Community composition and voter support for tax limitations: evidence from home-rule elections.

I. Introduction

Complaints about high taxes have stimulated taxpayer demands for explicit statutory or constitutional constraints on the taxing and spending powers of state and local governments. Although fiscal controls have long been a part of our system of government in the United States, the recent "Tax Revolt" movement commonly is thought to have its roots in the late 1970s with the approval of Proposition 13 in California. Voter discontent has produced a large number of new controls enacted during the last two decades. Why do voters need to impose explicit tax and expenditure constraints on the actions of their elected officials? A major objective of this paper is to evaluate, in a new setting, the rationale for the existence of tax limitations.

Direct evidence regarding the determinants of voter support for fiscal limitations is sparse. Despite the persistent clamoring for additional fiscal controls, there have been few empirical studies that address the question of why voters desire tax-and-spend limits. Most of the existing studies were conducted over a decade ago. Citrin [6], Courant, Gramlich, and Rubinfeld [8], and Ladd and Wilson [23] examined micro-based survey data to identify factors that led taxpayers to support state-imposed limitations on local governments. Although these three studies analyzed survey responses about tax limitation efforts in different states, the common findings indicate that support for the constraints comes from individuals who want reduced taxes but prefer the existing level of spending. The responses suggest that limitation supporters either are seeking increased municipal efficiency or, as suggested by Courant et al., are suffering from free-lunch illusion.

Using state-level data, Ladd [21] investigated the enactment by states of fiscal limits on local governments during the 1970s. She found evidence that high and rising aggregate local property tax burdens in a state increased the probability that the state subsequently imposed fiscal limitations on local governments. More recently, Skidmore and Alm [30] used state-level data over a longer time period to estimate the factors affecting the probability that tax or expenditure limitations were approved in statewide elections. They found mixed evidence on the relationship between high and rising property tax levels and the subsequent enactment of fiscal limits. Their results also suggest that fiscal limits were more likely to be approved by voters in high income states.

Unlike the other studies, this study analyzes the decisions made by some voters to remove state-imposed statutory constraints on the fiscal decisions of their local governments. Beginning in 1970, residents in many Illinois communities have been able to vote to obtain local home-rule authority instead of allowing their jurisdiction to be constrained by state statutory limits on local taxing and borrowing decisions. Unlike the other studies, this study investigates the role of community composition in influencing local voters' support for the state-imposed constraints. Buchanan and Tullock [5] argue that communities characterized by relatively heterogeneous populations are more likely than homogeneous communities to accept restrictive collective decision-making rules as a means of reducing the costs involved in collective choice. Using a data set of almost 300 communities, this study investigates whether or not state-imposed limits on local fiscal decisions are more likely to be preferred by voters in municipalities with relatively larger intra-community variation in socio-economic characteristics such as income and age. The findings are consistent with the Buchanan and Tullock argument. Moreover, this analysis tests hypotheses regarding the relationship between preferences for state-imposed limits and community median income, the level of property tax revenues, and the recent change in property tax revenues. Finally, this study examines the importance of political characteristics of the cities and villages in influencing voter decisions to delegate to the locality greater taxing and borrowing power.

II. Motivation for Fiscal Constraints Imposed on Local Governments

The existence of state government controls on local taxing power suggests that local governments collect more than the desired amount of tax revenues. Although the observed level of taxes is determined by the interaction of demand and supply factors, the explanations for why a local government may tax "too much" generally are discussed in terms of supply-side considerations. A principal-agent model can be used to explain why taxpayers desire constraints on the taxing power of their local government. Consider that government decision-makers perform their jobs as agents of the principal-voters. The voters employ the agents to produce government services, but the voters are not able to observe the quantity of services directly. The voters observe the tax bill for the production of government services, where the tax bill is the product of the quantity of government services and the per unit cost of producing that output. While the voters prefer that a particular amount of government services be produced at minimum cost, in the principal-agent framework the government officials have different preferences. In particular, the agents may prefer to produce at a higher cost and/or produce a greater amount of government services. Consequently, fiscal limitations can be imposed to prevent high levels of taxation that arise from excessive government output levels or from cost inefficiencies.(1)

To the extent that voter dissatisfaction with excessively high taxes is due to the production decisions of local officials whose behavior is difficult to monitor, the imposition of fiscal constraints could be welfare enhancing. Relying on the existence of unresponsive local officials as the explanation for state-imposed fiscal limits is somewhat unsatisfying, however, because students of fiscal federalism generally expect the state level of government to be even less responsive than the local level to the concerns of local voters.(2) An even more perplexing consideration involves the standard efficiency argument for subsidizing or restricting local government spending. Fiscal constraints may be efficient to the extent that they help internalize negative externalities generated by local government expenditures. However, many types of local expenditures are thought to generate positive externalities (e.g., education, water treatment and public safety) and commonly are subsidized rather than restricted by grants-in-aid to local governments.

The current analysis explores the role of demand-side factors, rather than supply-side factors, in influencing preferences for fiscal constraints. I investigate the possibility that constraints are motivated by the existence of within-community variations in demands for local public services. Although each resident of the community may have a different desired level of local taxes and expenditures, the political process translates individual preferences into a single community demand. Unless the community consists of residents with identical preferences for local public services, many voters will be dissatisfied with the particular level of spending chosen.(3) Although in a perfect Tiebout world taxpayers costlessly sort themselves into municipalities consisting of residents with homogenous preferences for government services, evidence suggests the existence of a large amount of intra-community variation in government service demands [27].

If fiscal limitations are motivated by voter dissatisfaction with the outcome of the collective decision-making process, then I expect that the limitations are preferred by residents of communities in which there exists substantial variation in individual preferences. This relationship between community composition and the willingness of voters to impose restrictive collective decision-making rules was explained by Buchanan and Tullock in The Calculus of Consent [5]. Buchanan and Tullock argue that a voter in a heterogeneous community is expected to prefer government constraints as a means of reducing the likelihood that the voter may be harmed by collective decisions that are favored by others but are not favored by that particular voter. Buchanan and Tullock refer to this anticipated harm as the "external costs" of collective decision-making, and they explain that these external costs will be lower in homogeneous communities.

The external costs argument suggests that voters in more homogeneous communities are less likely to impose statutory or constitutional restrictions on the actions of local fiscal decision-makers. Consider the commonly-used median voter model of the demand for public expenditures. That model states that (under certain conditions)(4) the community's demand for government expenditures is the median of the individual demands. The median preferred expenditure level will defeat any other level in a majority rule election. Importantly, however, the median spending level is not the utility-maximizing level for individuals with preferences greater or less than the median.

Because the median voter model of the demand for government spending takes into account individual rankings of various alternatives but not the intensity of preferences, it is possible that fiscal constraints can increase economic elfare. Two aspects of the pattern of individual demands determine whether or not constraints that reduce spending below the median level are welfare enhancing. First, constraints are more likely to increase welfare if the distribution of individual demands around the median is asymmetrical in a particular way. Specifically, the potential for welfare gains requires that in the case of individuals demanding more than the median amount, the difference between their preferred amount and the median is relatively small compared to the respective difference for below-median demanders. Alternatively, reductions in spending below the median level could increase welfare if the individuals with higher demands also tend to have more price-elastic demands for local expenditures.(5)

Recent research findings are consistent with the view that voters in heterogeneous communities have reason to be concerned about the external costs of collective decision-making. A recent study by Todo-Rovira [35] investigates the importance of within-community income dispersion on the demand for government spending. His findings suggest that the local political process aggregates individual preferences by placing greater weights on the expenditure preferences of high-income demanders. If this is true, then it is possible that fiscal limits are desired by a majority of residents for whom government spending may be too high.

Additional empirical studies find that individuals from opposite ends of the income distribution often have similar preferences for public expenditures (and that these preferences are different from those of median-income individuals.) Husted [19] and Temple and Porter-Hudak [34] find that demands for various categories of government expenditures are either U-shaped or inverted U-shaped functions of income. Brown and Saks [4] similarly found evidence that the demand for government expenditures on education is not a monotonic function of income. The imposition of fiscal constraints may reflect a desire by median-income voters in heterogeneous communities to thwart the formation of winning coalitions that potentially could arise between high-and low-income voters who might attempt to increase expenditures above the level preferred by those with incomes closer to the median.

III. Home Rule in Illinois

In order to identify the determinants of preferences for state-imposed limits on the taxing and spending powers of local governments, this study examines the decisions made by some Illinois municipalities to acquire home-rule status, where "home rule" refers to local autonomy from state control. The 1970 Illinois Constitution significantly altered the structure of state-local relations by automatically granting home-rule status to all municipalities with populations greater than 25,000.(6) An important feature of the 1970 Constitution is that municipalities with populations below 25,000 are able to obtain home-rule status by passing a referendum. The actions by these smaller communities are the subject of this current research, as this study examines local decisions to remove, rather than to enact, state-imposed fiscal constraints.

Although many states allow selected local governments some degree of local autonomy in certain fiscal or regulatory matters, the system of home rule in Illinois has been described as one of the most comprehensive in the nation.(7) In Illinois, the powers of home-rule are legion. The 1970 Constitution grants home-rule units the right to exercise any power not explicitly denied to local governments by the state. Local governments without home-rule status can only exercise powers explicitly granted to them, and must seek the approval of the state assembly to expand their existing powers. Home-rule governments have much greater fiscal and regulatory authority than do non-home rule governments. Communities in Illinois have used their regulatory power to limit the hours of liquor sales and to enact requirements regarding the licensing of sidewalk street vendors and burglar-alarm installers. Home-rule communities also have considerable latitude in making decisions regarding local government personnel issues.

The two most important aspects of home-rule authority for the purposes of this study are the fiscal powers that allow home-rule units to: (1) increase tax revenues by broadening the local tax base and raising tax rates, and (2) issue a larger quantity of municipal bonds. Although home-rule communities may not tax income or occupations unless authorized by the state, these local governments may impose any other type of local tax or service charge. Hence home rule communities are able to broaden their tax bases beyond the property tax traditionally relied on to include local taxes on specific goods such as cigarettes or alcohol and more generally taxes on retail sales. In addition, property tax rates chosen by these communities are not limited by state statutory constraint. Home-rule units also have considerable borrowing powers. Unlike non home-rule units, they are able to issue general obligation bonds without voter approval. Moreover, only home-rule governments issue industrial development revenue bonds.

In the more than two decades since the enactment of the 1970 Constitution, a majority of voters in approximately 50 small Illinois communities have chosen home-rule status. Given that there are over 1,000 municipalities in Illinois with populations below 25,000, it is clear that the vast majority of local residents are not interested in freeing their community from state control. Why do voters not value the expanded choice set that accompanies the adoption of home rule? Because the acquisition of home-rule status in no way prevents the municipality from behaving in the same way as a government that is not home rule, the unpopularity of home-rule status is perhaps strange at first glance. Viewed against the backdrop of the "Tax Revolt" that has been occurring over the last two decades, however, this unpopularity is more understandable.

IV. The Hypotheses and the Empirical Model

Decisions to retain or remove state government limits on the taxing power of local governments are analyzed using data on the adoption of home-rule status by small communities in Illinois. The primary hypothesis to be tested is that community composition plays an important role in influencing local voter preferences for state-imposed limits. I expect that communities consisting of residents with relatively heterogeneous preferences for government expenditures are more likely to prefer to retain the existing statutory controls. This hypothesis is consistent with the argument made by Buchanan and Tullock [5] on the relationship between community composition and the costs of various collective decision-making rules. The existence of widely varying demands for local public services may induce voters not to prefer the expanded choice set that home-rule status offers because of the potentially large external costs that could be incurred by residents who are not satisfied with the manner in which individual demands for the collective government services are transformed by the political process into a single community demand.

Preference variation is expected to play a significant role in influencing decisions by local voters to retain or to remove the fiscal limitations. The importance of variations in individual preferences in the positive analysis of local government structure also features prominently in recent work by Nelson [26]. Because studies of the demand for local government spending find that demand varies by residents' incomes and ages [13; 29], measures of intra-community variation in those variables are used in the empirical analysis to represent variation in individual demands.

Factors other than preference variation also may affect the decision to retain existing controls. Two fiscal variables that will be used in the empirical analysis are the existing level of property tax revenues and a measure of the growth in property tax revenues. The nature of the relationships between these two variables and the residents' choices between home-rule versus state rule cannot be predicted a priori. It may be that residents in high and rising property tax communities perceive home-rule as costly because they expect that the local government will use the additional taxing power to raise property taxes even higher or will increase other types of taxes without providing property tax relief. On the other hand, residents may prefer home-rule status because of the prospect of lower property taxes. The broader tax base that home rule makes possible may allow the local government to more efficiently finance the same or greater spending levels by adopting a different tax mix.

Another hypothesis concerning the home-rule versus state-rule choice involves the level of private incomes within a community. I expect that voters in higher-income communities are more likely to prefer home-rule status for two reasons. Given that numerous studies have found positive income elasticities of demand for local government expenditures, it is commonly thought that higher-income residents typically have greater willingness to pay for local public services. As a result, they are less likely to favor tax and spending limitations. The second reason concerns the borrowing limitations faced by communities that are not home rule. Recent studies have documented a positive relationship between private incomes and the level of borrowing undertaken by the jurisdiction. This relationship is observed for both general obligation and industrial development bonds [12; 25; 33].

The population of the community is included to assess its impact on the probability of adopting home-rule status. Population size may be related to the ability of local residents to easily monitor their elected officials in order to ensure efficient production. It is likely that preferences for fiscal controls are stronger in communities with inefficiently-produced government services. Davis and Hayes [9] use a monitoring explanation in their investigation of the relationship between community population and inefficiency in the production of local public services. They find evidence that is consistent with the existence of two competing relationships between population and efficiency. Smaller communities may have the ability to more closely monitor their elected officials due to the absence of a large bureaucracy. On the other hand, smaller communities may be less able to monitor the actions of their elected officials due to the frequent absence in smaller towns of a local newspaper or television station.

Finally, I expect that decisions by voters to allow a municipality to have greater taxing and borrowing power may depend on certain characteristics of the structure of local government. In particular, I investigate whether residents in communities with a manager form of government are more likely to vote to eliminate state controls on the fiscal behavior of their local municipality. To the extent that voters view city or village managers as being more highly trained than elected mayors, voters in the manager form of local government may be more willing to delegate greater fiscal authority to their community. But, on the other hand, voters may have less control over an appointed manager than they have over an elected mayor.

Home-rule adoption decisions also may depend on the method by which community representatives are selected. In the communities examined here, the local council members (often called councilmen, aldermen, or trustees) either are elected by districts or they are elected from the community at large. I investigate whether this feature of the local political structure is related to the decision to grant greater fiscal powers to the locality. Because it is often feared that at-large representation may result in a situation in which all representatives reside in the same part of the community [20], it is possible that voters in communities with at-large representation will be more likely to retain the existing state restrictions. Voters in communities with at-large representation may have stronger preferences for fiscal limits as a means of controlling elected officials who may not place the same weights on the concerns of residents from different neighborhoods. On the other hand, voters in communities with ward representation also may prefer state-imposed tax and borrowing limits. Because district lines may be drawn with the intent of increasing the political influence of minority groups, it is possible that voters in communities with ward representation prefer the externally imposed fiscal constraints as a means of ensuring that the local elected officials will not pursue policies that are not supported by a majority of residents. Consequently, the expected relationship is uncertain regarding preferences for controls on local taxing and borrowing power and the method of selection of the local council.(8)

The effects of these variables on the probability that a municipality will adopt home-rule status are estimated using a log-linear survival model. The dependent variable is the logarithm of the number of years that each locality retains the existing state controls once the option of home rule becomes available. Survival model analysis is preferable to the use of standard limited dependent variable models for two reasons. First, the survival model provides a convenient way of handling censored observations. Secondly, by focusing on the length of time before a home-rule adoption takes place, the survival model implicitly takes into account variations across jurisdictions in the intensity of individual preferences for eliminating or retaining the existing fiscal controls. The use of survival models is increasingly common in economics, especially in studies of unemployment, strikes, innovation, and migration. To date, public finance researchers have not made much use of these techniques in their studies of government behavior. Important exceptions, however, include recent analysis by Dubin, Graetz and Wilde [11] of the decisions of state governments to conduct income tax amnesties and also the work of Alm, McKee, and Skidmore [1] on the timing of the adoption of state lotteries.

To summarize the framework of survival model analysis, begin by assuming that the duration time t (the time that passes before home-rule adoption occurs) follows a particular distribution that is characterized by the cumulative distribution function F(t). Corresponding to F(t), for t [greater than] 0, is the probability density function f(t) = F[prime](t). While the density function reflects the probability that a government will retain the existing fiscal controls for a length of time t, of greater interest is the probability that a government will adopt home rule in time t, given that the government has not yet adopted it by time t - 1. This conditional probability is provided by the hazard function h(t) = f(t)/[1 - F(t)]. The denominator 1 - F(t) is the survivor function S(t).

Duration models incorporate information on duration [t.sub.i] for all observations i = 1,..., N. A dummy censor variable [c.sub.i] indicates whether or not each community has adopted home rule by the end of time T, the length of the observation period. In terms of duration times, [c.sub.i] = 1 if [t.sub.i] [less than or equal to] T and [c.sub.i] = 0 if [t.sub.i] [greater than] T. For the completed observations, P([c.sub.i] = 1, t = [t.sub.i]) = f([t.sub.i]). For the censored observations, P([c.sub.i] = 0) = P([t.sub.i] [greater than] T[where][X.sub.i]) = 1 - F(T) = S(T), where [X.sub.i] is a vector of covariates.

The log-likelihood for the sample of N independent observations is

ln L = [summation of] {[c.sub.i] ln[f ([t.sub.i])] + (1 - [c.sub.i]) ln[S(T)]} where i = 1 to N.

The form of the density function f([t.sub.i]) must be specified before parameter estimates can be obtained through use of the standard maximum likelihood techniques. The current research assumes that adoption times can be modeled using two commonly used distributions - the Weibull and the exponential. For the Weibull distribution, f(t) = [Mu]p[([Mu]t).sup.p-1] exp(-[([Mu]t).sup.p]). The exponential distribution is a special case of the Weibull assuming that p = 1. Explanatory variables contained in the vector [X.sub.i] are incorporated into the duration model through the equation [Mu] = exp(-[Beta][prime][X.sub.i]).

I specify the probability that a community that has not yet done so will adopt home rule conditional on a number of community-specific factors included in the vector [X.sub.i]. Vector [X.sub.i] includes:

1. INCOME VARIATION, a measure of the within-community variation in household incomes as represented by the coefficient of variation. For each community, information was obtained on the number of households in each of nine income classes. The standard deviation was computed using the grouped data. The coefficient of variation is equal to 100 times the standard deviation of household income divided by the mean.

2. AGE VARIATION, a measure of the within-community variation in individual ages as represented by one-minus the Herfindahl-Hirshman Index for age. The Herfindahl-Hirschman index is computed as 100 times the sum of the squared shares of the population in three age groups: less than 18 years old, 18-64 years old, and 65 and above.

3. PROPERTY TAXES, total local property taxes for each community.

4. PROPERTY TAX GROWTH, percentage change in property taxes per capita from 1964-1970.

5. MEDIAN HOUSEHOLD INCOME, median household income in each community.

6. POPULATION, total population of the community.

7. MANAGER, indicator variable equal to one if municipality has manager form of government.

8. AT LARGE, indicator variable equal to one if the representatives to the local council are elected from the community at large (rather than by districts or wards.)

Because Illinois communities automatically become home rule once they reach 25,000 in population, the sample consists of communities that never reach that size during the period 1970-1990. Research on the behavior of small cities and villages commonly is hampered by the lack of data. The available data impose two restrictions on the framework for the current study. First, it is not possible to obtain the needed demographic information on residents of the smallest communities. The decennial U.S. Census of Population contains information about characteristics of the population only for communities over 2,500 in size. Consequently, the sample is restricted to communities greater than 2,500 in 1970 that never reach 25,000 by 1990. After eliminating a few observations due to missing or inconsistent data, the sample consists of 287 small incorporated communities. The average 1970 population of the sample communities was 7,670.

The second data limitation concerns the availability of fiscal data, specifically information on local property tax burdens. Because the frequently used U.S. Census of Governments only provides fiscal information for communities greater than 10,000 in size, studies of local governments regularly exclude a large number of small cities and towns. This is a problem for the current [TABULAR DATA FOR TABLE I OMITTED] analysis, because limiting the sample to jurisdictions between 10,000 and 25,000 would unduly reduce the sample size. Fortunately, an alternative source of data for Illinois local governments is provided by the Statewide Summary of Municipal Finance published annually by the State of Illinois. The Statewide Summary, however, regularly frustrates Illinois researchers because the nature of the reports varies from year to year and because recent reports do not list property tax revenues. As a result, the conditional probability of retaining the fiscal controls over the period 1970-1990 is estimated using the 1970 values of the explanatory variables.(9)

Table I shows information on the means, standard deviations, and ranges for the explanatory variables for communities that have adopted home rule (HR) and those that have retained the existing fiscal controls (not HR). Examination of the mean differences between home rule adopters and nonadopters reveals that the adopters are characterized by less within-community variation in incomes and ages. This evidence supports the hypothesis that community composition is an important determinant of preferences of fiscal limitations. No statistically significant differences exist in the mean levels of property taxes or property taxes per capita. On average, home rule adopters have experienced greater property tax growth and have higher average household incomes.

The next section describes the results obtained by estimating a survival model of home-rule adoptions. The survival time, or duration of state rule, is measured by establishing 1970 as year zero. The duration times are the number of years that pass before each community becomes home rule, and here they range from one to twenty. The dependent variable is the logarithm of the durations. In the sample of 287 municipalities, 23 (or 8 percent) became home rule during the sample period. Communities not adopting home rule by 1990 are treated as censored observations.

V. Results

Survival model estimates of the conditional probability of retaining state controls on local fiscal authority once home-rule status is available are shown in Table II. As discussed in the previous section, estimation requires imposing a particular functional form on the distribution of duration times. I use two alternative specifications, the Weibull and the exponential distribution.(10) The sign of the estimated coefficient for each explanatory variable indicates the direction of the effect of the variable on the probability that a municipality will retain the existing controls on local taxing and borrowing power, given that it has not yet adopted home-rule status. I expect that communities in which there is greater variation in individual demands for local government spending are more likely to retain the existing state-imposed controls on local fiscal power. Given that studies often find that the demand for local government expenditures depends on age and income, variation in these two characteristics are used to represent within-community variation in demand.(11) The results in column (a) offer support for the hypothesis that community composition is related to voter support for tax limitations. The estimated coefficients on both INCOME VARIATION and AGE VARIATION have the expected positive signs. In particular, the results indicate that municipalities consisting of residents with greatly varying ages are more likely to retain the existing fiscal constraints. An increase in within-community AGE VARIATION is associated with an increased conditional probability that the existing controls will continue to be retained. The effect of income variation, however, is measured with a large standard error. While not shown here, similar results are obtained by using the standard deviation of income instead of the coefficient of variation.

The estimates in column (a) also show that the conditional probability of retaining fiscal limits is greater in communities that have higher levels of property taxes, holding population constant. This is consistent with the results of Ladd's 1978 study [21] that found that high levels of property taxes and increased past growth in these taxes were positively related to decisions by state governments to enact constraints on local governments in the 1970s. The estimated effect of recent past PROPERTY TAX GROWTH, however, is opposite to that found in Ladd's analysis. The negative coefficient found here has a plausible explanation. It appears that the rapid increases in property tax burdens per capita led residents to vote to adopt home-rule status as a means [TABULAR DATA FOR TABLE II OMITTED] of broadening the local tax base and potentially obtaining property tax relief. In comparing the findings here to other results in the literature, it appears that the effects of high and rising property tax levels on voter preferences for fiscal limitations may depending on the specific features of the fiscal controls under consideration. Consistent with Ladd's state-level findings, Skidmore and Alm [30] report that voters in states with rising property tax burdens are more likely to enact fiscal limitations. However, Skidmore and Alm observe that higher property tax levels are associated with a lower probability that voters will approve fiscal limitations.

As expected, the level of MEDIAN INCOME in the community has a negative effect on the conditional probability of retaining state controls. As discussed earlier, higher-income voters commonly are found to have a greater willingness to pay for government expenditures and are more willing to allow their community to issue municipal bonds. Ceteris paribus, voters in higher-income municipalities are more likely to prefer home rule and less likely to prefer the existing limits on local taxing and borrowing power.(12) Turning briefly to the POPULATION variable, there appears to be no discernible relationship between the population of the sample communities and the probability of adopting home rule.(13) A different result might have been observed, however, if municipalities under 2,500 in size were included. In 1970 there were over 1,000 incorporated municipalities in Illinois with fewer than 25,000 residents. While the sample used here consists of fewer than one-third of those communities, it accounts for almost half of all home-rule adoptions during 1970-1990. Although not indicated in Table II, it appears that very small communities are more likely to retain the existing controls on local government behavior.

Decisions by voters to retain or eliminate state controls on local government may depend on the political structure of each local government. Researchers analyzing the behavior of localities commonly investigate differences between the city manager form of government and governments led by elected mayors without the guidance of professional managers. While it is often hypothesized that manager-led local governments are run more efficiently and consequently would require less tax revenue to produce a given level of public output, recent evidence finds no efficiency differences between the two government types [15; 10]. The estimation equations in Table II include a dummy variable that is equal to 1 if the municipality has a manager form of government. While professional managers are fairly common in medium and large cities, they are less likely to be employed by smaller communities. As reported in Table I, just under ten percent of the sample communities have the manager form of government. Interestingly, the results indicate that communities that employ a manager are more likely to become home rule.

Also tested is a hypothesis about preferences for home-rule status and the community's method of selecting local council members. I investigate whether voters who reside in communities where council members are elected from the community at large will be more or less likely to adopt home rule status. Illinois communities with at-large representation typically have a "trustee-village" or "commission" form of government, whereas the "aldermanic-city" form of government ordinarily involves the election of representatives from wards or districts.(14) Although the estimated coefficient on AT LARGE has a positive sign, the results indicate that the nature of the selection process for community representatives is not significantly related to voter preferences for state controls on the taxing and borrowing power of local governments.

Column (b) contains the results of estimation with the exponential distribution. This distribution is a special case of the Weibull that results when the duration dependence parameter p is equal to one. Because the duration dependence parameter is estimated to be 0.97 in the Weibull equation shown in column (a), it is reasonable to impose the assumption that p equals one. As expected, the reduction in the number of estimated parameters has the effect of lowering the standard errors associated with some of the explanatory variables. The results in column (b) are almost identical to those reported in column (a). Finally, the results in column (c) are obtained from using a different specification of the property tax variable. Although in the first two columns the effect of higher total community property taxes is estimated while controlling for population, in column (c) property taxes per capita is included directly as an explanatory variable. While the sign of the property tax variable is consistent with the earlier results, the estimated coefficient is not as significantly different from zero.

In summary, the findings indicate that state-imposed limitations on local taxing and borrowing power are preferred by voters in age-heterogeneous, lower-income municipalities.(15) The results also provide some indication that the fiscal limitations are preferred by voters in high property tax communities in which the past growth in property taxes has been modest. Moreover, the results suggest that voters in communities run by professional managers may be more likely to vote to eliminate the existing state controls on local fiscal powers.

This study investigates the factors influencing decisions by voters to retain or eliminate limitations on the fiscal powers of local governments, and the results suggest that the potential endogeneity should be taken into account when investigating the effects of fiscal controls on the level or growth rate of government taxes and expenditures. A better understanding of the decision to impose fiscal constraints ultimately should improve our ability to model the impact of the Tax Revolt on the fiscal behavior of state or local governments.

VI. Conclusions

To identify the determinants of preferences for state-imposed limits on the taxing and borrowing powers of local governments, this study examines the decisions made by a number of communities to eliminate existing fiscal controls. The paper first outlines a set of factors that theoretically could lead a majority of residents to vote to eliminate existing state controls on the fiscal authority of their local government. The system of home rule in the state of Illinois provides a unique opportunity to test these hypotheses. By voting to adopt home-rule authority, Illinois residents are voting to remove state-imposed constraints on local fiscal decisions. Home-rule municipalities can raise property tax rates above the state statutory limit, broaden the tax base to include taxation of other activities (namely, retail sales), and issue industrial development bonds as well as general obligation bonds without conducting referenda.

The determinants of home-rule adoptions are examined with a sample of 287 small municipalities. The results indicate that the elimination of state controls clearly is not an attractive option for the majority of voters in most communities. Nevertheless, a majority of voters in almost two dozen of the sample cities and villages went against the tax limitation trend in the 1970s and 1980s and chose to remove existing state-imposed limits on the behavior of their local community. Despite the overwhelming popularity of the tax and borrowing limitations, the data contain enough information to identify a number of factors that affect preferences for home rule.

The empirical results confirm the importance of community composition as a factor influencing preferences for fiscal limitations. Limits on local fiscal behavior are more likely to be preferred by voters in municipalities with relatively larger intra-community variation in the ages of the residents. Because demands for local government expenditures commonly are found to vary across individuals in different age groups, the findings suggest that greater within-community variation in individual demands is associated with stronger preferences for state-imposed constraints on the fiscal behavior of local governments. The evidence provides a demand-side rationale for the existence of tax and borrowing limitations - they are sought by voters who view fiscal constraints as a device to protect themselves against the possibility of incurring large welfare costs due to the enactment of local fiscal policies that are preferred by other residents but are not preferred by them. These costs are referred to by Buchanan and Tullock [5] as the external costs of collective decision-making. Additional findings suggest that the fiscal controls are preferred by voters in low-income communities. Communities that have experienced larger increases in property tax revenues are more likely to adopt home rule status, most likely in an attempt to broaden the tax base beyond the property tax. Finally, the results suggest that residents of local governments that have city or village managers are more likely to vote to eliminate the existing controls.

An earlier version of this paper was presented at the 1993 meetings of the Southern Economic Association in Washington, D.C. I appreciate the helpful comments of Michael Nelson, Bill Sjostrom, and Ernest Zampelli.

1. Explanations for the existence of excessive levels of taxation are discussed by Rubinfeld [29] and Courant, Gramlich, and Rubinfield [8]. While the empirical work on the decision to enact limits is meager, theoretical discussions are contained in Courant and Rubinfeld [7], Toma and Toma [36], and Bell and Fisher [3]. The tax limitation movement of the late 1970s stimulated volumes of writings on the Tax Revolt, including Ladd and Tideman [22] and an entire supplemental issue of the National Tax Journal in 1979.

2. Giertz and McGuire [14] review traditional explanations for the existence of state-imposed limits on local governments. Hoyt and Toma [18] also discuss the paradox of having higher levels of government impose constraints on local governments, given that the greater competition among localities is likely to ensure that local governments are the most responsive to the preferences of their residents.

3. Although a system of Lindahl-type prices (in which prices are raised for voters with demands above the median and prices are lowered for voters with low demands) can generate complete agreement with a particular spending choice, it is unlikely that this type of price adjustment is undertaken in actual practice.

4. Fisher [13] discusses the conditions under which the median level of spending will defeat any alternative level of spending in a majority rule election. The conditions include single-peakedness of preferences and unidimensional choice.

5. This is shown graphically in Bell and Fisher [3].

6. Since the late 1800s, the city of Chicago had aggressively pursued local autonomy from state legislative control. The 1970 Constitution granted home-rule powers not only to Chicago but to other cities as well.

7. Much of the institutional material in this section on home rule in Illinois is from Banovetz and Kelly [2]. Hill [17] contains information on the number of states that allow some type of home-rule powers.

8. This study does not incorporate information on the method of selection of local school boards. Although school systems are financed by local property taxes, the boundaries of Illinois school systems are not coterminus with municipalities. Unlike most states, Illinois has no dependent school districts, which are school districts that are considered to be agencies of the municipalities or other levels of government [37].

9. If data permitted, it would be possible to incorporate time-varying covariates into the hazard model. Relying on beginning-of-spell values, however, is common [28]. Heckman and Singer [16] argue that there are identification problems inherent in using time-varying covariates in single-spell duration models such as the model used here.

10. Numerous specifications exist, and the choice among them depends on theoretical and empirical considerations. See Kiefer [24] for a discussion. For theoretical reasons, the Weibull is desirable because it allows for the conditional probability of adopting home-rule status to fall over time. This is consistent with what we expect - if positive net benefits exist from changing to home rule, then communities that adopt home rule are likely to do so earlier as opposed to later. An examination of the actual adoption times reveals that a slight majority of municipalities that became home rule over the 20 year period did so in the first 6 years. Because of the large number of censored observations, however, and because the rest of the home-rule adoptions were fairly evenly spread over the last 14 years of the sample period, the possibility that the conditional probability of retaining fiscal limits is time invariant cannot be ruled out. That calls for use of the exponential distribution.

11. Herfindahl-Hirschman indices for education and race also were constructed for use in the preliminary analysis. Because the index for education variation was highly correlated with the income variation index, it was not used here. A race index also is not included in the current analysis because this index varied little across communities. (The median percentage white in these communities was 97%.)

12. As further evidence that voter preferences for fiscal limits depend on the specific features of the limits under consideration, Skidmore and Alm [30] observe the opposite relationship between income and preferences for fiscal limits. In their analysis of voter approval of fiscal controls in statewide elections, they find that voters in high-income states are more likely to approve the limits. In the case of many of the statewide elections included in their sample, it is likely that redistributive concerns play an important role in enacting the fiscal limits.

13. The square of population also was used in preliminary regressions in order to capture any difference between the smaller and larger communities in the sample. An interactive variable between population and the growth in population also was used to see if large growing communities nearing 25,000 in size were more willing to walt until they automatically become home rule. No significant results were obtained.

14. Also note that the at-large communities have a fixed number of councilmembers (typically six or four) while the by-district communities typically elect two representatives from each ward where the number of wards is based on community population [20].

15. I investigated the robustness of the important finding that limitations on local government fiscal decisions are preferred by voters in age-heterogenous communities. The results in Table II indicate that home-rule adoptions during the period 1970-1990 were less likely to take place in localities with relatively high within-community variation in residents' ages as measured in 1970. A comparison of the age variation in 1970 with the age variation in 1990 for the sample of 287 communities reveals that the age variation varied little over time. The Spearman rank correlation coefficient indicates that the rankings of the communities in terms of age variation remained very stable over the 20 year period.

References

1. Alm, James, Michael McKee, and Mark Skidmore, "Fiscal Pressure, Tax Competition, and the Introduction of State Lotteries." National Tax Journal, December 1993, 463-76.

2. Banovetz, James and Thomas Kelty. Home Rule in Illinois: Image and Reality. Springfield, Ill.: Sangamon State University, 1987.

3. Bell, Michael and Ronald Fisher, "State Limitations on Local Taxing and Spending Powers: Comment and Re-evaluation." National Tax Journal, December 1978, 391-98.

4. Brown, Byron and Daniel Saks, "Spending for Local Public Education: Income Distribution and the Aggregation of Private Demands." Public Finance Quarterly, January 1983, 21-45.

5. Buchanan, James and Gordon Tullock. The Calculus of Consent: Logical Foundations of Constitutional Democracy. Ann Arbor: University of Michigan Press, 1962.

6. Citrin, Jack, "Do People Want Something for Nothing: Public Opinion on Taxes and Government Spending." National Tax Journal, June 1979, supplement, 113-30.

7. Courant, Paul and Daniel Rubinfeld, "On the Welfare Effects of Tax Limitations." Journal of Public Economics, December 1981, 289-316.

8. Courant, Paul, Edward Gramlich, and Daniel Rubinfeld, "Why Voters Support Tax Limitation Amendments: The Michigan Case." National Tax Journal, March 1980, 1-20.

9. Davis, Michael and Kathy Hayes, "The Demand for Good Government." The Review of Economics and Statistics, February 1993, 148-52.

10. Deno, Kevin and Stephen Mehay, "Municipal Management Structure and Fiscal Performance: Do City Managers Make a Difference?" Southern Economic Journal, January 1987, 627-42.

11. Dubin, Jeffrey, Michael Graetz, and Louis Wilde, "State Income Tax Amnesties: Causes." Quarterly Journal of Economics, August 1992, 1057-70.

12. Eberts, Randall and William Fox, "The Effect of Federal Policies on Local Public Infrastructure Investment." Public Finance Quarterly, October 1992, 557-71.

13. Fisher, Ronald. State and Local Public Finance. Burr Ridge, Ill.: Irwin, Inc., 1996.

14. Giertz, J. Fred and Therese McGuire, "State Aid and State-Imposed Controls and Local Fiscal Outcomes." Proceedings of the 83rd Annual Conference, National Tax Association-Tax Institute of America, 1991, 62-67.

15. Hayes, Kathy and Semoon Chang, "The Relative Efficiency of City-Manager and Mayor-Council Forms of Government." Southern Economic Journal, July 1990, 167-77.

16. Heckman, James, and Burton Singer, "Econometric Duration Analysis." Journal of Econometrics, January/February 1984, 63-132.

17. Hill, Melvin. State Laws Governing Local Government Structure and Administration. Athens, Georgia: University of Georgia Institute of Government, 1978.

18. Hoyt, William and Eugenia Toma, "State Mandates and Interest Group Lobbying." Journal of Public Economics, March 1989, 199-213.

19. Husted, Thomas, "Nonmonotonic Demand for Income Redistribution Expenditures: The Case of AFDC." Southern Economic Journal, January 1989, 710-27.

20. Illinois Municipal League. Illinois Municipal Handbook (1988-89 Edition). Springfield, Ill.: Illinois Municipal League, 1990.

21. Ladd, Helen, "An Economic Evaluation of State Limitations on Local Taxing and Spending Powers." National Tax Journal, March 1978, 1-18.

22. ----- and Nicolaus Tideman, eds. Tax and Expenditure Limitations. Washington: The Urban Institute Press, 1981.

23. Ladd, Helen and Julie Boatwright Wilson, "Why Voters Support Tax Limitations: Evidence from Massachusetts' Proposition 2-1/2." National Tax Journal, June 1982, 121-48.

24. Kiefer, Nicholas, "Economic Duration Data and Hazard Functions." Journal of Economic Literature, June 1988, 646-79.

25. Marlin, Matthew, "The Effectiveness of Economic Development Subsidies." Economic Development Quarterly, February 1990.

26. Nelson, Michael, "An Empirical Analysis of the Determinants of Local Government Structure in Metropolitan Areas in the U.S." Southern Economic Journal, October 1990, 443-57.

27. Pack, Howard and Janet Rothenberg Pack, "Metropolitan Fragmentation and Local Public Expenditures." National Tax Journal, December 1978, 349-62.

28. Rose, Nancy and Paul Jostow, "The Diffusion of New Technologies: Evidence From the Electric Utility Industry." Rand Journal of Economics, Autumn 1990, 354-73.

29. Rubinfeld, Daniel. "The Economics of the Local Public Sector," in Handbook of Public Economics, vol. II, edited by Alan Auerbach and Martin Feldstein. Amsterdam: North-Holland, 1987.

30. Skidmore, Mark and James Alm. "Voting on Tax and Expenditure Limitations." Mimeo, University of Colorado at Boulder, 1995.

31. State of Illinois. Illinois Counties & Incorporated Municipalities. Springfield, Ill.: Secretary of State, 1991.

32. -----. Statewide Summary of Municipal Finance in Illinois. Springfield, Ill.: Auditor of Public Accounts, 1964, 1970.

33. Temple, Judy, "Limitations on State and Local Borrowing for Private Purposes." National Tax Journal, March 1993, 41-52.

34. ----- and Susan Porter-Hudak, "Preferences for State Spending and Tax Policies: Evidence from Survey Data on the Role of Income." Economics & Politics, March 1995, 43-58.

35. Todo-Rovira, Adolf, "Empirical Analysis of the Provision of Local Public Goods: An Alternative to the Median Voter Model." Public Finance, 1991, 490-511.

36. Toma, Eugenia and Mark Toma, "Bureaucratic Responses to Tax Limitation Amendments." Public Choice, 1980, 223-35.

37. U.S. Department of Commerce, Bureau of the Census. 1987 Census of Governments, Compendium of Government Finances. Washington: Government Printing Office, 1990.

38. -----. Census of Population and Housing: Characteristics of the Population - Illinois. Washington: Government Printing Office, 1970.