w, ~_-- . — ________———. ....~... . 7‘ . _...... ' .. . ....' 7..“ .. -'n‘< ...{ “4.... u . p. , u. . v . v w a ‘ 7 v ‘ v > ‘. ‘ x ‘ I. . . ‘ .l 1 . ~. . . . , . . . , .2 5309/0? IIIIII III IIIIIIII II IIII II III III} 00627 4801 3Ll2IIKAK r F Michigan State University This is to certify that the dissertation entitled TAXES, PROPERTY TAX ABATEMENT, EXPENDITURE, AND THE COMPOSITION OF THE PROPERTY BASE IN COMMUNITIES WITHIN A METROPOLITAN AREA presented by Robert William Wassmer has been accepted towards fulfillment of the requirements for Ph . D . degree in Economics A, Major professor % Ronald C. Fisher Date Q/S 5 / ( unn...-ur .- .- r1 .n” . , I . 0-1 1 J 1-. -‘ fie _.-. (‘55,; <~_.- *-—-..~ «1‘. AA'Av—PA—u‘rV—F'v—vv—d :— -4-».—4' ,— -A,A ,4_.—~*5 . PLACE IN RETURN BOX to remove this checkout from your record. TO AVOID FINES return on or betore dete due. DATE DUE DATE DUE DATE DUE IL—I o I_________.I_______I_______ II_——_II——— I MSU Is An Affirmative Action/Equal Opportunity institution TAXES, PROPERTY TAX ABATEMENT, EXPENDITURE, AND THE COMPOSITION OF THE PROPERTY BASE IN COMMUNITIES WITHIN A METROPOLITAN AREA ‘ BY Robert William Wassmer A DISSERTATION Submitted to Michigan State University in partial fulfillment of the requirements for the degree of DOCTOR OF PHILOSOPHY Department of Economics 1989 qu'f; ABSTRACT TAXES, PROPERTY TAX ABATEMENT, EXPENDITURE, AND THE COMPOSITION OF THE PROPERTY BASE IN COMMUNITIES WITHIN A METROPOLITAN AREA BY Robert William Wassmer The simultaneous relationships between home value, local income tax rate, property tax rate, manufacturing property base, commercial property base, housing property base, local expenditure per capita, and manufacturing and commercial property abatements in a community within a metropolitan area is examined through theoretical and empirical analysis. Utility maximization yields equations representing median voter's demand for components of the local property base and local expenditure. Profit maximization yields equations representing the nonresidential property abatements necessarily offered by the median voter. The value of local homes, local employment, and local property and income tax rates are endogenously determined. The theoretical model yields ten simultaneous equations that are estimated from a panel data set taken from the Detroit Metropolitan Area for the period 1977 - 1987. Robert William Wassmer The regression analysis is unique because it accounts for the full endogeniety of local variables, the long-run adjustment toward desired property and necessary abatement values, and uses the first collection of community-specific abatement data. Regression results are used to conduct simulations that address six questions. (1) How do local fiscal variables, including property tax abatements, affect local property tax bases? (2) How do local median voter characteristics affect local property tax bases? (3) Why do communities offer property tax abatements? (4) What type of communities offer property tax abatements? (5) Besides the size of the local property base, what are the other effects of local property tax abatements? (6) Do property tax abatements work? Some important findings are that communities offer property tax abatements to decrease the effect of noncapitalized profit-reducing characteristics (high taxes and crime). The presence of a local property tax abatement program and local income taxes mitigate the negative impact that local property tax rates have been shown to have on the value of local property bases. Residents with greater income, age, and education demand smaller local property bases. An increase in nonresidential property abatements increases local property bases and net-property- tax revenue collected. Copyright by Robert William Wassmer 1989 This dissertation is dedicated to Sandra J. Wassmer, Robert O. Wassmer, and Dana I. Wu. It would not have been possible without my parents' and fiancee's unlimited support and encouragement. ACKNOWLEDGMENTS In preparing this dissertation I have received support and guidance from many people. I would like to specifically thank the following people. The dissertation's advising committee consisted of Professors Ronald Fisher, Charles Ballard, John Wo1fe, and Jeff Biddle. Professor Jeff Biddle provided helpful comments at the dissertation's defense. Professors John Wolfe and Charles Ballard provided encouragement and input from initial idea to final draft. Ronald Fisher supervised the dissertation committee and provided invaluable advice on the dissertation and encouragement throughout my education at Michigan State University. Through Mark Murray and William Greeley, of the Michigan Department of Commerce, I gained my first employment experience with state government. Professor John Anderson, then Michigan Deputy Treasurer of Taxation and Economic Policy, provided my second employment experience with state government. His research suggestion was the genesis of this dissertation. Howard Hiedeman and Howard Bunch, of the Michigan Department of Treasury, assisted greatly in my gathering the data necessary for this dissertation. I am indebted to Eastern Michigan University for providing employment while I completed this dissertation. At Eastern Michigan University I am particularly indebted to Professors John Anderson and Raouf Hanna for economic and statistical advice, and to the chairperson of the economics department, Professor Young-lob Chung, for providing summer office and computer use. Finally, Pat Redmon provided valuable advice on economic questions. Dana Wu provided valuable editorial suggestions. vi TABLE OF CONTENTS page LIST OF TABLESOOOOOOO0....O000......00.0.00000000000000001x LIST OF FIGURESOOOOOOOOCO0.000000000000000.00000000000000Xi INTRODUflIONOOOOCOOOOOOOO0.0I.OOOOOOOOOOOOOOOOOOOOOOOOOOO.1 1. LITEMWRE REVIEWOCCOOOOOOOOOOOCOOOOOOOOCOOOO00......0.4 A. FIRM INFLUENCE ON LOCAL NONRESIDENTIAL PROPERTY BASES B. COMMUNITY INFLUENCE ON LOCAL NONRESIDENTIAL PROPERTY BASES C. PROPERTY TAX ABATEMENTS AND NONRESIDENTIAL PROPERTY BASES D. IMPROVING THE ASSESSMENT OF THE INFLUENCE OF LOCAL FISCAL POLICY ON LOCAL PROPERTY BASES THEORETICAL MODELOOOOOOOOOOOOOOOOOOOOOOOOOO00.00.000.039 A. OVERVIEW B. LOCAL TAX RATES, MEDIAN VOTER'S HOME VALUE, AND MEDIAN VOTER'S DEMAND FOR LOCAL PROPERTY BASE AND EXPENDITURE C. MANUFACTURING PROPERTY ABATEMENTS NECESSARILY OFFERED BY MEDIAN VOTER D. COMMERCIAL PROPERTY ABATEMENTS NECESSARILY OFFERED BY MEDIAN VOTER 3. EMPIRICALANALYSISOOOOOOOOOOOOOOOOOOOOOOOOOOOO0.0.0.79 A. PRELIMINARIES B. DATA AND VARIABLES vii 3. APPENDIX A - APPENDIX B APPENDIX C APPENDIX D - TABLE OF CONTENTS (cont.) page EMPIRICAL ANALYSIS (cont.)............................79 C. REGRESSION RESULTS D. GRAPHICAL ANALYSES E. SIMULATIONS SUMMARY AND CONCLUSIONS..............................156 A. SUMMARY B. CONCLUSIONS VARIABLEVAI‘UESOOOOOOOOOOOO0.0.00.00000000170 EXCLUDED CITIESCOOOOOOOO00.0.00...00......185 VARIABLE DERIVATIONS......................186 VARIABLE SOURCESOOOOOOOOOOO0.0.00.00000000191 BIBLIOGRAPHYOOOOOOOOCOOO0....O0....0.0.0.000000000000000194 GENERAL REFERENCESOOOOOOOOOOOOOOOOOIOOOOO0.0...0.00.00.01.99 viii 4. 5. 6. 7. 10. 11. 12. 13. 14. 15. 16. LIST OF TABLES page Variable Definitions.................................95 Descriptive Statistics...............................97 Median Home Value Regression........................102 Effective Local Income Tax Regression...............107 Local Property Tax Rate Regression..................110 Percentage Of Residents Employed Locally Regression.1l4 Median Voter's Demand For Manufacturing Property RegreSSionOOOOOOO0.00...O0.00000000000000000000000.0116 Median Voter's Demand For Commercial Property RegrBSSion.OOOOOOOOOOOOO...00....0.0.0.0000000000000121 Median Voter's Demand For Housing Property RegreSSionO0.000000000000000000000.00.00.00.00000000124 Median Voter's Demand For Local Expenditure Per capita RegreSSionOOOOO0.00...0.0.00.000000000000130 Manufacturing Property Abatements Necessarily Offered By Median Voter Regression..................133 Commercial Property Abatements Necessarily Offered By Median Voter Regression..................137 Fiscal Simulation Results -- Full System............151 Fiscal Simulation Results -- No "l", "A", or "B"....153 Median Voter's Char. Simulation Results -- Full systemOOOOOOOOOOOOOOOOOOOOOO0.0.000000000000000154 Property Abatement Simulation Results -- Full system.OOOOOOOOOOOOOOOOOO000......0.0.00.000000155 ix LIST OF TABLES (cont.) page 17. Property Abatement Simulation Results -- up", ”1”, and ”R” constant.0.00.0.0.0.00000000000000155 18. Results of 10% Increase In Manufacturing Abatements.167 19. Results of 10% Increase in Commercial Abatements....168 2. 3. 4. 5. LIST OF FIGURES page Median Voter's Utility Maximizing Choice -- #1.......58 Median Voter's Utility Maximizing Choice -- #2.......58 Manufacturing Equilibrium -- #1.....................143 Manufacturing Equilibrium -- #2.....................146 Manufacturing Equilibrium -- #3.. ........ ...... ..... 147 xi INTRODUCTION This dissertation provides a theoretical and empirical analysis of the relationships between the local income tax, property tax, home value, property tax abatement, expenditure, and property tax base in a community within a metropolitan area. These analyses are to used to address six important questions. (1) How do local fiscal variables, including property tax abatements, affect local property tax bases? (2) How do local median voter characteristics affect local property tax bases? (3) Why do communities offer property tax abatements? (4) What type of communities offer property tax abatements? (5) Besides the size of the local property base, what are the other effects of local property tax abatements? (6) Do property tax abatements work? Because local policymakers have always been concerned with the relationship between local fiscal variables and the local property tax base, there is considerable research completed on the first question. The majority of this research concluded that local fiscal variables exerted little or no measurable impact on the value of local nonresidential property bases. This conclusion was drawn from theory that modeled intrametropolitan firm location as solely the outcome of profit-maximizing firms demanding 2 community sites. Researchers in the mid-1970's broadened their thinking to account for the influence that host communities have on intrametropolitan firm location. Recent empirical studies have reported that local fiscal variables exert a measurable impact on intrametropolitan firm location. Despite this, recent empirical studies have failed to agree on the size of the fiscal impact. This is attributable to a failure to apply theoretical and empirical advancements related to this topic. This dissertation is an attempt to correct this failure. It is unique in a few different ways. The theoretical and empirical analysis presented here takes into account the simultaneous nature of variable determination within a community. Community variables are modeled as jointly determined by the actions of both the community and firms. Particular attention is paid to a community's influence on these variables. The local property tax base is broken down to its commercial, housing, and manufacturing components. Each component is analyzed separately. The desired value of manufacturing property, commercial property, and housing property in a community; and the necessary level of manufacturing and commercial property abatement offered by a community: are modeled as long-run results that a community is expected to be moving toward, but has not necessarily reached. The empirical estimation uses the first collection of community-specific manufacturing and commercial property abatement values. The remainder of the dissertation proceeds as follows. 4A.review of previous literature relating to intrametropolitan 3 firm location and the determination of local nonresidential property bases is in Chapter 1. The review concludes with my own suggestions for improvement. These suggestions are the basis for the theoretical and empirical analyses presented. in Chapters 2 and 3. A summary of the dissertation and major conclusions drawn from it are contained in Chapter 4. CHAPTER 1 LITERATURE REVIEW The literature review is conducted chronologically and starts with a review of literature relating to the influence firms have on the determination of local nonresidential property bases in a metropolitan area. 1. A. FIRM INFLUENCE ON LOCAL NONRESIDENTIAL PROPERTY BASES The accepted approach to modeling the location choice of a firm is to assume that it chooses its production site to maximize expected profits. Assuming an imperfect short- run capitalization of profit differentials into land values, expected profits vary by location because the cost of inputs and the market for output vary by location. Taxes levied by a state or locality increase the cost of locating in that state or locality. State or local expenditure on services the firm uses in production reduce the cost of locating in that state or locality. A profit- maximizing firm's derived demand for a site is therefore greater, Q§£§11§_DQIIDQ§. for jurisdictions with relatively low taxes and/or relatively high expenditures on firm services. This simple extension of the profit-maximizing 4 c} fi (1 St 5 model of firm behavior illustrates how fiscal policy theoretically influences a firm's location choice. Economists have naturally questioned whether this influence should be of any concern to state and local policymakers. Floyd (1952) addressed this question by separating a manufacturing firm's location decision into two separate stages. In the "market stage," a firm chooses to locate in a region of a country, or a region of a state, based on its market characteristics. In the "site stage," a firm chooses a unique site to serve its predetermined market at the highest profit to the firm. Floyd contended that fiscal differences are usually too small to influence a manufacturing firm's market selection. In market selection, non-fiscal characteristics overwhelm the influence state and local taxation and expenditure have on expected profit. Floyd theorized that fiscal policy exerted its strongest influence during site selection. During site selection, non-fiscal characteristics that had overwhelmed the effects of local fiscal policy are largely constant. The Advisory Commission on Intergovernmental Relations (1967, p. 78) reiterated Floyd's reasoning in the following statement: The relative importance of the tax differential factor in industrial location decisions appears to increase as the location decision narrows down to a particular jurisdiction within a general region. ...[AJmong local governments within a state and especially within a metropolitan area, tax differentials exert discernable plant location pull. ...[I]n almost every metropolitan area there exists wide local property tax differentials - a cost consideration that can become a "swing" factor in the final selection of a particular plant location. an fi 6 Economic theory and casual empiricism support the idea that local fiscal policy can influence a firm's site selection. This idea has been tested by asking firm decision makers to rank location determinants. Mueller and Morgan (1962) randomly surveyed 239 new, relocating, or expanding Michigan firms. Firms were questioned on determinants of both their market and site choices. Labor cost had the largest reported effect on a manufacturing firm's market choice. In contrast, personal factors and historical accident were the reason most often given for site choice. Local tax concessions and inducements were ranked as the seventh most important determinant of firm site choice. Mueller and Morgan's study, and other survey research1 provide somewhat mixed and confusing evidence on the effects of fiscal policy on the determination of local nonresidential property bases. This is attributable to the two methodological flaws inherent in survey research. First, variables are ranked on a ordinal scale. There is no way to calculate the precise size of a variable's influence. Second, firms use surveys to lobby for lower taxes. This amplifies the relative importance of local fiscal policy to firm's location choice. An alternative to survey research is statistical analysis of more objectively collected data. Due (1961) published the first review of such analyses relating to state and 1Cornia, Testa, and Stocker (1978) and Wasylenko (1985) reviewed other survey research. t1 ta SL Pe Si 7 local tax influences on the location of industry. Due found no reported correlations between intezetete tax differentials and state employment growth rates. He concluded that tax effects are of no major importance to intezetete manufacturing location. However, Due (p. 171) did state that: ...in'some instances the tax element plays the deciding role in determining the optimum location, since other factors balance. This is most likely to be the case in the selection of a precise site in a metropolitan area. Due based this statement on informal observation. He called for a detailed investigation of the influence taxes have on intteetete firm location. Moses and Williamson (1967) were among the first researchers to relate a measure of intrametropolitan manufacturing firm location to a vector of local cost variables. They divided Chicago and its suburbs into 582 zones and examined 2000 firms that chose to locate in these zones between 1950 and 1959. Their dependent variable was the density of new firms in each zone. Their explanatory variables were proxies for land rent, wage rate, transportation accessibility, land availability, and a dummy variable equal to zero if more than half the zone was within the Chicago city limits. The dummy was said to control for tax and zoning differences between the central city and suburban areas. Only distance from the core and the percentage of a zone used for manufacturing were statistically significant. Vi Sc 19 Cl Re Do. f0] of rat Can 8 Moses and Williamson then divided the zones into north, west, and south samples and repeated the regressions for each sample. In all samples the regression fit improved. In the western sample the property tax and zoning dummy became statistically significant and negative. Because the western zone contained the highest percentage of white residents, Moses and Williamson recommended that future regressions contain explanatory variables reflecting residential composition. It is important to note that this early regression research concluded that local population composition (a variable that influences the host community's decision to supply manufacturing property) was a significant determinant of local firm density. Schmenner (1973) used regression analysis to examine the influence of local taxes on intrametropolitan manufacturing location. His community-specific dependent variables were firm density, change in firm density, and density of firms relocating from the central city. Schmenner's data came from two periods (1967-1969, 1969- 1971) and from four different metropolitan areas (Cincinnati, Cleveland, Kansas City, and Minneapolis-St. Paul). Regressions were run for each metropolitan sample and for a pooled sample. Local fiscal characteristics were accounted for by effective property and income tax rate as a fraction of the central city's effective property and income tax rate, per-pupil school expenditure as a fraction of the central city's per-pupil school expenditure, and the number 9 of years a local income tax had been in effect. Population and industry concentration were considered jointly determined and a two-stage least squares estimation was used. Schmenner found that population density, existing manufacturing firm density, and the presence of transportation facilities exerted the largest statistically-significant influence on all three measures of intrametropolitan firm location. In his dynamic regressions, the coefficient on the income tax variable was negative and statistically significant. In the same dynamic regressions, the coefficient on the property tax variable was never statistically significant. Schmenner believed that his results provided adequate evidence to reject the hypothesis that tax differentials do not influence intrametropolitan location. Others who have reviewed Schmenner's empirical results have not reached the same conclusion.2 Schmenner based his conclusion solely on results from the regression using a dynamic dependent variable. The tax coefficients from Schmenner's level regressions often had the wrong sign and were never statistically significant. Oakland (1978) pointed out that the statistically significant income tax coefficient occurred in only the pooled sample. Oakland attributable this to a dummy effect proxying for each city. After reviewing the evidence, I would agree with Oakland's (p. 23) conclusion that "Schmenner's work provides little evidence 2See Mestleman (1973) and Oakland (1978). 10 that tax considerations are important in intrametropolitan location decisions." Levin (1974) performed an econometric analysis of whether central city ”onerous tax burdens" (local property taxes less the value of local firm services) increased the flight of central city firms to the suburbs. Using ordinary least squares, she regressed the percentage of total manufacturing property found in the central city of 23 Michigan metropolitan areas against a measure of central city onerous tax burden, agglomeration economies, available market area, and production cost differences. Levin found that agglomeration economies and the ratio of central city to suburban income exerted a significant positive influence on the percentage of a metropolitan area's manufacturing activity conducted in its central city. Onerous tax burdens had no significant influence on the intrametropolitan dispersion of firms. Struyk and James (1975) reviewed the econometric studies previously described. They criticized this research for running regressions that were eg_nee and designed to measure long-run location equilibria that are never achieved. Struyk and James believed that an incremental process determines intrametropolitan firm location and research on this topic should be designed accordingly. Past regression models, were at best, reduced form proxies for a badly needed complete model of factors that influence ll intrametropolitan firm location. Struyk and James did not provide this complete model. Erickson and Wasylenko (1980) also conducted a regression study of local taxes and the site selection of firms moving from the Milwaukee central city to its suburbs. They criticized past research for its lack of an explicit theoretical model. Erickson and Wasylenko derived a general specification of firm demand for community sites based on cost minimization for manufacturing firms or profit maximization for commercial firms. They assumed that a community's supply of potential sites was perfectly elastic in the range demanded by firms. Erickson and Wasylenko's dependent variable was the number of firms in one industry that moved to a specific suburban city, divided by the entire number of firms in that industry that moved from Milwaukee to any suburban city. The dependent variable was calculated for a ten-year period beginning in 1964. Erickson and Wasylenko avoided the questionable assumption of long-run metropolitan location equilibrium by using a dynamic dependent variable. A weighted least squares logistic technique was used to regress their dependent variable against mid-period proxies for land price, wage rate, effective property tax rate, community provided firm services, and agglomeration economies. Erickson and Wasylenko included local measures of population density and per-capita income to proxy for market effects on commercial location. 12 In all of Erickson and Wasylenko's regressions, industry employment concentration and the available industry work force within a seven-mile radius exerted a significant positive influence on relocation. For the construction and wholesale trade industries, distance from the central city exerted a significant negative influence. For manufacturing, the percentage of land devoted to manufacturing, and the percentage of land vacant, exerted a significant negative and positive influence. The percentage of land vacant had a significant negative influence on the location of wholesale trade. With the exception of their finding that police and fire expenditure exerted a small negative influence on retail and wholesale firm location3, Erickson and Wasylenko found that local fiscal variables did not influence relocating firm's suburban site selection. The general conclusions drawn from this early empirical research is that non-fiscal characteristics such as local agglomeration economies, labor availability, and land prices fie influence intrametropolitan firm location. Local taxation and expenditure policies ge_net influence intrametropolitan firm location. 3This is possibly representative of higher crime and fire rates. 13 1. B. COMMUNITY INFLUENCE ON LOCAL NONRESIDENTIAL PROPERTY BASES Oakland (1978) reviewed some of the same literature described in Section A. He suggested that future research avoid replicating previous approaches that ignored the influence of host communities on the intrametropolitan location of firms. Communities influence firm location through zoning practices, by offering services to firms, and property tax reduction. Obvious anecdotal evidence to support this claim is the existence of ”bedroom communities" and "industrial enclaves" within the same metropolitan region. Oakland (p. 23) suggested: Before undertaking further tests, it seems imperative to make further'progress in model specification. ...[Ojne major advance over present models would be to incorporate the fact that industrial location is jointly determined by the behavior of firms and host communities. Concurrent work by White (1975) and Fischel (1974 and 1975) produced similar theoretical models of the influence communities have on intrametropolitan firm location. White and Fischel's models were both attempts to address problems with Tiebout's (1956) seminal model of local public service ' provision and residential location choice. In Tiebout's model, consumers reveal their preference for congestible local services by settling in a community whose population and size allow the community to provide the desired level of public services most efficiently. Under Tiebout's assumptions, the outcome of this "voting 14 with the feet" maybe a Pareto-efficient provision of local public goods. Tiebout implicitly assumed that benefit charges finance local public services. In reality, local property taxes largely finance local public services. Local property taxation cannot be considered a benefit charge for local public services. Considering this violation, the stability and efficiency of Tiebout's model breaks down. By locating a house in a community whose average housing value is greater than desired, a consumer gains a lower tax price per unit of local public service consumed. Hamilton (1975) addressed this problem by adding to Tiebout's model the assumptions that communities raise their revenue through property taxes and use "neutral fiscal zoning." Neutral fiscal zoning requires that each community set standards for residential entry such that property tax revenue from a new resident's home covers the cost of providing additional local service to it. With neutral fiscal zoning, the local property tax is transformed into a benefits pricing system. Both Tiebout and Hamilton assumed that communities are composed only of residents. White (1975) and Fischel (1974 and 1975) extended the Tiebout- Hamilton assumptions to a system in which both residents and firms locate in communities. Because they are similar, a summary of only White's theoretical model is provided. White demonstrated that if communities use a form of fiscal zoning, the introduction of firms into the Tiebout- Hamilton model not destroy its stability or efficiency. 9‘. PI Of eq co- fi: mix Se; rat Sela sit, 15 White developed a model of a community's willingness to supply firm sites. Communities are considered to have equal populations and are in a metropolitan region that sits on a plain without a central city. A firm desiring a site in this region chooses its location based on differences in local tax rates. In White's model, residential and industrial property are taxed at the same rate. Communities offer an equal level of locally-provided services to firms in any one industry. Industries desiring intrametropolitan location vary, but firms in an industry are homogenous. Firms operate in "footloose" competitive industries. Residents are the first to locate in a community. With Tiebout-Hamilton equilibrium, residents demand a uniform level of housing and local public service. A community's property tax rate equals the uniform value of residential public service provided per household divided by the zoned uniform value of housing. Firm entry is allowed only after residential equilibrium is established. Without firm externalities, a community using neutral fiscal zoning requires an entering firm to use a minimum amount of taxable property. The minimum property requirement is equal to the value of local services provided to the firm divided by the property tax rate. Firms produce output by employing land, local public services, and capital. Firms choose an intrametropolitan site based on the profit maximization of an industry-constant Cc he Si CO re< BeC CQm 16 _ Cobb-Douglas production function. A firm locates in a community whose residentially-determined tax rate is as close as possible to a calculated ratio of the production function's public service exponent. All firms in an industry subsequently seek a community with the same tax rate. Firms in an industry are expected to cluster in one city. White accounted for negative firm externalities by "pollution zoning." Pollution zoning consists of a community's zoning board calculating a social welfare function that gauges the median voter's willingness to trade a loss in environmental quality for a gain in "pollution- compensating transfers." Pollution-compensating transfers are property tax payments above the amount necessary to cover locally provided public services. By increasing the pollution compensating transfer a community receives per unit of environmental loss, White traced out a community's ”Offer” curve. White's Offer curve represented an increase in local firm sites (pollution level) as pollution- compensating transfers increase. The intersection of a community's positively sloping site supply curve, with the negatively sloping firms' demand curve for a community's sites, yields the combination of firms and property taxes a community seeks. Communities use this combination to set a schedule of required property use levels for firms in each industry. Because White only allowed firms to pay pollution- compensating transfers in the form of property taxes, required Fi VG th Th a f or all 4Ir 383 the 17 property consumption levels are the only way a community receives its desired pollution-compensating transfer.4 Required firm property consumption levels are equal to the sum of local service provided to the firm plus the industry- specific pollution compensation transfer, divided by the property tax rate. A community offers to supply sites to all firms that meet their minimum property consumption levels. If demand for environmental quality is income elastic, wealthy residential communities require a larger compensating transfer per unit of pollution. Wealthy residential communities offer a smaller and more inelastic supply of sites to environmentally damaging firms, than do poor residential communities. In White's system of pollution zoning, wealthier communities set a higher minimum firm property requirement. This level may be high enough to effectively zone out some or all firms. Kiefer (1974) made suggestions on how to improve Fischel's (1974) theoretical model that applied equally well to White's (1975) theoretical model.’ Kiefer believed that the theories of Fischel and White were too restrictive. They did not allow residents to change their location choice after firms had entered. To be consistent with Tiebout's original intentions, a dynamic theoretical approach that allows such movement is absolutely necessary. 4In reality there are other means of payment, including subdivision regulations, subdivision extractions, altering assessment values, user fees, etc. For a description of these, see Fischel (1985) pp. 23-25. a1 ir re av re ev. th4 To com lev ShOl 18 The theoretical models of White and Fischel revealed that, in some instances, communities rationally restrict their supply of sites to firms. Empirical research that followed these theoretical models has only begun to fully incorporated this implication. Fox (1981) began with the premise that an empirical analysis of firm response to local fiscal policy should not include localities that zone out industry. Previous researchers had attempted to estimate the demand for an average metropolitan community's firm sites. These regressions used a data set that contained observations on every community within a metropolitan area. Supply-side theory suggests that a community's property tax may encourage a larger than optimal number of firms to enter the community. To prevent this, communities zone out some firms and a community's observed property tax rate and corresponding level of firm activity represents supply conditions. It should not be used to estimate a demand equation. Fox applied his reasoning to a regression analysis involving 43 cities in the Cleveland Metropolitan Area. Twenty cities that had less than one percent of their property tax base devoted to manufacturing were considered to be effectively zoning it out. Using the remaining 23 cities, Fox regressed the percentage of a community's property tax base devoted to manufacturing against local property tax rate, land price, locally-provided firm services, highway and rail dummies, manufacturing capital-to-land ratio, and 19 population density.. He recognized that tax rate, land price, local business services, and capital to land ratio are endogenous to his model and appropriately used a two- stage estimation technique. He correctly pointed out that his interpretation of the regression results are only accurate if the communities are in long-run location equilibrium. For the sample of 23 communities that were believed to not zone out manufacturing, property tax rates and business service levels exerted a statistically-significant influence on manufacturing activity. The calculated elasticities were -4.43 and 2.78. Fox performed a similar regression that included the 20 communities thought to zone out manufacturing. In this regression, the tax variables were insignificant. If actions of the host community are taken into consideration, local fiscal variables can be shown to exert a statistically-significant and relatively large influence on intrametropolitan firm location. Following Fox's initiative, Wasylenko (1980) re- estimated the regression test originally performed in Erickson and Wasylenko (1980). Wasylenko revised his data set to exclude communities thought to zone out industry. His new results bolstered the findings of Fox. The non-fiscal explanatory variables that were previously statistically significant remained so. But whereas Erickson and Wasylenko found only one questionable relationship between community expenditure level and employment growth, Wasylenko also found a statistically-significant negative relationship 20 between local property tax rates and manufacturing and wholesale employment growth. Wasylenko believed his results changed because he accounted for a community's willingness to supply firm sites. Charney's (1983) regression study modelled the supply of community sites in a slightly different manner. Charney assumed that a profit-maximizing manufacturing firm makes an intrametropolitan location choice by calculating its expected profit level in each community. It then chooses the community that offers the highest expected profit. In Charney's model, all firms have the same production function and consequently wish to locate in the same community. This net-of-tax price of manufacturing land is bid up in some communities and bid down in others. This process continues until a manufacturing firm's expected profit levels are equal across all metropolitan communities. In Charney's model, the net-of-tax price of manufacturing land is a function of variables that influence demand for a community's manufacturing sites. Charney assumed that a community's supply of manufacturing sites is a function of the bid price by manufacturers, zoning restrictions, and determinants of other possible users' bids for land use. Charney calculated a supply price function by inverting her supply function. A community's manufacturing site market is considered in equilibrium when demand price equals supply price. Charney obtained the equilibrium level of community manufacturing ,21 sites by setting the two price functions equal and solving for the reduced form. In Charney's model, no equilibrium exists if other users are willing to pay more than manufacturers, or if communities zone out manufacturing firms. Consequently, Charney excluded from her regression analysis communities without manufacturing firms. There are two major problems with Charney's methodology. First, it assumed that community differences in firm profit levels are immediately and fully capitalized into land prices. Second, it did not explicitly model community zoning choices. Charney estimated her reduced form function. Her sample came from 110 out of a possible 126 zip code areas in the Metropolitan Detroit Area. The number of manufacturing firms that moved into a locality from 1970 to 1975 divided by the locality's land area was regressed against a vector of explanatory variables taken from 1970. The explanatory fiscal variables were the local property tax rate, local income tax rate, and local sanitation provision. The local property tax rate exerted a significant negative influence on location, while the local income tax rate and sanitation provision exerted no significant influence. The elasticity of firm location in regard to local property tax rates was -2.52. Dividing her sample between firms with a small, medium, and large number of employees, Charney found significant location elasticities with respect to local property tax rates of -.29, -1.77, and -2.22. These rising elasticities 22 were attributed to a complementarity between labor and capital use by firms. The greater amount of capital a firm employs, the greater its sensitivity to intrametropolitan property tax rate differentials. McGuire (1985) regressed the 1976 to 1979 change in local building permit values in the Minneapolis-St. Paul Metropolitan Area against proxies for differences in firm input costs. Like previous researchers, she excluded communities without firms. McGuire regressed her dependent variable against the following explanatory variables: (1) city property tax rate, (2) three dummies which control for city location in one of four concentric rings around the central city, (3) percentage of metropolitan labor in the city's county, and (4) city's per-capita income. Only the first three explanatory variables were statistically significant. The reported elasticity of local building permit values in respect to local property taxes was -2.06. McGuire tested the established practice of controlling for exclusionary zoning practices by excluding communities with no firm development. McGuire rightfully criticized this practice. Some communities may practice exclusionary zoning and still admit a few firms. If this occurs, observed community tax rates and the corresponding number of firms reflect local zoning practices and not intrametropolitan tax rate differentials. Calculating her original regression for 1981, she retrieved the residuals and used them as an 23 explanatory variable in a similar regression using 1976 values. The 1981 residuals were expected to capture community-specific zoning practices that could also explain location activity in 1976. McGuire found that the 1981 residual variable was statistically significant in the 1976 regression. This result demonstrated that community-specific characteristics, such as zoning practices, were being excluded from demand-type regressions. The regression research examined in Section A of this literature review included attempts by empirical researchers to estimate a structural equation of firm demand for an average metropolitan community's location sites. Section B of this literature review began by showing that theoretical researchers in the mid-1970's recognized that communities rationally limit their supply of firm sites. More recent research attempted to incorporate this finding into estimates of demand-side equations by excluding from their regression samples communities without firm activity. The problem with this simple methodology is that it limits a community's willingness to supply firm sites to one of two extremes. At the existing local property tax rate, a community allows a perfectly elastic supply of firm sites, or it zones out all firms. Charney loosened this extreme assumption by estimating a reduced-form equation for intrametropolitan firm location that incorporated both a structural demand and supply equation. Charney's methodology did not allow for an explicit measurement of supply-side influences. 24 McHone's (1986) research was the first attempt to econometrically estimate separate structural supply and demand equations. McHone's theoretical model of a community's willingness to supply firm sites was similar to White's and Fischel's. McHone's theoretical model of demand for a community's manufacturing sites was based on profit-maximizing firms choosing intrametropolitan locations to minimize costs. McHone's data set consisted of a 1970 cross-section of suburban Philadelphia communities. Manufacturing employment per capita was regressed against a vector of explanatory demand variables and a vector of explanatory supply variables. Because of the endogeniety of property tax rates, McHone used a two-stage estimation technique. McHone's regression results showed that transportation availability, distance to central city, and police/fire expenditures exerted a statistically significant positive influence on manufacturing firms' demand for community sites. Property taxes and library/park expenditures exerted a significant negative influence on demand. McHone also found that property taxes and total public expenditure exerted a significant positive influence on a community's willingness to supply firm sites. Median family income, residential tax base, and population density exerted a significant negative influence on site supply. McHone calculated the property tax elasticity of a community's supply of manufacturing sites and the manufacturing firms' f1 25 demand for community sites to be .55 and -.79. This property tax elasticity of demand was inelastic and significantly less than similar elastic measures calculated by Fox (- 4.43) and by McGuire (-2.06). McHone's results provided strong empirical evidence to support the contention that property tax differentials exert a statistically significant, but inelastic influence on both the demand and supply of intrametropolitan firm sites. White (1986) conducted an empirical study that utilized supply-side theory. White believed that California's 1978 passage of Proposition 13 provided an excellent laboratory in which to test the influence local property taxes have on firm location. Proposition 13 forcefully reduced California's local property tax rates by varying amounts. White posited that the passage of Proposition 13 had no affect on other community characteristics. White presented three models of intrastate firm location. White's models varied by the degree of influence localities were theorized to have on firm location. In all her models, White assumed that local property tax differentials were fully capitalized into local land values. She maintained that local property tax rates still exerted locational effects. White reasoned that a community's capitalized land price may rise high enough that it is no longer attractive to firms. Alternatively, a community's capitalized land price may fall to zero and still not compensate a firm for its high property taxes. Capitalization of high property 26 tax rates into land prices also result in firms substituting other production inputs for land. If high tax rates signal even higher taxes in the future, firms owning their own land may also wish to avoid a fully capitalized high tax jurisdiction. Any one of these occurrences would reduce firm demand for sites in a community with fully-capitalized high local property tax rates. Using a theoretical model similar to what she presented in 1976, White predicted that, following Proposition 13 California, cities containing polluting firms experienced a negative windfall. Such cities no longer received the higher property tax revenue that compensated them for their decreased environmental quality. White predicted that these cities would place pressure on polluting firms to clean up their production processes. Pressure came in the form of new building codes, new user fees, zoning variances, etc. White believed that this caused polluting firms to move out of previously-high-tax jurisdictions and into previously- low-tax jurisdictions. White tested her model's predictions using California county level data. She excluded counties without firms and divided her sample into manufacturing and commercial S.I.C. codes. White's regressed the 1977 to 1981 change in firms, employment, and payroll against the post-Proposition 13 change in property tax rates, a United States' measure of the dependent variable, rate of county labor increase before Proposition 13, and current property tax revenue as a f1 11' fi a1 Wa foj rel CQm‘ 27 percentage of total revenue. In all three commercial regressions, the property tax coefficient was statistically significant and negative. It was never statistically significant in the manufacturing regressions. The calculated elasticity for change in commercial firms, employees, and payroll with respect to change in local property taxes were -6.14, -9.21, and -15.05. Megdal (1986) offered an alternative explanation for White's surprising finding that commercial firms were more footloose and tax sensitive than manufacturing firms. Megdal reasoned that commercial firms were only following changes in residential location choices caused by Proposition 13. Megdal called for further empirical tests that would be derived from a theoretical model that accounted for both firm and residential location choices. Megdal also criticized White's work for using county level data that did not allow for intracounty city variation. Megdal pointed out that, following Proposition 13, most California communities instituted or increased user fees for business. These fees likely eradicated most incentives for communities to push firms out, and that this needed to be accounted for. Megdal also believed that White's five-year period of observation was too short to allow for full long-run location adjustment following the institution of Proposition 13. Ladd (1975 and 1976) performed a piece of indirectly related research. Ladd tested the hypothesis that the composition of a community's property tax base affects its IE 00 to 0&1 I'm ACc . 5 5 man: Pro; 28 local expenditure decision. Ladd believed that the non- residential size of the tax base influences local demand for municipal services in three ways: (1) reducing the perceived tax price of local services to residents, (2) increasing local expenditures on services provided to firms, and (3) increasing expenditures on local services necessary to mitigate firm's environmental damage. Ladd tested her hypothesis by estimating a standard demand function for local educational services in the Boston S.M.S.A in 1970. The residential tax price in her demand function equaled "1-aC-flM." "C" and "M" corresponded to the commercial and manufacturing portions of the locality's tax base. Ladd derived estimates of "a” and ”8.” If "a" and "8" both equaled one, the median voter's tax price for an additional dollar of public expenditure was perceived to be equal to the residential percentage of the local tax base. If "a" and "8" were both less than one, the median voter perceived his tax price to be greater than the residential portion of the local property tax base. This occurs if the firm portion of the property tax is shifted to local residents or an increased local property tax rate causes firm relocation and decreased future property tax revenue. Ladd calculated "a" to be .8 and ”B” to be .45. Accepting Ladd's logic, "-(1-a)" or "-.2" and "-(1-8)" or - .55 equals the elasticity of the local commercial and manufacturing property tax bases with respect to local property tax rates. r6 Ci 3C th. but Cit 29 Opposite to White (1986), Ladd found the more standard result that a metropolitan community's manufacturing tax base is more footloose than its commercial tax base. A manufacturing tax base is more likely to be driven out of a community by its high property tax rates. Oakland (1978, p. 27-28) pointed out an important caveat to Ladd's finding. Oakland stressed that Ladd's results only represent elasticities perceived by residents. These bear a questionable relationship to objective reality. Ladd's regression equations are also subject to a simultaneous. equations bias. Oakland's concluding point on Ladd's research is important to note: While a study of the determinants of a community's demand for public services is important in its own right, it cannot be expected to provide definitive evidence concerning the locational pattern of business activity. Such an approach, however, is an essential ingredient to a complete model for this purpose. One of the most recent papers on this subject is by Ladd and Bradbury (1988). Economic theory states that high city taxes may reduce city property tax bases through both a reduction in economic activity and/or negative capitalization into land values. Ladd and Bradbury used regression analysis to measure the relationship between city taxes and city property bases. Their theoretical model accounted for the four structural components that determine this relationship: (1) city tax base, (2) city balanced budget, (3) resident demand for public spending, and (4) city tax mix. The fact that a city's property tax rate is 30 endogenously determined by its property base was accounted for. Ladd and Bradbury estimate a reduced form equation representing determinants of a city's tax base. They are very careful to identify accurately the independent property tax variable. Their pooled data set is drawn for the years 1972, 1977, and 1982 from 86 large 0.8. cities whose 1970 population was greater than 300,000. The fact that observed property bases do not represent long-run equilibrium values is accounted for in their regression methodology. The long- run elasticity of a city's property base with respect to its property tax, overlying county and state taxes, and income tax was respectively calculated to be -.15, -.10, and -.07. A city's sales tax had a statistically insignificant influence on the city's property base. Ladd and Bradbury calculated that, over a five-year period, a city moves 46 percent of the way towards its long-run property base. As Ladd and Bradbury themselves state (p. 504): "...the study can be viewed as the first step in a larger and more ambitious study that would examine the effects of all major city taxes on each of a city's tax bases.” The review of the literature relating to the influence communities have on intrametropolitan firm location is now completed. Once the activity of a host community is taken into consideration, empirical research has shown that local taxation and expenditure policies fie influence intrametropolitan firm location. Unfortunately the size of an we th St 31 the reported influence varies. Possible explanations and solutions for this occurrence are discussed in Section D of this chapter. Before this is done, a review of the literature relating to the influence of property tax abatements on nonresidential property bases is presented in Section C. 1. C. PROPERTY TAX ABATEMENTS AND NONRESIDENTIAL PROPERTY BASES Because of the research described in Part B of this chapter, it is now generally accepted that differences in local fiscal characteristics exert some influence on firm location. In an attempt to utilize this finding, many local policymakers have persuaded state policymakers to allow them to grant property tax abatements to encourage nonresidential investment in their communities. There has been much debate, but little theoretical modeling or statistical evidence on the effectiveness of these abatement programs. Morgan and Hackbart (1974) were the first to use standard benefit-cost techniques to evaluate the effectiveness of tax abatements on a state-wide basis. Benefit was defined as increased state value added resulting from state tax exemptions. Cost was defined as loss of tax revenue resulting from state tax exemptions. Data from the period 1958-1961 was taken from seven states. Morgan and Hackbart found that five to ten percent of increased value added in a state had to be attributed specifically,to state tax 32 exemptions before the discounted benefit of the tax exemption exceeded the discounted cost. Coffin (1982) analyzed the influence of the city of Indianapolis's property tax abatement program on its economic development. The Indianapolis abatement program began in 1977 and is structured such that a 100 percent property tax reduction is given in the first year, falling to a five percent property tax reduction in the tenth and final year. Coffin calculated the average present value of a property abatement, as a fraction of an average total investment receiving a property abatement, as ranging from ".0786" for residential property to ".0188” for industrial property. He found no clear evidence that the tax abatement program reversed the relative decline of Indianapolis's Center Township district. McDonald (1983) presented a theoretical analysis of local inducements for business. Using a general model of urban production and input demand, he showed that municipal governments can benefit from fiscal inducements to firms, even if the firm would have located in the community without them. This occurs if, after a property tax reduction, the positive effects of increased land value and increased capital intensity of land use outweigh the negative effect of taxing at a lower rate. Using reasonable assumptions, McDonald found that, if the percentage property tax per full market value is nine or more times greater than the dc pe ab the fix tha atit: 33 real discount rate, the property tax elasticity of property tax revenue is negative. Wolkoff (1983) developed a systematic framework to assess the effectiveness of property tax abatements. Using a Jorgenson-type investment function and 1970's data from the city of Detroit, he found that on average a 50 percent reduction in property taxes increased firm investment by 2 to 5 percent. He concluded that this was too small of a response to justify the magnitude of property abatements the city of Detroit had offered in the 1970's. Wolkoff suggests that local policymakers differentiate between applicants and vary the size and length of abatement awards more. Wolkoff (1982) has also written a descriptive article on property tax abatements. Recently, Wolkoff (1987) has used game theory to analyze local property tax abatement offers as.a signalling game. A second benefit-cost analysis of property abatements was recently performed by Morse and Farmer (1986). Using 1980 data collected from the state of Ohio, Morse and Farmer derived the expected value criterion, that on average, 8.8 percent of all investment in communities using property abatements had to be attributable to the abatement in order that the community break even. In personal interviews, firm decision makers in Ohio reported to Morse and Farmer that 25 percent of their total investment was directly attributable to abatements. Morse and Farmer concluded tha in to ‘ prO] from alwa abat on b the lite chapt this‘ nonres eXtrem restri zoning‘ regress Practic (1975) . assumed ba8ed 0n 34 that, on average, local policymakers in Ohio acted rationally in offering property abatements. As this short review illustrates, literature relating to the influence of property abatements on nonresidential property investment is rather sparse. The evidence derived from this literature is somewhat contradictory and does not always directly address the influence of property tax abatements on nonresidential property bases. Some suggestions on how to improve property abatement research and all of the research reviewed in Sections A, B, and C of the literature review are given in Section D. 1. D. IMPROVING THE ASSESSMENT OF THE INFLUENCE OF LOCAL FISCAL POLICY ON LOCAL PROPERTY BASES The literature examined in previous sections of this chapter provide many insights as to how future research on this topic should proceed. Previous researchers have largely modeled local nonresidential zoning decisions as restricted to one of two extremes: (1) zoning out all firms, or (2) placing no restrictions on firm entry. To account for this dichotomous zoning, empirical researchers only excluded from their regressions localities without firms. Nonresidential zoning practices were more realistically accounted for in White's (1975) theoretical model. White's model of local zoning assumed that communities set entry requirements for firms based on residents' willingness to trade decreased local 35 environmental quality for increased local tax revenue. Firms unwilling to meet these requirements are zoned out. To account for the realistic zoning choices portrayed in White's (1975) theoretical model, future empirical research must do more than just restrict regression samples. Future empirical research should explicitly incorporate local zoning decisions into the theoretical models used to derive empirical tests.- Charney's (1983), McHone's (1986), and Ladd and Bradford's (1988) research are good examples of the initial direction of this incorporation. As Kiefer (1974) suggested, full incorporation calls for the development of a theoretical model that accounts for the political and institutional intricacies of local nonresidential zoning. Overlooked in even White's theoretical model is the fact that the compensating transfer paid by a firm to a community for environmental loss can be increased or decreased without changing the local property tax rate. A community can raise the pollution compensating transfer a firm pays through firm-specific user and development fees, subdivision regulations and extractions, and/or increased taxable property assessments. Communities alternatively lower the pollution- compensating transfer a firm pays without changing the local property tax rate. This occurs through property tax abatements, increased provision of firm services, and/or decreased taxable property assessments. These alternative forms of compensation need to be accounted for. The prevalence of local property tax abatement programs shows dnc Pre and Lad‘ tast firm “'0 36 that they exist, and their influence should be assessed. As the review of literature relating to property abatements showed, this assessment is far from complete. An equally important insight, first suggested by Struyk and James (1975), is that future empirical research cannot assume that existing firm location represents a long-run equilibrium. The existing pattern of firm location is the result of historical location decisions. These decisions were based on community characteristics that were likely different than observed now. Moving costs also prohibit established firms from moving to what would be a contemporaneous profit-maximizing location. It is erroneous just to regress current measures of intrametropolitan firm location against current community characteristics. To account for the influence local fiscal and non-fiscal characteristics have on firm location, future research needs to be patterned after dynamic regression tests similar to those performed by Erickson and Wasylenko (1980), Charney (1983), and Ladd and Bradford (1988). There is also a need to use separate theoretical models and empirical tests for manufacturing and commercial firms. Previous research has largely concerned itself with modeling and testing determinants of manufacturing firm location. Ladd (1975 and 1976), and Erickson and Wasylenko (1980) tested determinants of both manufacturing and commercial firm location and found that results differed between the two. This is evidence of the need for two separate models, 37 or at least two separate empirical tests. It is also essential that future research distinguish between manufacturing and commercial firms in community aspects of theoretical models. Equally important is the recognition that variables used in this type of empirical research are often simultaneously determined. In a regression using contemporaneous dependent and independent variables, independent variables such as property tax rate, income tax rate, local public service levels, property tax bases, etc. are definitely endogenously determined. Only about half of the previous researchers recognized this and did anything to account for it. Of the half who recognized it, most did not account for the full extent of endogeniety. Future researchers needs to consider possible endogeniety in their regression specifications and appropriately use a two-stage estimation technique. Results derived from empirical work in which this is not done are biased. The final and most important insight from the literature review is the need to construct a theory of intrametropolitan firm location that accounts for the many concurrent factors that influence it. Previous research moved closer to this I complete model by consolidating community supply-side considerations into purely firm demand-side models. The remaining factor not accounted for is residential location or the determination of local housing property bases. Tiebout (1956) recognized that local property tax rates and de ma 10 is 38 public service levels influence residential location choice. In turn, residential location choice influences local zoning decisions and the willingness of communities to supply firm sites. Through residential location choice, local fiscal policy has an indirect and most likely discernable influence on firm location. White (1974), Kiefer (1974), Fischel (1975), Struyk and James (1975), and Megdal (1986) all recognized this. They concluded their research with suggestions that future research develop a model that attempts to integrate the influence residential location has on firm location. It is my observation that this has not been done. To assess the true influence local fiscal policy has on intrametropolitan firm location, a major objective of future research should be the development of a theoretical model and an empirical test that accounts for both firm and residential location. The theoretical model would have to account for the joint determination of.a community's housing property base, manufacturing property base, commercial property base, local tax rates, and expenditure. Such a theoretical model is presented in the next chapter. CHAPTER 2 THEORETICAL MODEL Chapter 2 contains a theoretical model of the two sectors, community and firm, that simultaneously determine the rate of local taxation, home value, expenditure, and the composition of the property tax base in a metropolitan community. The chapter begins with an overview of the theoretical model and its assumptions. 2. A. OVERVIEW The metropolitan area is assumed to consist of a central city and a fixed number of independently governed surrounding suburban cities. City boundaries are constant and do not overlap. Residents live in the central city and all suburban cities. Firms reside in most communities, but not necessarily all. Communities primarily raise revenue by taxing residential and firm property at the same rate. Property tax rates vary across communities and are set to maintain a local balanced budget. For taxation purposes all property is assessed at a percentage of its full-market value. 39 40 Local assessment practices are equalized across the metropolitan area.5 Communities may also raise revenue by taxing residential and firm income. Income tax rates vary across communities and are endogenously determined in this model. Metropolitan communities use income and property tax revenue to provide services to residents and firms. A community provides each resident with a uniform amount of residential service. Firms within a community receive a uniform amount of locally-provided service per dollar of the firm's assessed property value. Uniform amounts of residential and firm services vary between communities in the metropolitan area. Locally-provided residential and firm services exhibit rivalness in consumption and are considered ”private goods" provided by a local public choice mechanism. The local public choice mechanism is the median-voter model.6 Local decision makers act "as if" there were a. referendum held on every tax, expenditure, and property tax base decision. These decisions are consequently modeled as being made by the local median voter.7 All metropolitan communities possess the legal ability to use local zoning 5For a description of the state of Michigan's equalization practices, see Courant (1982). 6Inman (1979) and Beaton (1983) provide reviews of the median voter model. 7For a discussion of the assumptions necessary for the median voter model to hold, see Mueller (1979) pp. 106-111. CC it ar. Clc 41 ordinances to restrict residential housing, manufacturing firm, and commercial firm entry.8 All firms are profit maximizers and operate in perfectly' competitive input and output markets. Firms are classified as either in manufacturing or commercial industries. Based on Floyd's (1952) theory of firm location and similar theories summarized in Weber (1986), at any time a fixed number of manufacturing and commercial firms desire to reside in the metropolitan area. These firms have made the metropolitan area their market choice, and wish to choose, or have already chosen a site to locate their taxable property. Manufacturing and commercial firms use three inputs in production: (1) a composite input consisting of taxable capital and land, (2) labor, and (3) locally-provided firm services. A distinct function represents the production technology of all manufacturing firms. A second distinct function represents the production technology of all commercial firms. The "typical manufacturing firm" sells its output to customers inside and outside the metropolitan area. Because of earlier patterns of development, a community closer to the metropolitan area's central city contains 8The justification for this presumption is Fischel's (1985) research. After extensively examining zoning practices in the United States, Fischel concluded that "...we may practically assume, without even investigating the existence of particular ordinances, that all general purpose governments have zoning,” (p. 23) and that "...[a] community can establish whatever standards it wants, so it is possible, with a little forethought (and sometimes just afterthought), to exclude most industrial and commercial activities" (p. 64). 42 more of the typical manufacturing firm's customers than one farther away. Transportation networks within a community reduce the cost of shipping manufacturing inputs and outputs. The closer a community is to the central city and the better the community's transportation network, the greater the post-shipping output price a manufacturing firm residing in the city receives. The "typical commercial firm" sells its output to a local market within the metropolitan area. The typical commercial firm's market is defined as the community it resides in and all communities that fall within a five-mile radius of its center.9 The presence of highways in the community increase the ease at which customers travel to the firm. The more attractive communities within a five- mile radius and the more extensive the community's highway network, the greater the revenue received by a commercial firm locating in the community. To reside in a community, a firm must pay local taxes. This includes a local property tax, and if levied, a local 9The size of the commercial firm's market is based on Hamilton's (1982) finding that the average worker in a medium- to large-sized urban area commutes 8.7 miles to work. The assumption in this model is that people are willing to drive as far to purchase a commercial firm's output as they are to work. Communities that fall within a five-mile radius of community "1" where determined by locating on a Metropolitan Detroit Area map the center of community ”1" and drawing a circle with a five-mile radius from this center. All communities that fall in this circle were included in the commercial firm's market.If a radius of 8.7 miles were used to calculate cities within the commercial firm's market, it would imply that people travel farther than 8.7 miles to shop. A radius of 5 miles was estimated to roughly satisfy Hamilton's finding. 43 income tax. A median voter will not alter local tax rates to encourage or discourage the entry of nonresidential property. The basis of this assumption is that the revenue fluctuations caused by such a policy would be too great. Instead, a median voter encourages the location of nonresidential property within the community by offering property tax abatements. Whereas a reduction in local tax rates affects all firms and residents, an abatement affects only the targeted marginal firm. A median voter can discourage the entry of nonresidential property by offering no abatements or by simply zoning it out. A median voter demands taxable nonresidential property from firms because of two benefits it provides. The first benefit is a reduced tax price for local expenditure. The second benefit is increased employment opportunities. These benefits are weighed against the environmental cost of increased nonresidential property. In demanding taxable property from firms a median voter's goal is the maximization of utility. Manufacturing and commercial firms are suppliers of taxable property to a community and its median voter. Firms supply taxable property to undertake production. In supplying taxable property, the goal of a firm is profit maximization.10 10Earlier research on this topic assumed that firms demand property from communities and communities supply property to firms. The role of demander and supplier is reversed in this research. The reason for this is that the theoretical model presented in Section B of this chapter explicitly yields an equation representing a local median voter's demand for firm property. 44 2. B. LOCAL TAX RATES, MEDIAN VOTER'S HOME VALUE, AND MEDIAN VOTER'S DEMAND FOR LOCAL PROPERTY BASE AND EXPENDITURE There is a direct relationship between a local median voter's utility maximization and his demand for the components of his community's property base and expenditure. A local median voter derives utility from the consumption of a composite private good, local expenditure per capita, and local environmental quality. All are considered normal goods. The composite private good and expenditure per capita are congestible "private" goods. Local environmental quality is a pure "public" good. The median voter chooses his consumption of the private good, the local level of expenditure per capita, and local environmental quality to maximize his utility subject to an environmental production function, income constraint, local property tax constraint, and local balanced-budget constraint. The median voter's utility function is represented as: 45 (1) 01 8 (Ban (ROM (X’H. where i - 1, 2, 3, ...number communities in metropolitan area, B s units of environmental quality -- UE > o, 033 < 0.11 R - units of local expenditure per capita --UR>O'URR o, Uxx <0, a, r, a a relative share corresponding good into total utility -- a + a + r = 1. The environmental production function relates local environmental quality to the dollar value of manufacturing, commercial, and housing property in the community13: 11”UE" represents the first-order partial derivative of ”U" with respect to "E". "UEE" represents the second-order partial derivative of "U" with respect to "E". This notation is used throughout this work. 12A unit of the composite private good and a unit of local expenditure per capita are valued at one dollar. 13Cobb-Douglas utility and production functions are used for ease of presentation and derivation. The choice of functional form has no significant impact, since utility maximization is only used to derive a general function that will be estimated empirically. 14 to 15. ta) 31;. hOu ACc but 46 (2) 21 = di (M1)“ (ci>° (ni)h. where d = constant representing community-specific factors that influence changes in local environmental quality caused by local property tax base,14 m, c, h = constants representing property-tax-base- specific factors that influence environmental quality -- constant across communities, M = value of manufacturing property -- EM < 0, Em <0, C = value of commercial property -- Ec < 0, ECC < 0, H a value of housing property -- EH < 0, EHH < 0. The environmental quality of one community can spillover to surrounding communities and affect their environmental quality. A local median voter is assumed to ignore the effects of their property base decisions on surrounding median voters. Substituting equation (2) into equation (1) results in the following: (3) vi = {d1 (M1)“ (ci)° (Hi)h}° (xi)' (R1)°. or (4) log Ui = ailogdi + aimilogMi + dicilogci + aihilogHi + ailogRi + Tilogxi. 14The community-specific constant would be positively related to community area and negatively related to community population. 15The assumptions that an increase in all three forms of taxable property reduce local environmental quality may not always hold. One can think of a blighted community where an increase in taxable property, especially commercial or housing, may effectively increase local environmental quality. Accounting for this alters the reported theoretical results but not the general functional form derived from them. 47 Equation (4) is a monotonic transformation of equation (3), and represents the same ordering of median voter preferences. To maximize equation (4), the median voter chooses the local dollar value of manufacturing, commercial, and housing property to allow in the community, his own consumption of the composite private good, and local expenditure per capita. These choices are subject to three constraints. Two of these are the median voter's income and the local property tax rate constraints: (5) Y1 = Xi + ini, and - T1 (6) p1 = -------------------------- , (Mi ' 31) + (C1 - A1) + Hi where median voter's income, units of composite private good, local property tax rate, value of median voter's house, local property tax revenue, value of manufacturing property abatements, value of commercial property abatements. >HWFB ru,1 = {PM 1 (K ,1)” (Lu 1)“) - (1 + p1)( ,1xu,1) I é}1>(P ,1 KM,15 - PLMLM,i + fi< ,1KM,1) it where i = 1, 2, 3, ...number communities in metropolitan area, (KM)2 (LM)“ - typical manufacturing firm's production function, E, u a relative share of corresponding input into total output -- E + u < 1, PM - post-shipping price per unit of typical manufacturing firm's output, =_f (cd 2 central city distance, tn = transportation networks) -- rcd > 0, rtn > 0, KM 2 units of composite manufacturing property -- land and capital, LM = units of manufacturing labor, p = local property tax rate per dollar of property, 1 = local income tax rate per dollar of property, f = value of firm services per dollar of propertyzo, Bf = value of property abatement to typical manufacturing firm, PPM = net-of—local-income-tax wage for a unit of manufacturing labor, (cont.) 19Decreasing returns to scale is necessary to guarantee that the second-order condition for profit-maximization is satisfied. 20Because the value of firm services (f) are a portion of total expenditure per capita (R), "f" is endogenously determined in the same way that "R" was determined. In simple functional form: f1 = f (A1, B1, N1, V1, Y1, 11, 91, a1, a1, 71). 64 P3“ - gross-of-local-property-tax price for a unit of composite manufacturing property, 8 = f (p, l, f, cd, tn, cm a number of crimes, 21 magg - manufacturing agglomeration economies) -- Fa < >O, r} < 0, off > 0, fat > 0, rcd > 0, cm < 0, magg >0. PKMKM - typical manufacturing firm's property value -- similar to "M" of Section 8, except for a single firm. Inputs used in manufacturing production are sold in metropolitan-wide competitive markets. The gross-of-local- property-tax price for manufacturing property (PKM) varies between communities because the difference in the profit the typical manufacturing firm earns in each community is capitalized into the price of its land component. Full capitalization occurs in the long run. Firms pay a local property tax (p) in proportion to the value of property used in production. This cost can be reduced by the community offering a property tax abatement to the typical manufacturing firm's property. The value of this abatement is measured as "Bf". ~ Labor does not necessarily reside in the community in which it is employed. Labor is considered perfectly mobile across communities in the metropolitan area and there is no cost to achieve this mobility. This forces the net-of- local-income-tax manufacturing wage (PLn) to be equal in all metropolitan communities. Firms have to pay a higher gross-of-local-income-tax wage to attract and employ labor 21The number of crimes is endogenously determined. In simple functional form: cm1 = f (C1, H1, M1, R1, N1, Y1, cd1, age1, y 1, age 1, area1, educ1, race1, pden 1). 65 in a city with a local income tax. Firms also bear the full cost of a local tax on firm income. The burdens of a local income tax are accounted for by the local income tax rate (1) the firm pays per dollar of manufacturing property. The local provision of firm services (police, fire, sewer, sanitation, and highway) enters the profit equation through variable ”f”. 'A firm necessarily uses these services to produce its output. If they are not provided by the community, the firm purchases them privately through the use of greater composite property and labor inputs. The profit equation is set up as if all firms purchase locally provided services privately. A local provision of these services is then measured as a reduction in production cost. To find its current profit-maximizing metropolitan site the typical manufacturing firm maximizes equation (36) with respect to "KM" and "LM". This yields the following first-order conditions:22 (37) pM(zKZ‘1)Lfl = (1 + p + 1 - £)pKM, (38) puxzmw‘l) = FL". The joint solution of equations (37 and 38) results in the following community-specific factor demand equations: u 6 * PKM(1 + p + l - f) (39) KM = ------------------------------------- uPKM(l+p+1-f) pus ~9IH 22To simplify the presentation of equations (37 - 42) the "i” subscript has been omitted. (40) *1“ = where 66 E 1 upKM(1 + p + 1 - f) n n = 2 + u - 1. PPM n Pun Substituting equations (39 and 40) into equation (36) results in an equation representing the maximum economic profit the typical manufacturing firm earns in any metropolitan community: pKM(1 + p + 1 - f) (41) *1“ = ------------------ 9M2 u p 2 a (1+p+1-f)PKM ----- E! ------ uPKM(l+p+1-f) F- 2 uP (l + p + l - f) 0 pg” --EE ............... ZPPM M Pu“ To derive the long-run supply of manufacturing property to one community an initial state is examined where: metropolitan area, (1) there were no manufacturing firms in the (2) the gross price of a unit of composite manufacturing property (PKM) was equal in all 67 communities and, (3) local property tax rates, central city distance, and transportation networks varied between communities. A group of identical, competitive, profit-maximizing manufacturing firms entered this state and desired to locate somewhere in the metropolitan area. To find the community that offered the profit-maximizing site, these firms substituted into equation (41) the appropriate community variables. In the initial state, manufacturing profits differed between communities because local property tax rates, transportation networks, and distance to the central city differed. Because manufacturing firms were identical, all initially desired to locate in the one community where profit was greatest. Competitive bidding drove up the price of manufacturing land in this community. The gross price for a unit of composite manufacturing property rose until the once optimal manufacturing community was no longer where the typical manufacturing firm maximized profit. This process continued until the economic profit the typical manufacturing firm earned in any metropolitan community was driven to zero. The initial group of manufacturing firms were consequently indifferent to where they located. After the initial entry of manufacturing property, zero economic profit in all metropolitan communities occurs only in the long run when all manufacturing property is again considered perfectly mobile. Imperfect 68 capitalization occurs in the short run because local characteristics which influence manufacturing profit are constantly changing. In the short run a community unattractive to manufacturing,23 whose lower profit potential is not fully capitalized into lower land prices, attracts no new manufacturing property. A median voter desiring new manufacturing property realizes this and directs the community to offer property tax abatements which increase the profit the typical manufacturing firm earns in the community. Local median voters may even offer abatements to new manufacturing property when local profit differentials are fully capitalized into local land values. The reason is based on White's (1986) observation on why perfectly capitalized local taxes influence firm location.24 1 If a median voter offers a property tax abatement to the typical new manufacturing firm he offers an abatement (*Bf) that brings the economic profit up to the long-run zero economic profit earned in other metropolitan communities. With an abatement offer less than "*Bf" new manufacturing property continues to avoid the community. The abatement offered to the typical manufacturing firm already residing in the community has been ignored. To have attracted this property in period "t", the local 23Because of relatively high local taxes, relatively poor transportation networks, far from the central city, etc. 24See Chapter I, p. 26 for a full description. 69 median voter would have offered the current "*Bft”. If in period ”t+1" community characteristics change such that "*Bft" causes the firm to earn less than a zero economic profit, the median voter does not necessarily have to offer the larger "*Bft+1" abatement. A firm residing in a community can earn a negative economic profit equal to its moving costs and find it optimal to stay in the short run. A median voter can exploit this by offering an abatement less than "*Bft11". Only in the long run, when all property residing in a community is perfectly mobile, does the abatement to the typical manufacturing firm residing in the community necessarily equal the current "*Bft".25 Manipulating equation (41) (replacing "*lm" with zero and solving for "*Bf") yields the local property abatement given to the typical manufacturing firm consistent with the firm earning zero economic profit: 25A second possibility is that a positive change in a community characteristic(s) allows the community to offer a lower level of abatement to the firm and it still earns a gero economic profit. If the community has promised "B t" to the firm for a certain period it may legally not be able to offer less. What the median voter does in this situation is reduce the value of locally provided firm services until "*8 t" is again the optimal level of abatement. 7O 7' _. z * f pKM(1 + p + 1 - f) n PPM (42) B = - P ------------------ —-- + PMz Pun [— .— u 1 (1+ +1 f)PKM PLME n PKM(l+p+1-f) n + p - ------------------------- uPKM(l+p+1-f) Pun E 1 uPKM(1 + p + 1 - f) a pFM H PF“ ------------------- —-- . If the local median voter desires new manufacturing property in the community, or existing manufacturing property to remain in the community in the long run, he necessarily offers "*Bf" to the typical manufacturing firm. To transform "*Bf" to the manufacturing abatements offered to all manufacturing property in long run, equation (42) must be multiplied by the long-run number of typical manufacturing firms in the community. Equation (42) is then equivalent to the following general form: (43) *31 = f (PM,ir PKM,1: Pi: 11: f1: *FM,1): where *FM = long-run number of typical manufacturing firms. In equation (43) the share of composite property (2) and labor (u) used by the typical manufacturing firm to produce its output, and the net-of—local-income-tax wage (PPM) for 71 a unit of labor have been excluded. These variables are invariant to communities in the metropolitan area. A basic form of equation (43) is equal to: (44) *B1 - r (p1, 11, £1, cd1, tn1, cm1, magg1, *M1) ??????? + where *M = long-run manufacturing property value, 'r(FM) " [*m>0. Equation (44) is a basic form of equation (43) because "PM”, "PKM", and "*FM" have been replaced by variables they are a function of. These variables are: ”cd”, "tn", ”cm", "magg", and "*M". The functional relationships between the replacement variables and "PM" and "PKM" are given in the list of variables under equation (36). "*M" is the only available proxy for the long-run number of manufacturing firms. The local property tax rate (p), local income tax rate (1), locally provided firm services (f), number of crimes (cm), and the long-run value of manufacturing property (*M) are endogenously determined. As shown in equation (14), the local property tax rate is a function of the value of manufacturing property and manufacturing property abatements. As shown in equation (12), the local income tax rate is a function of manufacturing property. As shown in the variable description under equation (36), locally provided firm services and number of crimes are also related to manufacturing property. As shown in equation (31), the value of manufacturing property is in part 72 determined by the value of manufacturing property abatements. Differentiating equation (42), with respect to each of the right-side variables in equation (44), yields for all variables, derivatives whose signs are indefinite.26 The expected relationship between "*M" and "*B" is positive. The relationship between the remaining right- side variables and "*B" is uncertain. The signs below equation (44) reflect this. A schedule of the total manufacturing property abatements a median voter finds necessary to offer to attract a given value of manufacturing property in the long run, can be derived by plugging into equation (42) the current levels of applicable variables and multiplying by the number of typical manufacturing firms. 2. D. COMMERCIAL PROPERTY ABATEMENTS NECESSARILY OFFERED BY MEDIAN VOTER A model, similar to the one just presented, is now used to derive the schedule of commercial property abatements a local median voter finds necessary to offer to attract a given value of commercial property in the long run. An equation representing the profit (13) the typical 26The first-order derivatives were not written out due to their length. To see that the signs of the derivatives are indefinite, observe that each of the relevant independent variables on the right-side of equation (42) enter with both a positive and negative sign. 73 commercial firm earns in each of the communities in the metropolitan area is equal to: (45) IC, 1 - {Q1Pc 1 Kc,1>° (Lc,11°} - (1 + f9111ch 1Kc 1) :A115(c,1Kc,1)-Pcbc,1+f(1 c,,1kc15 where i - 1, 2, 3, ...number communities in metropolitan area, (KC)’ (LC)e = typical commercial firm' s production function, 9, 9 = relative share of corresponding input into total output -- o + e < 1, q = relative demand for typical commercial firm's output -- q 2 1, (quantity sold)q > 0, a post-shipping price per unit of typical commercial firm' s output, a f (1m 8 positive local market characteristics, hn = highway network) -- rlm > 0, rhn > 0, RC = units of composite commercial property -- land and capital, LC 2 units of commercial labor, p 8 local property tax rate per dollar of property, 1 = local income tax rate per dollar of property, f = value of firm services per dollar of property, Af = value of property abatement to typical manufacturing firm, PLC = net-of-local-income-tax wage for a unit of commercial labor, PKC = gross-of—local-property-tax price for a unit of composite commercial property, a f (t, l, f, Af, lm, hn, cm a number of crimes, cagg a commercial agglomeration economies) “(t<°:f rrIf>°rAf>°Irlm>°v hn > 0 }< cd < 0 rcagg >0! PKCKC a commercial property value -- similar to "C" of Section B, except for a single firm. 74 The typical commercial firm's profit equation is equal in all but two respects to the typical manufacturing firm's profit equation (equation 36). One difference is that the price the typical commercial firm charges for its output also varies between communities. This occurs because the market for the commercial firm's output varies between communities. A second difference is that the quantity of output the typical commercial firm sells varies by the community the firm locates in. This is‘accounted for by "q” which measures the local demand for the typical commercial firm's output relative to other communities in the metropolitan area. All assumptions made earlier in regard to the manufacturing profit equation apply equally to the commercial profit equation. To find its current profit maximizing site in the metropolitan area, the typical commercial firm maximizes equation (45) with respect to "KC" and "LC". This yields the following first-order conditions:27 (46) qpc(4x°‘1)L9 = (1 + p + 1 - f)PKC, (47) qpcx°(eL°‘1) a PLC. The joint solution of equations (46 and 47) results in the following community-specific factor demand equations: 6 ch4 ; pKC(1 + p + 1 - r) (48) *KC 3 ------------------------------------- K 9? c(1 + p + 1 - f) ch0 ~ m lh‘ 27To simplify the presentation of equations (46 - 51) the "i" subscript has been omitted. (49) *Lc =- where e 8 Q + 9 - 1. 75 ech(1 + p + 1 - f) 4 6 PLC ch9 ~ 6 up: Substituting equations (48 and 49) into equation (45) results in an equation representing the maximum economic profit the typical commercial firm earns in any metropolitan community: (50) *1C - p (1+p+1-f)ch L pLC ...... PKC(1 + p + 1 - f) qPCO 9 PLc§ 6 epKC(1+p+1-f) 4 O 9 6 PLC 6 che .1 1 PLC 6 "" + Af. qPCO For the same reasons as given for manufacturing, in the long run a typical commercial firm is expected to earn a zero economic profit in any community in the metropolitan area. Replacing "*tc" with zero, and solving for ”Af" in equation (50), yields the local property abatement given 76 to the typical commercial firm consistent with the firm earning zero economic profit in the long run: '_ "-1 Q 6 - - L ' (51) *Af = - p Efgii-:-g-:-i---fl 6 3.8- E + qch che 9 1 PLCO e PKC(1+p+l-f) e + (1+p+1-£)pKC ------------------------- OPKC(1+p+l—f) che 4 1 FL 6pxc(1 + p + 1 ' f) 6 PLC 6 C ------------------- ---- e QPLC qpce If the local median voter desires new commercial property in the community, or existing commercial property to remain in the community in the long run, he necessarily offers ”*Af" to the typical commercial firm. To transform "*Af” to the commercial abatements offered to all commercial property in long run equation (50) must be multiplied by the long-run number of typical commercial firms in the community. Equation (51) is then equivalent to the following general form: (52) *A1 = f (Pc,1: PKc,1: Q1: P1: 11: f1: *Fc,1): where, *FC 8 long-run number of typical commercial firms in community. 77 In equation (52), the share of composite property (0) and labor (8) used by the typical commercial firm to produce its output, and the net-of—local-income-tax wage for a unit of labor (PLb), have been excluded. These variables are invariant to communities in the metropolitan area. A basic form of equation (52) is equal to: (53) *Ai = f (D1. 11. f1. qi. 1m1. hni. cmi. caggi. *c1) ? ? ? ? ? ? ? ? + where *C = long-run commercial property value, ‘ f ( Fc) " (*Fc > 0- Equation (53) is a basic form of equation (52) because "PC”, "PRC", and "*Fc" have been replaced by variables they are a function of. These variables include: 'cd", "tn”, "cm", "magg", and "*C". The functional relationships between the replacement variables and "PC" and ”PKC” are given in the list of variables under equation (45). ”*C" is the only available proxy for the long-run number of commercial firms. The local property tax rate, local income tax rate, locally provided firm services, number of crimes, and long- run value of commercial property are endogenously determined in the commercial property abatement equation (53). The reasons are similar to what was given for the manufacturing property abatement equation. Differentiating equation (51), with respect to each of the right-side variables in equation (53), yields for all variables derivatives whose signs are indefinite. The 78 relationship between "*C" and "*A" is positive. The relationship between the remaining right-side variables in equation (53) and "*A" is uncertain. The signs below equation (53) reflect this. A schedule of the value of commercial property abatements a median voter finds necessary to offer to attract a given value of commercial property in the long run can be derived by plugging into equation (51) the current levels of applicable variables and multiplying by the number of typical commercial firms. The two abatement schedules necessary to produce the median voter's demand for manufacturing property, commercial property, housing property, and residential service have now been derived. An empirical analysis of the complete system of equations derived in this chapter is presented in the next chapter. CHAPTER 3 EMPIRICAL ANALYSIS Chapter 3 contains an empirical analysis of the nine simultaneous equations derived in the theoretical model. The preliminaries necessary to perform the empirical analysis are presented first. III. A. PRELIMINARIES A set of nine equations represent the simultaneous determination of the value of the median voter's house, local income tax rate, property tax rate, manufacturing property, commercial property, housing property, expenditure per capita, manufacturing property abatement, and commercial property abatement in a community within a metropolitan area. The general form of these nine equations are repeated below: (7) V1 2 f (nc1, hc1, ac1, p1, 11, R1), (12) 11 = f (H1: C1: R1: R1: B1: A1: N1: P1: 91): (14) pi = (31) *M1 = (M1! Cir Hi: Rir Bit Ail Nil 11! 91): (Air Bi: Nil Via Y1, lit 91: “1! 01: 71)! (32) (Air Bil Ni: Vi: Yip lit 91! 01, 01, 71)! (33) *Hi = (41. 31, N1: V1: Y1. 11: 91: 01: 01. 71). I' O 1... fl qq—Dfifi (34) *R1 = (A1: Bil Nil Vi, Yir lit 91: air air '1): 79 80 (44) *B1 = f (p1, 11, f1, cd1, tn1, cm1, magg1, *M1), (53) *A1 = f (p1, 11, f1, q1, 1m1, hn1, cm1, cagg1, *C1). Some of the independent variables listed above refer to general community characteristics that observable local variables can proxy for. The substitution of observable variables for general community characteristics yields equations (54 - 62): (54) V1 8 f (mden1, cden1, hden1, hown1, cm1, race1, Y1. hci. cdi. hni: P1: 11: R1): (55) Ii = r (“it Cir Hit R1: Bil Ail Nil pi! 91): (56) pi = r (“it Cir Hit R1: 811 Ail Nil 11! 91)! (57) *M1 = r (A1, B1, N1, V1, Y1, 11, 91, age1, edUC1, area1, C1, H1, lemp1), (58) *C1 = f (A1, B1, N1, V1, Y1, 11, 91, age1, educ1, area1, H1, M1. lemp1), (59) *H1 = r (A1, B1, N1, v1, Y1, 11, 91, age1, educ1, area1, C1, M1, lempi): (60) It w H. II -fi (A1: Bit “1: V1, Y1: 11! 91: ageir educ1, area1, C1, Hi, M1, lemp1), (61) *B1 = f (p1, 11, f1, Cd1, hn1, wn1, cm1, magg1, *M1), (62) *A1 = f (p1. 11. f1: Y1: agei. pdeni: Ysg agesi. pden 1, hn1, cm1, cagg1, *C1). In equation (54), local neighborhood characteristics have been accounted for by manufacturing property value per square mile (mden), commercial property value per square mile (cden), housing property value per square mile (hden), percentage housing owner occupied (hown), 28Superscript "s” represents the average value of the variable for communities within a five mile radius of the center of community "i". For a description of how it is calculated see footnote #9. 81 number of crimes (cm), percentage residents nonwhite (race), and median income (Y). The percentage of local housing with more than one bath has been chosen to represent positive characteristics of the median voter's home (hc). The percentage of housing with more than one bath is expected to be a positive proxy for the size of local houses. Distance to central city (cd) and miles of divided highway (hn) are the variables chosen to represent the accessibility of the median voter's house. In equations (57 - 60), the median voter's taste for environment, local expenditure, and the private good is represented by distinguishing characteristics of the community's residents. These include: median age (age), median years of education (educ), and percentage of residents employed locally (lemp). As described in equation (63), "lemp" must be considered an endogenously determined variable: (63) lemp1 = f (M1, C1, N1) + + - The percentage of residents employed locally is expected to be positively related to either component of the community's nonresidential property base, and negatively related to the community's population. The greater the local population, £2£§Ii§.2§£122§. the greater the competition for a given number of local jobs, and the smaller the percentage of residents expected to be employed locally. 82 In the regression equations, "lemp" acts as a proxy for the likelihood that the median voter desires to be employed in the community of residence. A median voter is expected to demand greater nonresidential property if he desires to be employed locally. The median voter's taste for the choice variables is also influenced by local population (pop), square miles (area), and the current composition of its property tax base (C, M, and H) excluding the base being chosen. As described in equation (2), the marginal environmental damage experienced in allowing greater property into a community is expected to be positively related to community population and negatively related to community area. Environmental damage is also expected to be positively related to existing values of the prOperty tax base. The greater the existing bases, the greater the marginal environmental damage from allowing further property. In equation (61), highway network (hn) and whether the community is adjacent to a major water transportation source (wn) are observable replacements for local transportation networks that benefit manufacturing firms. In equation (62), the relative demand for the typical commercial firm's output and local market characteristics have been accounted for by six observable variables. Three of the variables are calculated for the community: median income (Y), median age (age), population density (pden). The remaining three variables are calculated as the average 83 of the same variables observed in communities within a five-mile radius of the center of community "i" (YS, ages, and pdens). Equations (54 - 63) are listed in regression form below29: (64) Vt = 80 + filmdent + fizcdent + fi3hdent + 84hownt + 35cmt + fisracet + fl7Yt + fishct +fl9Cdt + Biohnt + 311Pt + Bizlt + 513Rt + “t: (55) 1t = 314 + fiisut + fi16¢t + 517Ht + 318Rt + 3193t + fizoAt + 521Nt + fizzpt + 3239t + “t: (55) Pt = 824 + fizsut + 326Ct + 527Ht + stRt + 5293t + fiaoAt + 331Nt + 3321t + 3339t + “t: (57) *Mt = 534 + 335At + fiszt + 337Nt + 538Vt + 339Yt + 5401t + 5419t + 342399t + 343educt + 344areat + 545Ct + B46Ht + 347lemPt + "t: (53) *Ct = 348 + 349At + fisoBt + fisiNt + fiszvt + 553Yt + 3541t + fissgt + fissaget + 557edu°t + fisgareat + fingt + fiGOMt + Hellempt + fit, (59) *Ht = 562 + 363At + 364Bt + fissNt + 366Vt + 367Yt + fisslt + 3699t + 57oaget + 37ieduct + 3723reat + 373Ct + 374Mt + B7516mPt + “t: (70) *Rt = 376 + 377At + 37th + 379Nt + fisth + fieth + figzlt + 383gt + £84aget + fiesedUCt + figsareat + 387Ct + 5883t + 589Mt + figolempt + “tr (71) *3t = 591 + 392Pt + 3931t + 394ft + figscdt + £96h¥§ ++fifi7wnt + 898cmt + figgmaggt + _ 100 t t: * (72) At = 3101 + filozpt + 31031t + BlOéft + fiiosYts+ 3106399t8+ B107Pdent + fiioaY t + 510939e t + g1ioggen t + filllhnt + fiiizcmt + 51136399t + 114 t + “t: (73) lemPt = 3115 + 3115Mt + 3117Ct + BllBNt + “t 29To simplify the presentation of equations (64 - 91) the ”i" subscript has been omitted. The "t" subscripts refers to the current value of the variable. 84 Recall that given the right-side variables in equations (57, 58, and 59), the median voter calculates the optimal value of local manufacturing, commercial, and housing property. A problem in estimating the corresponding regression equations (67, 68, and 69) is that these optimal values are not necessarily observable. Consider a situation where the observed value of local manufacturing (commercial or housing) prOperty currently equals the median voter's optimal value. From this equilibrium situation, allow the median voter's taste for manufacturing property to either decrease or increase. This in turn decreases or increases the median voter's optimal value of local manufacturing property. The only way that the current value of local manufacturing property always equals the median voter's desired value is if manufacturing property can instantly leave or enter a community. This is not the case. "*M" and "*C" therefore represent optimal values that the median voter can only be expected to be moving toward. On the other hand, I assume that there are no restrictions that prevent the median voter from consuming the utility maximizing value of local residential service (*R) at all times. A similar estimation problem applies to equations (71 and 72). In these equations, "*8", "*A", "*M”, and "*C" represent the long run values of manufacturing and commercial property abatements and property. By definition these values are only observable in the hypothetical long 85 run and again represent an amount that the community can only be expected to be moving toward. The solution to these estimation problems is to alter regression equations (67, 68, 69, 71, and 72) using a stock-adjustment model.30 The stock-adjustment model assumes that the observed change in a variable, from period "t-l” to ”t”, is equal to a percentage of the total change to the optimal or long-run value of the variable. The stock-adjustment model, as applied to ”*M" and "*B", is represented in the following equations: (74) Mt - Mt;1= WM( - Mt-1), or (75) Mt= 6" Mt + (1 - t6 M)Mt-1r or (75) *Mt = Mt-1+ {(Mt ' MMt-1)/5M): (77) 3t ' Bt;1 = 5B( at: ‘ Bt-1): Or (78) gt 3 63 8t + (1 - OB)Bt-1, or (79) 3t = 3t. -1 + {(Bt - Bt-1)/63), where M = observed value of manufacturing property, *M = long-run value of manufacturing property, 6" - coefficient of manufacturing property adjustment -- O S 6M S 1, B = observed value of manufacturing property abatement, *B = long-run value of manufacturing property abatement, 63 = coefficient of manufacturing property abatement adjustment -- O S 5B S 1. Similar stock-adjustment equations apply to "*C", "*H," and "*A". Substituting equation (76) into (67), and solving for "Mt” yields: ‘ 30For the original derivation of the stock-adjustment Inodel, see Nerlove (1958). For a recent application to retail inventory investment, see Irvine (1981) . 86 (30) Mt ‘ 5Mfl34 + (5Mfi35)At + (5M336)Bt + (5M537)Nt + (5Mfi38)Vt + (5Mfi39>Yt + (5Mfi4o)1t + (5Mfi41)9t + (5Mfi42)398t + (5Mfi43)educt + (6Mfi44)areat + (6ufi45)Ct + (6nfl46)Ht + (6Mfi47)lempt + (1-6M)Mt_1 + 6uflt. Equation (80) contains all observable variables. An ordinary-least-squares estimation generates consistent estimates because ”ut” is assumed to satisfy the necessary conditions of the linear regression model and O 5 6M 5 1. A similar alteration of equations (67 and 68), results in regression equations that contain all observable variables: (81) Ct = 5cfi43 + (5cfi4 (5c352)Vt + ( (5cfiss)9t + ( (60653)areat (6C361)lempt (32) Ht = 611562 + (5H363)At + (5H364)3t + (5H565)Nt + (5H566)Vt + (5H567)Yt + (5H368)1t + (5H569)9t + (5HB7o>aget + (5H571)educt + (5H372)areat + (5Hfi73)ct + (5H374)Mt + (63675)lempt + (1-6H)Ht-1 + 6Hut, )At + (5c550)3t + (5c551)Nt + c553)Yt + (5c354)1t + c856)aget + (6C357)educt + (icfis9)fit + (5c560)Mt + 9 6 6 + + '6C)Ct_1 + 6Cut, The stock-adjustment model does not apply to equation (70). Equation (70) is equivalent to: (33) Rt ' 576 + 377At + 57th + 379Nt + fieth + 381Yt + flszlt + 383gt + 684aget + fiaseduct + 886areat + £37Ct + BBBHt + BB9Mt + figolempt + “tr Regression equation (71) contains two unobservable variables (*B and *M). The stock-adjustment model is applied by substituting equation (76) into (71), and then substituting the result into equation (78): (34) 3t ’ 53391 + (53392)Pt + (53393)1t + (53394)ft + (63395)Cdt + (63396)hnt + (63697)wnt + (63898)cmt + (63399)maggt + (535100/5M)(M ' Mt-l) + (5Bfi1oo)Mt-1 + (1’5B)Bt-1 + But: 87 A similar application of the stock-adjustment model to equation (72) results in: (35) At ‘ 5A§éOlf + (5Afi102)Pt + (5A31g3)1t+ <:.:1::%§s:: i151xz’::::sé 1”12:£:::1a;: . + (5A3110)Pden§ t + (5A 111>hnt + (5A5112)cmt + (5Afi113)0399t + (5Afi114/5c)(ct ' Ct-i) + (5A5114)Ct-1 + (1'5A)At-1 + 5Aut- Regression equations (80 - 85) contain all observable values and can be estimated. In these equations, the regression coefficients in parenthesis represent the short- run influence of the corresponding right-side variable on the dependent variable. Dividing these coefficients by the appropriate coefficient of adjustment (6), yields the long-run influence. In regression equations (80, 81, and 82), the coefficient of adjustment is equal to one minus the lagged dependent variable's coefficient. In equation (84), setting the regression coefficients on ”Mt-1" and "Bt-l" equal to constants (k1: j = 0, 1, and 2), and solving simultaneously, yields solutions for "6H”, "63" and "5100": (86) k0 = (JBfiloo/SM) <=> 5M= k1 / kg, (37) k1 = (5Bfi100) <=> 5100 = R1 / (1 ' R2): (88) k2 = (1 - 63) <=> 63 = 1 - k2. In equation (85), setting the regression coefficients on "Ct-1" and "At-1" in equation (85) equal to constants (kj: j I 3, 4, and 5), and solving simultaneously, yields solutions for "6c”, "6A" and "8114". (89) k3 - (6Afl114/6C) <=> 6C = R4 / R3, (90) k4 ' (5Afi114) <=> fl114 = R4 / (1 ' k5): 88 (91) k5 - (1 - 51) <=> 5A = 1 - k5. The simultaneous system to be estimated now consists of regression equations (64, 65, 66, 73, 80, 81, 82, 83, 84, and 85). In this system of ten equations there are 19 endogenous variables. These endogenous variables consist of the ten dependent variables: "Vt”, "1t", ”pt", “lempt", ”Mt”, "Ct", "Ht", "Rt", "Bt", ”At”: the change in some of the dependent variables: "Mt-Mt-1' and "Ct-Ct_1": the property densities: "cdent", "hdent", and "mdent": the rate of firm services: "ft": the number of crimes: "cmt": population: "Nt": and population density: "pdent". The change in the dependent variables are endogenous because the value observed in period "t" is endogenous. The property densities are endogenous because they are calculated using current values of the property base. The rate of local firm services and income tax are chosen by the median voter based on the composition of the property base, expenditure, and population. The number of crimes is negatively related to local expenditure on police protection and positively related to property tax base. Population and population density are positively related to the value of residential housing. The final thing to check before estimation is the identification of the ten simultaneous equations. A rank and order condition must be satisfied.31 The rank condition of identification is satisfied for all ten 315ee Gujarati (1988), pp. 573 - 591. 89 equations. The order condition of identification states that in a system of simultaneous equations containing 19 endogenous variables, 18 variables appearing in the system (endogenous or exogenous) must be excluded from each equation in order to "exactly" identify the equation. In all equations more than 18 variables are excluded. Each of the regression equations is over identified. A two-stage-least-squares estimation is appropriate.32 If you examine the manufacturing property and manufacturing property abatement regression equations (80 and 84) you notice that the coefficient of manufacturing property adjustment (6") appears in both. The calculated value of "SM" is required to be the same in both regression estimations. For this to occur, the two equations were estimated jointly.33 The commercial property and commercial property abatement regressions (equations 81 and 85) were also estimated jointly because the coefficient of commercial property adjustment (6C) appears in both regression equations. Before the estimation results are 32The two-stage estimation method is described in Gujarati (1988), pp. 603-605. For each regression equation it consists of first regressing all the independent endogenous variables on all exogenous variables in the system. The predicted values for the independent endogenous variables are then used in the estimation of the regression equation. 33The joint estimation was conducted in Micro T.S.P. - Version 6.0, using an iterative three-stage least squares technique. The restriction placed on the estimation was that the coefficient on "Mt-Mt-1" in regression equation (84), equal the coefficient on "Mt-1" in regression equation (84), divided by one minus the coefficient on "Mt-1" in regression equation (80). This linear restriction was derived from equations (86 - 88). 90 presented, a description of the data and the variables are given in Section B. 3. B. DATA AND VARIABLES The Detroit Metropolitan Area was chosen as the metropolitan area from which to gather data to estimate ~the ten simultaneous equations. The Detroit Metropolitan Area is representative of any large metropolitan area in the United States.34 The Detroit area makes up one of the largest concentrations of manufacturing property in the United States. It includes communities with many diverse mixtures of manufacturing, commercial, and residential housing property bases. A major reason for choosing the Detroit Area is the use of property tax abatements by localities. The state of Michigan has passed legislation that allow communities the right to grant manufacturing and commercial property abatements. Michigan's Public Act 198, enacted July 9, 1974 and titled "Michigan's Plant Rehabilitation and Industrial Employment Act,"35 was the first of this legislation. In effect today, a manufacturing property tax abatement can be granted at a community's discretion 34A possible unique characteristic of the Metropolitan Detroit Area is the way in which suburban residents shun their central city. This relationship will show up in some of the regression results reported in the next section. 35For a description of P.A. 198, see Wolkoff (1982) and Michigan Department of Commerce (1983). 91 for up to 12 years. For a new manufacturing facility the abatement is equal to one-half the normal property assessment. For a rehabilitated manufacturing facility the property assessment is frozen at the level before rehabilitation. Firms granted a manufacturing property abatement pay what is called an "Industrial Facilities Tax" instead of the standard property tax. Public Act 255 was later enacted on June 21, 1978 and titled "Michigan's Commercial Redevelopment Act."36 Revoked on December 31, 1985 this legislation was the commercial equivalent of P.A. 198. Firms granted a commercial property abatement pay what is called a ”Commercial Facilities Tax" instead of the standard property tax. In regards to the Michigan experience of communities granting property tax abatements, Wolkoff (1982, p. 289), observed: The flexibility and discretion that the [Michigan abatement] law allows have for the most part gone unused. Abatement applications are virtually always approved for the maximum period permitted by law. Wolkoff also observed that 84 percent of Michigan abatements went to new facilities, as opposed to the rehabilitation of existing facilities. An exception to this was the city of Detroit where abatements have predominantly gone for rehabilitation projects. 36For descriptions of P.A. 255 see Wolkoff (1982), Michigan Department of Commerce (1984), Michigan Department of Treasury (1984), or Citizen's Research Council (1986). 92 A panel data set was gathered because a single cross section of Metropolitan Detroit Communities resulted in too small a data set.37 The panel consists of three pooled cross sections beginning in 1977 and continuing in five year increments up to 1987. The choice of this period was defined by the availability of sector specific (manufacturing, commercial, housing) property values in Michigan and the tenure of Michigan's property abatement legislation. Five years is believed to be a period long enough to allow for substantial adjustment in local property and nonresidential property abatement values. The Detroit Metropolitan Area is defined as communities within Macomb, Oakland, and Wayne counties.38 The availability of data limited the metropolitan sample to communities with populations greater than 10,000 in 1980.39 The result is a sample of 47 communities for each of the three cross sections (1977, 1982, and 1987) or a total panel of 141 observations. The appropriateness 37For a description of panel data sets see Maddala (1977) pp. 322-331, or Pindyck and Rubinfeld (1981) pp. 252-261. 38This is the 1970 U.S. Census definition of the Detroit Standard Metropolitan Statistical Area (S.M.S.A.). An S.M.S.A. is the appropriate definition because it includes the central city and surrounding communities that are economically linked to the central city. Because data is drawn from the 1970's, the 1970 definition was chosen over the expanded 1980 definition which also included Lapeer, Livingston, and St. Clair counties. 39This resulted in a sample containing 65 percent of the possible communities. For a list of excluded communities, see Appendix B. 93 of pooling these regressions was tested. As described in the next section some cross-section dummies were added. To this point it has been sufficient to generally define the variable ”R” as units or dollars of local expenditure per capita. In the empirical analysis local expenditure per capita is calculated as expenditure on police, fire, sewer, sanitation, highway, and education divided by population. Because school districts overlap community borders in the Metropolitan Detroit Area, local education expenditure is calculated based on the percentage of the school districts tax base in the community. The value of local firm services (f) will be measured as local expenditure on police, fire, sewer, sanitation, and highway divided by the total local property tax base. This was chosen because the nature of these services makes it impossible with available data to separate the amount of each service going to residential and firm users. The value of the median voter's house and his income are measured as the median values observed in the community. Because observations were desired from 1977, 1982, and 1987, and some socio-economic variables were not available for these years, it was necessary to extrapolate some values from the 1970 and 1980 U.S. Census. All nominal dollar values have been placed in 1972 real dollars using the Detroit Consumer Price Index. Because the commercial property abatement law was not in effect, 94 the variable "A" consists of all zero observations for 1977. Table 1 contains definitions for all variables used in the regression estimation. Descriptive statistics relating to these variables are reported in Table 2. Appendix A contains a listing of the Metropolitan Detroit Communities in the data set and the value of all variables. Appendix B contains a listing of communities in Macomb, Oakland, and Wayne counties excluded from the data set. Appendix C contains a listing of formulas used to derive all variables. Each variable's source is reported in Appendix C. Endogenous Variables Exogenous Variables aget 95 Table 1 Variable Definitions Description value of commercial property abatement value of manufacturing property abatement value of commercial property value of housing property value of manufacturing property population value of local expenditure per capita median value owner-occupied house local property tax rate value of firm services per dollar property local income tax revenue per dollar property number major crime offenses population density value of commercial property density value of housing property density value of manufacturing property density percentage of residents employed locally five year change in "C" five year change in "M" Description median income grants and user charges per dollar property average "Y" in surrounding communities central city distance dummy if 1977 cross section dummy if 1982 cross section dummy if 1987 cross section percentage housing with more than one bath miles of divided expressway dummy if adjacent to major water five year lagged "A" five year lagged "B" five year lagged "C" five year lagged "H" five year lagged "M" median age Exogenous Variables 96 Table 1 (cont.) Variable Definitions Description dummy if city of Detroit dummy if in Macomb County dummy if in Oakland County dummy if in Wayne County square miles commercial agglomeration economies median years education percentage housing owner occupied manufacturing agglomeration economies percentage residents nonwhite average pop. density surrounding communities average median age in surrounding communities 97 Table 2 Descriptive Statistics Coefficient Variable Mean of Variation4o Maximum At 551,847 4.95 26,608,190 (49)"1 (1,587,967)42 (2.82) (26,608,190) Bt 10,496,705 2.56 143,873,700 (77) (19,221,238) (1.77) (143,873,700) ct 54,043,459 1.74 689,189,800 Ht 144,990,176 1.51 1,770,490,521 Mt 63,144,499 1.90 859,108,200 (135) (65,950,922) 1.85 859,108,200 Nt 64,086 2.63 1,285,351 Rt 359 0.27 693 vt 23,473 0.43 50,307 pt 0.065 0.13 0.090 ft 0.027 0.70 0.174 1t 0.004 4.75 0.151 (12) (0.052) (0.88) (0.151) cmt 5,580 3.56 152,962 pdent 5,003 0.41 10,537 cdent 4,258,909 0.58 14,929,390 hdent 14,238,877 0.60 42,270,760 mden 5,199,615 1.36 41,783,870 (135) (5,430,709) (1.32) 41,783,870 Minimum (17,396) 0 (19,348) 2,313,006 12,359,040 0 (51,454) 8,999 155 5,649 0.049 0.007 0 (0.018) 323 533 822,703 1,190,061 0 (16,079) 4oCalculated by dividing the standard deviation by mean. 41Number of non-zero values in previous variable. 42Calculated for only nonzero values of previous variable. Variable (79) wnt (33) (21) 98 Table 2 (cont.) Descriptive Statistics Mean 22.381 -5,565,694 -12,372,648 -12,922,544) 12,701 0.033 12,639 13.234 0.400 3.600 (6.427) 0.234 (1.000) 162,133 (1,088,610) 4,662,221 (16,434,328) 59,609,153 156,977,075 75,517,147 (78,873,467) 32.312 31.306 13.429 4,284,239 Coefficient of Variation Maximum Minimum ""332” 2:333- "$333 -8.17 88,905,170 —473,188,100 -5.34 89,409,830 -665,507,700 (-5.22) (89,409,830) (-665,507,700) 0.27 22,945 5,156 0.99 0.232 .008 0.16 20,787 8,014 0.37 25 4 0.46 0.870 0.070 2.51 61.600 0 (1.77) (61.600) (1.000) 1.81 1 0 (0.00) (1) (1) 8.74 16,717,450 0 (3.33) 16,717,450 (30,198) 3.68 110,290,900 0 (1.78) (110,290,900) (19,348) 2.12 1,162,377,885 2,423,922 1.84 2,596,522,245 13,845,440 2.27 1,524,615,828 0 2.21 (1,524,615,828) (53,640) 0.16 53.910 24.720 0.11 48.47 25.770 1.58 136.800 2.050 0.26 6,859,879 2,444,428 99 Table 2 (cont.) Descriptive Statistics Coefficient Variable Mean of Variation Maximum educt 12.760 0.08 17.700 hownt 72.718 0.17 96.320 maggt 4,853,854 0.53 12,831,450 racet 9.094 2.17 98.760 (123) 10.424 (1.99) (98.760) pdenst 5,082 0.22 7,900 Minimum 32.720 896,177 0 0.010 1,898 100 III. C. REGRESSION RESULTS The regressions were initially run as specified in equations 64 - 66, 73, and 80 - 85. The constant term in each regression was replaced with dummy variables for each of the three cross sections (d77, d82, d87), each of the three counties (dmac, doak, and dway), and for the city of Detroit (ddet).43 These dummies allowed for the control of year, county, and central city specific influences on each of the regression's intercept. With one exception there is no specific theory in regards to whether these dummies should or should not be included.44 Dummies were only included if they exerted a statistically-discernable effect on the regressions. If the Detroit dummy was not statistically significant at a 85 percent or better confidence level it was dropped. If the year or county dummies were not jointly statistically significant at a 85 percent confidence level they were also dropped.45 43It was possible to dummy all classes since a constant was not used in regression estimation. 44One exception is the local property tax regression where county dummies should be included. The reason to include them is that the local property tax rate used includes a county-specific levy that varies between Macomb, Oakland, and Wayne Counties. 45The F-test described in Pindyck and Rubinfeld (1981), p. 255 was used to test the joint statistical significance of the cross section and county dummies. The lower than generally used confidence level was chosen to minimize the chance of committing a "type II error" - not including a set of dummy variables when they should be included. The seriousness of this error was considered greater than a (continued...) 101 In regression estimation of this nature heteroskedasticity was expected to be a problem. A ”Park test” was used to test for the presence of heteroskedasticity in each of the regressions.46 If heteroskedasticity was detected, all observations were weighted by the variable that exhibited the largest t- statistic in the Park tests. A second Park test was then performed to insure that heteroskedasticity had been corrected. The two-stage regression results for each of the ten simultaneous equations are reported in Tables 3 - 12. Standard errors are recorded in parenthesis below the estimated coefficients. The elasticities are calculated using means. The "2-Tail Significance" refers to the two-tailed probability that the estimated coefficient is not significantly different than zero. One asterisk has been placed next to the calculated value if the estimated coefficient was statistically significant at the 85 - 99 percent confidence level. Two asterisks has been placed next to the calculated value if the estimated coefficient was statistically significant at the 99 percent or larger confidence level. 45(...continued) "type I error" - including a set of dummy variables when they should not be included. 46For a description of heteroskedasticity and the problems it causes, see Gujarati (1988), ch. 11. The ”Park Test” and the corrective method of weighted-least squares are also described in this chapter. 102 Table 3 Median Home Value Regression Dependent Variable: Vt Number of Observations: 141 No heteroskedasticity detected -- 2-stage least squares Independent ‘ 2-Tail Variable1 61 Elasticity1 Significance1 d77 3.0902 D+03 ----- 0.303 (3.0011 0+03) d82 2.9428 0+03 ----- 0.353 (3.1689 0+03) d87 1.5596 0+03 ----- 0.598 (2.9589 0+03) ddet 4.6990 0+04 ----- 0.023 (2.0674 0+04) dmac 3.1582 D+04 ----- 0.039 (1.5272 0+04) doak 3.3779 0+04 ----- 0.031 (1.5770 0+04) dway 3.2954 D+04 ----- 0.033 (1.5417 0+04) mdent -2.1370 0-04 -0.047 0.071* (1.2000 0-04) cdent -5.4470 0-04 -0.098 0.049* (2.8000 0—04) hdent 1.7140 0-04 0.103 0.017* (0.7200 0-04) hownt -2.1320 0+02 -0.660 0.000** (0.4951 0+02) cmt -4.4390 0-01) -0.106 0.004** (1.5335 0-01) racet -7.0400 0+01 -0.027 0.005* (2.5011 0-01) Yt 5.4177 0-01 0.293 0.149** (3.7520 0-01) hct 2.9712 0+04 0.506 0.000* (0.3928 0+04) cdt 1.8202 D+02 0.103 0.052* (0.9360 0+02) hnt 2.0114 D+02 0.031 0.141* (1.3667 0+02) pt -3.1852 0+05 -O.882 0.010** (1.2293 0+05 1t 7.9536 D+03 0.001 0.847 (41.0868 0+03) R 1.5704 0+01 0.239 0.020* (0.6730 0+02) R-squared a 0.931 F-statistic = 86.46 103 As recorded in Table 3, all three sets of dummy variables had a statistically-significant impact on the median home value in the average Metropolitan Detroit Community. Holding all other independent variables constant, the median value of the average Metropolitan Detroit home declined from 1977 to 1987. The biggest decline occurred between 1982 and 1987. An Oakland County location contributed most to home value. A Macomb County location contributed least. Qete21§_fietibn§, relative to other communities, a city of Detroit location contributed positively to home value. Twelve of the remaining thirteen non-dummy independent variables included in Table 3 were statistically significant.47 Community characteristics, as measured by local manufacturing (mden) and commercial property density (cden) reduced the median value of a community's homes. This supports the earlier theoretical assumption that an increase in a community's manufacturing and commercial property decreases its environmental quality, which decreases median voter's utility, and subsequently decreases the demand for, and the median value of homes in the community. The result that an increase in housing 47Throughout the interpretation of the regression results an 85 percent confidence level will be used to assess statistical significance. This was done because the results are used in a simulation analysis presented in Section E of this chapter. As described earlier, the possibility of committing a "type II error" in the simulation analysis was considered worse than committing a "type I error”. 104 property density (hden) increases the median value of homes does not support the earlier theoretical assumption that housing property reduces environmental quality. This possibility was noted earlier. It appears that Detroit area home buyers prefer to live in communities with a larger housing property density and bid up the median value of homes in such a community. Neighborhood characteristics measured by percentage of housing owner occupied (hown), number of crimes (cm), and percentage of nonwhite residents (race) decreased the median value of local homes. At first glance the sign on "hown" is somewhat surprising. An explanation may be that "hown" acts as a negative proxy as to the attractiveness of local owner-occupied homes for investment. Individual consumers purchase homes for consumption and investment purposes. Landlords purchase homes for purely investment purposes. Housing purchase for investment, as opposed to consumption, is done with greater consideration for rate of return. A higher rate of return is positively capitalized into the median value of a community's owner- occupied homes.48 The "hown" elasticity of median home value was the second largest. A one percent increase in the percentage of residents who own their home would reduce 48A second possible explanation is to consider two neighborhoods with identical distributions of housing values. Assume that the lowest value homes in each neighborhood are rental units. The measure of median home value excludes rental units. When comparing these two identical neighborhoods, the one with fewer rental units will have a lower median home value. 105 median home value by .66 percent. The negative coefficients on ”cm" and "race” are not surprising. In the eyes of Metropolitan Detroit home buyers, number of crimes and percentage of residents nonwhite are negative neighborhood characteristics. Median income (Y) is the final neighborhood characteristic included in the median home value regression. As expected there is a positive relationship between median income and median home value. Only people with high incomes can purchase expensive homes. Home buyers also prefer houses next to higher income neighbors. As expected the greater the percentage of a community's housing with more than one bath (he), the greater the median value of a home in the community. In regards to accessibility characteristics, miles of divided highway (hn) and distance from Detroit's central business district (cd) exerted a positive influence on median home value. According to standard urban theory the sign on "cd" is surprising. Understanding Metropolitan Detroiters' anomalous relationship with their central city, the negative sign is not surprising at all. The most interesting results reported in Table 3 come from the independent fiscal variables. As Tiebout (1956) predicted, and Oates (1969) and others have verified, local property taxes and expenditures were capitalized and accordingly lowered and raised the median value of local homes. The property tax elasticity was the largest 106 recorded. A one percent increase in the local property tax would decrease the median home value .88 percent. A one percent increase in local expenditure per capita would increase the median home value .24 percent. A concurrent one percent increase in local property taxes, to fund a one percent increase in local expenditure per capita, would on average decrease the median value of a local home.49 In the Metropolitan Detroit Area local income taxes had a statistically-insignificant impact on local home values. 49A possible explanation for this is that a property tax increase would directly affect all homeowners, whereas an increase in local expenditure may not affect homeowners if it is not spent on something they use. To check this explanation, different types of local expenditures could be substituted for total expenditure. I plan to do this in future research. 107 Table 4 Effective Local Income Tax Regression Dependent Variable: 1t Number of Observations: 141 No heteroskedasticity detected -- 2-stage ”Tobit" Normalized Independent ‘ ‘ Variable1 £1 Elasticity1 61 T-Ratio constant1 -5.8371 D-01 ----- -2.1721 D+01 -4.634 (0.4687 0+01) R1 -5.6712 D-05 -5.090 -2.1104 D-03 -0.755 (2.7960 0-03) B1 8.1849 0-10 2.148 3.0458 D-08 2.413** (1.2621 D-08) A1 -2.2426 D-09 -0.309 -8.3452 D-08 -1.041 (8.0153 D-08) N1 1.9963 D-08 0.319 7.4285 D-07 0.844 (8.7959 0-07) p1 7.4693 121.376 2.7795 D+02 4.316** (0.6440 0+02) g1 3.6817 D-01 3.037 1.3701 D+01 2.115* (0.6479 D+01) R-squared = 0.896 108 The regression results recorded in Table 4 were derived from a "Tobit" maximum-likelihood procedure.50 This was the only regression where this estimation technique was used. The reason for its use was that only 4 of 47 communities in the sample (Detroit, Hamtramck, Highland Park, and Pontiac) levied a local income tax. The income tax variable must be considered a limited- dependent variable whose determination is based on two decisions: (1) whether to levy a local income tax, and (2) the rate at which it should be levied. This makes Tobit the appropriate estimation process. As recorded in Table 4, none of the dummy variables had a statistically-significant impact on the effective local income tax rate. Three of the six independent variables included in the local income tax regression were statistically significant. The signs on the independent variables' coefficients could not be predicted in advance. The regression results showed that communities offering relatively large manufacturing property abatements (B) are likely to have large rates of effective local income tax. Perhaps this is due to the erosion of property tax base caused by greater property abatements. A one percent increase in manufacturing property abatements would result in a 2.15 percent decrease in the rate of local income tax revenue . 50For a description of Tobit, see Kmenta (1986), pp. 560- 564. The Tobit estimation was conducted using mainframe "Shazam”-Version 6.0. 109 The regression results also show a positive relationship between local property and effective local income tax rates. A one percent increase in the local property tax rate would result in a large 121 percent increase in the rate of effective local income tax. I believe this large elasticity can be explained by the fact that most communities adopt a local income tax after their property tax rate hits state or median voter mandated limits. The percentage increase in local income tax revenue from going from no local income tax to having one is infinite. Hence, the large elasticity caused by an increase in the property tax rate. In Table 4, the rate of grants/user charges had a positive impact on the effective rate of local income tax. This is most likely due to the chosen sample and reflects characteristics of the communities levying a local income tax. 110 Table 5 Local Property Tax Rate Regression Dependent Variable: pt Number of Observations: 141 No heteroskedasticity detected -- 2-stage least squares R-squared = 0.477 F-statistic Independent ‘ 2-Tai1 Variable1 61 Elasticity1 Significance1 ddet ~1.9251 0-01 ----- 0.000 (0.4760 0-01) dmac 5.5524 0-02 ----- 0.000 (0.4209 0—02) doak 5.5389 0-02 ----- 0.000 (0.3700 0-02) dway 5.6575 0-02 ----- 0.000 (0.3710 0-02) Mt -4.9300 0-11 -0.047 0.001** (1.5250 0-11) ct -6.9280 0-12) -0.006 0.657 (15.6200 0-12) Ht -9.3930 0-11 -0.210 0.000** (1.8120 0-11) Rt 2.5710 0-05 0.142 0.009** (0.9774 0-05) Bt 7.5060 0-11 0.012 0.122* (4.8530 0-11) At -2.1390 0-09 -0.018 0.000** (0.6055 0-09) Nt 3.1540 0-07 0.311 0.000** (0.6313 0-07) 1t 1.6686 0-01 0.010 0.037* (0.7999 0-01) gt 3.3065 0-02 0.017 0.237 (2.7968 0-02) 9.75 111 As reported in Table 5, the county dummies and the city of Detroit dummy had a statistically-significant impact on local property tax rates. The county dummies were expected to be significant because the measure of the local property tax rate contains a county specific levy. From 1977 to 1987, this levy was highest in Wayne county and lowest in Oakland County. Qete:1e_ne;1nne, the city of Detroit had lower property tax rates than other communities in the sample. Seven of the remaining nine independent variables included in Table 5 were statistically significant. The signs of the coefficients on the property base variables (C, M, and H) could not be predicted in advance. The signs of the coefficients on "M" and "H" were negative in the regression analysis. A one percent increase in the average community's manufacturing property base would decrease their local property tax rate .05 percent. A one percent increase in the housing property base would decrease the local property tax rate .21 percent. The housing elasticity is the second highest recorded. Perhaps a reason for the difference in the size of the elasticities is that a community provides greater services after a one percent in manufacturing property, than after a one percent increase in housing property. The expense of greater firm services mitigates the effect an increase in the manufacturing property base has on lowering property tax rates. 112 As predicted in the theoretical model, a one percent increase in local expenditure (R) would increase the local property tax rate. Also as predicted, a one percent increase in manufacturing property abatements (B) would increase the local property tax rate. Opposite to what was predicted in the theoretical model, an increase in commercial property abatements (A) would decrease the local property tax rate. An increase in population (N) would increase the local property tax rate. This was predicted in theoretical model. The population elasticity was the largest. A one percent increase in population would increase the local property tax .31 percent. Communities with a higher rate of local income tax (1) were shown to have a higher property tax rate. This complementarity was not predicted in the theoretical model. A possible explanation for it was given after Table 4. In comparing the statistically-significant regression results for the local income tax (Table 4) and for the local property tax (Table 5), local expenditure per capita only exerted a positive influence on the property tax rate. The coefficients on the manufacturing property abatement variable were positive in both regressions. The manufacturing property abatement elasticity of the property tax rate is much smaller than the manufacturing property abatement of the local income tax rate. Commercial property abatements only exerted a negative influence on 113 local income tax rates. Population only exerted a positive influence on local property tax rates. Again, it is interesting to note the complimentarity between the local property and income tax instruments. The property tax elasticity of the local income tax was ten thousand times greater than the local income tax elasticity of the property tax. Again, the reason for this is the cycle of local income tax adoption described earlier. The rate of grants/user charges only exerted a positive influence on the local income tax. 114 Table 6 Percentage of Residents Employed Locally Regression Dependent Variable: lempt No heteroskedasticity detected -- 2-stage least squares Independent Variable1 £1 1.9987 (0.1569 1.9505 (0.1458 1.9203 (0.1418 1.2602 (0.3948 1.2490 (0.1829 4.4820 (1.6240 -1.6180 (0.4436 0+01 0+01) 0+01 0+01) 0+01 0+01) 0+02 0+02) 0-07 0-07) D-08) D-08) 0-04 0-04) R-squared = 0.583 Number of Observations: 141 2-Tail Elasticity1 Significance1 ""IIIIZ" """3'333" ----- 0.000 ----- 0.000 ----- 0.001 0.352 0.000** 0.108 0.006** -0.463 0.000* F-statistic = 31.22 115 As recorded in Table 6, only the cross-section dummy variables had a statistically-significant impact on the percentage of residents employed locally. Holding other explanatory variables constant, the percentage of residents employed locally was greatest in 1977, and continually declined through 1987. The coefficient on the Detroit dummy was positive and indicated, eetetie_netinne, that a larger percentage of Detroit residents were employed locally. I All of the remaining non-dummy independent variables included in the percentage of residents employed locally regression were statistically significant. The calculated signs on the independent variables' coefficients match the predictions made in the theoretical model. Both the manufacturing (M) and commercial property base (C) exerted a positive influence on the percentage of residents employed in Detroit area communities. The manufacturing property elasticity was the three times larger than the commercial property elasticity. If the median voter desires to attract nonresidential property for employment reasons, these results show he would do better to attract manufacturing property. The regression results also showed that a one percent increase in local population would result in a 4.63 percent decrease in the percentage of residents employed locally. This was the largest elasticity-calculated. 116 Table 7 Median Voter's Demand For Manufacturing Property Regression Dependent Variable: Mt Number of Observations: 141 Heteroskedasticity detected -- 2-stage l.s. weighted by "pden" Indep. ‘ S-R ‘ L-R 2-Tail Var.1 S-R B1 E1as.1 L-R £1 E1a8.1 Signif. d77 -3.5191 0+07 --------------- 0.265 (3.1546 0+07) d82 -3.0638 0+07 --------------- 0.341 (3.2167 0+07) d87 -2.6052 D+07 --------------- 0.422 (3.2423 0+07) dmac 1.5043 D+08 --------------- 0.003 (0.5024 D+08) doak 1.4809 D+08 --------------- 0.004 (0.5094 D+08) dway 1.5210 D+08 --------------- 0.003 (0.5157 D+08) At 5.6754 0.050 1.4552 D+01 0.128 0.007** (2.1221) Bt 5.0226 0-01 0.083 1.2878 0.212 0.000** (1.0679 0-01) Nt —3.9268 0+02 -0.399 -l.0068 D+03 -1.023 0.000** (0.7859 0+02) Vt 9.6675 0+02 0.359 2.4788 0+03 0.921 0.085* (5.6191 0+02) Yt -l.4639 0+03 -0.294 -3.7535 0+03 -0.754 0.500 (2.1692 0+03) 1t 5.8403 D+08 0.037 1.4975 0+09 0.095 0.117* (3.7303 D+08) t -1.0085 0+07 -0.005 -2.5859 0+07 -0.013 0.929 (11.2967 0+07) aget -7.4371 0+05 -0.381 -l.9069 D+06 -0.977 0.082* (4.2720 0+05) educt -9.0753 D+06 -1.834 -2.3270 D+07 -4.703 0.003** (3.0929 D+06) areat -8.524O 0+04 -0.018 -2.1856 0+05 -0.046 0.649 (18.7041 0+04) ct -l.1186 0-01 -0.096 -2.8682 0-01 -0.246 0.004** (0.4070 0-01) Ht 2.6200 0-01 0.602 6.7179 0-01 1.544 0.000** (0.4623 0-01) lempt 8.1651 0+05 0.289 2.0936 D+06 0.741 0.010** (3.1858 0+05) Mt- 0.6093 --------------- 0.000** (0.0293) R-squared a 0.979 6M (coefficient of manufacturing F-statistic = 292.67 property adjustment) - .39 117 For the median voter's demand for manufacturing property regression only the year and county dummies were statistically significant. Holding the remaining independent variables constant, the demand for manufacturing property was greatest in 1987 and least in 1982. Manufacturing property demand was greatest in Wayne County and least in Oakland County. The coefficient of manufacturing property adjustment (6”) was calculated to be .39. Over a five year period it was estimated that an average community in the Detroit Metropolitan Area moved 39 percent of the total way towards its median voter's desired manufacturing property base. Based on this statistic, eetetie_netitn§, in a given year it would take approximately a 30 year adjustment process for the average Metropolitan Detroit Community to reach 95% of the median voter's desired manufacturing property base. Eleven of the fourteen remaining non-dummy independent variables were statistically significant. Because of income and substitution effects, the signs of the regression coefficients on "A", ”B", ”N”, ”V”, ”Y”, ”1", and ”g” could not be predicted in advance. The regression results show that a positive relationship existed between c0mmercial property abatements (A), manufacturing property abatements (B), median value of local homes (V), local income taxes (1) and the median voter's demand for manufacturing property. The commercial and manufacturing 118 property abatement elasticities were very similar. A one percent increase in commercial and manufacturing property abatements would increase the average median voter's demand for manufacturing property .05 and .08 percent in the short run, and .13 and .21 percent in the long run.51 The regression results also show that a one percent increase in population (N) would decrease the median voter's demand for manufacturing property .40 percent in the short run and 1.02 percent in the long run. This elasticity was the third largest. The predictions derived from the theoretical model in regards to ”age”, "educ", ”area", "C", "H”, and "lemp" were that variables expected to have a positive influence on median voter's taste for environmental quality (a) and local expenditure (0) should have a negative and positive influence on median voter's demand for manufacturing property. The regression results show that an increase in median age (age) and median education (educ) would decrease the median voter's demand for manufacturing property. The education elasticity was the largest. A one percent increase in a community's median education would decrease the median voter's demand for manufacturing 51The short-run elasticities reported in Table 7 are the percentage change in manufacturing property demand that occurs after five years, after a one percentage change in the respective independent variable. Ninety-five percent of the long-run elasticity can be considered the percentage change in the dependent variable after 30 years, after a one percent change in the respective independent variable. 119 property 1.83 percent in the short run and 4.70 percent in the long run. The regression results also show that an increase in existing commercial property would decrease the demand for manufacturing property. An increase in existing housing property would increase the demand for manufacturing property. The housing property elasticity is the second largest. A one percent increase in housing property would increase manufacturing property demand .60 and 1.54 percent in the short and long run. Perhaps a reason for the differing signs on the commercial and housing property coefficients is that a median voter considers commercial and manufacturing property ”bad”, and housing property "good" when making utility calculations.52 The regression results show that if the median voter experiences an increase in the "bad" property tax base (commercial), he demands less of the other ”bad“ property base (manufacturing). If the median voter experiences an increase in the ”good" property base (housing), he is willing to accept more of one of the other "bad" property bases (manufacturing). The percentage of residents employed locally (lemp) exerted a positive influence on the average Metropolitan Detroit median voter's demand for manufacturing property. 52Evidence for this proposition was also reported in Table 3 where local commercial property density, manufacturing property density, and housing property density respectively decrease, decrease, and increase the median home value. 120 A one percent in increase in "lemp" would result in a .29 and .74 percent increase in manufacturing property demand in the short and long run. 121 Table 8 Median Voter's Demand For Commercial Property Regression Dependent Variable: Ct Number of Observations: 141 Heteroskedasticity detected -- 2-stage l.s. weighted by R-squared = 0.964 II pden N Indep. Var.1 S-R B1 ddet -5.8232 D+08 (2.7623 D+08) dmac 1.0925 D+08 (0.4850 D+08) doak 1.0122 D+08 (0.4830 D+08) dway 1.0401 D+08 (0.4841 D+08) At 9.2020 (1.8552) 3t -l.4124 D-01 (1.2527 D-Ol) Nt 1.3714 D+02 (3.7419 D+02) Vt 2.0779 D+03 (0.4130 D+03) Yt -1.9185 D+03 (0.6979 D+03) 1t 1.5250 D+09 (0.3998 D+09) 9t -3.1594 D+07 (1.2187 D+07) aget 7.4366 D+05 (4.9973 D+05) educt -l.2362 D+07 (0.3453 D+07) areat -1.8288 D+05 (1.9815 D+05) Mt 1.1609 D-02 (5.8012 D-02) Ht 4.1258 D-03 (111.8501 D-03) lempt -3.7487 D+04 (35.8315 D+04) Ct-l 0.9406 (0.0385) S-R 0.094 -0.027 0.163 0.903 -0.451 0.113 -0.019 0.445 -2.919 -0.045 0.014 -0.011 -0.016 1.5337 -2.3540 2.2857 3.4632 -3.1975 2.5417 -5.2656 1.2394 -2.0603 -3.0480 1.9345 -6.8763 -6.2478 D+03 D+05 D+04 D+10 D+08 D+07 D+08 D+06 D-01 D-02 D+05 L-R 2-Tail Elas.1 Signif. 1.567 -0.450 2.717 15.050 -7.517 1.883 0.317 7.417 -48.650 -0.750 0.233 -0.183 0.267 0.260 0.714 0.000** 0.006** 0.000** 0.795 0.137* 0.000** 0.356 0.841 F-statistic - 194.16 6C (coefficient of commercial property adjustment) = .06 122 As reported in Table 8, only the county dummies and the city of Detroit dummy had a statistically-significant impact on the average median voter's demand for commercial property in the Metropolitan Detroit area. getetie perigee, from 1977 to 1987 the demand for commercial property was least in Macomb County and greatest in Oakland County. Holding other independent variables constant, median voters in the city of Detroit had a lower demand for commercial property. The coefficient of commercial property adjustment was .06. Over a five year period the average community in the Detroit Metropolitan Area moved 6 percent of the total way towards its desired commercial property base. Based on this statistic, a thirty year adjustment process would result in the average community reaching approximately 27 percent of its desired commercial property base. To reach 95 percent of its desired commercial property base, it would take approximately 225 years. This slow adjustment must be considered when interpreting the large long-run elasticities in Table 8. Seven of the remaining fourteen independent variables included in the median voter's demand for commercial property regression were statistically significant. Because of income and substitution effects the signs of the coefficients on "A", "B", "N", "V”, "Y”, "1", and "g" could not be predicted in advance. The regression results show that an increase in commercial property 123 abatements (A), median home value (V), and the local income tax rate (1) would increase the median voter's demand for commercial property. The median home value elasticity of commercial property demand was the second largest. A one percent increase in ”V” would increase commercial property demand .90 percent in the short run and 15.05 percent in the long run. The regression results also show that an increase in median income (Y) would all decrease the median voter's demand for commercial property. The predictions derived from the theoretical model in regards to "age”, "educ", "area", ”M", "H", and ”lemp" were that variables expected to have a positive influence on median voter's taste for environmental quality (a) and local expenditure (0) should respectively have a negative and positive influence on median voter's demand for commercial property. The regression results showed that median age (age) exerted a positive influence on commercial property demand, while median education (educ) exerted a negative influence. The education elasticity was the largest calculated. A one percent increase in local median education would decrease commercial property demand 2.92 percent in the short run and 48.65 percent in the long run. Recall that the long-run change would take over 225 years to complete. 124 Table 9 Median Voter's Demand For Housing Property Regression Number of Observations: 141 2-stage l.s weighted by ”pden" Dependent Variable: Ht Heteroskedasticity detected -- Indep. ‘ S-R ‘ L-R 2-Tail Var.1 S-R B1 Elas.1 L-R B1 Elas.1 Signif. ddet -1.5920 D+09 --------------- 0.000 (0.2224 D+09) At 4.5017 0.017 1.8007 0+01 0.068 0.056* (2.3514) at -2.3939 0-01 -0.017 -9.5756 0-01 -0.068 0.230 (1.9934 0-01) Nt 1.3825 D+03 0.611 5.5300 D+03 2.444 0.000** (0.3482 0+03) Vt 2.9230 D+03 0.473 1.1692 0+04 1.892 0.000** (0.4284 0+03) Yt -2.6702 0+03 1-0.234 -1.0681 0+04 -0.936 0.009** (1.2542 D+03) 1t 2.0060 D+09 0.055 8.0240 D+09 0.220 0.002** (0.6351 D+09) gt -5.4299 D+08 -0.124 -2.1720 D+09 -0.496 0.000** (1.4875 D+08) aget 1.7053 D+05 0.038 6.8212 0+05 0.152 0.773 (5.9187 D+05) educt -4.1632 D+06 -0.366 -1.6653 0+07 -1.464 0.034* (1.9639 D+06) areat 3.2451 D+04 0.003 1.2980 D+04 0.012 0.891 (23.6478 D+04) ct -2.4458 0-01 -0.091 -9.7832 0-01 -0.364 0.005** (0.8798 0-01) Mt -1.2120 0-01 -0.053 -4.8480 0-01 -0.212 0.215 (0.9775 0-01) lempt 6.6012 0+05 0.102 2.6405 D+06 0.408 0.272 (6.0106 D+05) Ht_1 0.7529 --------------- 0.000** (0.1111) R-squared = 0.975 F-statistic = 346.03 5c (coefficient of housing property adjustment) = .25 125 As shown in Table 8, only the city of Detroit dummy had a statistically-significant impact on median voter's demand for housing property. Holding other independent variables constant, median voters in the city of Detroit had a lower demand for housing property. The coefficient of housing property adjustment was calculated to be .25. Over a five year period an average community in the Detroit Metropolitan Area moved 25 percent of the way towards its desired housing property base. Based on this statistic, a thirty year adjustment process would result in the average Metropolitan Detroit community moving approximately 88 percent of the way toward its desired housing property base. It would take approximately 45 years for the community to move 95 percent of the way toward its desired housing property base. Again, this must be considered when interpreting the long-run coefficients. Nine of the remaining fourteen non-dummy independent variables included in the median voter's demand for manufacturing property regression were statistically significant. Again, because of income and substitution effects it was impossible to predict the signs on "A", ”B”, "N”, "V", "Y”, "1", and "g” in advance. The regression results show that an increase in commercial property abatements (A), population (N), median home value (V), and local income taxes (1) would increase the median voter's demand for housing property. The population and median 126 home value elasticities of housing property demand were the first and second largest. A one percent increase in population would increase housing property demand .61 percent in the short run and 2.44 percent in the long run. A one percent increase in median house value would increase housing property demand .47 percent in the short run and 1.89 percent in the long run. The regression results also show that an increase in median income (Y) would decrease the median voter's demand for housing property. Again, in regard to "age", "educ", "area", "C", "M", and ”lamp", the predictions derived from the theoretical model were that variables expected to have a positive influence on median voter's taste for environmental quality (a) and local expenditure (0) would respectively have a negative and positive influence on median voter's demand for housing property. The regression results show that median education (educ) and existing commercial property (C) exerted a negative influence on the median voter's demand for housing property. The education elasticity was the third largest. A one percent increase in median education would decrease median voter's demand for housing property .37 percent in the short run and 1.46 percent in the long run. The negative sign on existing commercial property supports the hypothesis put forth earlier. An increase in the "bad" property base (commercial) decreases the demand for any of the other property bases (housing). 127 Comparing the coefficients of property base adjustment reported in Tables 7, 8, and 9, ”6M" (.39) was greater than "‘H" (.25), which was greater than ”5C" (.06). These concur with Ladd's (1976) finding that a community's manufacturing property base is more footloose than its commercial property base. A theoretical justification for this is that manufacturing firms are less tied to a market within a metropolitan area than commercial firms. They are therefore freer to move in and out of a community. Ladd and Bradbury (1988) calculated a five-year coefficient of total property base equal to 46 percent for a similar period. This larger figure was most likely due to the fact that the data set used to calculate it included large 0.8. cities with 1970 populations greater than 300,000. More dynamic growth occurred in these cities, than cities in the Detroit Metropolitan Area. In comparing the statistically-significant property base results reported in Tables 7, 8, and 9, the coefficient on the commercial property abatement variable was positive in all three regressions. The calculated short-run commercial property abatement elasticity was greatest for commercial property and least for housing property. In the same tables, the manufacturing property abatement variable was only positive in the demand for manufacturing property regression. An increase in population would result in a decrease in manufacturing property demand and an increase in housing property 128 demand. The reasons for these differing signs could be that an increase in population increases the marginal environmental damage from manufacturing property which causes the median voter to demand less of it. On the other hand, an increase in population could also increase the demand for commercial firms, for the increased population to purchase goods and services at. In Tables 7, 8, and 9, median home value increased the median voter's demand for all three components of the property tax base. As the theoretical model showed, this is attributable to the fact that median voters with a greater home value face a greater marginal tax price for increased local expenditure (see equation 9). An increase in any of the property tax bases lowers this marginal tax price. Median income exerted a negative influence on the median voter's demand for commercial and housing property. The short-run income elasticity was larger for commercial property. The local income tax had a positive influence on median voter's demand for manufacturing and housing property. The short-run income tax elasticity was larger for commercial property. The rate of grants/user charges only had a negative effect on median voter's demand for commercial and housing property. The only effect of median age was to reduce the median voter's demand for manufacturing property and increase the demand for commercial property. The reason for this may be that older people are more concerned about the negative 129 environmental and health consequences of increased manufacturing property than younger people. Older people may also desire the increased services that a larger commercial property tax base provides. The education elasticity was negative for all forms of the property bases. Qete11§_net1hne, better educated median voters desire smaller property tax bases. A reason for this could be that a higher education increases one's taste for environmental quality. The short-run education elasticity of property base demand was greatest for commercial property and least for housing property. The percentage of residents employed locally had a positive effect on manufacturing property base demand, but no effect on commercial property base demand. The regression results provided some evidence that median voters demand manufacturing for its increased employment opportunities. 130 Table 10 Median Voter's Demand For Local Expenditure Per Capita Regression R-squared = 0.392 F-statistic = Dependent Variable: Rt Number of Observations: 141 No heteroskedasticity detected -- 2-stage least squares Independent ‘ 2-Tail Variable1 61 Elasticity1 Significance1 d7? 1.3170 D+02 ----- 0.178 (0.9784 D+02) d82 1.4510 D+02 ----- 0.161 (1.0361 D+02) d87 2.1347 D+02 ----- 0.049 (1.0822 D+02) At -2.4390 D-06 -0.004 0.793 (9.2750 D-06) Bt -5.7230 D-07 -0.017 0.505 (8.5860 0-07) Nt 1.2090 D-04 0.022 0.645 (2.6240 0-04) Vt 2.3520 D-03 0.154 0.240 (2.0000 0-03) Yt 2.0976 D-03 0.074 0.808 (8.6236 0-03) It 2.7852 D+03 0.031 0.044* (1.3830 0+03) gt -1.2737 D+03 -0.117 0.071* (0.7042 0+03) aget -6.8620 -0.618 0.000** (1.8232) educt 2.5627 0+01 0.911 0.026* (1.1525 0+01) areat 5.1505 D-01 0.019 0.744 (15.7881 D-Ol) Ct 1.9450 D-07 0.029 0.377 (2.2030 0-07) Mt 7.5780 0-07 0.133 0.121* (4.8930 0-07) Ht 5.5810 0-07 0.225 0.040* (2.7130 0—07) lempt 2.3090 0.143 0.341 (2.4250) 5.00 131 In Table 10, only the year dummies were statistically significant. getetie_nezitne, the median voter's demand for local expenditure per capita was lowest in 1977, increased in 1982, and increased again in 1987. Six of the remaining fourteen non-dummy independent variables were statistically significant. The regression results show that local expenditure per capita was positively related to the rate of local income tax (1) and negatively related to the rate of local grants/user charges (9). The positive coefficient on "1” was predicted in the theoretical model. The negative coefficient on ”g" was not. A possible explanation for the unexpected negative coefficient on "g" is that a community with a large local expenditure per capita receives a smaller rate of state and federal grants. The predictions derived from the theoretical model regarding "age", "educ", "area", ”C", "M", "H", and "lemp" were that variables expected to have a positive influence on median voter's taste for environmental quality (a) and local expenditure (0) should respectively have a negative and positive influence on median voter's demand for local expenditure per capita. The regression results show the coefficient on median age as negative. The age elasticity of local expenditure was the second largest. A one percent increase in median age would reduce demand for local expenditure 6.18 percent. This is most likely due to the large education component of local expenditure. Older 132 folks are less likely to have children attending local schools. Demand for local expenditure was positively related to median education. The education elasticity was the largest recorded. A one percent increase in a community's median education would increase the median voter demand's for local expenditure .91 percent. Again, this was most likely related to the large education component of local expenditure. Better educated residents demand more expenditure on local education. Finally, a one percent increase in the average Metropolitan Detroit community's manufacturing or housing property tax base would increase local expenditure per capita .13 and .23 percent. These positive elasticities were most likely due to the decrease in the median voter's marginal tax price for local expenditure following an increase in the manufacturing or housing property tax base. 133 Table 11 Manufacturing Property Abatements Necessarily Offered By Median Voter Regression Dependent Variable: 3t Number of Observations: 141 Heteroskedasticity detected -- 2-stage l.s. weighted by "pden" Indep. ‘ S-R ‘ L-R 2-Tail Var.1 S-R B1 Elas.1 L-R 61 Elas.1 Signif. dmac -4.6071 D+07 --------------- 0.000 (1.2638 D+07) doak -5.0705 D+07 --------------- 0.000 (1.2213 D+07) dway -4.8165 D+O7 --------------- 0.000 (1.3241 D+07) pt 8.2194 D+08 5.089 2.9355 D+09 18.175 0.000** (2.1959 D+08) 1t 1.0157 D+08 0.039 3.6275 D+08 0.139 0.613 (20.1094 D+08) ft -4.1584 D+08 -1.070 -1.4851 D+09 -3.821 0.022* (1.8119 D+08) Cdt 2.4819 D+05 0.313 8.8639 D+05 1.118 0.227 (2.0528 D+05) hnt -1.5632 D+05 -0.054 -5.5828 D+05 -0.193 0.588 (2.8861 D+05) wnt 6.0700 D+06 0.135 2.1679 D+07 0.482 0.032* (2.8337 D+06) cmt 4.4133 D+02 0.235 1.5762 D+03 0.839 0.088* (2.5871 D+02) maggt -4.2155 D-01 -0.195 -1.5055 -0.696 0.275 (3.8579 D-01) Mt-Mt-l 0.4210 -------------------- Mt-1 0.1641 --------------- 0.000** (0.0189) Bt_1 0.7172 --------------- 0.000** (0.0785) R-squared = 0.939 F-statistic = 151.49 5B (coeff. of manuf. property abatement adjustment) = .28 £100 (coeff. on long-run manufacturing property) = .585 134 As recorded in Table 11, only the county dummies were statistically significant in the regressiOn representing the manufacturing property abatements necessarily offered by the local median voter. Holding other independent variables constant, manufacturing property abatements were largest in Macomb County and smallest in Oakland County. The coefficient of manufacturing property abatement adjustment was calculated to be ".28”. Over a five-year period, the average Metropolitan Detroit community moved 28 percent of the way toward its necessary level of manufacturing property abatement. Based on this statistic it would take approximately a-45 year adjustment process, cetetis pazibns, for the average community to reach 95 percent of necessary property abatements it has to offer in a given year. This also indicates it would take 45 years to attain 95 percent of the value cited for each long-run elasticities. In Table 11, six of the remaining eleven independent variables were statistically significant. Due to scale and substitution effects the signs on all the remaining independent variables, except the positive sign on the long-run value of manufacturing property (*M), could not be predicted in advance. The regression results show that a one percent increase in the local property tax rate (p) would increase the value of necessarily offered manufacturing property abatements 5.09 percent in the short run and 18.18 percent in the long run. A one percent 135 increase in the rate of locally provided firm services (f) would decrease the value of manufacturing property abatements 1.07 percent in the short run and 3.82 percent in the long run. A one percent increase in the number of crimes (cm) would increase the necessary value of manufacturing property abatements .24 percent in the short run and .84 percent in the long run. The regression results also show that a community adjoining a major water source offers larger manufacturing property abatements. Although these results could not be predicted in advance, they do make intuitive sense. An increase in a variable that increases the typical manufacturing firm's cost (p and cm), and reduces its profit, forces the median to offer greater manufacturing property abatements. An increase in a variable that decreases the typical manufacturing firm's cost (f), and increases its profit, allows the median to offer less manufacturing property abatements. The positive sign on ”wn" could be attributed to the fact that communities that border major water sources have longer established manufacturing property tax bases, whose manufacturing firms would exit without the community offering greater manufacturing property abatements. These are important results because they provide evidence that fiscal variables affecting firms are not fully capitalized into local land values in the short- 136 run and force the median voter to offer a certain level of manufacturing property abatements. This was the basis of the theoretical model of the firm presented earlier. These results also provide evidence of a rationality to the manufacturing abatement process in Metropolitan Detroit. Communities appear to offer manufacturing property abatements to offset the lower profit a manufacturing firm in the community would earn without them. In Table 11, statistics relating to the coefficient on "Mt'Mt-l" are not reported because the coefficient was calculated by formula from the restrictive system estimation. The calculated coefficient on the long-run value of manufacturing property was ”.585".53 The positive coefficient was predicted in the theoretical model. 53See equation (87) for derivation. 137 Table 12 Commercial Property Abatements Necessarily Offered By Median Voter Dependent Variable: At Number of Observations: 141 No heteroskedasticity detected -- 2-stage least squares Indep. ‘ S-R ‘ L-R 2-Tail Var.1 S-R p1 Elas.1 L-R B1 Elas.1 Signif. dmac -1.4838 D+07 --------------- 0.002 (0.4722 0+07) doak -1.4395 D+07 --------------- 0.003 (0.4889 D+07) dway -1.5576 D+07 --------------- 0.002 (0.5003 0+07) pt 2.1029 D+08 24.769 7.0097 D+09 825.633 0.001** (0.6497 D+08) 1t 5.8381 D+08 4.232 1.9460 D+10 141.067 0.123* (3.7851 D+08) ft -1.4819 D+08 -7.250 -4.9397 D+09 -24l.667 0.016* (0.6174 D+08) Yt 1.1039 D+02 2.541 3.6797 D+03 84.700 0.238 (0.9351 D+02) aget 6.6274 D+04 3.881 2.2091 D+06 129.367 0.085* (3.8501 D+04) pdent 3.3509 D+02 3.038 1.1169 D+04 101.267 0.149* (2.3223 D+02) yet 8.1925 D+01 1.876 2.7308 D+03 62.533 0.468 (11.2819 D+Ol) agest 2.1007 D+04 1.192 7.0023 D+05 39.733 0.670 (4.9348 D+04) pdenst 6.0040 D+01 0.553 2.0013 D+03 18.433 0.759 (19.5869 0+01) hnt 8.5484 D+04 0.558 2.8495 D+06 18.600 0.250 (7.4326 D+05) cmt 1.0914 D+02 1.104 3.6380 D+03 36.800 0.026* (0.4908 D+02) caggt -6.2931 D-Ol -4.886 -2.0977 D+01 -162.867 0.120* (4.0459 0-01) ct-ct-1 0.0350 -------------------- ct-1 0.0021 --------------- 0.191 (0.0015) At_1 0.9716 --------------- 0.001* (0.3233) R-squared = 0.735 F-statistic = 20.02 6A (coeff. of comm. property abatement adjustment) = .03 8107 (coeff. on long-run commercial property) = .075 138 As reported in Table 12, only the county dummies were statistically significant in the regression representing the commercial property abatements necessarily offered by the median voter. Holding other independent variables constant, commercial property abatements were largest in Oakland County and smallest in Wayne County. The coefficient of commercial property abatement adjustment was calculated to be ”.03". Over a five-year period, the average Metropolitan Detroit community moved 3 percent of the way toward its necessary level of commercial property abatement. Based on this statistic it would approximately take a 350 year adjustment process, ggtg11§_pgribg§, for the average community to reach 95 percent of necessary commercial property abatements it has to offer in a given year. This also indicates it would take 350 years to attain 95 percent of the value cited for each long-run elasticities. This long adjustment process must be considered when interpreting the very large long- run elasticities recorded in Table 12. In Table 12, eight of the remaining fifteen non-dummy independent variables were statistically significant. Due again to scale and substitution effects, the signs on the all remaining independent variables, except the positive sign on the long-run value of commercial property (*C), could not be predicted in advance. The regression results show that an increase in the local property tax rate (p), local income tax rate (1), median age (age), population 139‘ density (pden), and number of crimes (cm) would increase the value of necessarily offered commercial property abatements. The calculated property tax elasticity was by far the largest. A one percent increase in the local property tax would increase commercial property abatements 24.77 percent in the short run and 825.63 percent in the long run.54 The regression results also show that an increase in the value of locally provided firm services (f) and commercial agglomeration economies (cagg) would decrease the value of necessarily offered commercial property abatements. The firm service elasticity was the second largest. A one percent increase in locally provided firm services decreases commercial property abatements 7.25 percent in the short run and 241.67 percent in the long run. Although these results could not be predicted in advance, most again make intuitive sense. Variables that increase the typical commercial firm's cost (p, cm, or 1), or reduce its market (age), decrease the profit the typical commercial firm earns in the community and forces the median to offer greater commercial property abatements in the long ratements was approximately seven times larger for commercial property. The measure of local 54This extremely large long-run elasticity can be attributed to slow long-run adjustment process and the fact that communities offering commercial abatements in 1982, moved from offering zero commercial abatements in 1977. crime abate was 1 than econ: on C1 pr0p: expe manu resu slow manu This are 140 crime was statistically significant in both property abatement regressions. The short-run crime elasticity was five times larger for commercial property abatements than manufacturing property abatements. Agglomeration economies only exerted the expected negative influence on commercial property abatements. These results show that necessarily offered commercial property abatements are more responsive to variables expected to influence it, than necessarily offered manufacturing property abatement (5 to 7 times). The results also show commercial property abatements are much slower to adjust to their necessary long-run values than manufacturing property abatements (approximately 9 times). This corresponds with the finding that commercial firms are less footloose than manufacturing firms. l.) 141 3. D. GRAPHICAL ANALYSES The simultaneous relationship between the long-run value of local manufacturing property abatements and local manufacturing property can be illustrated in a graph. Regression equation (80) represented the median voter's demand for manufacturing property. The graphical representation of this equation will be referred to as the ”M” Demand Schedule. Regression results for the "M" Demand Schedule were recorded in Table 7. The long-run coefficient on manufacturing property abatements was "1.28". Regression equation (84) represented the manufacturing property abatements a median voter necessarily offers in the long run. The graphical representation of this equation will be referred to as the "B” Necessary Schedule. Regression results for the ”B" Necessary Schedule were recorded in Table 11. The long- run coefficient on manufacturing property was "0.59". The "M" Demand Schedule and "B" Necessary Schedule are represented in the two-dimensional graph in Figure 3. For simplicity the schedules are shown to be linear. All variables, except ”8" and "M", are held constant. 142 As the regression results indicated, the "B” Necessary Schedule is steeper than the "M" Demand Schedule.55 ”M" Demand Local Schedule Manuf. Abat. .. NB" >BN "B” Necessary Schedule *B M1 >Mo Local Manuf. Prop. Value .. I'M" Figure 3 Manufacturing Equilibrium -- #1 55For the "M” Demand Schedule, the regression results indicated that a one dollar increase in manufacturing property abatements caused a long-run $l.28 increase in manufacturing property value (ggtgri§_p§;ibu§). The inverse of this result is that a one dollar increase in manufacturing property value results in a $.78 increase in manufacturing property abatements. For the ”B" Necessary Schedule, the regression results indicated that a one dollar increase in manufacturing property value resulted in $.59 increase in manufacturing property abatements. The "M" Demand Schedule is therefore steeper than the ”B" Necessary Schedule. 143 In Figure 3, the long-run equilibrium values of local manufacturing property abatements and local manufacturing property value are given as ”*B” and "*M". These are the values that a community displaying the ”B" Necessary Schedule and the "M" Demand Schedule can be expected to be moving toward if the following adjustment process occurs. As described earlier in the theoretical model, competitive forces created by both firms and communities in a metropolitan area cause a community to offer to a given value of manufacturing property the value of manufacturing property abatements specified on the ”B" Necessary Schedule. If local manufacturing property value (>Mo), is greater than the long-run equilibrium value (*M), these competitive forces cause the community's median voter to offer ">BN” manufacturing property abatements to it. At ">BN" the corresponding value of local manufacturing property (>M1) demanded by the median voter is less than the current value of manufacturing property (>Mo). The median voter reduces the local value of manufacturing property through restrictive zoning ordinances. This process repeats itself until the community reaches a manufacturing property value of "*M". Only at "*M" are the values of manufacturing property and manufacturing property abatements on each of the schedules equal to each other. 144 If local manufacturing property value ((M0), is smaller than the long-run equilibrium value (*M), metropolitan-wide competitive forces again cause the community's median voter to offer "<8N" manufacturing property abatements to it. At "<3N" the corresponding value of local manufacturing property ((M1) demanded by the median voter is greater than the current value of manufacturing property ((M0). The median voter increases the local value of manufacturing property through nonrestrictive zoning ordinances. This process repeats itself until the community reaches a manufacturing property value of "*M". Only at "*M" are the values of manufacturing property and manufacturing property abatements on each schedule again equal to each other. The equilibrium shown in Figure 3 is only one of three possible long-run manufacturing equilibriums observed in a community within a metropolitan area. The three equilibriums consist of the following combinations: (1) positive manufacturing property value and positive manufacturing property abatements, (2) positive manufacturing property value and zero manufacturing property abatements, (3) zero manufacturing property value and zero manufacturing property abatements. The schedules yielding positive manufacturing property value and zero manufacturing property abatements are shown in Figure 4. 145 Local Manuf. Abat. .... "all ”M" Demand Schedule "B" Necessary * / Schedule B = 0 / 0 *M Local Manuf. Prop. Value .. ”MM Figure 4 Manufacturing Equilibrium -- #2 In Figure 4, the "B" Necessary Schedule equals zero manufacturing property abatements at all values of manufacturing property. It is not necessary for the median voter to offer manufacturing property abatements to attract manufacturing property. In Figure 4, the "M" Demand Schedule intersects the horizontal axis at ”*M". The long- run equilibrium occurs at a "*M" manufacturing property value and zero manufacturing property abatements. The final long-run equilibrium of zero manufacturing property value and zero manufacturing property abatements is shown in Figure 5. 146 Local Manuf. Abat. -- "B" "M" Demand Schedule ”B" Necessary ,,.——— Schedule 8* =- 0 0 = M* Local Manuf. Prop. Value -. "MOI Figure 5 Manufacturing Equilibrium -- #3 In Figure 5, the "M" Demand Schedule is above the ”B" Necessary Schedule at all positive values of manufacturing property. This can occur as shown, or if the "B” Necessary Schedule always equal to zero. As long as either of these situations occur the two schedules intersect at a negative value of manufacturing property. A negative value of manufacturing property is not possible. The observed long- run value of manufacturing property is zero with zero manufacturing property abatements offered to it. The simultaneous relationship between the long-run value of commercial property abatements and commercial property in a community can also be illustrated in a graph. In this chapter regression equation (81) 147 represented the median voter's demand for commercial property. Regression results for the "C" Demand Schedule were reported in Table 8. The long-run coefficient on commercial property abatements was "153.37”. Also in this chapter, regression equation (85) represented the commercial property abatements a median voter's necessarily offers in the long run. Regression results for the "A” Necessary Schedule were reported in Table 12. The long- run coefficient on commercial property was ".030”. These regression results indicate that the "C" Demand Schedule would also be steeper than the "A" Necessary Schedule.56 Given this finding, the graphical analyses conducted in Figures 3, 4, and 5 apply equally well to the community sector. 56For the "C" Demand Schedule, the regression results indicated that a one dollar increase in commercial property abatements caused a long-run $153.37 increase in commercial property value (ggtgzi§_pg:1bg§). The inverse of this result being that a one dollar increase in commercial property value results in a $.007 increase in commercial property abatements. For the "A" Necessary Schedule, the regression results indicated that a one dollar increase in commercial property value resulted in a $.075 increase in commercial property abatements. The "A" Necessary Schedule is again steeper than the "C" Demand Schedule. PLEASE NOTE: Page(s) missing in number only; text follows. Filmed as mceived. 148 U-M-I 149 2. E. SIMULATIONS Both the regression and graphical analyses were informative in their own right, but alone cannot answer the six questions posed at the beginning of this research.58 Simulations are performed in order to answer these questions. The simulations involve changing a selected variable and tracing its effect on endogenous variables five years after the change.59 An example of such a simulation is to find the full effect of an exogenous, one-percent increase in the average local property tax rate on the average community's manufacturing property tax base. The "M" Demand and "B” Necessary Schedules in Figure 3 will be used as a graphical aid to understand this example. Regression results show that an exogenous increase in a community's property tax rate would increase the manufacturing property abatements necessarily offered by the median voter. In Figure 3, the "B" Necessary Schedule 58See Chapter 1, page 1. 59The calculation of short-run effects was chosen over the calculation of long-run effects because only the short run was fixed in the regression analysis. The long run could not be fixed and was calculated to vary widely among endogenous variables. This variation ran from a minimum 30 years for manufacturing property demand, to a maximum 350 years for necessarily offered commercial property abatements. Short-run simulation values represent the result of an exogenous change in a variable after five years, (ggtgris paribug). These should be of more interest to policymakers than a long-run simulation value that could conceivably represent the change after 350 years. This is especially true considering that a property tax abatement is only granted for a twelve-year period in Michigan. Some of the long-run simulation results will be discussed in the next chapter. 150 would shift to the left. Regression results also show that an increase in a community's property tax rate would have more than one effect on the median voter's demand for manufacturing property. The effects were an increase in the necessarily offered manufacturing and commercial property abatements (see Tables 11 and 12), a decrease in the median value of homes (see Table 3), and an increase the effective local income tax rate (see Table 4). These effects would increase, decrease, and increase the median voter's demand for manufacturing property (see Table 7). The "M" Demand Schedule would shift to the right if the two increases are larger than the decrease. The new equilibrium would occur at a larger value of manufacturing property and manufacturing property abatements. The precise change in the average community's manufacturing property value can be derived using the regression elasticities. A one percent increase in the property tax rate increases the necessary manufacturing property abatements 5.09 percent in the short run. Manufacturing property value increases .41 (5.09*.08) percent from this change. A one percent increase in the property tax rate also increases the necessary commercial property abatements 24.77 percent, decreases median home value .88 percent, and increases the effective local income tax rate 121.38 percent. The percentage changes in manufacturing property value from these changes would be 1.24 (24.77*.05), - .32 (-.88*.36), and 4.85 (121.38*.04). The full, short- run, effect of a one percent increase in a community's property tax rate is a 6.18 (.41+1.24-.32+4.85) percent 151 increase in its manufacturing property value. The results from 18 fiscal simulations are given in Table 13. represent the percentage change in the respective property tax base, five years after a one-percent increase in the variable listed in the first column. Table 13 The numbers recorded in the body of the table Fiscal Simulation Results -- Full System VARIABLES PROPERTY BASES Manufacturing Commercial Housing p 6.18 15.25 6.67 £50 -0.36 -0.46 -0.01 1 0.27 0.51 0.12 R 0.09 0.22 0.11 A 0.05 0.09 0.01 B 0.16 0.24 0.12 A somewhat surprising finding was that an exogenous increase in the local property tax rate would increase the value of 60A one percent increase in firm services (f) was considered to also represent a one percent increase in total expenditure A one percent increase in "R" was considered The simulation results for "R" per capita (R). to have no effect on "f". represent the result of a one percent increase in local expenditure used only by residents. 152 both the manufacturing and commercial property tax base. This is opposite to what previous researchers had found and requires further explanation. Tracing through the simulation, an exogenous increase in the property tax rate is followed by an increase in both property tax abatements and the effective local income tax rate. These increases cause the increase in manufacturing and commercial property bases. If property tax abatement programs did not exist, for most communities in the Metropolitan Detroit Area a property tax increase would be followed by a decrease in manufacturing and commercial property value. The reason is that 91 percent of the communities in the sample did not have a local income tax. Without a local income tax, an increase in property tax rates could have no effect on the effective local income tax rate. To prove this assertion, a second fiscal simulation was run under the assumption of no property tax abatement programs and no local income tax. The results are recorded in Table 14. 153 Table 14 Fiscal Simulation Results -- No "l", "A", or ”B" VARIABLES PROPERTY BASES Manufacturing Commercial Housing p -0.32 -0.79 -0.42 f 0.09 0.22 0.11 R 0.09 0.22 0.11 As earlier researchers had found, the property tax elasticity of local property bases is now negative and inelastic. The local expenditure elasticity of local property bases is now positiVe and inelastic. A third simulation was run to calculate the effects of a one percent increase in a local median voter's characteristic on observed components of the average community's property base. The results are given in Table 15. 154 Table 15 Median Voter's Char. Simulation Results -- Full System VARIABLES PROPERTY BASES Manufacturing Commercial Housing Y 0.01 -0.19 -0.05 V 0.55 0.90 0.39 age -0.42 0.45 -0.02 educ -1.77 -2.92 -0.10 Characteristics of a community's median voter had no effect on necessarily offered property abatements or effective local income tax rates. Other effects of local property abatements are derived by simulating an exogenous, one-percent increase in manufacturing and commercial property abatements and checking its impact on local property and income tax rates, total expenditure per capita, median value of homes, and percentage of residents employed locally. The results of these simulations are recorded in Table 16. The numbers in the body of the table represent the percentage change in the respective variable, five years after a one-percent increase in the property abatement listed in the first column. 155 Table 16 Property Abatement Simulation Results é- Full System PROPERTY VARIABLES ABATEMENT p l R V lemp A 0.02 2.18 0.01 -0.03 0.03 0.03 3.61 0.01 -0.01 0.01 The simulation recorded in Table 16 allowed all endogenous variables to fully adjust. A local policymaker would also want to know the effect of holding local variables under his control (local property tax rate, income tax rate, and expenditure per capita) constant. The result of such simulations are recorded in Table 17. Table 17 Property Abatement Simulation Results -- "p", "1", and "R" constant PROPERTY VARIABLES ABATEMENT M C H V lemp A 0.05 0.09 0.01 0.01 ‘ 0.03 B 0.08 0.00 0.00 0.00 0.01 All the simulation results presented in the section will be used to derive the conclusions presented in the next and final chapter. CHAPTER 4 SUMMARY AND CONCLUSIONS This chapter contains a summary of the research and answers to the six important questions the dissertation set out to answer. 4. A. SUMMARY This research provides a theoretical and empirical analysis of the simultaneous relationships between the local income tax, property tax, home value, property tax abatement, expenditure, and composition of the property tax base in a community within a metropolitan area. The dissertation began with a summary of previous literature relating to the influence of local fiscal policy on firm location. This summary traced the literature from early research that only concerned itself with firm influence on local nonresidential property bases, to recent innovations that began to incorporate the influence communities have on their nonresidential property bases. I observed a general failure to apply many theoretical and empirical advancements related to this topic. This is especially true as it relates to the influence communities have on their nonresidential property bases. I attempted to correct this failure by 156 157 developing a theoretical model that accounts for the fact that variables are determined by the joint actions of~ communities and firms. The theoretical model also accounts for the fact that desired values of local manufacturing, commercial, and housing property bases, and the necessary level of manufacturing and commercial property abatements are long-run values that a community is moving toward. The theoretical model is an attempt to comprehensively account for all factors that determine the influence of local fiscal variables on the composition of a community's property tax base. Ten simultaneous equations from the theoretical model were estimated using weighted, two-stage least squares: restricted, weighted, two-stage least squares: and weighted, two-stage "Tobit". Data were used from 47 communities in the Detroit Metropolitan Area (Macomb, Oakland, and Wayne Counties) for the years 1977, 1982, and 1987. This data contained the first collection of community-specific manufacturing and commercial property abatement values. A "Nerlove” stock-adjustment model accounted for the long- run adjustment of dependent variables. The simultaneous relationship between local manufacturing (commercial) property abatements and manufacturing (commercial) property value was then described graphically. System simulations based on the regression results were also conducted. The results of the simulations are used to derive the conclusions presented in the next section. 158 4. B. CONCLUSIONS The conclusions derived from the work described in Chapters 2 and 3 are given in response to the six important questions in the dissertation's introduction. (1) How do local fiscal variables, including property tax abatements, affect local property tax bases? In this research, an exogenous increase in local property and income taxes insxeeses local property tax bases. An exogenous increase in local expenditure on firm services geezesses local property tax bases. Previous researchers have reported a negative relationship between local property tax rates and local manufacturing property bases, and a positive relationship between local expenditure on firm services and local manufacturing property bases. A reason for this discrepancy may be that this research addressed this question differently than previous researchers. The simultaneous nature of variable determination, the long-run adjustment process, local income taxes, and local property tax abatements are accounted for here. Upon further examining the simulations, the positive relationship between the local property tax rate and local property tax bases are attributable to the presence of property tax abatement programs and the local income tax. An exogenous property tax increase causes the median voter to necessarily offer greater nonresidential property 159 abatements.61 Both the property tax rate increase and subsequent increase in abatements, increase the total property taxes paid by the median voter. The property tax rate increase does this directly. The abatement increase does it indirectly by reducing the property tax base and making it necessary for further property tax rate increases. To mitigate the increase in property taxes paid, the median voter desires greater taxable property in the community. In addition, the exogenous increase in the property tax rate causes the median voter to substitute local income tax revenue for property tax revenue, and raise the local income tax rate. An increase in the local income tax rate, increases the median voter's demand for all local property tax bases. An increase in any property tax base now brings the added benefit of increased income tax revenue. The result of the increase in median voter demand for local property bases is the calculated increase in nonresidential property following a property tax increase. The simulation results recorded in Table 14 controlled for the impact of property tax abatement programs and the local income tax. The result was a negative relationship between the local property tax rate and all local property tax bases, and a positive relationship between local 61A property tax increase also decreases the average median voter's home value. Through a drop in assessed property value, this decrease would decrease the property taxes paid by the median voter and decrease his demand for taxable property. In the simulation, the decrease in demand was smaller than the increase in demand. 160 expenditure on firm services and all local property tax bases. Controlling for ”A”, ”B", and ”l", the long-run elasticity of the manufacturing, commercial, and housing property bases with respect to the property tax rate were simulated to be "-.81” (30 yrs.), ”-13.27" (350 yrs.), and ”-1.67" (45 yrs.).62 So if abatements were not available, and localities did not have the option of substituting income taxes for property taxes, the results are similar to those reported by earlier researchers.63 The results in Table 13 also show that an exogenous increase in local manufacturing property abatements increases local manufacturing property bases, and an exogenous increase in local commercial property abatements increases all local property bases. The reason that an increase in commercial property abatements increases manufacturing property is that a rise in "A" would increase median voter's demand for local manufacturing, commercial, and housing property bases. The increase in the housing property base increases manufacturing property demand, while the increase in the commercial property base decreases manufacturing property demand. The two increases in manufacturing property demand are greater than the decrease. The result is greater manufacturing property in the community. 62The years in parenthesis represent the time it would take to reach 95 percent of this long-run elasticity. 63Previous researchers calculated the long-run elasticity of local manufacturing property bases with respect to local property tax rates to range from -4.43 (Fox, 1981) to -.15 (Ladd and Bradbury, 1988). 161 The presence of a local property abatement program and local income taxes mitigate the effect that local property taxes have on local property bases. The burden of a property tax increase falls less on firms because communities offer property abatements, or raise their income tax after an increase in property tax rates. (2) How do local median voter characteristics affect local property tax bases? The simulation results recorded in Table 15 show that the characteristics of a community's median voter (median income, median value of homes, median age, and median education) influence the determination of all local property tax bases. A one-percent increase in the average community's median income decreases its commercial and housing property bases and increases its manufacturing property base. An exogenous, one-percent increase in the median value of local homes increases all local property bases. This is most likely due to the fact that an increase in "V" would increase the median voter's marginal tax price for local expenditure. If the marginal tax price increases, the median voter demands larger property bases to compensate. The simulation results in Table 15 also show that an increase in a community's median age would decrease the community's manufacturing and housing property bases, and increase its commercial property base. The decrease is apparently due to increased environmental concerns as age 162 increases. The increase in commercial property apparently exists because the demand for services provided by greater commercial property is greater than their environmental concerns. The simulated increase in median education resulted in a reduction in all property bases. After five years the negative response in the manufacturing and commercial property bases are both elastic (-1.77 and -2.92). The elastic reduction is apparently due to increased environmental concerns as education increases. This research suggests that the characteristics of a community's median voter affect the value and composition of local property tax bases. Generally, the value of local property bases decrease as resident's income, age, and education increase. The value of local property bases increase as the value of local homes increase. (3) Why do communities offer property tax abatements? There has not been much, if any, research on this question. The theoretical and empirical analyses conducted provide some answers to this question. Communities offer property tax abatements because a community's profit-reducing characteristics are not necessarily capitalized into lower land prices in the short run. If such a community desires to attract new firms, it is forced to offer an off-setting property tax abatement or 163 other incentive.64 The fact that communities have the option of granting property tax abatements may magnify this situation. A community's propensity to offer property tax abatements is another profit-influencing characteristic that in the short run may not be capitalized into local land prices. As White (1986) pointed out, even with full capitalization, local land prices may fall to zero and still not be low enough to off-set relatively high local property taxes. At the opposite extreme, local land prices may rise high enough that they are no longer attractive to nonresidential buyers. Also, high capitalized property taxes may be a signal for even higher taxes in the future. If any of these occur, a community desiring to retain its nonresidential property base, or attract new non-residential property, would again be forced to provide an off-setting property tax abatement. As Wolkoff (1982).pointed out, in the Detroit Metropolitan Area property abatements in suburban communities have gone mainly to new facilities, while Detroit city abatements have gone mainly to the rehabilitation of existing facilities. The suburbs have faced the short-run problem of local profit reducing characteristics not being capitalized into lower nonresidential land prices. Detroit has faced 64If there is no allowance for property tax abatements the community would offer an equivalent off-setting incentive to the firm. These could include community-provided firm services, community-backed financing packages, or even community-purchased land or capital. 164 the long-run problem of capitalized land prices for firms falling to zero, or near zero, and still not being low enough to compensate for relative profit-reducing characteristics (high local taxes and crime). (4) What type of communities offer property tax abatements? If the theory that communities offer property tax abatements to off-set noncapitalized characteristics is correct, communities with characteristics that reduce profits should offer greater property tax abatements and communities with profit-increasing characteristics should offer less property tax abatements. The regression results recorded in Tables 11 and 12 support these contentions. Communities with higher local property taxes and more crime offer greater abatements. Communities that provide more local services to firms and that have greater agglomeration economies offer fewer-abatements. These results provide some of the first evidence that communities offer property tax abatements to mitigate the effect of noncapitalized, profit-reducing characteristics. At least in the Detroit Metropolitan Area, the granting of manufacturing and commercial abatements appears to follow this pattern. 165 (5) Besides the size of the local property tax base, what are the other effects of local property tax abatements? An exogenous increase in manufacturing or commercial property abatements increases the average Metropolitan Detroit community's manufacturing, commercial, and housing property bases. The effects of the same simulated increase in property abatements on local income tax rate, local expenditure per capita, median home value, and percentage of residents employed locally are recorded in Tables 16. ‘ An exogenous increase in local manufacturing or commercial property abatements increases the local property tax rate, local expenditure per capita, and percentage of residents employed locally by inelastic percentages (g .03). The same increase in abatements increases the local income tax rate by elastic percentages (2.18 and 3.61) and decrease the median value of local homes by inelastic percentages (.03 and .01). If local property tax rates, income tax rates, and expenditure are held constant, the simulation results in Table 17 show that an increase in commercial abatements decreases the median home value to a lesser degree (from - .03 to -.01). An increase in manufacturing property abatements know has no effect on median home value (from - .01 to .00). The effect of an increase in property tax abatements on percentage of residents employed locally remains the same. If it were possible for a community to grant property tax abatements without the cost - that is 166 without raising other revenues or lowering expenditures - then obviously the community would be better off. The offering of property tax abatements affects more than a community's property bases. When a community offers a manufacturing or commercial property abatement its property and income tax rate, expenditure per capita, median value of homes, and percentage of residents employed locally all change. (6) Do Property abatements work? One way to address this question is to simulate the short-run effects of a ten-percent increase in manufacturing or commercial property abatements on the average Metropolitan Detroit community during the period 1977 to 1987. By reporting the effects in originally measured units, it should be easier for the local policymaker to reach a conclusion as to whether abatements work. All dollar figures are reported in 1972 real dollars. The effects of a ten percent increase in manufacturing property abatements ($1,049,750 in new abated taxable property value) for the average Metropolitan Detroit community are listed in Table 18. 167 Table 18 Results of 10% Increase In Manufacturing Abatements (1) The community's property tax rate increases by .3 percent or a $.195 increase in the property tax paid per $1000 dollars of taxable property. - (2) The community's effective local income tax rate increases 36 percent or a $1.44 increase in the local income tax paid per $1000 dollars of property. (3) The median value of local homes decreases by .1 percent or $23.47. (4) The increases in the tax rates and the decrease in the value of local homes result in the average resident paying $1.58 more in annual property taxes and $33.71 more in annual income taxes. The total tax increase is $35.29. (5) The annual local expenditure per capita increases by .1 percent or $.36. (6) The percentage of residents employed locally increases 0.02 percent from an average base of 22.38 percent. (7) The manufacturing property tax base increases by $1,010,312. The commercial property tax base increases by $1,297,043. The housing property tax base increases by $1,739,882. This results in a total property tax base increase of $4,047,237 and an increase in yearly property tax revenue of $263,151. Considering that the abatement cost is $64,438 in lost property tax revenue, the net- property-tax-revenue gain to the community ($198,713) is approximately one percent of total property tax revenue before the 10 percent increase in manufacturing abatements were granted. The similar effects of a ten percent increase in commercial property abatements ($55,185 in new abated taxable property value) for the average Metropolitan Detroit community are listed in Table 19. 168 Table 19 Results of 10% Increase In Commercial Abatements (1) The community's property tax millage rate increases by .2 percent or a $.135 increase in the property tax paid per $1000 dollars of taxable property. (2) The community's effective local income tax rate increases by 22 percent a $.87 increase in the local income tax paid per $1000 dollars of property. (3) The median value of local homes decreases by .3 percent or $70.42. (4) The increases in the tax rates and the decrease in the value of local homes result in the average resident paying $.71 less in annual property taxes and $18.89 more in annual income taxes. The total tax increase is $18.18. (5) The annual local yearly expenditure per capita increases by 1 percent or $.36. (6) The percentage of residents employed locally increases 0.06 percent from an average base of 22.38 percent. (7) The manufacturing property tax base increases by $315,723. The commercial property tax base increases by $486,391. The housing property tax base increases by $144,990. This results in a total property tax base increase of $1,065,265 and an increase in yearly property tax revenue of $69,386. Considering that the abatement cost is $3587 in lost property tax revenue, the net-property-tax-revenue gain to the community ($65,799) is approximately .4 percent of the total property tax revenue before the 10 percent increase in commercial abatements were granted. A local policymaker can examine Tables 18 and 19 and see that, in the simplest sense, property tax abatements have worked. An increase in manufacturing or commercial property abatements are shown to respectively increase the average community's manufacturing and commercial property base and the property tax revenue collected. Allowing communities to grant property tax abatements provides them 169 with an instrument that has been used to effectively overcome local profit-reducing characteristics. The results leave open to some debate whether property tax abatements are completely beneficial to communities offering them. The offering of abatements must be paid for by higher property and income tax rates which reduce the median value of local homes. Local taxpayers benefit from increased local employment and local business property that increases local tax revenue, but pay in these other ways. Local policymakers should incorporate this finding into the calculus used to decide if an abatement should be offered. A local increase in manufacturing and commercial property abatements also causes a slight increase in local expenditure per capita. The measure of local expenditure used here lumps all categories of community expenditure together. The increase in this variable could be due to an increase in services provided to the larger nonresidential property tax base. Local expenditure on residential services could fall. Property tax abatement programs have been criticized for unnecessarily giving away property tax bases, reducing local revenues, and subsequently decreasing local expenditure on residential services. In the future I plan to test this assertion by dividing the local expenditure variable into measures of residential and firm expenditures. APPENDICES County City lane lane 1 Hacoob East Detroit 2 Fraser 3 Hount Clenens 4 Roseville 5 St. Clair Shores 6 Sterling Heights 7 Harren 8 Oakland Berkley 9 lirninghaa 10 Clavson 11 Farnington 12 Ferndale 13 Hazel Park 14 Hadison Heights 15 Hovi 16 Oak Park 17 Pontiac 18 Royal Oak 19 Southfield 20 Troy 21 Hayne Allen Park Dearborn 1: IR 23 3: £3 #31:: £3 58 68 23 58 28 52 83 23 22 2% 58 8% =3 8% 23 Si 23 :3 Dearborn Heights Detroit Ecorse Garden City Crosse Pointe Fares Crosse Pointe Park Crosse Pointe Hoods Hantranck Harper Hoods Highland Park lnhster Lincoln Park Livonia Helvindale Plynouth River Rouge Riverviev Ronolus Southgate laylor Trenton Hayne Hestland Hoodhaven Iyandotte OOOOOOOOOOOOOOOOOOOOOOOO00000900000000OOOOOOOOO Aqqpxarudioc Ii OOOOOOOOOOOOOOOOOOOOOOOO60°00000000000000000000 319643 44787 0 0 173303 104543 0 0 186036 418910 0 16717448 30198 34575 503392 0 0 94341 O 1181:1683113 \kaliuas A 87 434224 393914 247338 0 354511 160395 1614931 6813437 0 182871 1413115 0 26608188 84656 391009 444565 0 0 800563 0 17396 0 0 3513028 0 1313468 0 0 4914081 579867 1416808 0 253925 0 0 242575 A 77-72 °°°°°°°°°°°°°600000000000OOOOOOOOOOOOOOOOOOO00° 82-77 107244 0 224338 0 0 722259 0 0 0 319643 44787 0 0 173303 104543 0 0 186036 418910 0 16717448 30198 34575 503392 0 0 94341 900000 172795 1995513 77698 256159 269881 A 87-82 419458 169372 355239 121900 0 311556 0 0 -84405 0 0 434224 74270 202552 0 354511 -12908 1510388 6813437 0 -3166 994205 0 9890740 54458 356434 City lane 1 East Detroit 2 fraser 3 Hoant Clesens 4 Ioseville 5 St: Clair Shores 6 Sterling Heights 7 Harren 8 Dertley 9 Iirninghas 10 Clauson 11 Farsington 12 ferndale l3 Hazel Part 14 Hadison Heights 15 Iovi 16 Dal Part 17 Pontiac 18 Royal Cat 19 Southiield 20 troy 21 Allen Park 22 Dearhorn 23 Dearhorn Heights 24 Detroit 25 Ecorse 26 Garden City 27 Grosse Pointe Fares 28 Grosse Pointe Park 29 Crosse Pointe Hoods 30 Hantranct 31 Harper Hoods 32 Highland Park 33 lntster 34 Lincoln Part 35 Livonia 3 Helvindale 37 Ply-oath 3 liver longs 39 Iivervien 44 hassles 41 Soothgate 42 laylor 43 lrenton 4' him 45 Hestland 46 Hoodhaven 47 mite N N 00°00°°°°°°°°°°°OOOOOOOOOOOOOOOOOOOO00000000000 157]. Ikpqpelictiac 11. (c30r1t:.) lla::ieflalra \haliuss 8 8 8 77 82 87 0 123112 234653 424239 4412957 12288560 0 153206 1004499 0 548900 809480 0 0 0 16806751 90757112 104097581 0 32382386 30577475 0 0 0 0 0 0 0 0 0 0 0 0 371633 538791 1739425 0 183680 240981 0 247059 453532 0 0 241611 0 1197865 2223451 26590368 56185240 “279860 0 2084449 4895378 0 0 945087 0 0 6159981 0 0 69294 0 91420442 143873705 0 0 0 14908907 110290900 128542139 0 19348 46060330 215110 142460 384722 0 0 0 0 0 0 0 0 0 0 8307507 8396436 0 0 0 0 1969570 4210691 0 101541 179350 1154250 722259 2826905 29048618 87018724 94039155 0 1466224 2163268 0 0 193772 0 132640 730341 0 0 364272 17055742 26486748 47760718 0 65660 366283 0 1391672 3744040 0 22353149 37152046 0 865397 33897990 0 2103846 4348619 0 0 2995980 2273522 4851135 8170618 77-72 0 424239 0 0 E 9. § OOOOEOOOOO °°°°°°§ u. g °N8 215110 0000000 2273522 8 8 82-77 87582 123112 111541 3988718 7875603 153206 851292 548900 260580 0 0 73950362 13340469 32382386 '1804910 0 0 0 0 0 0 0 0 167158 1208634 183680 57382 247059 206473 0 241611 1197865 1025586 29594872 30094620 2084449 2810929 0 945087 0 6159981 0 69294 91420442 52453263 0 0 95381993 18251239 19348 46040983 -72650 242262 0 0 0 0 0 0 8307507 88929 0 0 1969570 2241121 101541 77809 431991 2104646 579701“ 7020431 1466224 697044 0 193772 132640 597701 0 364272 9431006 21273970 65660 300623 1391672 2352368 22353149 14798897 865397 33032598 2103846 2244773 0 2995980 2577613 3319484 City Hose 1 East Detroit 2 Fraser 3 Haunt Clenens 4 Roseville 5 St. Clair Shores 6 Sterling Heights 7 Harren 8 Dertley 9 Dirninghan 10 Clavson 11 Farnington 12 Ferndale 13 Hazel Park 14 Hadison Heights 15 Iovi 16 Oak Park 17 Pontiac 18 Royal Bat 19 Southlield 20 Troy 21 Allen Park 22 Dearhorn 23 Dearborn Heights 24 Detroit 25 Ecorse 26 Garden City 27 Grosse Pointe Farns 28 Grosse Pointe Park 29 Grosse Pointe Hoods 30 Hantranct 31 Harper Hoods 32 Highland Park 33 lntster 34 Lincoln Park 35 Livonia 36 Helvindale 37 Plynooth 38 River Rouge 39 Riverviev 40 Ronulus 41 Southgate 42 Taylor 43 Trenton 44 Hayne 45 Hestland 46 Hoodhaven 47 Hyandotte 24876102 38706981 45148339 40738225 142542597 10173438 58494277 19026647 24070088 28079565 17866720 42698208 26211203 40598471 93688544 97396557 172 inppxaruiiac £1 <3:cn1t:.) \harisflblea \haliuas 22668171 9815460 18364568 42669535 39196292 89870864 135935117 8237880 37475656 17146275 20469220 20008779 14416684 35082756 29357957 23630115 66426758 81175341 20827708 12334384 18501428 41219610 38170409 84428234 128923375 7554323 42776144 13023462 17186802 15025585 12574426 36340125 49896826 22579602 57123445 62059216 20276127 14107461 18211182 45224126 41451096 102482417- 160265208 9353827 54511717 12654996 22044268 17030838 11541237 46088707 70956883 22349830 61484066 67776199 360386846 310142183 306723594 395628761 121925737 131057817 162940468 174104458 33054982 23899895 23512169 20163447 42236254 9402843 23873851 9764543 4575671 14217114 17582098 37559533 48321683 26787671 41837352 118085663 13323278 17247936 7467020 14882530 31311785 47733111 88968709 19829846 21656127 95825905 12656361 30661100 134209842 140805771 162763165 187124745 29680378 31240053 29083170 6061490 16825568 6783106 2884540 10881917 11215250 30325883 32557817 18714935 29377265 94357817 10486884 12844526 4452536 16348164 28794614 38285414 85335257 16039699 19513466 76503218 11176425 21698461 5023632 15523873 7212410 2423922 10367214 7333080 23207879 16630970 13223047 24161830 1162377885 689189787 475270567 459592143 4804932 15093358 6034717 2313006 10177830 7814196 24483692 19641924 13857820 23332614 89140753 110168378 8608886 14384264 3510068 14095119 33309564 30750886 68964653 17794922 18647010 69635916 11524776 23068568 7321374 15866299 2899954 11427366 31778879 31856483 67187235 16736767 17349114 78155515 10284978 18308550 77-72 -133762 ~4468083 ~6511534 3962554 '5952047 49132639 '6607480 71935557 -21018621 -1880372 -3600868 -8070785 -3450036 -7615452 3146754 -16968355 -27261786 -16221216 -50244663 9132081 -9155087 6595929 -12555877 -473188099 -3341353 -7048282 -2981437 ~1691131 -3335197 -6366848 -7233650 -15763866 -8072736 '12460087 -23727846 '2836395 -4403410 -3014484 1465634 ~2517171 '9447697 '3633452 '3790147 -2142661 -19322688 -1479936 -8962639 C 82-77 ~7011742 -683558 5300488 -4122812 -3282418 '4983194 -1842258 1257369 20538869 ~1050514 -9303312 '19116125 -3418589 31882651 '3N8 21957394 1§%fi '213919220 -1037858 '1301696 429304 -460618 -514703 '3882170 '7118004 -15926848 -5491889 '5215435 -5217065 -1877998 1539737 -942468 -2253044 4514951 ~7534528 '16370604 1755223 '866456 '686 “348350 1370107 0 87:82 -551581 1773077 -290247 4004515 3280687 18054183 31341833 1799504 11735573 -368466 4857465 2005252 -1033188 9748583 21060057 '229771 4360621 5716983 88905167 11163990 -3348722 24361580 -2156882 '15678424 -218700 -430515 -1177693 -110916 -189384 481116 1275813 3010955 634774 '829216 21027625 -1287512 1482036 -610114 -2667753 -1530685 1105597 -1777418 -1058155 ~1297897 ~ 8519599 '1239798 ~4760017 City lane 1 East Detroit 2 Fraser 3 Hount Cleaens 4 Roseville 5 St. Clair Shores 6 Sterling Heights 7 Harren 8 Berkley 9 Dirninghan 10 Clavson 11 Parnington 12 Ferndale 13 Hazel Part 14 Hadison Heights 15 Hovi 16 Oak Park 17 Pontiac 18 Royal Cat 19 Southiield 20 Troy 21 Allen Part 22 Dearhorn 23 Dearhorn Heights 24 Detroit 25 Ecorse 26 Garden City 27 Grosse Pointe Fares 28 Crosse Pointe Park 29 Grosse Pointe Hoods 3D Hantranck 31 Harper Hoods 32 Highland Park 33 lntster 34 Lincoln Park 35 Livonia 36 Helvindale 37 Plynouth 38 River Rouge 39 Riverviev 44 Ronulus 41 Southgate 42 Taylor 43 Trenton 44 Hayne 45 Hestland 46 Hoodhaven 47 Hyandotte H 72 123084513 34290839 41546039 139761013 234812596 246328669 466218035 54516714 113545716 43919734 36110775 72642916 54334046 87620672 35267805 109415923 117630612 221061467 253871819 170011027 136498294 325624551 254429307 2596522245 24177876 102285016 87204035 74242785 121935072 33357565 67382811 33044460 78591087 121966373 400991903 31407148 37666438 18996328 34909251 56170903 91736472 188827843 76508651 52039028 207312888 23402918 93226731 173 ltpupenasiioc It. (czarrt:,) Tlauzizflalea \haliuas H 77 119882567 42265157 43555677 133583494 255613222 322373923 508301125 53535054 118477308 38925703 40573216 62080797 45585328 81544795 69894657 86294343 95237423 202163330 256046210 240096262 121179248 340063603 221929906 1770490521 23827866 92123809 83131102 63341920 108943326 24918178 57695999 27165981 63870581 119354033 381643984 26443973 36700119 19581808 38372364 52577796 87097653 157456698 75758828 44955180 189301105 28740350 83866404 H 82 106661863 37229863 38844282 116897217 244617927 363461895 442958968 56539030 143535826 42785172 43827144 56454894 40029459 79459409 92277633 81888492 93036539 221877023 263877942 306931820 125956949 354212062 245135640 1331840396 20643692 99279193 107816268 84772657 133646938 19626080 '63068249 19409214 58343191 118455931 431404193 24895550 41313209 13845443 43676443 50814528 87285122 157751350 77232819 45412134 204086667 33442807 87497682 H 87 105609696 38727861 37817911 118343166 248106851 368504942 439803751 53338211 155490327 42152910 43083432 52098361 37843844 77187181 102793877 69422574 86881911 212861511 231520237 324569968 102081783 291057053 199510432 1136250835 16397170 83819168 112017508 79693894 122603299 17062044 55217601 15412113 48200634 92866526 402191474 20455283 36519543 12359044 36457150 41652144 72295158 124329863 62824944 38432788 167136837 27675191 72318073 77-72 '3201946 7974319 2009638 '6177519 20800626 76045254 42083090 -981660 4931592 -4994030 4462441 -10562118 -8748718 '6075878 34626851 '23121580 -22393189 -18898136 2174391 70085235 -15319046 14439052 ‘32499402 -826031724 -350009 -10161207 -4072933 -10900865 -l2991746 -8439387 -9686812 -5878478 '14720506 -2612340 '19347919 -4963175 '966319 585481 3463113 73593107 -4638819 ~28071145 -749823 -7083848 -18011783 5337432 '9360327 82-77 -13220704 -5035294 -4711395 -16686277 '10995295 41087972 -65342157 3003976 25058518 3859469 3253929 '5625903 -5555870 '2085386 22382977 -4405851 -2200884 19713693 7831733 66835558 4777701 14148459 23205735 438650125 -3184174 7155384 24685167 21430737 24703612 -5292098 5372250 -7756767 -5527389 -898101 49760210 -1548423 4613090 '5736365 5304079 -1763268 187469 294652 1473991 456954 14785561 4702458 3631278 87-82 -1052167 1497998 ~1026371 1445949 3488924 5043047 -3155217 -3200819 1 1954501 -632262 -743713 -4356533 -2185615 '2272228 10516244 -12465918 -6154628 '9015512 -32357705 17638147 '23875166 -63155009 -45625208 -195589561 -4246522 -15460025 4201240 -5078763 -11043639 -2564036 -7850648 -3997100 -10142557 -25589405 '29212719 -4440268 -4793665 -1486399 -7219294 -9162385 -14989964 '33421487 '14407875 -6979346 -36949830 -5767616 -15179610 1J7le .Apnpezmdioc Al (czaru:.) karisflalea Vhilinas City H H lane 72 77 1 East Detroit 3081242 2533388 2 Fraser 19756785 16524228 3 Hount Cleaens 24701589 14431188 4 Roseville 33042122 28331449 5 St. Clair Shores 10004719 7028869 6 Sterling Heights 194699727 147637682 7 Harren 423795654 296179970 8 8ertley 2468275 2033263 9 Dirninghan 5244263 3565968 10 Clavson 6064091 5065339 11 Farnington 6651216 4155609 12 Ferndale 32495845 21637845 13 Hazel Part 10487249 7498414 14 Hadison Heights 58915371 51407703 15 Hovi 15192250 17760663 16 Oak Part 28559095 18689574 17 Pontiac 304123096 217332427 18 Royal Oat 26171945 19136659 19 Southfield 31262664 23738685 20 Troy 103855689 87986485 21 Allen Part 30260928 26985439 22 Dearhorn 488000008 296017744 23 Dearhorn Heights 8587671 5202595 24 Detroit 1524615828 859108172 25 Ecorse 143421824 100472998 26 Garden City 2593535 3142490 27 Grosse Pointe Paras 0 0 28 Grosse Pointe Park 0 0 29 Grosse Pointe Hoods 67283 54872 30 Hantranct 66946355 35925778 31 Harper Hoods 428382 300665 32 Highland Part 89375422 40020007 33 Intster 6107187 4194743 34 Lincoln Park 8618381 6478755 35 Livonia 262451117 203539780 36 Helvindale 18802946 17308496 37 Plynouth 16406176 11508601 38 River Rouge 161529853 106548877 39 Riverviev 23131000 15262854 40 Ronulus 75779419 67841153 41 Southgate 4730919 4033585 42 Taylor 38771066 36060203 43 Trenton 179467592 137536362 44 Hayne 53971437 42833872 45 Hestland 6919578 5578290 46 Hoodhaven 62672500 42857261 47 Hyandotte 56388319 45779286 H 82 2094288 24846036 13884241 33890973 7471335 198333596 302795098 1170020 3118933 4900131 3599231 15364651 5532252 50607206 27459991 17803718 209059859 16033482 12346028 100466022 23752119 385427577 4556141 570277197 69543253 2945898 0 0 53640 28095054 239562 28666388 3159353 5117075 261054712 14700372 8615485 75575159 13500762 78939048 3601978 31929304 108600273 37388127 10168608 29282652 24064978 H 87 17104728 37967698 8606795 220523132 296047012 1406139 3616504 4583111 3613503 15046655 5226547 62649010 25097110 17045568 227358402 18422127 14501270 137429184 24653817 378124459 3417383 508125533 82651992 2920207 0 0 51454 19480288 442598 33505424 3526601 8169296 245380889 13418046 6699426 49145555 9094863 99813224 4473079 31041936 104090608 68551347 13453480 37351487 24027587 H 77772 710270402 74710673 72975850 747062045 7127615684 7435012 71678295 7998752 72495607 710858000 72988834 77507668 2568413 79869521 786790669 77035287 77523979 715869204 73275489 7191982264 73385076 7665507656 742948826 548955 0 0 712410 731020577 7127718 749355415 71912444 72139626 758911337 71494450 74897575 754980976 77868146 77938266 7697334 72710863 741931230 711137565 71341289 719815238 710609032 H 82777 50695913 6615128 7863243 7447036 7165208 7556377 76273194 71966162 7800497 9699329 7885856 78272569 73103177 711392657 12479537 73233320 89409833 7646454 7288830975 730929744 7196592 0 0 71233 77830725 761102 711353619 71035390 71361680 57514933 72608124 72893116 730973718 71762092 11097895 7431607 74130899 728936089 75445745 4590319 713574610 721714308 87782 331352 10947508 3220487 4076725 1135460 22189537 76748086 236119 497572 7317020 14272 7317996 7305705 12041804 72362882 7758150 18298543 2388645 2155242 36963162 901699 77303118 71 138758 762151664 13108739 725691 0 0 72186 78614766 203036 4839036 367248 3052221 715673824 71282325 71916059 726429604 74405899 20874176 871102 7887368 74509665 31163220 3284871 8068835 737391 175 Appendix A (cont.) 1681:1683112 \REJJJEKS City R R R V V lane 77 82 87 77 82 1 East Detroit 1032 1344 1863 22982 17819 2 Fraser 955 1098 1419 25748 28243 3 Hount Clenens 1138 1213 1609 19058 18665 4 Roseville 895 1107 1392 19081 18257 5 St. Clair Shores 1028 1251 1627 23449 22878 6 Sterling Heights 936 849 1530 31972 33327 7 Harren 1028 1174 1733 24017 23176 8 Berkley 1140 1202 1425 19922 19420 9 Dirninghan 1342 1812 2500 37504 39922 10 Clavson 940 1133 1430 22411 22564 11 Farnington 1192 1538 2204 34230 34265 12 Ferndale 1109 1232 1497 15021 13089 13 Hazel Park 1049 1167 1407 14241 12640 14 Hadison Heights 1304 1525 2151 20295 19439 15 Hovi 1084 1109 1656 37781 41570 16 Oak Park 1518 1653 1919 21871 19136 17 Pontiac 1228 1244 1548 13870 12165 18 Royal Oak 1309 1548 1951 22541 22231 19 Southiield 1429 1854 2386 34789 31999 20 Troy 1066 1238 1863 38689 44066 21 Allen Park 1116 1574 1673 23880 23990 22 Dearborn 1505 1833 1820 24483 24417 23 Dearhorn Heights 1104 1250 1489 23463 23192 24 Detroit 1070 1194 1430 12350 9667 25 Ecorse 1352 1300 1401 12401 10954 26 Garden City 995 1290 1542 21338 21537 27 Grosse Pointe Farns 1317 1664 2055 46230 45931 28 Grosse Pointe Park 1317 1664 2055 43672 45630 29 Grosse Pointe Hoods 1317 1664 2055 37959 38591 30 Hantranck 1165 1200 1442 8380 6841 31 Harper Hoods 1163 1525 1857 22884 21487 32 Highland Park 1266 1112 1372 10685 8383 33 lntster 1092 1151 1407 15806 14021 34 Lincoln Park 973 1017 1143 17668 16875 35 Livonia 1109 1369 1803 31115 32490 36 Helvindale 1100 1300 1832 15579 14485 37 Plynouth 1009 1069 1300 30945 30085 38 River Rouge 1447 1170 1463 11967 9711 39 Riverviev 1182 1817 1498 29519 27781 40 Ronulus 981 1225 1681 19470 19713 41 Southgate 1004 1228 1380 20982 21110 42 Taylor 981 1188 1472 18957 18171 43 Trenton 1175 1545 1748 29049 29854 44 Hayne 1162 1232 1532 19417 18381 45 Hestland 1144 1266 1602 22712 22448 46 Hoodhaven 1086 996 1299 31029 32559 47 Hyandotte 1069 1304 1528 17755 17296 0.06522 0.05727 0.05870 0.07021 0.06363 0.06115 0.05952 0.06861 0.06983 0.06222 0.05885 0.07436 0.06423 0.06947 0.05878 0.05855 0.06045 0.05937 0.06008 0.07488 0.05984 0.06224 0.06543 0.06768 0.06299 0.06095 0.05834 0.08195 0.06700 0.05738 0.06455 0. 06722 0.06431 0.06005 0.06165 0. 06397 0.06841 0.07158 0.06344 0.07313 0.06935 0.06071 0.07093 P 82 0.06244 0.06475 0.07418 0.06041 0.06113 0.05755 0.05948 0.06665 0.05276 0.06059 0.05287 0.07419 0.07244 0.05945 0.05251 0.07577 0.07061 0.06009 0.06023 0.05440 0.06301 0.04872 0.05554 0.07887 0.06002 0.06773 0.05004 0.05413 0.05030 0.06652 0.05715 0.08343 0.07385 0.05947 0.06018 0.06779 0.06431 0.05859 0.06971 0.06226 0.06954 0.07129 0.06491 0.07418 0.07124 0.06377 0.06977 p 87 0.06680 0.07293 0.07720 0.06713 0.06042 0.05947 0.06021 0.07143 0.05447 0.06714 0.05360 0.07681 0.07253 0.05622 0.05825 0.08223 0.08162 0.06937 0.06055 0.05156 0.07099 0.05538 _ 0.05764 0.08208 0.07672 0.07090 0.05457 0.05935 0.05573 0.08754 0.06175 0.09027 0.07753 0.08112 0.06010 0.08505 0.06892 0.07319 0.06480 0.06977 0.06900 0.07643 0.06569 0.07828 0.07974 0.07303 0.07128 City 1 lane 77 1 East Detroit 0.02368 2 Fraser 0.01842 3 Hount Clenens 0.07992 4 Roseville 0.02245 5 St. Clair Shores 0.01875 6 Sterling Heights 0.01814 7 Harren 0.01950 8 Derkley 0.03194 9 Dirninghan 0.01914 10 Clavson 0.01461 11 Farnington 0.01547 12 Ferndale 0.03065 13 Hazel Park 0.03204 14 Hadison Heights 0.02085 15 Hovi 0.01207 16 Oak Park 0.02542 17 Pontiac 0.03874 18 Royal Oat 0.01960 19 Southiield 0.01383 20 Troy 0.01400 21 Allen Park 0.01953 22 Dearborn 0.01412 23 Dearborn Heights 0.02534 24 Detroit 0.06448 25 Ecorse 0.02094 26 Garden City 0.02171 27 Grosse Pointe Ferns 0.02140 28 Grosse Pointe Park 0.01823 29 Grosse Pointe Hoods 0.01060 30 Hantranck 0.02794 31 Harper Hoods 0.02039 32 Highland Park 0.04677 33 lnkster 0.03756 34 Lincoln Park 0.02702 35 Livonia 0.01550 36 Helvindale 0.02211 37 Plynouth 0.02006 38 River Rouge 0.01875 39 Rivervieu 0.01778 40 Ronulus 0.01435 41 Southgate 0.01645 42 Taylor 0.01926 43 Trenton 0.01645 44 Hayne 0.01964 45 Hestland 0.02682 46 Hoodhaven 0.01416 47 Hyandotte 0.02602 1 82 176 Appendix A (cont. ) 1laflii£83113 16111133. 1 87 0.02620 0.01793 0.03073 0.02672 0.02562 0.01671 0.01659 0.01810 0.01732 0.01624 0.01299 0.03180 0. 03677 0.02133 0.00958 0. 02593 0.02679 0.02039 0.01345 0.01523 0. 01383 0.01440 0.01455 0.08824 0.02010 0.02790 0.01052 0.01495 0.00714 0.02115 0.01767 0.03127 0.03248 0.02756 0.01220 0.02516 0.01944 0.01838 0.01521 0.00703 0.01822 0.02013 0.01693 0.02113 0.02153 0.01249 0.02391 0.03277 0.02677 0.03987 0.05055 0.03012 0.02077 0.02335 0.03106 0.01938 0.02415 0.01920 0.04521 0.05577 0. 02543 0.01595 0.04086 0.03327 0.02718 0.01874 0.01561 0.02755 0.01847 0.03860 0.17438 0.01828 0.03183 0.02014 0.02561 0.02098 0.05980 0.03287 0.09251 0.05519 0.05590 0.01996 0.03641 0.02628 0.04017 0.02118 0.01612 0.04341 0.03667 0.01986 0.02502 0.03099 0.02320 0.02653 0.00000 0.00000 0.02378 0.00000 0.00000 0 . 00000 0.00000 0.00000 0.00000 0.03867 0.00000 0.00000 0.00000 0.00000 0.00000 0.02323 0.00000 0 . 02054 0.00000 '6 ° °°§§§§§§§§§§° § 2 5% 0.00000 0.00000 0.00000 0.00000 0.00000 0.00000 0.00000 0.00000 0.00000 0.00000 0.00000 0.00000 0.00000 0.00000 0.00000 0.00000 0.01969 0.00000 0.00000 0.00000 0.00000 0.00000 0.00000 0.11523 0.00000 0.00000 0.00000 0.00000 0.00000 0.01844 0.00000 0.03649 0.00000 0.00000 0.00000 0.00000 0.00000 0.00000 0.00000 0.00000 0.00000 0.00000 0.00000 0.00000 0.00000 0.00000 0.00000 0.00000 0.00000 0.00000 0.00000 0.00000 0.00000 0.00000 0.00000 0.00000 0.00000 0.00000 0.00000 0.00000 0.00000 0.00000 0.00000 0.02786 0.00000 0.00000 0.00000 0.00000 0.00000 0.00000 0.15094 0.00000 0.00000 0.00000 0.00000 0.00000 0.03597 0.00000 0.11208 0.00000 0.00000 0.00000 0.00000 0.00000 0.00000 0.00000 0.00000 0.00000 0.00000 0.00000 0.00000 0.00000 0.00000 0.00000 0.02227 0.01940 0.08709 0.02425 0.02014 0.01673 0.02361 0.02840 0.01830 0.01857 0.03176 0.02845 0.03721 0.02171 0.01462 0.02273 0.12781 0.01881 0.01442 0.01216 0.01662 0.01900 0.01961 0.11964 0.01404 0.02719 0.00979 0.01128 0.00808 0.04065 0.01662 0.05898 0.05167 0.02838 0.01839 0.02020 0.01763 0.01604 0.03317 0.02041 0.01726 0.02340 0.00973 0.02054 0.02396 0.00859 0.13049 9 82 0.01987 0.01816 0.03473 0.02925 0.02386 0.01539 0.01635 0.01399 0.01410 0.01392 0.00933 0.03107 0.03165 0.01970 0.03703 0.02629 0.12456 0.02837 0.01207 0.01551 0.01262 0.01509 0.01193 0.16539 0.00986 0.02030 0.01172 0.00898 0.01195 0.03878 0.01035 0.05964 0.03695 0.02857 0.01431 0.02262 0.01325 0.01923 0.02072 0.01070 0.01795 0.02436 0.00948 0.01528 0.02203 0.00905 0.16710 9 87 0.02668 0.02087 0.05643 0.03890 0.03416 0.01928 0.02400 0.03382 0.03824 0.02591 0.01878 0.04456 0.05223 0.03514 0.02015 0.04083 0.11640 0.03775 0.02273 0.02530 0.02675 0.02287 0.03649 0.23223 0.03540 0.03275 0.01727 0.01845 0.01974 0.08533 0.02508 0.06775 0.06000 0.08585 0.01878 0.05168 0.04469 0.03810 0.08229 0.02374 0.03531 0.03715 0.01772 0.02676 0.03807 0.01954 0.02965 City nagg lane 77 1 East Detroit 2261447 2 Fraser 3398160 3 Hount Clenens 2896807 4 Roseville 2559730 5 St. Clair Shores 2705497 6 Sterling Heights 5687172 7 Harren 3448557 8 Berkley 3393240 9 Birninghan 3823546 10 Clavson 4537360 11 Farnington 3371671 12 Ferndale 6282214 13 Hazel Park 7257565 14 Hadison Heights 4563711 15 Hovi 3685626 16 Oak Park 4621098 17 Pontiac 2686610 18 Royal Oak 4735780 19 Southfield 2918723 20 Troy 4135587 21 Allen Part 10783664 22 Dearhorn 11104172 23 Dearborn Heights 4293557 24 Detroit 10106348 25 Ecorse 7930430 26 Garden City 4342780 27 Grosse Pointe Farns 2306431 28 Grosse Pointe Park 2306431 29 Grosse Pointe Hoods 1826248 30 Hantranck 7427177 31 Harper Hoods 2370376 32 Highland Park 7402610 33 lnkster 3726130 34 Lincoln Part 11277361 35 Livonia 1995625 36 Helvindale 12724920 37 Plynouth 2196127 38 River Rouge 8703528 39 Riverviev 9967429 40 Ronulus 2009914 41 Southgate 8962682 42 Taylor 3400699 43 Trenton 4339694 44 Hayne 896177 45 Hestland 4101833 46 Hoodhaven 6799195 47 Hyandotte 12831452 177 Itpqpenaciiic £1 (k:cr1t:.) 1681:1283113 189111883 3085785 3184156 6398873 3456040 2924007 3728522 4156910 3278384 5280223 6026614 4020791 4470632 3636702 3139563 4228336 2741619 4333782 8405688 8172441 4911450 9247289 6602409 4724548 2025089 2025089 1931458 5889471 2205908 6984716 4253083 8863727 1598140 9893756 2977844 6931504 7122766 1801949 6560662 3319381 3020110 928235 4801836 5121701 9691195 2593223 4028054 4851477 3538257 3660163 7642950 3957826 3100726 4130567 4552413 fi%%7 5040829 5748610 4210243 4169473 3883192 4196311 4522685 2602704 4690497 7645538 7351867 4670418 8534136 5255517 5261543 1963319 1963319 1992466 5951856 2188018 7164788 4776639 7965045 1380346 9021962 2798536 7369336 7859922 2851903 6966889 3253752 2951973 1029481 5457063 4663267 8733118 792437 2302427 1629650 5694170 2726696 7396792 569252 3493378 11319397 1664057 895799 2686610 3696635 12257464 431751 6280031 37914339 528150 0 0 17148 17524770 117908 13800002 676571 1098094 5741602 6531508 5352838 41783873 3508702 1938319 584578 1524744 18840598 7138979 277527 6543093 9537351 410645 6211509 3560062 3567471 644081 5418951 8802183 508704 693096 2227332 1439692 4152608 2048982 7442236 918394 3423792 11179672 1418892 487985 3139563 3298905 16194436 376541 4205584 25756760 535618 0 0 17303 14047527 95825 10237996 517927 882254 7501572 5653989 4307742 31489650 3139712 2255401 545754 1370356 15083371 6231354 513566 4575414 5469313 3934476 732493 6008805 8631108 573934 803668 2083232 1417060 3959646 1900563 9014246 804395 3186088 11841583 1601924 547218 4196311 3377235 15657327 283600 3714368 31189431 490791 0 0 16079 9502579 173568 11553594 568807 1384626 6921887 5063414 3116012 19272767 2090773 2851806 648272 1312555 14258987 1 1425224 669327 5702517 5005747 3120 2970 3420 1650 6776 178 Mix A (cont.) Variable Values City area area area cagg cagg lane 77 B2 87 77 82 1 East Detroit 5.10 5.10 5.10 3834705 3395125 2 Fraser 4 4.10 4.10 4.10 3887182 3734504 3 Hount Clenens 3.90 3.90 3.90 3150095 3207223 4 Roseville 9.65 9.65 9.65 4554233 4105839 5 St. Clair Shores 11.75 11.75 11.75 5086115 4577216 6 Sterling Heights 36.70 36.70 36.70 3851695 4242790 7 Harren 34.30 34.30 34.30 4123952 3751415 8 Berkley 2.45 2.45 2.45 6389621 5882188 9 Birninghaa 4.50 4.50 4.50 5855573 5421728 10 Clavson 2.20 2.20 2.20 6074840 5430256 11 Farsington 2.55 2.55 2.55 5086023 4790620 12 Ferndale 3.80 3.80 3.80 6508184 5196819 13 Hazel Park 2.75 2.75 2.75 5650007 4269785 14 Hadison Heights 6.95 6.95 6.95 5718507 4756682 15 Hovi 31.20 31.20 31.20 5344430 4627236 16 Oak Park 5.35 5.35 5.35 6859879 5437613 17 Pontiac 19.20 19.20 19.20 4001765 4975281 18 Royal Oak 11.50 11.50 11.50 6049684 5310860 19 Southiield 26.50 26.50 26.50 5556099 4990688 20 Troy 32.75 32.75 32.75 5773368 5351668 21 Allen Park 7.30 7.30 7.30 3908853 3430353 22 Dearhorn 24.15 24.15 24.15 3374639 2772495 23 Dearhorn Heights 12.05 12.05 12.05 3863769 3489117 24 Detroit 136.80 136.80 136.80 6166529 4722821 25 Ecorse 2.65 2.65 2.65 4065786 3651502 26 Garden City 5.95 5.95 5.95 3361605 3122215 27 Grosse Pointe Farns 2.65 2.65 2.65 4773772 4004800 28 Grosse Pointe Park 2.15 2.15 2.15 4947774 4232551 29 Grosse Pointe Hoods 3.20 3.20 3.20 4565606 4044681 30 Hantranck 2.05 2.05 2.05 6147158 4298867 31 Harper Hoods 2.55 2.55 2.55 3563159 3240699 32 Highland Park 2.90 2.90 2.90 5980910 4813403 33 lnkster 6.20 6.20 6.20 3171823 3124080 34 Lincoln Park 5.90 5.90 5.90 3821073 3451547 35 Livonia 35.45 35.45 35.45 5097597 4843034 36 Helvindale 2.65 2.65 2.65 3749653 3457410 37 Plynouth 2.15 2.15 2.15 2469602 2526093 38 River Rouge 2.55 2.55 2.55 4210939 3836603 39 Riverviev 4.35 4.35 4.35 3515179 3198450 40 Ronulus 35.00 35.00 35.00 3229654 2842729 41 Southgate 6.90 6.90 6.90 3275144 3021238 42 Taylor 23.65 23.65 23.65 3802142 3570007 43 Trenton 7.30 7.30 7.30 3828381 3435684 44 Hayne 6.00 6.00 6.00 2757759 2444428 45 Hestland 20.10 20.10 20.10 2982371 2949729 46 Hoodhaven 6.55 6.55 6.55 3926559 3571315 47 Hyandotte 4.80 4.80 4.80 3484022 2987504 3611868 4455888 4926245 5015234 4070441 6685233 6215075 6509047 5917737 5981906 4846578 5377320 5876261 6203987 5316167 6234816 5873548 6309321 3336059 2586573 3812651 5280731 3457948 33 19029 4199050 4370683 4259553 4696747 3344449 5467244 3233025 3263266 5678287 3315996 3090102 3693944 2958122 2853885 2685140 3496835 3093853 2470429 3120144 3238350 2756385 4708864 4421713 3335855 2448797 3963123 3362400 8327924 7793761 8027145 5265468 5242431 5047879 940960 4416844 3459727 7058725 11703479 4001765 3273958 5830467 2463102 5037937 2287355 2827827 2559663 1341647 3400599 5470854 11892503 11226834 3018538 4979197 2661716 3957315 5974198 1746092 3758199 822703 5548611 3608256 2197219 3252244 3806130 1706324 4520513 3008386 4743956 4271462 3248545 2300497 3758699 3083397 9505810 5919756 6739922 3954101 4572518 5228795 1599257 4220486 2975179 5396454 11574475 4975281 3220845 6739676 2592535 3474200 1895710 2609054 2721664 1127406 3239754 3577112 9101129 5734817 2132749 4095225 2514549 3248636 6690355 1376497 3240257 951702 4456650 2916053 2437661 3107835 3464473 1759508 4805952 cden 87 3975711 3440844 4669534 4686438 3527753 2792436 4672455 3817888 12113715 5752271 8644811 4481799 4196814 6631469 2274259 4177538 3202295 5893583 14929387 5316167 2762116 7748437 2413541 3359592 1813182 2536699 HNMZ 1075817 3180572 381 1803 9601448 6773077 2235132 3954680 3107712 2762783 7379674 1137237 2626981 907968 4616882 2840898 2292708 2891519 3888334 1570226 3814281 179 Appendix A (cont.) Variable Values City educ educ educ lane 77 82 B7 1 East Detroit 12.05 12.30 12.56 2 Fraser 12.40 12.57 12.75 3 Hount Clenens 12.22 12.53 12.84 4 Roseville 12.21 12.36 12.52 5 St. Clair Shores 12.34 12.44 12.54 6 Sterling Heights 12.54 12.64 12.74 7 Harren 12.24 12.34 12.44 8 Berkley 12.46 12.53 12.61 9 Birninghan 15.22 16.06 16.95 10 Clavson 12.43 12.55 12.68 11 Farnington 12.80 12.80 12.80 ' 12 Ferndale 12.31 12.46 12.62 13 Hazel Park 11.61 12.29 13.02 14 Hadison Heights 12.31 12.46 12.62 15 Hovi 12.75 13.00 13.26 16 Oak Park 12.64 12.74 12.84 17 Pontiac 11.68 12.42 13.21 18 Royal Oak 12.68 12.88 13.09 19 Southiield 12.91 13.06 13.21 20 Troy 13.16 13.57 14.00 21 Allen Park 12.41 12.56 12.72 22 Dearhorn 12.41 12.56 12.72 23 Dearhorn Heights 12.34 12.44 12.54 24 Detroit 11.77 12.34 12.95 25 Ecorse 11.17 12.11 13.15 26 Garden City 12.24 12.34 12.44 27 Grosse Pointe Fares 15.50 16.56 17.70 28 Grosse Pointe Park 14.15 . 14.58 15.02 29 Grosse Pointe Hoods 13.81 14.49 15.20 30 Hantranck 10.62 11.47 12.41 31 Harper Hoods 12.36 12.43 12.51 32 Highland Park 11.90 12.42 12.96 33 lnkster 12.05 12.30 12.56 34 Lincoln Park 11.90 12.42 12.96 35 Livonia 12.64 12.74 12.84 36 Helvindale 11.92 12.23 12.54 37 Plynouth 12.67 12.89 13.11 38 River Rouge 12.11 11.93 11.75 39 Riverviev 12.46 12.53 12.59 g 40 Ronulus 11.93 12.57 13.25 41 Southgate 12.31 12.46 12.62 42 Taylor 12.05 12.30 12.56 43 Trenton 12.47 12.69 12.91 44 Hayne 12.14 12.41 12.70 45 Hestland 12.34 12.44 12.54 46 Hoodhaven 12.57 12.79 13.03 47 Hyandotte 11.87 12.44 13.05 hden 77 23506386 10308575 11168122 13842849 21754317 8784031 14819275 21851042 26328291 17693501 1591 1065 16337052 16576483 11733064 2240213 16129784 4960282 17579420 9662121 7331184 16599897 14081309 18417420 12942182 8991648 15482993 31370227 29461358 34044789 12155209 22625882 9367580 10301707 20229497 10765698 9978858 17069823 7679141 8821233 1502223 12622848 6657789 10377922 7492530 9417965 4387840 17472168 hden 82 20914091 9080454 9960072 12113701 20818547 9%5“ 12914256 23077155 31896850 19447806 17187115 14856551 14556167 11433008 2957616 15306260 4845653 19293654 9957658 9371964 17254377 14667166 20343207 9735675 7790073 16685579 40685384 39429143 41764668 9573698 24732647 6692832 9410192 20077277 12169371 9394547 19215446 5429586 10040562 1451844 12650018 6670247 10579838 7568689 10153566 5105772 18228684 hden 87 20707784 9445820 9696900 12263541 21115477 10041007 12822267 21770698 34553406 19160414 16895463 13710095 13761398 11106069 3294676 12976182 4525100 18509697 8736613 9910533 13983806 12052052 16556882 8305927 6187611 14087255 42270758 37066927 38313531 8322948 21653961 5314522 7774296 15740089 11345317 7718975 16985834 4846684 8380954 1190061 10477559 5257077 8606157 6405465 8315266 4225220 15066265 City lane Inn 77 180 leap Y Y 1 East Detroit 2 Fraser 3 Hount Clenens 4 Roseville 5 St. Clair Shores 6 Sterling Heights 7 Harren 8 Berkley 9 Birninghaa 10 Clavson 11 Far nington 12 Ferndale 13 Hazel Park 14 Hadison Heights 15 Hovi 16 Oak Park 17 Pontiac 18 Royal Oak 19 Southlield 20 Troy 21 Allen Park 22 Dearhorn 23 Dearhorn Heights 24 Detroit 25 Ecorse 26 Garden City 27 Grosse Pointe Farns 28 Grosse Pointe Park 29 Grosse Pointe Hoods 30 Hantranck 31 Harper Hoods 32 Highland Park 33 lnkster 34 Lincoln Park 35 Livonia 36 Helvindale 37 Plynouth 38 River Rouge 39 Riverviev 40 Ronulus 41 Southgate 42 Taylor 43 Trenton 44 Hayne 45 Hestland 46 Hoodhaven 47 Hyandotte 13.6 14.1 43.3 19.2 16.9 19.7 36.8 10.8 22.9 14.4 18.6 15.1 16.5 17.5 16.8 12.7 67.0 19.7 31.6 28.1 17.1 47.5 10.4 59.2 31.6 12.8 16.9 7.5 11.0 22.2 15.1 23.2 14.4 19.6 30.7 12.6 23.8 27.7 14.7 - 26.0 14.8 22.3 32.5 25.5 15.4 12.5 29.5 0...... coo-.0000... MU‘OMQOONNUONM‘NMNOM 88 :3 E: 86 23 :3 5: 86 5: =5 23 :3 23 53 S: 85 ES 83 :2 ~o .Apnperniiac £1 (cxucn:.) 1881:1883112'101111683 s 1 pden 87 77 82 87 77 13.6 12910 10424 12601 5585 14.1 13526 11048 14430 5840 43.3 9999 8220 10666 4828 19.2 12560 9826 12570 5692 16.9 14331 11635 14666 5952 19.7 15894 12734 16636 3864 36.8 14077 11392 14573 5263 10.8 13118 10722 13921 5889 22.9 17171 13241 19611 5205 14.4 13936 11478 15701 5514 18.6 15403 11760 17358 3980 15.1 10747 8494 11339 7070 16.5 10579 8126 10889 7524 17.5 12833 10092 14207 6290 16.8 15793 12876 18537 3740 12.7 13505 10500 12904 7132 67.0 9674 7449 11556 1898 19.7 13416 10903 14267 5886 31.6 16345 12289 15807 6048 28.1 17664 14350 20787 5281 17.1 15640 12889 14539 5740 47.5 13239 10491 12553 5776 10.4 14757 12004 14110 5370 59.2 8742 6904 8266 7363 31.6 9708 7765 9045 5534 12.8 14754 12106 14431 4530 . 16.9 22036 18168 22945 7017 7.5 18596 14099 18725 6945 11.0 19297 15430 22410 6375 22.2 7419 5602 6481 7900 15.1 12996 10637 12075 6712 23.2 6740 5156 6202 6700 14.4 11069 8894 10821 4076 19.6 12571 9802 11286 5314 30.7 17340 14292 17499 4941 12.6 12062 9387 11332 5982 23.8 12740 10155 12375 2705 27.7 8703 6452 7902 6093 14.7 16250 12921 14893 5053 26.0 12271 9598 12099 4488 14.8 14620 11583 13503 4935 22.3 13060 10424 12659 4884 32.5 16073 12890 15114 4236 25.6 12334 10522 11992 4376 15.4 13435 10851 13405 4110 12.5 15780 12603 15381 4550 29. 5 11542 8967 10431 4803 82 5220 4537 5178 5419 3762 4972 5529 4912 5150 3672 6498 6927 5803 3504 6570 2047 5469 5584 4948 5270 5290 4993 4995 4219 6002 5695 7196 5978 6091 3692 4803 4452 5335 2551 4506 4283 4472 4544 4062 4112 4347 87 4810 5102 4905 3628 4836 5412 4779 5047 3683 6281 6696 5688 3377 6397 2056 5366 5445 4799 5168 5181 4857 6590 4928 4126 6115 5647 5475 7046 5731 5979 3601 4725 4377 5274 2601 5369 4372 4149 4300 4397 3769 3739 4001 4306 City Y lane 77 1 East Detroit 13616 2 Fraser 13295 3 Hount Clenens 14078 4 deeville 14719 5 St. Clair Shores 14228 6 Sterling Heights 14525 7 Harren 11716 8 Berkley 13494 9 Birninghan 13811 10 Clavson 13774 11 Farnington 14555 12 Ferndale 12013 13 Hazel Park 11178 14 Hadison Heights 13183 15 Hovi 16372 16 Oak Park 11828 17 Pontiac 17664 18 Royal Oak 13398 19 Southfield 14099 20 Troy 14554 21 Allen Park 11825 22 Dearborn 11812 23 Dearborn Heights 13097 24 Detroit 10467 25 Ecorse 12835 26 Garden City 12898 27 Grosse Pointe Farns 14421 28 Grosse Pointe Park 14913 29 Grosse Pointe Hoods 15358 30 Hantranck 10177 31 Harper Hoods 15319 32 Highland Park 11662 33 lnkster 14092 34 Lincoln Park 12575 35 Livonia 13996 36 Helvindale 11996 37 Plynouth 15523 38 River Rouge 12542 39 Riverviev 13624 40 Ronulus 12931 41 Southgate 13744 42 Taylor 13539 43 Trenton 14250 44 Hayne 13224 45 Hestland 13681 46 Hoodhaven 14309 47 Hyandotte 13187 181 lAqipxzruiiJc It. (cnarut. ) \erisflalea 18111153 (I (I (I 82 87 77 82 87 11 11 2260 3057 2719 15 15 606 870 836 20 20 1902 1737 1650 13 13 2483 4223 4805 14 14 3173 3477 3567 17 17 3893 4954 5395 11 11 9210 10507 10487 14 14 640 678 559 17 17 851 1154 1254 15 15 406 469 543 19 19 471 479 518 10 10 1642 2050 2221 10 10 1822 2620 2255 12 12 1778 2339 2703 25 25 822 1274 1887 12 12 1604 2131 2509 25 25 8525 9238 8590 14 14 3328 4067 4035 15 15 5336 7692 7665 18 18 3004 3697 4719 9 9 1610 1728 1510 9 9 7066 7613 7661 12 12 3802 4388 3540 4 4 123748 152962 138411 8 8 1483 1573 1338 15 15 1841 1990 1548 9 9 490 646 485 7 7 619 908 722 10 10 478 546 428 5 5 2081 2693 2895 10 10 1759 1959 2097 6 6 5128 1361 4461 14 14 2948 2897 2722 9 9 3418 4116 3590 18 18 4508 5189 5482 8 8 1715 1261 1068 22 22 509 443 323 6 6 980 1355 1195 13 13 710 639 397 18 18 2133 2492 2469 12 12 2143 2249 2211 14 14 4028 4528 5782 15 15 1006 883 614 18 18 1728 1856 1939 17 17 4767 4810 5072 17 17 509 643 629 10 10 2296 2035 1504 City lane 3 a 3' a 23 1112 Itpqpennciiic 11 (32(1112.) Vkrrisfla1ea 16111833 I a Q N E! N ‘6 ii :3 1 East Detroit 2 Fraser 3 Hount Clenens 4 Roseville 5 St. Clair Shores 6 Sterling Heights 7 Harren 8 Berkley 9 Birninghan 10 Clavson 11 Farnington 12 Ferndale 13 Hazel Park 14 Hadison Heights 15 Hovi ‘ 16 Oak Park 17 Pontiac 18 Royal Oak 19 Southiield 20 Troy 21 Allen Park 22 Dearhorn 23 Dearhorn Heights 24 Detroit 25 Ecorse 26 Garden City 27 Grosse Pointe Farns 28 Grosse Pointe Park 29 Grosse Pointe Hoods 30 Hantranck 31 Harper Hoods 32 Highland Park 33 lnkster 34 Lincoln Park 35 Livonia 36 Helvindale 37 Plynonth 38 River Rouge 39 Rivervieu 40 Ronulus 41 Southgate 42 Taylor 43 Trenton 44 Hayne 45 Hestland 46 Hoodhaven 47 Hyandotte '8'8 '8'8 NM . I 181:; o 53 05 .. C as . ouooounoooo-NQNO-o—noooosso—owoman—omunowoooonumooo— . C I 53 53 53 53 55 58 58 2; 55 53 53 53 83 58 53 29 53 33 53 53 23 53 53 SB 53 25 53 2; 23 29 58 53 23 28 23 53 23 53 53 53 2; 58 E; 58 53 53 23 _ —O°°°N”N°-—‘—0°°°°° 88388888888888888838883388838888888 O C 53 a o o I g o s c a o s s s I o 0 cu. O I 5’ :7 . 888883888888888888888883833888888838883838888 3.00 0.00 ~—coo—coo—uooo—oococo—nowv—o—oooooooooooococoa—coco -ooo—°o°—_o°°—ococo—no—u—o—ooooooooooococoo—oooo p990033050090330009mo 2' 85 86 53128 28 :3 56 33 5: 55 85 55 86 63 5: :3 23 23 35 55 35 86 -coo-—coo—nooo—ooooo—uo-o—Oo—booooooooooccoco-oooo 0.00 0.00 17.67 1.07 0.20 0.17 0.17 0.01 0.24 0.02 0.00 0.44 0.08 0.49 15.81 0 In no 8882 0.07 3.60 2.60 0.28 0.07 0.00 18.72 1.12 0.20 0.22 0.22 0.01 0.34 0.02 0.00 0.54 0.12 0.84 0.20 14.36 39.18 0.12 10.90 1.08 0.46 0.00 0.24 66.88 39.50 0.12 0.11 0.12 0.24 12.84 0.00 89.62 59.50 0.60 00 ‘2 0.48 0.00 39.10 0.48 18.46 . 00" 1.92 0.12 6.10 2.60 0.48 0.12 0.85 0.17 0.68 0.00 21.11 1.02 2.72 0.17 8.68 2.60 On“ 0.17 City 1 East Detroit 2 Fraser 3 Hoont Clenens 4 Roseville 5 St. Clair Shores 6 Sterling Heights 7 Harren 8 Berkley 9 Dirninghan 10 Clavson 11 Farnington 12 Ferndale 13 Hazel Park 14 Hadison Heights 15 Iovi 16 Oak Park 17 Pontiac 18 Royal Oak 19 Southiield 20 Troy 21 Allen Park 22 Dearhorn 23 Dearhorn Heights 24 Detroit 25 Ecorse 26 Garden City 27 Grosse Pointe Farns 28 Grosse Pointe Park 29 Grosse Pointe Hoods 30 Hantranct 31 Harper Hoods 32 Highland Park 33 lnkster 34 Lincoln Part 35 Livonia 36 Helvindale 37 Plynooth 38 River Rouge 39 Itiverviev 40 Ronulus 41 Southgate 42 7ay1or 43 7renton 44 Hayne 45 Hestland 46 Hoodhaven 47 Hyandotte 13 qupxzruiiJc It.1(cxxnt:.) \Lalziadalxa \haJJdeus 33.96 36.10 41468 37033 30.05 34.18 14170 14233 29.32 30.63 23927 18845 29.32 32.36 56572 52785 35.86 40.51 85582 73450 29.25 32.37 99403 108482 31.93 35.51 168795 156131 30.25 31.61 18990 18152 35. 08 37 . 28 22827 20965 31.99 35.96 15818 14513 45.29 53.00 11118 10577 29.38 29.33 26890 25550 28.41 29.43 20810 20645 29.16 32.06 36002 34609 30.82 33.61 16644 23023 32.77 34.18 32221 30716 25.81 26.33 78920 73156 33.09 35.31 76055 68390 36.89 39.87 76133 73311 31.65 34.01 62173 67031 37.88 41.84 37231 32418 36.64 37.00 93145 86544 34.36 38.26 75560 64702 28.58 28.29 1285351 1138717 28.77 30.20 14459 13956 29.86 34.00 37996 33811 39.10 38.85 14686 10254 33.00 32.76 13006 13761 39.47 41.39 20825 18219 35.68 34.12 20947 20071 48.47 53.91 17423 15461 27.02 26.30 30558 25733 27.74 30.09 31005 33786 31.24 33.09 46471 43533 33.79 38.53 110584 101366 30.89 32.87 11877 11934 32.19 34.42 11286 9745 33.04 36.15 12870 9578 31.50 35.73 13447 14117 26.46 27380 24249 34.83 33953 30679 28.73 79344 73796 39.54 28470 21680 31.46 20115 20520 30.87 89702 81533 30.01 15285 10807 32.77 35910 32526 3833323: 33333333 29232 31221 70499 65640 72810 67330 30808 86389 61137 1123911 13267 32210 8999 14347 15995 18345 14522 25592 31429 42679 100334 10954 10039 11793 13971 24063 30318 72101 21043 21120 81104 11261 31056 30.02 333333333333333 333333223333323 31.29 184 Aggpxanuiisc 11. (cxacnt. ) ‘Véuziaflalxa \halxuas City do do do do do do dv du dv hc hc hc Hane 77 82 87 77 82 87 77 82 87 77 82 G7 1 East Detroit 1 1 1 0 0 0 0 0 0 0.40 0.39 0.38 2 Fraser 1 1 l 0 0 0 0 0 0 0.52 0.51 0.50 3 Hount Clenens 1 1 1 0 0 0 0 0 0 0.28 0.27 0.26 4 Roseville 1 1 1 0 0 0 0 0 0 0.29 0.30 0.31 5 St. Clair Shores 1 1 1 0 0 0 0 0 0 0.44 0.46 0.48 6 Sterling Heights 1 1 1 0 0 0 0 0 0 0.60 0.65 0.71 7 Harren 1 1 1 0 0 0 0 0 0 0.48 0.48 0.49 8 Sertley 0 0 0 1 l 1 0 0 0 0.26 0.28 0.29 9 8irninghan 0 0 0 1 1 1 0 0 0 0.55 0.57 0.58 10 Clavson 0 0 0 1 l l 0 0 0 0.31 0.32 0.33 11 Earnington 0 0 0 1 1 1 0 0 0 0.59 0.54 0.50 12 Ferndale 0 0 0 l 1 1 0 0 0 0.17 0.17 0.17 13 Hazel Part 0 0 0 1 1 1 0 0 0 0.12 0.13 0.14 14 Hadison Heights 0 0 0 1 1 1 0 0 0 0.26 0.26 0.26 15 Hovi 0 0 0 1 1 1 0 0 0 0.53 0.59 0.65 16 Oak Park 0 0 0 1 l 1 0 0 0 0.54 0.54 0.55 17 Pontiac 0 0 0 1 1 1 0 0 0 0.17 0.17 0.18 18 Royal Oak 0 0 0 1 1 1 0 0 0 0.34 0.36 0.38 19 Gouthfield 0 0 0 1 1 1 0 0 0 0.64 0.61 0.58 20 Troy 0 0 0 1 1 1 0 0 0 0.59 0.68 0.77 21 Allen Park 0 0 0 0 0 0 1 1 1 0.44 0.46 0.48 22 Dearborn 0 0 0 0 0 0 1 1 1 0.42 0.44 0.47 23 Dearhorn Heights 0 0 0 0 0 0 1 1 1 0.43 0.45 0.46 24 Detroit 0 0 0 0 0 0 l 1 1 0.22 0.23 0.24 25 Ecorse 0 0 0 0 0 0 1 1 1 0.19 0.20 0.21 26 Garden City 0 0 0 0 0 0 1 l 1 0.28 0.31 0.35 27 Grosse Pointe Farns 0 0 0 0 0 0 1 1 1 0.87 0.86 0.85 28 Grosse Pointe Park 0 0 0 0 0 0 1 1 1 0.65 0.66 0.67 29 Grosse Pointe Hoods 0 0 0 0 0 0 1 1 1 0.76 0.76 0.76 30 Hantranck 0 0 0 0 0 0 1 1 1 0.11 0.09 0.07 31 Harper Hoods 0 0 0 0 0 0 1 1 1 0.36 0.36 0.36 32 Highland Park 0 0 0 0 0 0 1 l 1 0.19 0.20 0.21 33 lnkster 0 0 0 0 0 0 1 1 1 0.19 0.20 0.21 34 Lincoln Part 0 0 0 0 0 0 1 1 1 0.23 0.25 0.27 35 Livonia 0 0 0 0 0 0 1 1 1 0.65 0.68 0.72 36 Helvindale 0 0 0 0 0 0 l 1 l 0.16 0.16 0.16 37 Plynouth 0 0 0 0 0 0 l l 1 0.47 0.49 0.51 38 River Rouge 0 0 0 0 0 0 1 1 l 0.20 0.20 0.19 39 Riverviev 0 0 0 0 0 0 1 1 1 0.46 0.51 0.55 40 Ronulus 0 0 0 0 0 0 1 1 1 0.23 0.28 0.32 41 Southgate 0 0 0 0 0 0 1 1 1 0.33 0.36 0.39 42 Taylor 0 0 0 0 0 0 1 1 1 0.27 0.32 0.36 43 Trenton 0 0 0 0 0 0 1 1 1 0.55 0.55 0.54 44 Hayne 0 0 0 0 0 0 1 1 l 0.29 0.31 0.33 45 Hestland 0 0 0 0 0 0 1 l 1 0.32 0.34 0.35 46 Hoodhaven 0 0 0 0 0 0 1 1 l 0.44 0.51 0.57 47 Hyandotte 0 0 0 0 0 0 1 1 1 0.31 0.24 0.18 Center Line Memphis New Baltimore Richmond Utica QAKLAED_QQQEIX Bloomfield Hills Farmington Hills Huntington Woods Keego Harbor Lake Angelus Lathrup Village Northville Orchard Lake Village Pleasant Ridge Rochester Rochester Hills South Lyon Sylvan Lake Walled Lake Wixom Appendix B Excluded Cities 185 W Belleville Flat Rock Gibraltor Grosse Pointe Rockwood Appendix C Variable Derivations The variables were calculated for each of the 47 communities listed in Appendix A for each of the years listed. Variable deflatort At Derivation For Year "t" 1972, '77, '82, '87: (1972 base Detroit Consumer Price Index for all items)t / 100. 1972, '77, '82, '87: { (sum of state-equalized assessed value commercial property abatements through year) / deflator 1t- 1972, '77, '82, '87: ( (sum of state-equalized assessed value manufacturing property abatements through year) / deflator }t° 1972, '77, '82, '87: ( ("A" + state-equalized assessed value commercial property) / deflator 1t- 1972, '77, '82, '87: ( (state-equalized assessed value residential housing property) / deflator 1t- 1972, '77, '82, '87: { ("8" + state-equalized assessed value manufacturing property) / deflator 1t- 1977: {1976 population + .25(1980 population - 1976 population)}. 1982: (1982 population). 1987: {1986 population + .25(1986 population - 1982 population)}. 1977, '82, '87: { (police + fire + sewer + sanitation + highway expenditure) + (number of public students * average expenditure on public student) / (deflator * "N") )t. average expenditure on public student = { (first school district expenditure per pupil * (first school district property base / community property base) ) + ... + (last school district expenditure per pupil * (last school district property base / community property base) ) ) 186 Variable Vt Pt pdent cdent hdent mdent lempt 187 Appendix C (cont.) Variable Derivations Derivation For Year "t" 1977: ( (1970 median value owner-occupied home + .7(1980 home value - 1970 home value) ) / deflator }. 1982: { (1980 median value owner-occupied home + .2(1980 home value - 1970 home value) ) / deflator }. 1987: { (1980 median value owner-occupied home + .7(1980 home value - 1970 home value) ) / deflator ). 1977, '82, 87: (property taxes per $1 of state-equalized property value)t. 1977, '82, '87: ( ( (police + fire + sewer + sanitation + highway expenditure) / deflator ) / "C" + "H" + "MID 711' 1977, '82, '87: { (local income tax revenue / deflator) / "C" + "H" + "M" 1t- 1977, '82, '87: (number of convicted murder, rape, robbery, assault, burglary, larceny, and motor vehicle theft offenses)t. 1977, '82, '87: ("N" / "area”)t. 1977, '82, '87: ("C" / "area")t. 1977, '82, '87: ("H" / "area")t. 1977, '82, '87: ("M" / "area")t. 1977, '82, '87: (1980 number residents worked in area of residence / 1980 number residents employed)t. 1977: { (1969 median household income + .8(1979 household income - 1969 household income) ) / deflator ). 1982: { (1979 median household income + .3(1979 household income - 1969 household income) ) / deflator ). Variable Cdt d7? d82 d8? hct hnt wnt 188 Appendix C (cont.) Variable Derivations Derivation For Year ”t" 1987: ( (1979 median household income + .8(1979 household income - 1969 household income) ) / deflator }. 1977, '82, '87: { (state and federal intergovernmental revenue + misc. general revenue) / deflator 1t- 1977, '82, '87: { ( ("Y" for first community within five mile radius of community's center) + ("Y" for second community within five mile radius) + ... + ("Y" for last community within five mile radius) ) / (number of communities within five mile radius) 1t: 1977, '82, '87: (miles from center of city to Detroit's central business district)t. 1977, '82, '87: (equals "1" if 1977 cross section, equals "0" if not)t. 1977, '82, '87: (equals "1" if 1982 cross section, equals "0" if not)t. 1977, '82, '87: (equals "1" if 1987 cross section, equals "0" if not)t. 1977: { 1970 percentage housing with more than one bath + .7(1980 percentage more than one bath - 1970 percentage more than one bath) }. 1982: { 1980 percentage housing with more than one bath + .2(1980 percentage more than one bath - 1970 percentage more than one bath) ). 1987: { 1980 percentage housing with more than one bath + .7(1980 percentage more than one bath - 1970 percentage more than one bath) }. 1977, '82, '87: (miles of divided expressway)t. 1977, '82, '87: (equals "1" if adjacent to Lake St. Clair, Detroit River, or Lower Rouge River: equals "0" if not)t. Variable ddet dmac doak dway agest areat caggt educt 189 Appendix C (cont.) Variable Derivations Derivation For Year "t" 1977: ( 1970 median age + .7(1980 median age - 1970 median age) }. 1982: { 1980 median age + .2(1980 median age - 1970 median age) }. 1987: ( 1980 median age + .7(1980 median age - 1970 median age) }. 1977, '82, '87: (equals "1" if city of Detroit, equals "0" if not)t. 1977, '82, '87: (equals "1" if in Macomb County, equals "0" if not)t. 1977, '82, '87: (equals "1" if in Oakland County, equals "0" if not)t. 1977, '82, '87: (equals "1" if in Wayne County, equals "0" if not)t. 1977, '82, '87: ( ( ("age" for first community within five mile radius of community's center) + ("age" for second community within five mile radius) + ... + ("age" for last community within five mile radius) ) / (number of communities within five mile radius) 1t: 1977, '82, '87: (square miles)t. 1977, '82, '87: { ( ("C" for first community within five mile radius of community's center / area) + ("C" for second community within five mile radius / area) + ... + ("C" for last community within five mile radius / area) ) / (number of communities within five mile radius) 1t- 1977: { 1970 median education + .7(1980 median education - 1970 median education) }. 1982: { 1980 median education + .2(1980 median education - 1970 median education) }. Variable maggt racet pdenst 190 Appendix C (cont.) Variable Derivations Derivation For Year "t" 1987: ( 1980 median education + .7(1980 median education - 1970 median education) ). 1977: { 1970 percentage homes owner occupied + .7(1980 % homes owner occupied - 1970 8 homes owner occupied) ). 1982: ( 1980 percentage homes owner occupied + .2(1980 % homes owner occupied - 1970 % homes owner occupied) }. 1987: ( 1980 percentage homes owner occupied + .7(1980 % homes owner occupied - 1970 8 homes owner occupied) ). 1977, '82, '87: ( ( ("M" for first community within five mile radius of community's center / area) + ("M" for second community within five mile radius / area) + ... + ("M" for last community within five mile radius / area) / (number of communities within five mile radius) 1t- 1977: { 1970 percentage nonwhite + .7(1980 % nonwhite - 1970 % nonwhite) }. 1982: { 1980 percentage nonwhite + .2(1980 % nonwhite - 1970 % nonwhite) ). 1987: ( 1980 percentage nonwhite + .7(1980 % nonwhite - 1970 % nonwhite) }. 1977, '82, '87: { ( ("pden" for first community within five mile radius of community's center) + ("pden" for second community within five mile radius) + ... + ("pden" for last community within five mile radius) ) / (number of communities within five mile radius) 1t: Variable deflatort At CtrHtth Nt Appendix D Variable Sources Source For Year "t" 1972, '77, '82, '87: Detroit Chamber of Commerce (1988). Esongmie.£a2t_ngek. Detroit. Michigan. ‘ 1972, '77, '82, '87: Michigan Property Tax Commission (1988), "P.A. 255 Handwritten Log," Michigan Department of Treasury, Lansing, Michigan. 1972, '77, '82, '87: Michigan Property Tax Commission (1988), "P. A. 198 Handwritten Log," Michigan Department of Treasury, Lansing, Michigan. 1972, '77, '82: Michigan Office of Community Development (1988), "L.U.C.I. -- Local Unit Computerized Information," Michigan Department of Commerce, Lansing, Michigan. 1987: Michigan Property Tax Commission (1988), "Analysis for Equalized Variations,” Computer Printout, Michigan Department of Treasury,' Lansing, Michigan. 1977. '82. '87: ui2higan.§fafisfisal.bbstraet (1979, 1987), edited by David Veraway, Wayne State University Press, Detroit, Michigan. 1977, '82 Police, Fire, Sewer, Sanitation, and Highway Expenditure: U.S. Census Bureau (1979, 1984), Census of governments, ”Finances of Municipalities and Township Governments,” U.S. Department of Commerce, Washington, D.C. 1987 Police, Fire, Sewer, Sanitation, and Highway Expenditure: Michigan Local Unit Bureau (1988), "Annual Local Unit Fiscal Report for Counties, Cities, Villages, and Townships for the Fiscal Year between July 1, 1986 and June 30, 1987," Michigan Department of Treasury, Lansing, Michigan. 191 Variable vtIYt' agate areat Pt lempt 9t 192 Appendix D (cont.) Variable Sources Source For Year "t" 1977, '82, '87 Public Education Expenditure: Michigan Office of Community Development (1988), "L.U.C.I. -- Local Unit Computerized Information," Michigan Department of Commerce, Lansing, Michigan. 1982 School District Property Base: Michigan Department of Education (1982), "S.E.V By Unit - Program R0903", computer printout, Lansing, Mich. 1977, '82, '87: U.S. Census Bureau (1977, 1983), Cifx_and_§29nfx_natabeek. U.S- Department of Commerce, Washington, D.C. 1977, '82, '87: Property Tax Commission (1977, 1982. 1987). SLate_Equalized_1alnafiens_and Axerage_lax_nafe_2afa. Hichigan Department of Treasury, Lansing, Michigan. 1977, '82, '87: See source listing for ”Rt". 1977, '82, '87: Michigan Property Tax Commission (1977, 1982, 1987), "Information on Michigan Cities Levying an Income Tax," typewritten sheet, Michigan Department of Treasury, Lansing, Michigan. 1977, '82, '87: U.S. Department of Justice (1978, 1983, 1988), ifO m -- Usss, Federal Bureau of Investigation, Washington, D.C. 1977, '82, '87: U.S. Census Bureau (1982), I U.S. Department of Commerce, Washington, D.C. 1977, '82: U.S. Census Bureau (1979, 1984), ansgs sf Qovsrnmengs, "Finances of Municipalities and Township Governments," U.S. Department of Commerce, Washington, D.C. Variable cdt,hnt, hct, educt, hownt, racet 193 Appendix D (cont.) Variable Sources Source For Year "t" 1987: Michigan Local Unit Bureau (1988), "Annual Local Unit Fiscal Report for Counties, Cities, Villages, and Townships for the Fiscal Year between July 1, 1986 and June 30, 1987,” Michigan Department of Treasury, Lansing, Michigan. 1977, '82, '87: Michigan State Highway Commission (1977, 1983, 1986), "Michigan Official Transportation Map," scaled from "Detroit and Vicinity," Michigan Department of Transportation, Lansing, Michigan. 1977, '82, '87: U.S. Bureau of Census (1972, 1982) . WW Mishigan , U.S. Department of Commerce, Washington, D.C. BIBLIOGRAPHY AND GENERAL REFERENCES BIBLIOGRAPHY Advisory Commission on Intergovernmental Relations (1967), I Washington, D. C., April. Ball, M. J. (1973), "Recent Empirical Work on the Determinants of Relative Housing Prices,” Uzbsn_fi;§g1ss, 213-233. Beaton, W. P. (1983), "The Demand for Municipal Goods," in W. edited by W. P. Beaton, Center for Urban Policy Research, New Brunswick, New Jersey. Charney, A. H. (1983), "Intrametropolitan Manufacturing Location Decisions and Local Tax Differentials,” Jsgzns; W123. 184-205- Citizen's Research Council of Michigan (March 1986), "Council Comments: The Use of Property Tax Abatements in Michigan 1974 - 1983," No. 959, Lansing, Michigan. Coffin, D. A. (1982), "Property Tax Abatements and Economic Development in Indianapolis," g;gg§h_sng_ghsngs, 18-23. Cornia, G. C., W. A. Testa and F. D. Stocker (1978), figsgs; ves E o , Academy for Contemporary Problems, Columbus, Ohio. Courant, P. (1982), "The Property Tax," in nishigsn;s_fiisss1 MW. edited by H- E- Brazer and D. S. Laren, University of Michigan Press, Ann Arbor, Michigan. Due, J. F. (1961), "Studies of State and Local Tax Influences on Location of Industry," fissisgs1_1sx41sgznsl, 163-173. Erickson, R. A. and M. Wasylenko (1980), "Firm Relocation and Site Selection in Suburban Municipalities," gggznsl WM: 69-85. 194 195 Fischel, W. A. (1974), "Fiscal and Environmental Considerations in the Location of Firms in Suburban Communities: A Non-Technical Digest," in Bgssssgings 2f_the_SixtY_aexenth_Ann2al.£2nferense.2n_1axatien. National Tax Association-Tax Institute of America, 632-656. ------ (1975), "Fiscal and Environmental Considerations in the Location of Firms in Suburban Communities," in zis2al_zenins_and_Land_u_e_£2nfrols. edited by E- So Mills and W. E. Oates, Lexington Books, Lexington, Massachusetts. ------ (1985). The_Economigs_2f_22ning_Laws. Johns Hopkins University Press, Baltimore, Maryland; Floyd. 3. 8.. Jr- (1952). Effe2t__2f_IaxafiQn_2n_Indnstrial , The University of North Carolina Press, Chapel Hill, North Carolina. Fox, W. F. (1981), "Fiscal Differentials and Industrial Location: Some Empirical Evidence," U:bsn_§;u§1ss, 105-111. Gujarati, D. M. (1988) B§§1§_EQQan§EIiQ§. McGraw-Hill Company, New York, New York. Hamilton, B. W. (1975), "Zoning and Property Taxation in a System of Local Governments, U;ng_§;gg1ss, 205-211. ------ (1932), "Wasteful Commuting." lgurnal_of_zglitisal Eggngmx. 1035-58- Kmenta. J- (1988). Elsmen:§_2f_fisgn2mefriss. 2nd Edition. Macmillan Publishing Company, New York, New York. Inman, R. P. (1979), "The Fiscal Performance of Local Governments: An Interpretive Review," in Cn;zsnt_1ssgss in_u:bsn_£ssnsmiss, edited by P. Mieszkowski and M. Straszheim, Johns Hopkins University Press, Baltimore, Maryland. Irvine, O. F. (1981), ”Retail Inventory Investment and the Cost of Capital." Amerisan.§2222mis_8exiex. 633-648- Ladd, H. F. (1975), "Local Education Expenditures, Fiscal Capacity, and the Composition of the Property Tax Base," National_max_lenrnal , 145-158. ------ (1976), ”Municipal Expenditures and the Composition of the Local Property Tax Base," in Land_nse_and_£ublie_zelisx. University of Wisconsin Press, Madison, Wisconsin. 196 Ladd, H. F. and K. L. Bradbury (1988), ”City Taxes and Property Tax Bases." Hafignal_Tax_lenrnal. 503-524- Lea, M. L. (1978), ”Local Public Expenditure Determination: A Simultaneous Equations Approach," in Ezssssgings_gf V " I National Tax Association - Tax Institute of America, 131-138. Levin, S. G. (1974), "Suburban-Central City Property Tax Differentials and the Location of Industry: Some Evidence." .Land_nsgn2miss 380- -386- Maddala, G. S. (1988), Macmillan Publishing Company, New York, New York. McDonald, J. F. (1983), "An Economic Analysis of Local Inducements for Business." 22urnal_ef_nrban_fisen9miss. 322-333. McGuire, T. J. (1985), "Are Local Property Taxes Important in the Intrametropolitan Location Decisions of Firms: An Empirical Analysis of the Minneapolis-St. Paul Metropolitan Area." i2nrnal_2f_nrban_322n2mies. 226-234- McHone, W. W. (1986), "Supply-Side Considerations in the Location of Industry in Suburban Communities: Empirical Evidence from the Philadelphia S.M.S.A.," Lang Eggngmiss. 64-73. Megdal, S. B. (1986), Comment on White's "Property Taxes and Firm Location: Evidence from Proposition 13,” in figndiss in_§tate_and_Lesal_Einanss. edited by 8- Rosen. N.B.E.R., University of Chicago Press, Chicago, Illinois. Mesteleman, S. (1983), Comment on Schmenner's "City Taxes and Industry Location." in 2reseedings_ef_tns_§ixty: sixth Annual Qonfegencs 9n Tanatisn, National Tax Association - Tax Institute of America, p. 532-538. Michigan Department of Commerce (1983), "A Guide to Michigan's Plant Rehabilitation and Industrial Development Districts Law of 1974," Office of Business and Community Development, Lansing, Michigan. ------ (1984), "Report on Commercial Redevelopment Act for 1983,” Office of Business and Community Development, Lansing, Michigan. Michigan Department of Treasury (1984), "Tax Expenditure Analysis of Public Act 255," Taxation and Economic Policy Office, Lansing Michigan. 197 Morgan, W. E. and M. M. Hackbart (1974), "An Analysis of State and Local Industrial Tax Exemption Programs,” S92:hern.£29n2mis_lgurnal. 200-205. Morse, G. W. and M. C. Farmer (1986), ”Location and Investment Effects of a Tax Abatement Program,” assigns; Tax_lgnrnal. 229-236. Moses, L. and H. F. Williamson, Jr. (1967), ”The Location of Economic Activity in Cities," Bsyisy, 211-222. Mueller, D. C. (1979), Ennlis_gnsiss, Cambridge University Press, New York, New York. Mueller, E. and J. N. Morgan (1962), "Location Decisions of Manufacturers." in Ea2ers_and_£reseedings_ef_the_ ssyenty:E2urtn_Annnal_Meeting_2f_Amsrisan_Eegn2mis Assssiatien. 2044217- Nerlove, M. (1958), "Distributed Lags and Demand Analysis For Agricultural and Other Commodities,” Aggisnlgnzsl fisngngsk_1111, U. S. Department of Agriculture, Washington, D. C. Oakland, W. H. (1978), ”Local Taxes and Intraurban Industrial Location: A Survey," in et , edited by G. F. Break, University of Wisconsin Press, Madison, Wisconsin. Oates, W. (1969), "The Effects of Property Taxes and Local Public Spending on Property Values: An Empirical Study of Tax Capitalization and the Tiebout Hypothesis,” 19urnal_ef_zelitisal_322n2my 957 - 971- Pyndyck, R. S. and D. L. Rubinfeld (1981), EQQDQEELI1£_HQ§£1§ and.£22n_mis_£2resasts McCray-Hill Book Company. New York, New York. Rosen, S. (1974), "Hedonic Prices and Implicit Markets," 12nrnal_9f_zelifisal_fisenemy. 34-55- Schmenner, R. W. (1973), "City Taxes and Industry Location, " in cee t - 9n_1sxs§isn, National Tax Association - Tax Institute of America, 528-532. Struyk, R. J. and F. J. James (1975), Inggsnstxgnglitan 0 ‘ a 01' ‘ ‘ I 1° ' 0 ‘;= 0‘ 91:10‘, Lexington Books, Lexington, Massachusetts. Tiebout, C. (1956), "A Pure Theory of Local Public Expenditure." 12nrnal_gf_£211tical_zsgn2my. 416-424. 198 Wasylenko, M. (1980), "Evidence on Fiscal Differentials and Intrametropolitan Firm Location,“ Lsnd_fissngm1ss, 339 - 349. ------ (1985), ”Fiscal Tax Policy and Industry Location: A Review of the Evidence," in szsssgings_gfi I National Tax Association - Tax Institute of America, 222-228. Weber. M- J- (1975)..1ndustrial_122atign. Scientific Geography Series, Sage Publications, Beverly Hills, California. White, M. J. (1975), "Firm Location in a Zoned Metro Area," in Eissal_z2ning_and_Land_nse_antrels. edited by E.S. Mills and W. E. Oates, Lexington Books, Lexington, Massachusetts. ------ (1986), ”Property Taxes and Firm Location: Evidence from Proposition 13." in S1udies_in_§tate_and_Lesal Einsnss, edited by S. Rosen, N.B.E.R., University of Chicago Press, Chicago, Illinois. Wolkoff, M. (1982), "Tax Abatement as an Incentive to Industrial Location." in _Misniganls_£issal_and Essnsnis_§§rns§nzs, edited by H. E. Brazer and D. S. Laren, University of Michigan Press, Ann Arbor, Michigan. ------ (1983), ”Chasing a Dream: The Use of Tax Abatements to Spur Urban Economic Development," mimeo, Department of Economics, University of Rochester, New York. ------ (1987), "Economic Development as a Signalling Game,” mimeo, Department of Economics, University of Rochester, New York. GENERAL REFERENCES Aronson, J. R. and E. Schwartz (1973), "Financing Public Goods and the Distribution of Population in a System of Local Governments, usgisnsl_1sx_193:nn1, 137-155. Bahl, R. (1980), "The Impact of Local Tax Policy on Urban Economic Development." urban_§2nsertium_nnllstin. U. S. Department of Commerce, Washington, D. C., September. Benson, B. L. (1986), "Do Taxes Matter? The Impact of State and Local Taxes on Economic Development,” Essnsmis Deyelonment_£2mmentary. 13-17. ------ and R. N. Johnson (1986), "The Legged Impact of State and Local Taxes on Economic Activity and Political Behavior." Esonomie_lnguiry. 389-401- Brazer, H. E., D. S. Laren and F. Y. Sung (1982), "Elementary and Secondary School Financing, " in M12h1§§fl.§.£1§2§1 and_Esgn2mis_§trusturs edited by H- E- Brazer and D. S. Laren, University of Michigan Press, Ann Arbor, Michigan. Brown, H. J., J. R. Ginn, F. J. James, J. F. Rain and M. R. Straszheim (1972), m i ’ al Mode 3 U ° .008 is! 0! i “-S C 00‘ ,8 .1. 0 ’1! o. or, N. B. E. R., Columbia University Press, New York. Carlton, D. W. (1979), "Why New Firms Locate Where They Do: An Econometric Model." in Interregional_uoyements_and , edited by W. C. Wheaton, Urban Regional_§royth Institute, Washington, D. C. ------ (1983), "The Location and Employment Choices of New Firms: An Econometric Model with Discrete and Continuous Endogenous Variables," Eeyiey_2f_E22n2mis§_and Stafisfiss. 440-449- Fisher. R- C- (1988). Sfats_and_Losal_£ublis_Einanse. Scott. Foresman, and Company, Glenview, Illinois. 199 200 Fisher, R. C. and J. E. Kohlhase (1982), "Fiscal Problems and Policies of the Cities, " in uisnigsn_s_fiisgs1_nn§ , edited by H. E. Brazer and D. S. Esonomig_§frnsture Laren, University of Michigan Press, Ann Arbor, Michigan. Fox, W. F. (1978), "Local Taxes and Industrial Location, Publis_£inanss_nuarterly. 93-114- Goldberg. H- and P- C- Chinloy (1984). Urban_Land_E£9nomies. John Wiley and Sons, New York, New York. Grieson, R. E., W. Hammovitch, A. M. Levenson and R. D. Morgenstien (1977), "The Effect of Business Taxation on the Location of Industry," Jgnrn§l_gf_flrban Esonomiss. 170-185- Helms, L. J. (1985), ”The Effect of State and Local Taxes on Economic Growth: A Time Series Cross Section Approach.” EsyieE_2f_Esonom12§_and_§fafi§tiss. 574-532. Henderson, J. V. (1983), "Industrial Bases and City Sizes," American.£sen9mig_8eyisr. 164-168. Hirsohman. A- 0- (1970). 88111.221se1_and_Loyalty. Harvard University Press, Cambridge, Massachusetts. Holtz-Eakin, D. (1986), "Unobserved Tastes and the Determination of Municipal Services,” Ennisnsl_2nx Journal. 527-532- Killingsworth, M. (1983), sto; supply, Cambridge University Press, New York, New York. Kiefer, D. W. (1974), Comment on Fischel's ”Fiscal and Environmental Considerations in the Location of Firms in Suburban Communities," in gsngxsls, edited by E. S. Mills and W. E. Oates, Lexington Books, Lexington, Massachusetts. Leone, R. F. (1972), "The Role of Data Availability in Intrametropolitan Workplace Location Studies," Annals of_E22n2m12_and_§2§ial_ueasuremsnf. 171-182. Manson, D. M., M. Howland and G. E. Peterson (1984), "The Effect of Business Cycles on Metropolitan Suburbanization," Essnomis Gsogrsnny, 71-80. McGuire, T. J. (1983), "Firm Location in a Tiebout World, Journal_gf_Bsgional_§2iense. 211-222- 201 Mieszkowski, P. (1972), "The Property Tax: An Excise or a Profits Tax?” l9urnal_of_£ublig_822n9m12§. 73-96- Mills, E. S. and B. W. Hamilton (1989), QIRQDIEQQanigfi. Fourth Edition, Scott, Foresman, and Company, Glenview, Illinois. Mills, E. S. and W. B. Oates (1975), ”The Theory of Local Public Services and Finance: Its Relevance to Urban Fiscal and Zoning Behavior," in nss_§nn;:sls, edited by E. S. Mills and W. E. Oates, Lexington Books, Lexington, Massachusetts. Orbel, J. M. and T. Uno (1972), "A Theory of Neighborhood Problem Solving: Political Action vs. Residential Mobility.” Ameri2an_Politisal_ssiense_neyier. 471-489- Palumbo, G. and P. Hutton (1987), "On the Causality of Intraurban Location." J2urnal_2f_nrban_zsgn9miss. 69-79. Papke, L. (1986), "The Location of New Manufacturing Plants and State Business Taxes: Evidence from Panel Data," in o e d ev Qn_1sxs;isn, National Tax Association - Tax Institute of America, 44- -55. Schuler, R. E. (1974), "The Interaction between Local Government and Urban Residential Location," Anszisnn EQQani§_B§!1§!. 682-700- Steinnes, D. N. (1977), "Causality and Intraurban Location,” I9urnal_2f_nrban_fisonomiss. 69-79- Thomas, A. R. (1982), "Industrial Revenue Bonds," in u1shiganLs_Eis2al.snd.§sonomis_§trnefure. edited by H.E. Brazer and D. S. Laren, University of Michigan Press, Ann Arbor, Michigan. Thompson, W. R. (1982), "Industrial Location: Causes and ConsequenceS.” in Mishiganls_rissal_and_nsgnomis finznsgnzs, edited by H. E. Brazer and D. S. Laren, University of Michigan Press, Ann Arbor, Michigan. Tybout, R. A. and J. M. Mattila (1977), "Agglomeration of Manufacturing in Detroit," u , 1-16. Wassmer, R. W. (1987), "The Supply and Demand of Manufacturing and Commercial Development Within Communities in the Metropolitan Detroit Area," Department of Economics, Michigan State University, East Lansing, Michigan. 202 Wasylenko, M. (1981), "The Location of Firms: The Role of Taxes and Fiscal Incentives, " in finance... W. edited by R- Bab1.Sa9e Publications, Beverly Hills, California. Wendling, W. R. (1981a), ” ” n , The W. E. UpJohn Institute for Employment Research, Kalamazoo, Michigan. ------ (1981b), snd_1ngns;zy, The W. E. UpJohn Institute for Employment Research, Kalamazoo, Michigan.