legitimately move away from a "separate facilities" philosophy to a "mixed land use" philosophy without lowering property values.

6. Racial considerations exert important influences on urban prop­ erty market. The hypothesis that the presence of a higher proportion of ethnic minorities in a neighborhood tends to de­ press value of homes there is given additional endorsement.

7. This study fails to corroborate the relationship between property values and noise or crime. This is due probably to the compara­ tively poor quality of data which are used to quantify these two

29 variables.

29

The nonconcordant date of data for property value and for noise and crime has been discussed in Chapter 3.

100

4. Further Discussions on Both Zoning and Land Use Models

A. The estimation results of this chapter and the conclusions concerning the empirical relationship between neighborhood externalities and property values will surely not go unchallenged. It seems appro­ priate to attempt to anticipate probable objections to them.

(1) The first objection may be based upon the observation that the variables representing the physical characteristics included in the regression equation (number of rooms, age, plumbing conditions, basement) do not entirely capture all dimensions of structure quality. Obviously, this is true. Clearly many secondary characteristics have been omitted, such as the type of building material, the lot size, the existence or nonexistence of carports, garages, fireplaces, swimming pools, etc.

Certainly these housing characteristics do influence the decision of home buyers, but they do not appear in the regression equation. These missing variables are important in their own right. Also they are prob­ ably highly correlated with many of the included variables. If this is the case, then the omissions of these variables would likely have the result that the coefficient estimates presented in Table 4-1 are the

30 products of biased estimations. An implied hypothesis is that perhaps the conclusiveness of the above empirical findings has been misrepre­ sented. Unfortunately, because additional physical quality variables were simply not available, it was not possible to test this implied hypothesis explicitly.

30

J. Johnston (1972), p. 169.

101

(2) It may be the case that the coefficient estimators in

Table 4-1 are not in fact biased. In his 1970 article, Harold Brodsky stated that there are alternative levels of aggregation in land value analysis and that the variables which play an important role at one level may be insignificant at another.

3

"'' It is possible that a variable may influence price at the city block level, but not at the community level. It may be true that brick houses command a higher price, ceteris paribus, than wood houses. However, if each neighborhood in the sample has a similar proportion of brick houses then building material will not be an important factor in explaining price differences between neighbor­ hoods (it is hoped this is the case in this study). The point here is that aggregation probably eliminates some variables, and the secondary physical characteristics that have been omitted in this study appear to

2 be likely candidates. The relatively high R obtained for equations 1 through 6 is additional evidence that no important variables have been omitted.

(3) Another objection is that the regression equation did not include average neighborhood income among the independent variables.

One can argue that the value of the residential property in a given neighborhood is influenced by the income of its buyers; thus, income must be included in the equation. This objection is, however, not

32 valid. The true value of a home does not in fact depend upon the income of the buyer. Instead, it is determined by the quality of the

31

H. Brodsky, Land Economics, August 1970, pp. 236-237.

32

See Ball (1973), p. 224, for a critique of this objection.

102

property and its neighborhood environment. If a house is large, and of high quality, it would tend to command a high price and be generally purchased by a household of high income. But it is incorrect to assume that there must be a direct causal relationship between price and owner income.

Griliches warns against the use of variables which are not direct characteristics of the commodity, such as the purchaser's income in ex-

33 plaining house prices.

Freeman III agrees with the above comments because the hedonic price technique seeks to explain house prices in terms of the house's own characteristics. He states that "since income is a characteristic of households rather than of housing, the logic of the technique dictates that income of the purchaser not be included in the regression equa­ tion."

34

Ball also argues that in no way can income be treated as entering the hedonic price equation directly. Instead, the influence of income is indirect, via its effects on the valuation of characteristics. The value of the coefficient derived for the mean family income variable might not have reflected the effects of differences in household income on prices, but rather could have been acting as a surrogate for other

35 variables. Furthermore, according to Ball, a particular problem re­ sulting from the inclusion of income is the relationship between it and

33

Griliches (1971), p. 5.

34

Freeman III (1979), p. 202.

35

Ball (1973), p. 224.

the other explanatory variables, which creates considerable multicol-

103

linearity problems.

B. The coefficients given in Table 4-1 are space and time spe­ cific. It is not correct to assume that they can be applied to some urban area other than Tucson without sufficient consideration of differ­ ences in the housing stock and neighborhood quality between the two areas. The time constraint also applies to intertemporal comparisons within the City of Tucson itself. The housing conditions and the landuse patterns of an urban area change over time, thus the implicit prices for various components of the housing bundle will change with them.

C. Although 20,478 acres of the land area of Tucson are zoned nonsingle-family uses, only 8,252 acres are actually occupied by various nonsingle-family uses, such as multiple-family residential, commercial and industrial. (See Table 4-4.) For example, within the City of

Tucson, 11,624 acres are zoned multiple-family while in actuality only

3,436 acres, or 29.6 percent, are inhabited by multiple-family homes.

The ratio of actual acreage of land use to acreage zoned commercial and industrial is 59 percent and 49 percent, respectively. On an aggregate basis, i.e., all categories of nonsingle--family uses combined, this ratio is around 40 percent. In other words, the actual nonsingle-family land use picture in Tucson in the period under study is not one which can be characterized as having an enormous amount of apartments, commer­ cial and industrial activities. The negative sign obtained for the coefficients of the zoning variables and the positive sign of the land use coefficients can partially be explained by the discrepancy in the

104

Table 4-4. Zoning and Land Use in the City of Tucson (Nonsinglefamily Uses)

Land Zoned by

Categories in

Acres (1)

Actual Land Use in Acres (2)

% of

(2) (1)

Multiple-family

Commercial

Industrial

Total

11,264.22

4,592.65

4,261.28

20,478.15

3,435.61

2,721.49

2,094.98

8,252.08

29.6%

59.3%

49.2%

40.3%

Source: Appendix A and Chapter 3 of this study. Note that data in column (1) are those which read horizontally from Appendix A.

Data in column (2) are those which read vertically from the table presented in Appendix A. acreages of land zoned and actually used in each category. Presumably the actual amount of land uses seems to approximate the optimal level of land-use admixtures, while the proportions of land zoned for various uses deviate from the point of optimum integration of residential and nonresidential activities.

D. The empirical results of this chapter have demonstrated that neighborhood externalities played an important role in determining the price of single-family homes in a sizeable urban area (Tucson) in a particular year (1970). In themselves these results are not particularly surprising. Similar conclusions have been reached by a few other empiri­ cal studies, in a different context (Stull, Lafferty and Freeh III, etc.). It is fortunate that this study did not have to use proxy vari­ ables to quantify various neighborhood land use characteristics, as many

36

Note again that the zoning coefficients are not significant.

105

37

earlier studies did. Also it did not have to use complicated composite variables derived from statistical technique of factor analysis and

38 principal component analysis. What is unique in this study is that it does more than simply confirm the significance of neighborhood ex­ ternalities. Indeed, the study shows that, for the first time, it is possible to take account of both the advantages and disadvantages of neighborhood externalities in order to measure their net effects. Equa­ tion 6 is particularly interesting in this regard and summarizes the main conclusions of this study which appears to be the first ever to have uncovered beneficial effects of mixed land use.

37

See, for example, E. Brigham (1965); R.N.S. Harris et al.

(1968); W. Oates (1969); H. Brodsky (1970). The variables they usually chose were average or median income as proxies for neighborhood charac­

38

See, for example, Kain and Quigley (1970); A.T. King (1973).

Essentially, this technique provides a way to combine a number of inde­ pendent variables. Some critics argued that no particular economic meaning should be attached to the components obtained, for they are merely statistical artifacts. For a discussion about the validity and usefulness of this technique, see M.G. Kendall, A Course in Multivariate

Analysis, New York, Hafner Publishing Co., 1968.

CHAPTER 5

MIXED LAND USE AND POLICY IMPLICATIONS

1. Introduction

The empirical findings of Chapter 4 have shown that improved accessibility to employment centers or shopping areas increases the value of the residential property because negative externalities (con­ gestion, noise, etc.) are more than offset by the benefits accruing from locational advantages gained by proximity.

Neighborhood industrial and commercial activity provides easy accessibility to employment opportunities, shopping, and recreational activities for nearby residents. However, some degree of nuisance is likely to be incurred. In modern city life, there is always the neces­ sity of giving up something to get something else. Environmental quality is therefore a relative matter--a question of choices and tradeoffs. In the case of mixed neighborhood land use, where activities of differing nature can be accommodated on the same land area, this problem of trade­ offs is of particular importance.

In section 2, the question of mixed land use shall be discussed concomittantly with that of accessibility and transportation. Then, section 3 shall consider some of the more important policy implications in the light of the findings of this study. Finally, section 4 will close with a few comments on future research issues.

106

107

2. Mixed Land Use

The impact of new lifestyles and economic growth on urban landuse patterns has been very important. With the increase in population, in automobile transportation and in economic activities, many cities

(including Tucson) attempted to cope with this problem by using their open land. The centers of the cities—which were historically the base for cultural needs, employment opportunity, living and leisure—lost ground. The surrounding neighborhoods became blighted and their func­ tions widely separated. People moved to the suburbs. The result has been urban sprawl.

Urban zoning practices must bear some responsibility for encour­ aging this expansion, according to Clawson, because "its" (zoning prac­ tice) emphasis upon the separation of different land uses, and the scale at which this was applied, have also created the sprawled city and forced citizens to travel to jobs."'

1

' The application of this land-use separa­ tion philosophy, indeed, has been too extensive, resulting in inefficient spatial arrangement of economic activities. This can be illustrated by considering a city divided into two separate areas: a nuisance-free residential area (zone 1) and another area where economic and other activities are permitted (zone 2). The costs of land use separation will then be compared with the benefits of mixed land use.

(a) Costs of Land Use Separation

One could reasonably believe that the costs incurred by using the

"separate facilities" scheme would be extremely important.

108

(1) First, there is the commuting cost, in terms of time and money. When economic activities (commerce, industry) are spatially segregated from the residential areas according to the land-use separa­ tion philosophy, travel from home (zone 1) to work (zone 2) would take up significant amounts of time for each household to perform its daily tasks.

(2) Second, the amount of energy resources expended for travel between workplace, shopping areas and living place would increase, as would the costs of energy use. In the City of Tucson, in the year under study, 92 percent of the households depended on the automobile to go to

2

work. The dependence of so many people on this means of transporta­ tion—and on oil--would cause a very vulnerable situation in the face of potential energy supply interruptions or oil price increases. During the past year, a number of detailed studies that examine U.S. energy problems have been published. They did not promise that high-cost energy can be avoided, but only that there are better ways (e.g., conservation

3 measures) to adapt to it. The philosophy of land-use separation which encourages urban expansion does not seem to promote such conservation policies.

(3) Another effect of increased volume of automobile traffic re­ sulting from long distance travel between home and workplace is the

2

1970 Census of Population and Housing, Tucson, Arizona, SMSA.

3

See

:

(1) Energy Future: Report of the Energy Proiect at the

Harvard Business School. New York: Random House, 1979; (2) Energy: The

Next Twenty Years, A Report sponsored by the Ford Foundation. Cambridge:

Ballinger, 1979; and (3) Energy in America's Future: The Choices Before

Us, a Study by the staff of the RFF. Baltimore: The Johns Hopkins

University Press, 1979-

109

aggravation of air pollution in cities. Pollution is the harmful alter­ ation of urban environment by its own residents. Transportation accounts for about 75 percent of air pollution in Tucson, it is recalled. The costs incurred by the degradation of air quality can be quite important, because atmospheric pollutants create health hazards, corrode metals, rot masonry, damage crops, injure livestock, etc. One way of dealing with the principal source of metropolitan air pollution, the automobile, may be the lowering of the number of automobiles operating at any given time in the city. Obviously, the policy of land-use separation is not conducive to the reduction of travel loads.

(4) The decrease in vehicular traffic would also "decongest" the

4 urban transportation system, especially during the rush hours. But ex­ tensive traveling between residential and non-residential areas by urban commuters aggravates the congestion problem. Urban zoning ordinances which advocate the "separate facilities" scheme are responsible for the increase in the use of the automobile as the principal means of trans­ portation and all the costs incurred by traffic congestion.

(5) Finally, land use separation would result in a regrettable loss of urban diversity, or a lack of visual heterogeneity, because of the sc'.meness and dreariness of the city."' Richard Dober, of the American

Institute of Planning, criticizing this land-use philosophy, stresses that "In the name of health and safety, the physical surrounds have been

4

It is unrealistic to speak about the elimination of congestion, so long as most businesses choose to close at approximately the same hour of the day.

5

Jacobs (1961), pp. 222-238, passim.

110

sterilized, a result often expressed in a single-color land use map, an abstract document that almost denies human aspirations and existence.

6

Symbolically and otherwise, we need not live a one-color life. ..."

(6) Despite the relatively high costs involved in the spatial segregation of residential from non-residential land uses, the x^hilosophy of land use separation predominates contemporary urban zoning practices.

The primary objective of these practices is supposedly to provide pro­ tection to the single-family residential area on the ground that a change in character (e.g., an intrusion of commercial establishment) will have a direct, substantial, and adverse impact on property values. The find­ ings of the present study, however, did not give empirical support to this presumed negative relationship as they indicate that there is no decrease in residential property values for low levels of non-residential activity. All the above arguments would lead to the conclusion that in the case of land use separation the benefits are relatively smaller than the costs that have to be incurred.

(b) Benefits of Mixed Land Use

In recent years, some authors vehemently criticize the land-use separation doctrine and suggest a new approach often referred to as

7

"mixed land use." A promising aspect of this approach is its ability to create, on relatively small areas of land, an environment that has many advantageous characteristics, such as proximity of economic

0

Procos (1976), p. v.

7

See, for example, J. Jacobs (1961), D. Procos (1976).

Ill

activities, increased educational or recreational opportunities, and social interaction.

(1) Proximity to industrial and commercial activity provides, as discussed earlier, accessibility to jobs and shopping. Its main ad­ vantage is decreased travel. Consequently, much of the cost incurred by traffic, congestion, commuting time, air pollution, etc. can be avoided g or eliminated. While urban developments disperse and commuting costs rise due to land use separation, a reduction of the transportation burden can be achieved through mixed land use. Transportation can indeed be replaced by appropriate land use combination. Restrictions in the avail­ ability of energy for transportation require a careful consideration of more intensive utilization of space. Inflation and high gasoline costs would certainly place tremendous pressure on urban societies. Mixed land use deserves to be studied seriously as it can be used to eliminate unnecessary transportation and reduce wasteful commuting costs. More­ over, the basic task of land use is to serve human activities. To the extent that many such activities can take place in the same spatial location, the designation of that location as the exclusive setting for

. . 9 an arbitrarily limited number of these does not protect property values.

(2) Another advantage of mixed land use is that it can be used to facilitate social interaction. Land use separation, on the contrary,

Q

For the discussion of household as a consumer of time, see Gary

Becker, "A Theory of the Allocation of Time," Economic Journal, September

1965.

9 .

Richard Muth (1969) theorizes that the savings in land at lower densities in the household budget is offset by increases in transporta­ tion costs.

112

encourages social withdrawal. A community, where mixed land use accepts the juxtaposition of diverse social and economic activities, offers its residents both a place to live and a place to work, and provides the necessary amenities (education, medical care) as well as the opportuni­ ties for social contacts, associations, etc.

(3) Of course, certain mixed land uses can be a source of ex­ ternal diseconomies if their operations are not in conformance with prescribed performance or design standards. This dissertation has found, however, that property values are not adversely affected by low levels of non-residential activity. The inference is that, in the context of the present study, the benefits of mixed land use are comparatively greater than the costs of external diseconomies. Similar results are also found in a recent study by Li and Brown.'

1

'

0

(4) Finally, it is worth noting that the National Commission on

Urban Problems (NCUP), in its Report to the U.S. Congress and the Presi­ dent, has recommended that States should enact legislation enabling lo­ calities to prevent wasteful and scattered development and to facilitate work opportunities near homes."'"''"

Li and Brown (1980) studied the effects of proximity to non­ residential activities on housing prices. Reasoning that proximity pro­ vides both benefits (accessibility) and detriments (externalities), they found that: (a) in the case of proximity to commercial establishments, accessibility dominates the externality, i.e., benefits are greater than detriments; and (b) in the case of proximity to industry, housing prices decline by about $1,400 for each doubling of distance; other things equal, a house located about 600 yards from industry commands the highest premium.

"'""'"NCUP (1968), pp. 235-253. The National Commission on Urban

Problems was appointed by President Johnson in 1967 and chaired by Sena­ tor Paul Douglas to carry out the purposes defined in Section 301 of the

Housing and Urban Development Act of 1965.

113

It was certainly a judicious recommendation since as more industries, more commercial centers, and more households can be accommodated on the same neighborhood, the peak load of automobile traffic in many urban areas will become smaller, thereby reducing the costs of transportation in the process.

Having completed the discussion of the costs and benefits of land use separation and mixed land use as well as the empirical results of this study regarding the problem of neighborhood externalities, it is now necessary to consider the implications of these findings for public policy which relates to this problem.

3. Policy Implications

(a) First, in the light of the conclusion regarding mixed land use, and if commercial and industrial activities can make good neighbors then a municipal zoning plan should provide, among other things, the following:

(1) Nearness of work places to residential areas to create a community geography in which spare time can be devoted to the family and recreation rather than driving or riding on buses;

(2) A land saving policy should be adopted: only as much land as is needed for urban development should be so used and the rest can be preserved for recreation, farming, and open space. This notion is often referred to as a "more compact living" concept. Also, it is time to change the view that land is little more than a commodity to be exploited and traded. There is a need for a land ethic that regards land as a resource which, improperly used, can have the same ill effects as the

114

pollution of air and water, and which therefore warrants the same pro­ tection.

(b) Until now, the predominant practice in zoning is to segre­ gate each use of group of similar uses into separate districts. But another more promising approach, as mentioned in Chapter 2, is to con­ trol the mix of land uses through the utilization of design cr.ii^ria aimed at eliminating undesirable externalities by controlling noise, ugliness, traffic generation, landscaping, etc. In this way many uses formerly thought to be incompatible and blighting may be brought into harmonious relationships. City planners, concerned with the problem of accommodating mixes of land uses at close quarters, should contemplate the creation of a new zoning classification which can be called "mixed use district." In this district, it is possible to economize in the use of land and other resources (energy, in particular) by clustering hous­ ing, job, shopping, recreation, etc. Through the convenient location of homes, shops, and work places, the volume of travel required to perform the daily tasks of the households can be drastically reduced.

(c) The central business district can be rezoned and rebuilt not for office and commercial uses alone, but as living communities that have a variety of housing choices (single-family homes, townhouses, apart­ ments, condominiums, etc.), convenient neighborhood shopping and nearby recreation facilities. The effect will be a 24-hour-a-day living com­ munity. Also, the commercial areas of the city that often make poor neighbors can be clustered with cultural activities, services, and

115

recreation, in the kind of multipurpose neighborhood centers mentioned earlier.

(d) Clean and well designed industrial facilities can be good neighbors as well as essential economic supports. This will make it practical to introduce residential land uses close to job opportunities.

In other words, the practice of mixed land use would bring industries and jobs closer to where people live. However, some neighborhood resi­ dents may easily get "scared" by the mention of "industrial" and they may react accordingly. Therefore it would be advisable for the zoning board to create a new zoning classification which might be called "parkoffice," dropping any connection with the word "industrial."

(e) Zoning can be a poor tool for achieving an optimum choice in living and working environments because there are localized vested in­ terests in retaining obsolete zoning patterns which perpetuate or promote economic, social, and housing-type segregation. This is true when the planners attempt to introduce the new concept of mixed land use (Tucson, in recent time, has several examples of this kind: Kino Area, General

Instruments Corporation, Midvale Farms). Thus, some kinds of institu­ tional reform (e.g., decentralization of land-use control) seem to be necessary. Without such reform most of the innovations that have been mentioned here would be unimplementable.

4. Future Research Issues

This study has now completed the limited range of tasks set out in Chapter 1. Having come this far, it is reasonable to

"^See p. 34 of this paper.

116

ask about what lies ahead. Since the Ph.D. dissertation is a restricted art form by its very nature, it is inevitable that in the present study many interesting topics on the economics of land use control were either inadequately treated or ignored in the main text. It seems now appro­ priate to redress this involuntary deficiency by indicating at least briefly what some of these topics are so that they might be analyzed fruitfully by future researchers.

It is a generally accepted rule of empirical research that no single study can be truly conclusive since there is always the possibil­ ity that there are furtive factors lurking behind the scenes. The gen­ eral results given in Chapter 4 appear to be fairly solid. Nevertheless, their validity is particularly strong only when the land- use parameters are within certain low ranges studied in this research (see Table 4-2).

It is not practical to infer from the present study about the effects of externalities on property values over higher ranges of land use. Surely it would be interesting, in future studies, to analyze the relationship between neighborhood land uses and residential housing price at high ranges of various nonsingle-family uses.

It would also be desirable for some future researchers to con­ sider the impact of neighborhood externalities on the market value of properties other than single-family homes. As noted previously, multifamily housing became a large factor in the housing market. In the period since 1957, multifamily housing has accounted for a growing share

12

of private housing starts. Solid statistical evidence on the

12

In 1957, multifamily housing starts in the U.S. accounted for

10.2 percent of total starts; in 1971, they accounted for 41.6 percent

(see Schafer, 1974, pp. 126-127).

relationship between land use externalities and market value of multifamily property would be of obvious usefulness to city planners and municipal government officials.

Turning to the issue of mixed land use, it seems fair to say that the discussions in Sections 2 and 3 above are much more suggestive than definitive. The main reason for this is that the topic being con­ sidered is entirely novel, not resembling something previously known or discussed in the economics literature. It would be interesting for future researchers to assess the effects of mixed land use on at least two areas of concern. The first is the issue of savings in transporta­ tion, whose importance was only slightly touched upon in Section 2. The other is the preservation of agricultural land. In both these two cases, the sharing of land or space by a mix and multiplicity of land uses should become a major aspect of the solution to a pressing problem.

This is the problem of efficient allocation of land to various uses as a hedge against a loss of scarce natural resources—a loss that is in­ creasingly being seen as irreversible.

APPENDIX A

118

Table A-l. Land Use in Acres by Zoning—City of Tucson.

Zoning

Category

Vacant

Land

Total

Developed

Single-

Family

Multi-

Family

Commer­ cial

Indus­ trial

Public/

Semi-P.

Washes/ Acres

Medians Net Area

SR a

UR

SH

Sub-Total 2,903.

426.53

565.28

99.181

317.71

229.92

547.63

15.

10. ,86 2,652.68

.02

1,208.48

— — —

33.87

317. ,86 3,895.03

RX-1

RX-2

R-l

MH

5,279.

138.69

230.13

555.77

133.52

218.01

,289.35

106.66 439.

Sub-Total 6,878. ,747.54 712,

132.

215.

HHP

R-2 2,104.

0.55

6,406.81 3

0.55

R-3

R-4

307. 1,190.52

745.11

R-5

PR

12.24

0.34

Sub-Total

2,908. 8,355.57 3

357.32

181.69

0.16

659.

— —

— —

235.

188.

2.

741.76

934.79

847.34

— —

2.15

,553. .70 8,510.88

117. .05 1,498.49

2.

3. 1,236.91

15.45

0.34

,866. ,07 11,264.22

B-l

B-2A

932.32

767. 1,575.02

B-2

236.11

B-2H

141.48

B-3

67.34

MU

29.53

Sub-Total

1,610. 2,981.80

P-I

1-1

1-2

102.49

1,841,

2,049.54

57.39

Sub-Total

2,051. 2,209.42

98.19

12.35

1.84

26.32

303.94

47.08

78.31

125.39

87,

,50

143.14

198,

22.10

25, .72

331.

29, 284.

— —

31, 291.

65.

3,

7.

,34

36, .39

13, .39

1,675.45

2,342.19

298.57

156.18

— —

68.71

51.55

171. ,40 4,592.65

TOTAL a

Includes blocks surrounding the immediate fringe area of the City of Tucson.

*

312.67

.97 131. 3,891.22

57.39

.97 131. 4,261.28

Source: Interim Concept Plan, Department of Community Development, City of Tucson, 1973, pp. 6-13.

APPENDIX B

CITY OF TUCSON ZONING DISTRICT

NARRATIVE SUMMARIES

SR,

UR,

RESIDENTIAL & URBAN RANCH

A low density residential area with horses permitted in SR, UP.

RX-1, (Urban Ranch) and RX-1 districts. Riding stables, guest ranches,

RX-2 veterinary hospitals, golf courses, etc., are permitted.

R-l RESIDENTIAL-SINGLE FAMILY

Primarily for the use of single family (SF) residences. Schools, churches and public buildings are permitted. Restricted home occupations and private automobile parking is permissible.

Second dwelling on lots over 10,000 sq. ft.

R-2

R-3

LOW DENSITY RESIDENTIAL-MULTIPLE DWELLING

Multi-family (MF) residences are permitted with the primary re­ striction that there will be at least 3,000 sq. ft. of lot area for each dwelling unit. Single detached residences can be built on a minimum of 5,000 sq. ft. F.-l uses are permitted.

HIGH DENSITY RESIDENTIAL-MULTIPLE DWELLING

Primarily for apartment houses and dwelling courts. Social, wel­ fare and R-2 uses are permitted.

R-4

R-5

MH

MHP

OFFICES, LIMITED RESEARCH AND DEVELOPMENT USES

R-3 uses are permitted.

HIGH DENSITY RESIDENTIAL/COMMERCIAL

High density, high-rise multiple residential with compatible commercial and professional uses permitted. Buildings may be built to a height of 30 stories.

MOBILE HOMES

Mobile (trailer) homes will only be permitted in an "MH" district.

R-l uses are permitted. On individual lots there must be at least 7,000 sq. ft. per trailer and within mobile home courts there must be at least 2,000 sq. ft. per mobile home.

MOBILE HOME PARK ZONE

Primarily for mobile home parks where spaces are rented to occu­ pants. Schools, parks, churches, public utility and municipal uses are permitted. One single family residence per park, social and recreation centers, private recreation uses are allowed as accessory uses.

120

121

RV

PR

TRAVEL TRAILER-RECREATIONAL PARK ZONE

A travel trailer park is the only permitted principal use. Three mobile homes per park as accessory uses and the accessory uses of "MHP" are allowed.

PARKING RESIDENTIAL ZONE

The purpose of this zone is to provide off-site off-street park­ ing at or below grade to areas in residential zonos to serve buildings and uses which cannot feasibly provide on-site parking.

B-l LOCAL BUSINESS

The most restrictive of the commercial zones. Limited to retail sales. Items that are produced on the premises must be- sold on the premises. R-4 uses are permitted.

B-2, GENERAL & INTENSIVE BUSINESS

B-2A, Retail business with wholesale, warehousing, repairing and amuse-

B-2H, ment enterprises permitted. Limited manufacturing is permitted.

B-3 Restricted residential uses are permitted.

P-I PARK INDUSTRIAL

The most restrictive of the industrial zones. Intended to be used as a buffer between the industrial and residential area.

Administrative, manufacturing and wholesale activities that can be carried on in an unobtrusive manner are permitted. Limited retail sales are permitted when incidental to primary use.

1-1, LIGHT & HEAVY INDUSTRIAL

1-2 Commercial, industrial and manufacturing uses. Dwelling uses are not permitted except for caretakers residences. Within 1-2 buffers are required to protect nearby residential areas.

Source: City of Tucson, Arizona, Zoning Ordinance, No. 3038, Adopted

Sept. 11, 1967.

APPENDIX C

DEPENDENT AND INDEPENDENT VARIABLES

Description, Notation and Sources of Data

This is taken directly from the U.S. Bureau of Census, 1970 U.S. Census of Population and Housing, Census Tracts, Tucson, Arizona, SMSA (here­ inafter cited as 1970 Housing Census), Table H-l.

MEDIAN NUMBER OF ROOMS (Notation: ROOMS)—

The basic source is the 1970 Housing Census (Table H-l). The Census provides a series for the median number of rooms in all year-round housing units.

PROPORTION OF OLD HOUSING UNITS (Notation: AGE)—

The 1970 Housing Census (Table H-2) provides the distribution of all year-round housing units among six age-of-construction categories for each census tract (1939 or earlier, 1940 to 1949, 1950 to 1959, 1960 to

1964, 1965 to 1968, and 1969 to 1970). In this study, housing units are divided, for convenience, into two groups: new and old. Dwelling units bu.ilt in 1959 or earlier are considered "old." The number of old houses divided by the total number of housing units gives the proportion of old housing units in the neighborhood.

PROPORTION OWNER-OCCUPIED DWELLING UNITS LACKING SOME OR ALL PLUMBING

FACILITIES (Notation: PLUMBING)--

This is taken directly from the 1970 Housing Census (Table H-l).

The 1970 Housing Census (Table H-2) provides the number of single-family homes with basement. This number divided by the total number of singlefamily homes gives the percentage of single-family homes with basement.

ROAD DISTANCE TO PLACE OF WORK (Notation: DISTANCE)—

This series is obtained by measuring the road distance from the geograph­ ical center of each and every census tract to the center of all other tracts of Tucson, using the Map of Tucson and Vicinity Census Tracts,

1970 Housing Census. This measure, denoted d^. (where i = origin and

122

123

, is then weighted by the ratio —where E is total

E employment in the city and e, is local employment for tract j. It is

52 e.

3

A. = E d. . (-

1

), i = 1, ...

1 j=i ^

Employment figures are taken from Employment Survey by Traffic Analysis

Zone, Tucson Area Transportation Planning Agency, 1973, hereinafter cited as TATPA Survey.

PROXIMITY TO LOCAL EMPLOYMENT (Notation: PROXIMITY)—

This is the ratio of local employment to the number of single-family homes. The denominator of this ratio is taken from the 1970 Housing

Census. The numerator, total local employment, is given by TATPA Survey.

PROPERTY TAX RATE (Notation: TAXRATE)—

Property tax rates are obtained from the Pima County, Arizona, 1969-1970

Annual Report and 1970-1971 Adopted Budget, Schedule 38, page 53. Addi­ tional information (e.g., improvement district tax rate) are provided by

Pima County Assessor's Office and Property Management Department.

EXCESS TAX BURDEN (Notation: EXCESTAX)--

Same source of data as above. Using the formula specified in Chapter 3,

(EXCESTAX = 1 +

T i "

T

1

T

1 census tract. This is an index ranging from 1.000 to 1.502.

READING ACHIEVEMENT TESTS (Notation: SCHOOLS)—

Reading achievement tests are used as a proxy variable in the absence of a direct measure of the local school quality. Note that the first reading scores in the State of Arizona and Tucson were given in January

1971. These test scores are obtained from the Office of the Pima County

School Superintendent.

PERCENT OF NEGRO IN NEIGHBORHOOD'S POPULATION (Notation: BLACK)—

This is taken directly from the 1970 Housing Census (Table P-l).

PERCENT OF PERSONS OF SPANISH LANGUAGE OR SURNAME (Notation: HISPA)—

PERCENT OF ETHNIC MINORITY IN NEIGHBORHOOD (Notation: MINORITY)—

This is simply the arithmetic sum of BLACK and HISPA.

PERCENT OF OCCUPIED HOUSING UNITS WHICH ARE CROWDED (MORE THAN 1.01

This is taken from the 1970 Housing Census (Table H-l).

PERCENT OF CIVILIAN LABOR FORCE UNEMPLOYED (Notation: UNEMPLOYED)--

This is taken from the 1970 Housing Census (Table P-3).

124

PERCENT OF FAMILIES WITH INCOME BELOW POVERTY LEVEL (Notation: POORS)—

This is taken directly from the 1970 Housing Census (Table P-4).

CRIME AGAINST PERSONS (Notation: CRIMEPERS)—

This is taken from the Monthly Crime Statistics, Department of Police,

City of Tucson. Data are available for the period August-December 1972.

It is expressed as a rate per 1,000 population.

CRIME AGAINST PROPERTY (Notation: CRIMEPROP)—

Same source of data as above. This type of crime is expressed as a rate per 100 residential single-family homes.

AUTOMOBILE TRAFFIC NOISE (Notation: AUTONOISE)—

Data for this variable are taken from V.B. Conley and M.R. Bottaccini,

Daytime Noise Environment in Tucson, Arizona, EES Report No. 40, College of Engineering, University of Arizona, July 1973.

AIRCRAFT NOISE (Notation: AIRNOISE)--

Same source of data as above.

PROPORTION OF LAND IN THE NEIGHBORHOOD DEVOTED TO MULTIPLE-FAMILY USES

(Notation: MULFAM if actual land use; and MF if zoning)—

The source of all the land use variables used in this study is the City of Tucson, Department of Community Development, Land Use, Zoning, and

Census Data by Census Tract, Spring 1973. This report, published in

1973, describes the 1971 land use situation. The assumption made in this study is that the city land use patterns changed very little between

1970 and 1971.^ The denominator of the proportion multiple-family vari­ ables (and of all the following land use variables) is the total land area of the census tract net of washes and medians. The numerator is the acreage of land devoted to multiple-family uses—defined to include land zoned R2, R3, R4, R5, MHP, PR (see Appendix B for a summary of these and other zoning classifications).

PROPORTION OF LAND IN THE NEIGHBORHOOD DEVOTED TO COMMERCIAL USES (Nota­ tion: COMMERCE if actual land use,- COMMCIAL if zoning)--

Commercial uses, as defined in this study, include the following land uses: Bl, B2, B2A, B2H, B3, RV.

PROPORTION OF LAND IN THE NEIGHBORHOOD DEVOTED TO INDUSTRIAL USES

(Notation: INDUSTRY if actual land use; INDTRIAL if zoning)—

This category includes land zoned P-I, II, 12.

PROPORTION OF LAND IN THE NEIGHBORHOOD DEVOTED TO PUBLIC AND SEMI-PUBLIC

USES (Notation: PUBLIC)--

This category includes land devoted to religious, medical, cultural ac­ tivities (churches, hospitals, museums); schools; universities, public buildings.

City of Tucson Planning Department officials concurred with the assumption above.

P L E A S E N O T E :

C o l o r e d M a p o n p a g e 1 2 5 w i l l n o t r e p r o d u c e w e l l i n x e r o g r a p h y .

F i l m e d i n t h e b e s t p o s s i b l e w a y .

U N I V E R S I T Y M I C R O F I L M S .

f .N

:

.

/

r'

• h

>-"

:

5

/'

—V:

• .

o

< o

• sS :

-•

w

5 c

•X

n

O

TO

o z

>

o

o

:

<3

z

>

C O R O N A D O N A T I O N A L F O R E S T

E X H I B I T 1

S u n r i s e D r i v e

I V ;

y i

)

\ '

I

J

J l

R -I

• /

' - ' O

Los R o alt's tl d.

V l j

L

i n

(INDUSTRIAL

•:£s:tr&ct1ve

AGRICULTURAL

•HLES

Sp

N

O,

I

I—:

§S«J-

iiiij

'"•-r vX.;.'.;.

4502

4503

26

4501

27

20

28

2901

2902

34

35

30

33 01

40 04

40 01

3302

4002

4003

4005

4007

4006

4402

25

23

24

22

4102

38

4301

BOUNDARY SYMBOLS

Census Tract Boundaries

39

37

4101

4102

4102

Exhibit 2. Tucson and Vicinity—Map of 52 Census Tracts

36

BIBLIOGRAPHY

Adelman, Irma and Griliches, Zvi. "On an Index of Quality Change,"

Journal of the American Statistical Association, Sept. 1961, pp. 535-548.

Alonso, William. Location and Land Use. Cambridge, Mass.: Harvard

University Press, 1964.

Arrow, Kenneth J. "The Organization of Economic Activity: Issues

Pertinent to the Choice of Market versus Nonmarket Allocation," in Haveman, R.H. and Margolis, J. (eds.), Public Expenditures and Policy Analysis, Chicago: Markham Publishing Company, 1970, pp. 59-73.

Bailey, Martin J. "Note on the Economics of Residential Zoning and

Urban Renewal," Land Economics, August 1959, pp. 288-292.

Bailey, Martin J. "Effects of Race and of other Demographic Factors on the Values of Single-Family Homes," Land Economics, May 1966, pp. 215-220.

Bailey, Martin J., Muth, R.F., and Nourse, H.O. "A Regression Method for Real Estate Price Index Construction," Journal of the American

Statistical Association, December 1963, pp. 933-942.

Ball, Michael J. "Recent Empirical Work on the Determinants of Relative

House Prices," Urban Studies, Vol. 10, No. 2, June 1973, pp. 213-233.

Barlowe, Raleigh. Land Resource Economics. 3rd edition, Englewood

Cliffs, New Jersey: Prentice-Hall Inc., 1978.

Becker, Gary. "A Theory of the Allocation of Time," Economic Journal,

Sept. 1965, pp. 493-517.

Bogue, D.J. "Discussion," in Perloff, H.S., Wingo, L. Jr. (eds.),

Issues in Urban Economics, Baltimore: The Johns Hopkins Press,

1968, pp. 413-419.

Brigham, Eugene E. "The Determinants of Residential Land Values," Land

Economics, Nov. 1965, pp. 325-334.

Brodsky, Harold. "Residential Land and Improvement Values in a Central

City," Land Economics, August 1970, pp. 229-247.

127

128

City of Tucson, Arizona, Planning Department, Industrial Patterns and

Trends, Executive Summary, March 1976.

City of Tucson, Arizona, Planning Division. Interim Concept Plan,

Tucson, Vol. 1, Community Inventory and Analysis, July 1973.

City of Tucson, Arizona, Planning Division. Zoning Ordinance, No. 3038, adopted Sept. 11, 1967.

City of Tucson, Arizona, Department of Police, Monthly Crime Statistics,

1972.

City of Tucson, Arizona, Department of Community Development. Land Use,

Clawson, Marion. "Environment and Land Use," in Perloff, H.S. (ed.).

Agenda for the New Urban Era. Chicago: American Society of Plan­ ning Officials, 1975, pp. 123-134.

Coase, R.H. "The Problem of Social Cost," Journal of Lav.' and Economics,

1960, pp. 1-44.

Conley, V.B. and Bottaccini, M.R. Daytime Noise Environment in Tucson,

Arizona. EES Series Report No. 40, College of Engineering, Univer­ sity of Arizona, July 1973.

Court, Andrew T. "Hedonic Price Indexes with Automotive Examples," in

The Dynamics of Automobile Demand, New York: General Motors

Corporation, 1938.

Crecine, John P., Otto A. Davis, and John D. Jackson, "Urban Property

Markets: Some Empirical Results and Their Implications for Munici­ pal Zoning," Journal of Law and Economics, Vol. 10, October 1967, pp. 79-99.

Daniels, Charles B. "The Influence of Racial Segregation on Housing

Prices," Journal of Urban Economics, April 1975, pp. 105-122.

Davis, Otto A. The Economics of Municipal Zoning. Ph.D. Dissertation,

University of Virginia, 1960.

Davis, Otto A., and Whinston, Andrew B. "The Economics of Urban Re­ newal," Law and Contemporary Problems, VoJ. 26, No. 1, Winter 1961, pp. 105-117.

Davis, Otto A. "Economic Elements in Municipal Zoning Decisions," Land

Economics, Vol. 39, Nov. 1963, pp. 375-386.

Davis, Otto A., and Andrew B. Whinston. "The Economics of Complex Sys­ tems: The Case of Municipal Zoning," Kyklos, Vol. 17, Fasc. 3,

1964, pp. 419-445.

129

Edelstein, Robert. "The Determinants of Value in the Philadelphia Hous­ ing Market: A Case Study of the Main Line, 1967-69," Review of

Economics and Statistics, Aug. 1974, pp. 319-328.

Ellickson, Robert C. "Alternatives to Zoning: Covenants, Nuisance

Rules, and Fines as Land Use Controls," The University of Chicago

Law Review, Vol. 40, Summer 1973, pp. 681-781.

Energy Future: Report of the Energy Project at the Harvard Busine

School. New York: Random House, 1979.

Energy: The Next Twenty Years, A Report sponsored by the Ford Founda­ tion. Cambridge: Ballinger, 1979.

Energy in America's Future: The Choices Before Us, A Study by the staff of the RFF. Baltimore: The Johns Hopkins University Press, 1979.

Freeman, III, A. Myrick. "The Hedonic Price Approach to Measuring De­ mand for Neighborhood Characteristics," in Segal, D. (ed.), The

Economics of Neighborhood, New York: Academic Press, 1979, pp. 191-217.

Gautrin, J.F. "An Evaluation of the Impact of Aircraft Noise on Property

Value with a Simple Method of Urban Land Rent," Land Economics,

February 1975, pp. 80-86.

Gordon, Colin G., William J. Galloway, B. Andrew Kugler, and Daniel L.

Nelson. Highway Noise, NCHRPR 117, National Research Council,

Washington, D.C., 1971.

Grether, D.M., and Mieszkowski, Peter. "Determinants of Real Estate

Values," Journal of Urban Economics, 1974, pp. 127-146.

Grether, D.M., and Mieszkowski, Peter. "The Effects of Nonresidential

Land Uses on the Prices of Adjacent Housing: Some Estimates of

Proximity Effects," Working Paper No. 163, California Institute of

Technology, Social Sciences, 1977.

Griliches, Zvi. "Hedonic Price Indexes for Automobile: An Econometric

Analysis of Quality Changes," The Price Statistics of the Federal

Government, Princeton, New Jersey: National Bureau of Economic

Research, 1961.

Griliches, Zvi. "Hedonic Price Indexes Revisited: Some Notes on the

State of the Art," Proceedings of the American Statistical Associa­

Griliches, Zvi. "Hedonic Price Indexes Revisited," in Griliches, 2vi

(ed.), Price Indexes and Quality Change, Cambridge: Harvard

University Press, 1971.

Grossman, Howard J. "Apartments in Community Planning," Urban Land,

1966, pp. 3-6.

130

Haig, Robert. "Toward an Understanding of the Metropolis: The Assign­ ment of Activities to Areas in Urban Regions," Quarterly Journal of

Economics, May 1926, pp. 402-434.

Harris, R.N.S., Tolley, G.S., and Harrell, C. "The Residential Site

Choice," Review of Economics and Statistics, May 1968, pp. 241-247.

Hoch, Irving. "The Three-Dimensional City: Contained Urban Space," in

Perloff, H.S. (ed.), The Quality of the Urban Environment, Balti­ more: The Johns Hopkins Press, 1969, pp. 75-135.

Jacobs, Jane. The Death and Life of Great American Cities. New York:

Random House, 1961.

Johnson, J. Econometric Methods, 2nd edition. New York: MacGraw-Hill

Book Company, 1972.

Kain, John F., an John M. Quigley. "Measuring the Value of Housing

Quality," Journal of the American Statistical Association, Vol. 65,

No. 330, June 1970, pp. 532-548.

Kain, John F., and John M. Quigley. Housing Markets and Racial Discrimi­ nation. New York: National Bureau of Economic Research, 1975.

Kendall, M.G. A Course in Multivariate Analysis. New York: Hafner

Publishing Co., 1968.

King, A. Thomas. Property Taxes, Amenities, and Residential Land Values.

Cambridge, Mass.: Ballinger Publishing Co., 1973.

King, A. Thomas, and Mieszkowski, P. "Racial Discrimination, Segrega­ tion, and the Price of Housing," Journal of Political Economy,

May-June 1973, pp. 590-606.

Kish, L., and J. Lansing. "Response Errors in Estimating the Value of

Homes, Am. Stat. Assoc. Journal, Sept. 1954.

Lafferty, R.N. and Freeh III, H.E. "Community Environment and the Market

Value of Single-Family Homes: The Effect of the Dispersion of Land

Uses," Journal of Law and Economics, October 1978, pp. 381-394.

Lancaster, K.J. "A New Approach to Consumer Theory," Journal of Politi­ cal Economy, April 1966, pp. 132-157.

Lapham, Victoria. "Do Blacks Pay More for Housing," Journal of Political

Economy, Nov.-Dec. 1971, pp. 1244-1257.

131

Laurenti, Luigi. Property Values and Races. Berkeley, California:

University of California Press, 1960.

Leary, Robert'M. "Zoning," in Goodman, W.I. (ed.). Principles and

Practice of Urban Planning. Washington, D.C.: The International '

City Management Association, 1968.

Li, M.M., and Brown, H.J. "Micro-Neighborhood Externalities and Hedonic

Housing Prices," Land Economics, May 1980, pp. 125-141.

Lin, Steven A.Y. (ed.). Theory and Measurement of Economic Externali­

Manvel, Allen. Local Land and Building Regulation. Washington, D.C.:

The National Commission on Urban Problems, Research Report No. 6,

1968.

Marshall, Alfred. Principles of Economics. London: MacMillan and

Co., 8th edition, 1930.

Maser, Steven M., William H. Riker, and Richard M. Rosett. "The Effects of Zoning and Externalities on the Price of Land: An Empirical

Analysis of Monroe County, New York," Journal of Law and Economics,

April 1977, pp. 111-132.

Mills, Edwin S. "Economic Analysis of Urban Land-Use Controls," in

Mieszkowski, P. and Straszheim, M. (eds.). Current Issues in Urban

Economics, Baltimore: The Johns Hopkins University Press, 1979, pp. 511-541.

Mishan, E.J. The Economic Growth Debate. London: Allen and Unwin,

Ltd., 1977.

Muller, Thomas. Economic Impacts of Land Development: Employment,

Housing, and Property Values. Washington, D.C.: The Urban Insti­ tute, 1976.

Muth, Richard F. Cities and Housing. Chicago: University of Chicago

Press, 1969.

National Commission on Urban Problems (NCUP). Building the American

City. Washington, D.C.: U.S. Government Printing Office, 1968.

Nourse, Hugh O. "The Effect of Public Housing on Property Values in St.

Louis," Land Economics, Nov. 1963, pp. 433-441.

Oates, Wallace E. "The Effect of Property Taxes and Property Values:

An Empirical Study of Tax Capitalization and the Tiebout Hypothe­ sis," Journal of Political Economy, Nov.-Dec. 1969, pp. 957-971.

132

Office of the Pima County School Superintendent. Reading Achievement

Tests, Jan. 1971, Memorandum from the Pima County School Superin­ tendent, July 20, 1979.

Parzen, Emanuel. Modern Probability Theory and Its Application. New

York: John Wiley and Sons, 1960.

Patterson, T.W. Land Use Planning Techniques of Implementation. New

York: Van Nostrand Reinhold Co., 1979.

Peterson, George E. "The Influence of Zoning Regulations on Land and

Housing Prices," Land Use Center Working Paper No. 1207-24.

Washington, D.C.: The Urban Institute, 1974. edition, 1920.

Pima County, Arizona, Board of Supervisors. Annual Report 1969-1970 and Adopted Budget 1970-1971.

Pima County, Arizona. Pima County Air Pollution Emissions Inventory,

AQ-129, April 1980.

Polinsky, A.M. and Shavell, S. "The Air Pollution and Property Value

Debate," Review of Economics and Statistics, February 1975, pp. 100-104.

Procos, Dimitri. Mixed Land Use: From Revival to Innovation. Stroudsburg, Pennsylvania: Dowden, Hutchinson and Ross Inc., Community

Development Series, Vol. 25, 1976.

Ridker, Ronald G. and Henning, John A. "The Determinants of Residential

Property Values with Special Reference to Air Pollution," Review of

Economics and Statistics, May 1967, pp. 246-257.

Romanos, Michael C. Residential Spatial Structure. Lexington, Mass.:

Lexington Books, 1976.

Rueter, Frederick H. "Externalities in Urban Property Markets: An

Empirical Test of the Zoning Ordinance of Pittsburgh," Journal of

Law and Economics, Vol. 16, October 1973, pp. 313-349.

Schafer, Robert. The Suburbanization of Multifamily Housing. Lexington,

Mass.: Lexington Books, 1974.

Schnare, Ann B. "Racial and Ethnic Price Differentials in an Urban

Housing Market," Urban Studies, June 1976, pp. 107-120.

Siegan, Bernard H. "Non-Zoning in Houston," Journal of Law and Economics,

Vol. 13, April 1970, pp. 71-148.

133

Silberberg, Eugene. The Structure of Economics: A Mathematical Anal­

Small, Kenneth A. "Air Pollution and Property Values: Further Com­ ments," Review of Economics and Statistics, Feb. 1975, pp. 105-107.

Smith, V. Kerry. "Residential Location and Environmental Amenities: A

Review of the Evidence," Regional Studies, No. 1, 1977, pp. 47-61.

Stein, Clarence S. Towards New Towns for America. Cambridge, Mass.:

The Massachusetts Institute of Technology Press, 1966.

Stone, R. Quality and Price Indexes in National Accounts. Paris: The

Organization for European Economic Cooperation, 1956.

Straszheim, M.R. "Housing Market Discrimination and Black Housing Con­ sumption," Quarterly Journal of Economics, Feb. 1974, pp. 19-43.

Stull, William J. "Community Environment, Zoning, and the Market Value of Single-Family Homes," Journal of Law and Economics, October 1975, pp. 535-557.

Tideman, T.N. Three Approaches to Improving Urban Land Use. Unpublished

Ph.D. Dissertation, The University of Chicago, 1969.

Tiebout, Charles. "A Pure Theory of Local Expenditures," Journal of

Political Economy, October 1956, pp. 416-424.

Tucson Area Transportation Planning Agency. Employment Survey by Traffic

Analysis Zones, Arizona State Government, Department of Transporta­ tion, 1973.

U.S. Department of Commerce, Bureau of the Census. United States 1970

Census of Population and Housing, Census Tracts, Tucson, Arizona,

SMSA, No. PHC(1)-218, Washington, D.C.: Government Printing Office,

January 1972.

Wheaton, William C. "Monocentric Models of Urban Land Use: Contribu­ tions and Criticisms," in Mieszkowski, P. and Straszheim, M. (eds.),

Current Issues in Urban Economics, Baltimore: The Johns Hopkins

University Press, 1979, pp. 107-129.

Williams, Norman Jr. The Structure of Urban Zoning. New York: Buttenheim Publishing Corporation, 1966.

Wilson, James Q. (ed.). The Metropolitan Enigma. Cambridge, Mass.:

Harvard University Press, 1968.

Yinger, John. "Prejudice and Discrimination in the Urban Housing Mar­ ket," in Mieszkowski, P. and Straszheim, M. (eds.). Current Issues in Urban Economics, Baltimore: The Johns Hopkins University Press,

1979, pp. 430-468.

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