SEARCH

SEARCH BY CITATION

Abstract

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. REVENUE DIVERSIFICATION
  5. REVENUE COMPLEXITY
  6. MODEL ESTIMATION
  7. REGRESSION RESULTS
  8. DISCUSSION
  9. CONCLUSION

This paper examines (1) whether revenue diversification leads to greater instability as represented by revenue volatility, and (2) whether revenue complexity produces fiscal illusion as represented by increased public expenditures. These questions are answered by analyzing panel data on municipal governments between 1970 and 2002. The findings suggest that fiscal illusion does not occur among municipal governments, but revenue diversification does influence levels of volatility. However, the way in which municipalities diversify is important for achieving revenue stability. When diversification is considered in isolation, both tax and nontax diversification reduce revenue volatility. When diversification and complexity are considered simultaneously, the statistical effect of nontax diversification disappears. But, when a tax revenue structure is both diversified and complex, the likely outcome is greater revenue volatility rather than stability.


INTRODUCTION

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. REVENUE DIVERSIFICATION
  5. REVENUE COMPLEXITY
  6. MODEL ESTIMATION
  7. REGRESSION RESULTS
  8. DISCUSSION
  9. CONCLUSION

Dating back to medieval times in Europe and colonial times in America, the property tax is perhaps the oldest form of taxation used by contemporary government. State governments began to decrease their reliance on property taxation during the Great Depression. Conversely, local governments have remained heavily dependent on property tax revenue—seen as the most stable and arguably the most visible tax source—to finance public service provision. However, the property tax revolts initiated by California's Proposition 13 forced limits to be placed on property tax growth across the nation. In response, local governments began to diversify their revenue structures away from reliance on property taxation. In some instances, this also led to greater complexity in local government revenue structures.

Oates1 suggested that the theoretical premises of revenue diversification and revenue complexity represent two competing hypotheses explaining government revenue structures. As presented in the extant literature, complexity and diversification are different and unique constructs. Revenue diversification is often advocated as a strategy underlying effective fiscal management. In this context, financial managers diversify their revenue structures (but do not necessarily make them more complex) to decrease the instability of the overall tax structure2 and better prepare governments for economic downturns and fiscal crises.3 On the other hand, revenue complexity is often associated with pursuits of government expansion. Under this veil, public managers attempt to increase the complexity of revenue structures, thereby decreasing the aggregate visibility of government financing to its citizens, creating fiscal illusion, and advancing self-interested opportunities for increasing public expenditures.1 This expansion likely leads to an overproduction or inefficient level of public goods and services.

Applying these two theoretical premises to explain revenue structures at the local level of government presents a unique scenario. Local government diversification requires a greater reliance on tax and nontax revenue sources possibly less stable than the property tax. Thus, a potential consequence of local government revenue diversification is greater revenue volatility rather than stability. Moreover, the devolution of power from the federal government to lower levels of government has instilled upon local governments greater responsibility for direct public service provision. This added responsibility has created an inherent need for budget expansion at the local level. Thus, financial managers might feel compelled to pursue revenue complexity strategies to exploit opportunities for increasing public expenditures. Therefore, it is possible that local governments' increasing reliance on revenue sources alternative to the property tax has resulted in both greater instability in revenue structures and increased expenditures due to fiscal illusion.

The purpose of this paper is to examine the consequences of revenue diversification and revenue complexity at the local level of government. In particular, this paper seeks to determine (1) whether revenue diversification leads to greater instability as represented by revenue volatility, and (2) whether revenue complexity produces fiscal illusion as represented by increased public expenditures. These questions are answered by analyzing panel data on municipal governments during the 1970–2002 time period. The next two sections provide a brief discussion of the extant literature on revenue diversification and revenue complexity. The paper then proceeds with an explanation of the econometric models and regression results. Finally, the implications of the findings are discussed.

REVENUE DIVERSIFICATION

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. REVENUE DIVERSIFICATION
  5. REVENUE COMPLEXITY
  6. MODEL ESTIMATION
  7. REGRESSION RESULTS
  8. DISCUSSION
  9. CONCLUSION

During the Great Depression, there was a significant increase in property tax delinquencies4 coupled with a decline in property values and property tax revenues.5 As a result, states began to adopt both sales and income taxes to reduce or replace their dependence on the property tax.4 However, local governments remained heavily dependent on property taxation.6 Within a few years after the Great Depression, the property tax essentially became a “local tax” and was primarily used to finance education and municipal services.4 Between the Great Depression and World War II, the property tax dominated local public finance and was the only major tax source for local governments.7 Local governments were so heavily dependent on property tax revenue that the scope of public service provision was largely determined by the property tax base.7

The tax revolts significantly changed the fiscal environment for local governments because property taxation was the primary target of the antitax sentiment. Leading up to the passage of California's Proposition 13, the United States Advisory Commission on Intergovernmental Relations (ACIR) issued a report endorsing the use of income and sales taxes for local governments to promote more balanced revenue structures.8 The commission advocated the tradition of strong local government and encouraged states to authorize the use of local sales and income taxes.8 The commission justified its new stance on tax diversification on the basis of (1) the increasing unpopularity of the property tax and perception that sales and income taxes were more equitable and palatable, and (2) the belief that local diversification allowed for greater local autonomy and reduced pressure on state policymakers to impose new taxes to support local governments.8 The commission also advocated adoption of user charges for local governments in cases where (1) beneficiaries of a service could be readily identified, (2) fees could reduce waste, (3) the service would benefit an individual more than the community, (4) fees could be easily collected, and (5) the fee seemed generally equitable.8

The combination of changing viewpoints on local government finance and flurry of tax and expenditure limitations emerging from the tax revolts influenced local governments to diversify their revenue structures away from property taxation toward other tax and nontax revenue sources. Subsequent studies on diversification questioned the motivations underlying this trend, as well as the implications of more diverse revenue structures. Shannon3 discovered that governments are motivated by specific goals in selecting a particular mix of taxes. Tax revenue diversification is used to accentuate the goals of equity, efficiency, administrative and compliance simplicity, revenue adequacy, and public acceptability.9 For example, a diversified revenue base promotes equity by capturing revenues from individuals who can avoid some taxes but not others.5 However, Ladd and Weist10 concluded that revenue balance for its own sake does not necessarily achieve the goal of a good revenue structure. Although balanced revenue structures are less distortionary and more equitable, balance does not promote revenue adequacy or competitiveness of the tax structure relative to surrounding jurisdictions.10 In addition, revenue balance is linked to greater centralization of authority in state government, inherently undermining local autonomy.10 Specifically, property tax reductions have a high probability of adversely affecting local governments if they are imposed by the state.9

Empirical work on the implications of state-level revenue diversification has generally found that diversification decreases revenue volatility even while controlling for changes in state economic activity and the composition of state revenues,11 that greater overall volatility in the revenue structure can be attributed to instability within individual tax sources,2,12,13 and that revenue diversification leads to better fiscal performance.14 At the local level, finance managers advocate diversification because it increases flexibility and stability and often leads to better management of environmental and fiscal stress.15 In particular, diversification away from property taxation can lead to an improved financial position; however, local governments with stability as their primary goal should select revenue portfolios with greater dependence on stable sources like property taxes.16

REVENUE COMPLEXITY

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. REVENUE DIVERSIFICATION
  5. REVENUE COMPLEXITY
  6. MODEL ESTIMATION
  7. REGRESSION RESULTS
  8. DISCUSSION
  9. CONCLUSION

The concept of fiscal illusion was first introduced by Puviani when he defined the term “as erroneous representation in our mind of phenomena that are by the force of circumstances of the most different natures.”17 One aspect of fiscal illusion involves circumstances in which informed taxpayers rationally accept taxes or expenditures they would otherwise dispute because they do not understand the elements involved in taxing and spending.18 The link between fiscal illusion and revenue complexity was first offered by Wagner,19 who attributed fiscal illusion to perception. Wagner19 argued that a taxpayer's perception of the price of public outputs is influenced by the methods used to extract resources from citizens to finance public outputs. As revenue structures become more complex, it becomes more difficult for citizens to develop accurate perceptions of the price of public outputs. As a result, citizens underestimate the cost of government and support increases in public expenditures.

Empirical work examining the implications of revenue complexity suggests that increased complexity in state tax structures results in a higher level of taxation and revenue adequacy.14,20 Other findings, however, suggest no relationship between revenue complexity and fiscal illusion in the form of increased demand for public expenditures21 or increased levels of real tax revenues and expenditures22 at the state level. At the local level, diversification can increase slack resources, thereby weakening the relationship between complexity and revenue growth.15 However, local governments that prefer higher revenue growth can achieve that goal with portfolios relying more heavily on high-growth business and personal income taxes.16 Other studies have found that tax and intergovernmental complexity leads to local government overspending23 and greater tax effort24 as a result of the fiscal illusion created by imperfect tax and consumption information. For example, of the 101 largest U.S. cities adopting a local sales or income tax between 1963 and 1990, forty-two cities implemented small but significant decreases in property taxes accompanied by large increases in spending.25 However, there is evidence to suggest that fiscal illusion is not universal, but rather has a varied impact on government spending across sectors.26

MODEL ESTIMATION

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. REVENUE DIVERSIFICATION
  5. REVENUE COMPLEXITY
  6. MODEL ESTIMATION
  7. REGRESSION RESULTS
  8. DISCUSSION
  9. CONCLUSION

To examine the consequences of revenue diversification and revenue complexity at the local level of government, two econometric models are offered. The first model examines the impact of revenue diversification on stability as represented by revenue volatility. The second model assesses the extent to which revenue complexity produces fiscal illusion as represented by increased public expenditures. As mentioned earlier, the extant literature conceptualizes complexity and diversification differently and uniquely, as well as attributes distinct outcomes to each premise. For these reasons, revenue diversification and revenue complexity are measured and empirically tested individually in this analysis. However, recognizing that these are not mutually exclusive concepts, the linkage between the two premises and presumed outcomes is explored in the discussion section after the initial regression results are offered.

Both econometric models use panel data of municipal governments between 1970 and 2002. Data were obtained from the Census of Governments Survey conducted by the U.S. Census Bureau. As required by law under Title 13, United States Code, Section 161, a census of all governments within the United States is conducted at five-year intervals.27 The Census of Governments Survey covers three major subject fields—government organization, public employment, and government finance. The U.S. Census Bureau also collects this information from a sample of governments in the years between each five-year census.

This analysis consists of all general-purpose municipal governments with populations greater than 25,000 included in the government finance portion of the Census of Governments Survey between 1970 and 2002.28 Owing to sampling in the non-census years, each municipality is not observed every year during the time period. The final data set consists of an unbalanced panel with 28,185 total observations. Although some municipalities are observed each of the 33 years under analysis, the number of municipalities observed in a given year ranges from the minimum of 372 to the maximum of 1,243. The unit of analysis is each individual municipal government each year it is observed.

Revenue Diversification and Instability

To determine whether revenue diversification leads to more or less revenue stability, municipal government revenue volatility is estimated as shown in the following equation:

  • image(1)

In this equation, revenue volatility for municipality i in year t (RVit) is modeled as a function of prior revenue volatility (RVit−1), revenue diversification (RD), fiscal capacity (FC), revenue policy (RP), and economic conditions (EC). All nondichotomous independent variables are lagged one year to overcome problems with endogeneity.29Table 1 provides a description of all variables.

Table 1.  Variables and Data Sources
VariableDescription and data source
Dependent variables
Per capita expendituresTotal general expenditures less capital outlay expenditures divided by population; Source: U.S. Census Bureau.
Revenue volatilityAbsolute value of residuals from general revenue growth trend regression model; Source: U.S. Census Bureau.
Revenue complexity and diversification
Tax revenue sourcesNumber of sources tax revenue is derived from; Source: U.S. Census Bureau.
Nontax revenue sourcesNumber of sources general charges and miscellaneous general revenue are derived from; Source: U.S. Census Bureau.
Tax diversificationHirschman–Herfindahl Index (HHI) measure of four tax revenue categories: property tax, sales and gross receipts tax, income tax, and other tax; Source: U.S. Census Bureau.
Nontax diversificationHirschman–Herfindahl Index (HHI) measure of three own-source revenue categories: total tax, general charges, and miscellaneous general revenue; Source: U.S. Census Bureau.
Fiscal capacity
Debt burdenTotal long-term debt outstanding divided by population; Source: U.S. Census Bureau.
Debt service burdenTotal debt service expenditures divided by total general revenue; Source: U.S. Census Bureau.
Tax leverageTotal general current operations expenditures divided by total tax revenue; Source: U.S. Census Bureau.
Tax visibilityPercentage of total tax revenue derived from property, sales, and income tax revenue; Source: U.S. Census Bureau.
Demand for services
PopulationMunicipal population; natural log values used for analysis; Source: U.S. Census Bureau.
Cost of providing services
State-local tax burdenState-local taxes as a percentage of income; Source: The Tax Foundation.
Intergovernmental aid
Per capita federal IGRTotal federal intergovernmental revenue divided by population; Source: U.S. Census Bureau.
Per capita state IGRTotal state intergovernmental revenue divided by population; Source: U.S. Census Bureau.
Revenue policy
Sales tax dummyDichotomous variable coded 1 if the municipality derives revenue from a local sales tax, and 0 otherwise; Source: U.S. Census Bureau.
Income tax dummyDichotomous variable coded 1 if the municipality derives revenue from a local income tax, and 0 otherwise; Source: U.S. Census Bureau.
Economic conditions
State unemploymentUnemployment rate for state corresponding to the municipality; Source: Bureau of Labor Statistics.
U.S. unemploymentUnemployment rate for the United States; Source: Bureau of Labor Statistics.

Using Carroll and Stater's30 approach, revenue volatility is defined as the extent to which actual revenue differs from expected revenue. This approach builds upon the work of White,2 but develops an adapted measure of volatility that allows for variation to occur both across units of analysis as well as over time. To measure deviations in actual revenue from expected revenue, a revenue growth trend regression model was first estimated as shown in the following equation:

  • image(2)

In this equation, the natural log of total general revenue for municipality i in year t (Rit) is modeled as a function of a time variable indicating the year (t) and a series of n−1 dichotomous variables identifying each municipality in the data set (i). From this regression equation, absolute deviations of the residuals serve as the dependent variable of revenue volatility. These values represent differences between actual general revenue for municipality i in year t and predicted general revenue for municipality i in year t based on the municipality's unique expected growth trend in total general revenue.31 Greater values of this dependent variable represent greater revenue volatility. A one-year lag of this variable is also included in the econometric model as an independent variable to capture the potential influence of prior revenue volatility on current volatility. This approach is important for capturing the incremental nature of public budgeting and the likely effect of prior-year decisions concerning a government's overall revenue structure on the volatility of its current revenue structure.

Two variables are used to measure revenue diversification. Both measures are calculated on the basis of the Hirschman–Herfindahl Index (HHI). Several definitions of revenue diversification have been offered throughout the extant literature.32 Among the various definitions, the most popular approach for measuring revenue diversification is the HHI.14,15,33,34 This approach calculates a diversification score ranging from 0 to 1 based on how evenly balanced a government's total revenue is among selected revenue categories.35 Increasing values of the HHI score represent greater levels of diversification. While the HHI approach is commonly used and is an accepted method for measuring revenue diversification, three important caveats should be noted. First, the HHI measure implies each unit of analysis (i.e., municipality) is equivalent in its ability to diversify its revenue structure.33 Second, the HHI measure assumes each entity utilizes all of the revenue categories selected for the calculation.33 Third, the HHI measure assumes an equal reliance on each revenue source is possible.

When using HHI to examine revenue diversification among municipal governments, these assumptions should be made with caution. Municipalities vary considerably in their abilities to diversify their revenue structures, primarily because the availability of alternative revenue sources is largely determined by the state. Of course, it should be expected that home rule versus Dillon's rule designations affect municipalities' abilities to diversify. However, even among home rule municipalities, diversification is at least partially affected by the existence of local options for sales and income taxes, which are granted by the states and often coupled with stipulations that further restrict access to those revenue sources. These restrictions place caps on local tax rates and/or limitations on increases in tax rates, which inhibit municipalities from using these alternatives to generate sufficient revenue to supplant their dependence on the property tax. Therefore, even if municipal governments have alternative revenue sources at their disposal, it is implausible to expect municipalities to generate equal amounts of revenue from these other sources compared with the property tax. That is not to say municipal governments have not diversified their revenue structures since the 1970s; municipalities have decreased their reliance upon property tax revenue. However, property tax revenue still accounts for a significant portion of own-source revenue for municipal governments. Therefore, using a single HHI measure that assumes balance among property taxes and the other selected revenue categories could underestimate and inappropriately measure diversification at the local level of government. To avoid adhering to strict assumptions and to account for variability in revenue structures among municipal governments, two different measures of revenue diversification are used for this analysis. These measures were developed to correspond most appropriately with trends in municipal uses of revenue that have occurred since the property tax revolts.36

The first variable (tax diversification) measures diversification only within a municipality's tax structure.37 This variable more appropriately captures diversification when nontax revenue sources are unavailable (or available on a very limited basis) and diversification is more likely to occur within the tax structure. This HHI score is based on the tax revenue categories of property, sales and gross receipts, income, and other.38 The proportion of total tax revenue generated from each category is used to determine the level of diversification. Increasing values of this HHI score are indicative of municipalities with greater balance among taxes within their revenue structures.

The second variable (nontax diversification) measures diversification that occurs mainly outside of a municipality's tax structure. This variable most appropriately captures diversification when local options for sales and income taxes are unavailable (or strictly limited) but nontax revenue sources like user charges and fees are more accessible and likely to garner greater diversification. This HHI score uses the following revenue categories: total taxes, general charges, and miscellaneous general revenue.39 The proportion of total own-source revenue generated from each category is used to determine the level of diversification. Increasing values of this HHI score capture municipalities that might still be somewhat dependent on property tax revenue but have diversified their revenue structures outside of taxes by implementing a more direct fee-for-service type of financing for public services. Municipalities that have diversified their revenue structures both within and outside of their tax structures would score high on both HHI measures.

Revenue diversification is typically attributed to a reduction in the instability of the overall tax structure.2 As a result, diversification might better prepare governments for economic downturns and fiscal crises.3 At the local level in particular, diversification has been shown to lead to improved stability and management of fiscal stress.15 However, Berg, Marlin, and Heydarpour16 caution that local governments with stability as their primary goal should select revenue portfolios with greater dependence on stable sources like the property tax. Diversification at the local level of government involves movement away from property taxation and requires a greater reliance on tax and nontax revenue sources possibly less stable than the property tax. Therefore, a potential consequence of municipal government revenue diversification is greater revenue volatility rather than stability. As such, I hypothesize that a higher level of revenue diversification will lead to greater revenue volatility.

In equation (1), the variables used to control for the influence of a municipality's fiscal capacity on its level of revenue volatility are debt burden, debt service burden, and tax leverage. Debt burden is measured as total long-term debt outstanding divided by total population. Debt service burden is measured as total debt service expenditures divided by total general revenue. These two variables account for the longer-term solvency of a government.40 In addition, repayment of debt typically represents a mandatory expenditure for governments.41 Tax leverage is measured as total operating expenditures divided by total tax revenue. This variable assesses the extent to which a government has the capacity to increase resources if a particular revenue source generates an unexpected deficit.40 Specifically, the tax leverage factor indicates the amount that taxes would need to increase to support an expenditure increase. A higher tax leverage value indicates a higher tax increase would be required to support an expenditure increase.

Revenue policy is captured using two dichotomous variables: sales tax dummy and income tax dummy. These two variables are coded with values of 1 if the municipality generates revenue from sales and income taxes, respectively, and 0 otherwise. Two variables are used to control for the influence of economic conditions. The first variable (state unemployment rate) measures the annual unemployment rate for the state corresponding to the municipality. All municipalities within a single state receive the same value for this variable in a given year. The second variable (U.S. unemployment rate) measures the annual unemployment rate for the United States. All municipalities observed within a particular year receive the same value for this variable. Finally, a variable measuring the natural log of a municipality's population is included to control for the size of the jurisdiction.

Revenue Complexity and Fiscal Illusion

To determine whether revenue complexity leads to fiscal illusion, municipal government per capita expenditures are estimated as shown in the following equation:

  • image(3)

In this equation, per capita expenditures for municipality i in year t (EXPit) are modeled as a function of incrementalism (EXPit−1), revenue complexity (RC), fiscal capacity (FC), demand for services (D), cost of providing services (C), and intergovernmental aid (IA). All independent variables are lagged one year to overcome problems with endogeneity.29 Again, Table 1 provides a description of all variables.

The dependent variable (EXP) is per capita general current expenditures, which equals total general expenditures minus capital outlay expenditures divided by total population. General expenditures include all amounts of money paid out by a government during its fiscal year other than for debt retirement, investment security purchases, loan extensions, agency or private trust transactions, and amounts transferred to funds or agencies of the same government; it also excludes expenditures classified as liquor store, utility, or insurance trust.42 A one-year lag of this variable is also included in the econometric model as an independent variable to capture the influence of incrementalism in municipal budgeting. Again, this approach is important for capturing the likely effect of prior-year decisions concerning a government's level of service provision on its current level of expenditures.

Two variables are included in the econometric model to measure revenue complexity. The first variable (tax revenue sources) measures the total number of sources from which a municipality derives tax revenue. The second variable (nontax revenue sources) measures the total number of sources from which a municipality derives general charges and miscellaneous general revenue. Based on the Census of Governments classification system for government revenues, there are 19 possible tax sources and 25 possible nontax sources. Among the municipalities in this analysis, a maximum of 18 tax sources and 21 nontax sources were used by a single municipality to report revenue in a single year between 1970 and 2002.

Wagner19 suggested that citizens' perceptions of the price of public outputs are affected by the methods used to extract resources from them to finance public outputs. A complex revenue structure consists of multiple bases, and complexity of a government's overall revenue structure increases as the number of devices used to extract revenue from citizens increases.19 As a revenue structure becomes more complex, it becomes more difficult for citizens to develop accurate perceptions regarding the price of public outputs.19 This leads citizens to underestimate the cost of government and results in an expansion in public expenditures.19

This analysis closely follows Wagner's19 conceptualization by measuring revenue complexity on the basis of the number of devices used to extract revenue from citizens (i.e., number of tax and nontax sources). However, an important aspect of fiscal illusion is the visibility of a government's overall revenue structure to its citizens, which might not necessarily decrease as the number of revenue sources increases. For example, if a government changes its fiscal policy from deriving 100 percent of revenue from property taxes to deriving some portion of total revenue from a local option income tax, the aggregate visibility (and therefore complexity) of the government's overall revenue structure might not change although the number of tax sources increased from one to two. In such a case, a simple measure of the number of tax sources might overestimate the aggregate complexity of the government's overall revenue structure. However, an examination of the municipalities under analysis suggests the measures employed here do appropriately capture revenue complexity of municipal government revenue structures overall, because aggregate visibility does appear to decrease as the number of revenue sources increases.

Of the 1,243 municipalities in this analysis, 32 municipalities generated tax revenue from only one source at least once during the time period. In all instances when only one tax source was used (52 instances among the 32 municipalities), revenue was derived from property taxes. A total of 345 municipalities generated tax revenue from two sources at least once during the time period, many of which utilized two tax sources during several years under analysis. Among these 345 municipalities, there were 4,509 total instances during the 33 years of this study in which revenue was generated from only two tax sources. Nearly all of these instances (4,495) involved the property tax as one of the two tax sources. However, revenue was derived from both property tax and income tax in only seven of these instances, from both property tax and sales tax in only 15 instances, and never derived from both sales tax and income tax when two tax sources were used. This suggests that the overwhelming majority of instances in which municipalities generated revenue from only two tax sources involved property taxes coupled with sources other than the two other most visible taxes.22 Moreover, in the 5,114 instances in which three tax sources were used, tax revenue was derived from property, sales, and income taxes only five times. These cross-tabulations lend credence to the revenue measures used for this analysis by suggesting that municipal tax structures do indeed become less visible in the aggregate (and therefore more complex) as the number of revenue sources increases. This added complexity potentially creates fiscal illusion and leads to higher expenditures. As such, I hypothesize that a greater number of tax and nontax revenue sources will lead to higher per capita expenditures.

Four variables are used to control for the influence of a municipality's fiscal capacity on per capita expenditures over time. Debt burden, debt service burden, and tax leverage are measured the same as for the revenue volatility model shown in equation (1). Governments with less fiscal capacity might be expected to have lower expenditures because of resource constraints limiting their levels of public service provision. The fourth variable (tax visibility) measures the percentage of total general tax revenue derived from the three most visible sources: property, sales, and income.22 Greater visibility of a government's overall tax structure should limit opportunities for creating fiscal illusion to expand service provision and increase expenditures.

The demand placed upon municipal governments to provide services is represented by a variable measuring the population of the jurisdiction.43 Natural log values are used for analysis. A larger population might represent a greater demand for public service provision.

The cost of providing public services is represented by a variable measuring the state-local tax burden. State-local tax burden equals the amount of state-local taxes as a percentage of income. Higher levels of state-local tax burden might be indicative of higher costs of providing services because municipal governments primarily finance public services through tax revenue.

Finally, two variables are included to control for the influence of intergovernmental revenue on municipal expenditures. These two variables are the per capita amounts of federal and state intergovernmental revenue. Higher levels of intergovernmental revenue represent greater potential for public service provision and likely lead to higher per capita expenditures.

REGRESSION RESULTS

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. REVENUE DIVERSIFICATION
  5. REVENUE COMPLEXITY
  6. MODEL ESTIMATION
  7. REGRESSION RESULTS
  8. DISCUSSION
  9. CONCLUSION

Before running both regression models, several tests were conducted to determine the most appropriate estimation method for the data. The Modified Wald test for groupwise heteroskedasticity revealed heteroskedasticity in both econometric models; however, the Wooldridge test for autocorrelation in panel data indicated serial correlation was not problematic for either model. In addition, Hausman's specification test indicated that the random effects estimator would not be appropriate for the data. Based on these findings, both models are estimated using fixed effects regression and semirobust standard errors. The standard errors are clustered on the variable identifying each municipality to account for variation in the number of observations of municipalities over time due to sampling in the non-census years. Finally, all nondichotomous independent variables are lagged one year to overcome endogeneity.29

Table 2 provides descriptive statistics for all variables. As shown in Table 2, the municipalities under analysis annually spend an average of US$808.53 per capita excluding capital expenditures. However, with a standard deviation of US$604.87, per capita spending varies considerably among municipalities. Washington, DC, and New York City have the consistently highest per capita expenditure levels during the time period. Municipalities with the lowest levels of per capita expenditures can be found in California and Utah. In terms of volatility, actual general revenue deviates from predicted general revenue an average of US$3.41 in natural log terms. Coincidentally, Washington, DC, and New York City consistently maintain the most revenue volatility, as well as the highest levels of per capita expenditures. Moreover, municipalities with the least revenue volatility can again be found in California and Utah, as well as the states of Oklahoma, Georgia, and Indiana. Based on this information, revenue volatility appears to be correlated with municipal expenditures. However, the pairwise correlation coefficient between the two variables is only 0.2877.

Table 2.  Descriptive Statistics
VariableMeanStandard deviationMinimumMaximum
  1. N=28,185.

Per capita expenditures$808.53$604.87$2.56$23,080.60
Revenue volatility (Ln)$3.41$1.91$0.00$12.03
Tax revenue sources4.111.43018
Nontax revenue sources8.642.42021
Tax diversification0.51730.272800.9932
Nontax diversification0.73940.197101
Debt burden$1,278.32$1,732.40$0.00$64,100.31
Debt service burden7.90%8.04%0.00%212.10%
Tax leverage131.50%80.76%5.09%5047.20%
Tax visibility80.37%17.15%0.00%100.00%
Population107,685310,71825,0008,008,278
State-local tax burden10.08%1.01%6.56%15.32%
Per capita federal IGR$64.78$119.91$0.00$3,528.72
Per capita state IGR$181.09$257.45$0.00$6,883.76
Sales tax dummy0.76760.422401
Income tax dummy0.10630.308201
State unemployment6.18%2.05%2.20%18.00%
U.S. unemployment6.28%1.38%4.00%9.70%

The variables of greatest interest in Table 2 are the two measuring revenue complexity and the two representing revenue diversification. According to Table 2, the municipalities under analysis maintain an average of four tax and nine nontax revenue sources. With standard deviations of 1.43 and 2.42, respectively, there is less variation in the number of tax and nontax sources from which municipalities generate revenue. The city with the single greatest number of tax sources is Washington, DC, which used between 11 and 18 taxes to generate revenue throughout the time period. New York City also ranks notably high by generating revenue from at least nine tax sources each year under analysis. In terms of nontax revenue, New York City ranks at the top by using at least 17 nontax sources each year during the time period. Washington, DC, ranks second by utilizing at least 15 nontax sources every year. The city of Baltimore also ranks high by generating revenue from between 13 and 19 nontax sources throughout the time period.

Both revenue diversification variables are measured on a 0–1 scale. As can be seen from Table 2, the municipalities under analysis are more diversified among their nontax revenue sources than within their tax structures. Mean values for tax and nontax revenue diversification are 0.52 and 0.74, respectively. Washington, DC, Birmingham, and Auburn consistently have highly diversified tax structures during the time period. In terms of nontax revenue, the most diversified municipalities can be found in the states of Georgia, Mississippi, and California.

Table 3 provides the regression results for the revenue volatility model shown in equation (1). All variables except for debt service burden, sales tax dummy, and state unemployment achieve statistical significance at the 95 percent confidence level.44 The effect of each independent variable on revenue volatility should be considered as above and beyond the effect of prior-year volatility due to the inclusion of the lagged dependent variable as an independent variable. In addition, all statistically significant variables except for the two diversification variables exhibit positive signs, indicating an increase in any of those variables leads to greater revenue volatility. The model is significant overall at the 99 percent confidence level and explains 97.25 percent of the overall variation in revenue volatility among municipalities over time.

Table 3.  Regression Estimates for Revenue Volatility (Ln)
VariableCoefficienttP>|t|
  1. Note: Regression results are reported using two-way fixed effects (not shown) and robust standard errors clustered on the group variable. All nondichotomous independent variables are lagged one year.

Prior year volatility (Ln)0.742035.450.000
Tax diversification−0.0730−2.570.010
Nontax diversification−0.0590−2.260.024
Debt burden0.00002.000.045
Debt service burden0.00020.190.852
Tax leverage0.00022.540.011
Sales tax dummy0.00820.880.379
Income tax dummy0.23463.990.000
State unemployment−0.0019−1.190.235
U.S. unemployment0.11924.470.000
Population (Ln)0.16906.180.000
Constant−2.6055−6.250.000
Number of observations=24,852; number of groups=1,266
Observations per group (minimum=1; average=19.6; maximum=32)
F=25391.98; Prob.>F=0.0000; overall R2=0.9725

Table 3 shows that prior-year revenue volatility has the single greatest influence over current volatility. According to the regression results, a 1 percent increase in prior-year volatility leads to an average increase in current revenue volatility of 74.2 percent over time. This finding suggests that municipalities tend to experience persistent instability in their revenue structures rather than encountering revenue volatility as an isolated event within a single fiscal year. Moreover, the large influence of population on revenue volatility suggests that municipalities serving larger constituencies have greater tendencies to experience instability in their revenue structures. According to the regression results, a 1 percent increase in municipal population leads to an average increase in revenue volatility of 16.9 percent over time. The magnitude of this effect is third largest of all variables in the model.

Table 3 reveals consistent findings with respect to the revenue diversification variables. Contrary to expectations, both tax and nontax revenue diversification variables are shown to decrease revenue volatility. Although these variables do not exhibit the strongest influences over revenue volatility, the magnitudes of these effects suggest the variables are important determinants. Because these variables are measured on a 0–1 scale, a one-unit change in revenue diversification can be thought of as a change of one percentage point. Based on the regression results, a one-percentage-point increase in tax revenue diversification (moving from the mean HHI score of 0.52 to 0.53) leads to an average decrease in revenue volatility of 7.3 percent over time. Similarly, a one-percentage-point increase in nontax revenue diversification (moving from the mean HHI score of 0.74 to 0.75) would result in an average decrease in revenue volatility of 5.9 percent over time. These findings suggest that revenue diversification both within and outside of the tax structure does produce the desired outcome of reducing revenue volatility among municipal governments. However, the insignificance of the sales tax dummy and the positive effect of the income tax dummy suggest that the way in which municipal governments diversify their tax structures might be important for achieving revenue stability. Table 3 shows that generating revenue from a local option sales tax has no statistically significant effect on revenue volatility. Moreover, municipalities deriving revenue from a local option income tax will experience average increases in revenue volatility over time of 26.45 percent.45 Based on these findings, municipalities might be cautious about diversifying their tax structures through the use of local options for sales and income taxes if the desired outcome is greater revenue stability. However, adopting a revenue diversification strategy that excludes these two visible tax sources might produce greater complexity in the revenue structure overall and inadvertently create fiscal illusion for a municipality. The connection between these two concepts and potential outcomes will be explored further in the discussion section.

Turning to the analysis of revenue complexity, Table 4 provides the regression results for the expenditure model shown in equation (3). All of the variables except for population, per capita federal intergovernmental revenue, and both revenue complexity variables achieve statistical significance at the 95 percent confidence level. Again, the effect of each independent variable on per capita expenditures should be considered as above and beyond the effect of prior-year expenditures due to the inclusion of the lagged dependent variable as an independent variable. The model is significant overall at the 99 percent confidence level and explains 81.06 percent of the overall variation in per capita expenditures among municipalities over time.

Table 4.  Regression Estimates for Per Capita Expenditures
VariableCoefficienttP>|t|
  1. Note: Regression results are reported using two-way fixed effects (not shown) and robust standard errors clustered on the group variable. All independent variables are lagged one year.

Prior-year expenditures0.47504.340.000
Tax revenue sources−3.1097−0.450.654
Nontax revenue sources0.40040.250.803
Debt burden0.02924.260.000
Debt service burden−2.1544−2.950.003
Tax leverage−0.9062−2.190.029
Tax visibility−1.3764−2.550.011
Population (Ln)−25.017−1.310.191
State-local tax burden−5.4720−2.050.040
Per capita federal IGR−0.0295−0.860.388
Per capita state IGR0.48013.970.000
Constant789.012.540.011
Number of observations=24,726; number of groups=1,251
Observations per group (minimum=1; average=19.8; maximum=32)
F=504.74; Prob.>F=0.0000; overall R2=0.8106

Table 4 reveals that prior-year expenditures have a sizable influence on current expenditure levels, which provides evidence of incrementalism in municipal budgeting. According to the regression results, a US$1 increase in prior-year per capita general expenditures leads to an average increase in current per capita expenditures of US$0.48 over time. Aside from the influence of incrementalism, a municipality's debt service burden and tax visibility, as well as the state-local tax burden exhibit the largest influences over per capita expenditures. Increasing values of each variable lead to decreases in per capita municipal expenditures over time.

Perhaps the most striking finding from Table 4 is that both revenue complexity variables fail to achieve statistical significance at any conventional levels. Neither an increase in the number of tax sources nor an increase in the number of nontax sources will lead to a change in per capita expenditures among municipalities over time. As noted earlier, Wagner19 defined a complex revenue structure as consisting of multiple bases. In addition, complexity of a government's overall revenue structure increases as the number of devices used to extract revenue from citizens increases.19 The variables measuring the number of tax and nontax revenue sources directly accounts for the number of ways in which municipalities extract revenue from taxpayers. However, the regression results reveal that an increase in either of these variables will not lead to an increase in municipal expenditures. These findings suggest that revenue complexity does not create fiscal illusion at the local level of government. It might be the case, however, that the implications of revenue complexity depend on its interrelationship with revenue diversification. This idea is explored in the next section.

DISCUSSION

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. REVENUE DIVERSIFICATION
  5. REVENUE COMPLEXITY
  6. MODEL ESTIMATION
  7. REGRESSION RESULTS
  8. DISCUSSION
  9. CONCLUSION

The extant literature on revenue diversification and revenue complexity tends to conceptualize these two concepts differently, as well as associate unique outcomes to each premise. Following this predisposition, the analysis presented here aimed to further our understanding of these unique constructs and their associated outcomes as they apply to local government. In doing so, the preceding analysis revealed that revenue diversification does achieve the intended outcome of reducing revenue volatility for municipal governments, while revenue complexity does not lead to government expansion in the form of increased expenditures. The question that remains, however, is whether revenue diversification and revenue complexity truly represent mutually exclusive concepts leading to different outcomes. Revenue complexity and revenue diversification do not appear to be sufficient conditions for each other. A diversified revenue structure does not necessarily imply complexity if the diversification strategy focuses on achieving balance among visible revenue sources. Likewise, a complex revenue structure utilizing obscure revenue sources does not necessarily achieve balance often considered necessary for diversification. Nonetheless, revenue diversification and revenue complexity are not mutually exclusive concepts. In some instances, diversification could result in greater complexity within a revenue structure, and vice versa. In such cases where diversification and complexity intersect, it is unclear what the expected outcome should be. To address this question, the revenue volatility and expenditure models shown in equations (1) and (3) were reestimated with both diversification (tax diversification and nontax diversification) and complexity (number of tax sources and number of nontax sources) variables included in both models, as well as with interactions between them. The tax interaction variable interacts tax diversification and the number of tax sources, while the nontax interaction variable interacts nontax diversification with the number of nontax sources. The fixed effects regression estimates for both models are shown in Table 5. Both models are statistically significant at the 99 percent confidence level and explain 81.59 and 97.28 percent of the variation in per capita expenditures and revenue volatility, respectively, among municipalities over time.

Table 5.  Regression Models with Interaction Terms
VariableExpenditure modelVolatility model
CoefficienttCoefficientt
  • *, **, ***

    Statistical significance at the 10%, 5%, and 1% levels, respectively.

  • Note: Regression results are reported using two-way fixed effects (not shown) and robust standard errors clustered on the group variable. All nondichotomous independent variables are lagged one year.

Prior year expenditures0.49855.35***  
Prior year volatility (Ln)  0.741035.30***
Tax diversification−81.054−1.34−0.1974−2.91***
Tax revenue sources−0.3309−0.03−0.0183−1.68*
Tax interaction9.55160.680.03531.99**
Nontax diversification−1.1597−0.030.05140.65
Nontax revenue sources5.34701.280.01181.62
Nontax interaction−4.9283−1.02−0.0139−1.56
Debt burden0.02804.33***0.00001.96**
Debt service burden−2.7981−4.30***0.00020.29
Tax leverage−1.1866−2.60***0.00022.57***
Tax visibility−0.6126−1.35  
Population (Ln)−33.539−1.72*0.16526.14***
State-local tax burden−8.1614−2.98***  
Per capita federal IGR−0.0277−0.85  
Per capita state IGR0.47364.44***  
Sales tax dummy  0.02001.65*
Income tax dummy  0.24344.07***
State unemployment  −0.0019−1.20
U.S. unemployment  0.12064.51***
Constant862.532.83***−2.6221−6.31***
F=483.76F=23171.00
Prob.>F=0.0000Prob.>F=0.0000
Overall R2=0.8159Overall R2=0.9728

The expenditure model in Table 5 reveals that fiscal illusion in the form of increased expenditures simply does not occur at the local level of government. Commensurate with the preceding findings, both the number of tax sources and number of nontax sources fail to reach statistical significance at any conventional level. In addition, neither tax diversification nor nontax diversification display a statistically significant relationship with per capita expenditures. Finally, both the tax and nontax interaction variables also fail to achieve statistical significance in Model 5. These findings clearly reveal that revenue complexity and revenue diversification cannot be associated with the presumed outcome of government expansion when referring to municipal governments. Based on all of the findings presented in this paper, incrementalism, debt levels, tax capacity, and state aid have the greatest impacts upon municipal expenditures over time.

The regression results for the revenue volatility model presented in Table 5 are less consistent and therefore even more compelling. Although nontax diversification had a statistically significant and negative effect upon revenue volatility in the preceding analysis, the variable fails to reach statistical significance in the regression model shown in Table 5. In addition, neither the number of nontax revenue sources nor the nontax interaction displays a statistically significant relationship with revenue volatility. What is most interesting about this finding is that, based on the descriptive statistics presented in Table 2, municipalities are both more diversified outside of their tax structures and more complex among nontax revenue sources. While the findings presented in this paper suggest this added complexity will not necessarily produce fiscal illusion and lead to government expansion, a diversification strategy focusing on nontax revenue sources is likely to have little effect in reducing volatility within a municipality's revenue structure. However, the positive effect of the income tax dummy and insignificant effect of the sales tax dummy in Table 5, which are consistent with the preceding results, suggest that revenue stability might best be achieved with a diversification strategy utilizing tax sources other than sales and income taxes. But, such an approach might incorporate less visible taxes into the revenue structure, thereby creating more overall complexity in the pursuit of greater diversification. And, even more problematic are the empirical findings related to the interaction between tax diversification and complexity. According to the regression results presented in Table 5, a tax structure that is both diversified and complex (i.e., tax interaction) will likely be more volatile over time. While tax diversification and tax complexity individually lead to a reduction in revenue volatility among municipalities over time, the reverse effect occurs when the two concepts intersect. Although the descriptive statistics in Table 2 show the municipalities under analysis are less diversified within their tax structures and less complex among tax sources, an overwhelming majority generating revenue from more than one tax source seem to be utilizing property taxes coupled with less visible sources than sales and income taxes. It is perhaps more likely, therefore, for municipal revenue structures to exhibit characteristics of both diversification and complexity. While this may not lead to budget expansion resulting from fiscal illusion, the combination might produce an outcome related to revenue volatility that is completely opposite of intentions for local governments.

CONCLUSION

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. REVENUE DIVERSIFICATION
  5. REVENUE COMPLEXITY
  6. MODEL ESTIMATION
  7. REGRESSION RESULTS
  8. DISCUSSION
  9. CONCLUSION

The purpose of this paper was to examine the consequences of revenue diversification and revenue complexity at the local level of government. In particular, this paper sought to determine (1) whether revenue diversification leads to greater instability as represented by revenue volatility, and (2) whether revenue complexity produces fiscal illusion as represented by increased public expenditures. These questions were answered by analyzing panel data on municipal governments between 1970 and 2002. The findings suggest that fiscal illusion does not occur among municipal governments. The regression results revealed that neither an increase in the number of tax sources nor an increase in the number of nontax sources will lead to an increase in municipal expenditures. This finding remained consistent when tax and nontax diversification, as well as two interaction terms, were also included in the econometric model.

To the contrary, the findings also revealed that revenue diversification does influence levels of revenue volatility among municipal governments. However, the way in which municipal governments diversify their revenue structures is important for achieving the desired outcome of revenue stability. When diversification is considered in isolation, both tax diversification and nontax diversification are likely to reduce revenue volatility. However, when diversification and complexity are considered simultaneously, the statistical effect of nontax diversification disappears. Moreover, the regression results revealed that when a tax revenue structure is both diversified and complex, the likely outcome is greater revenue volatility rather than stability. Overall, these findings imply that revenue diversification as a financial management strategy to insulate governments from economic downturns and fiscal crises should encompass more visible tax sources (but not necessarily sales and income taxes) that minimize aggregate complexity of the tax structure to successfully accomplish this goal. If municipalities are not granted the ability to pursue diversification in this sense, the implications might be opposite of intentions as revenue volatility might increase rather than decrease. Although revenue complexity does not likely produce the presumed outcome of fiscal illusion and government expansion at the local level of government, the findings from this study suggest that complexity should nonetheless be minimized to achieve the desired outcome of revenue stability associated with revenue diversification.

Footnotes
  1. 1. Wallace E. Oates, “On the Nature and Measurement of Fiscal Illusion: A Survey,” in Studies in Fiscal Federalism, ed. Wallace E. Oates (Aldershot, UK: Edward Elgar, 1991): 431–448.

  2. 2. Fred C. White, “Trade-Off in Growth and Stability in State Taxes,” National Tax Journal XXXVI, no. 1 (1983): 103114.

  3. 3. John Shannon, “State Revenue Diversification—The Search for Balance,” in The Quest for Balance in State-Local Revenue Structures, ed. Frederick D. Stocker (Cambridge, MA: Lincoln Institute of Land Policy, Tax Policy Roundtable, Property Tax Papers Series TPR-16, 1987): 9–37.

  4. 4. Glenn W. Fisher, “Some Lessons from the History of the Property Tax,” Assessment Journal 4, no. 3 (1997): 4047.

  5. 5. Holley H. Ulbrich, “Nonproperty Taxes,” in Local Government Finance, eds. John E. Petersen and Dennis R. Strachota (Chicago: Government Finance Officers Association, 1991): 113–133.

  6. 6. Institute of Property Taxation, “The Property Tax: History and Economic Impact as a Background to Modern Management Techniques,” in Property Taxation, 2nd ed., ed. Jerrold Janata (Washington, DC: Institute of Property Taxation, 1993): 3–94.

  7. 7. Lang Cantrell, “Some Basic Modifications of American Property,” The Journal of Finance 9, no. 4 (1954): 427428.

  8. 8. Advisory Commission on Intergovernmental Relation (ACIR), “Local Government Reorganizational Issues,” in The Challenge of Local Government Reorganization, Washington, DC: U.S. Government Printing office.

  9. 9. John H. Bowman, “Recent Changes in Property Taxation and Their Implications for Balance in State and Local Revenue Systems,” in The Quest for Balance in State-Local Revenue Structures, ed. Frederick D. Stocker (Cambridge, MA: Lincoln Institute of Land Policy, Tax Policy Roundtable, Property Tax Papers Series TPR-16, 1987): 71–105.

  10. 10. Helen F. Ladd and Dana R. Weist, “State and Local Tax Systems: Balance among Taxes vs. Balance among Policy Goals,” in The Quest for Balance in State-Local Revenue Structures, ed. Frederick D. Stocker (Cambridge, MA: Lincoln Institute of Land Policy, Tax Policy Roundtable, Property Tax Papers Series TPR-16, 1987): 39–69.

  11. 11. Craig L. Johnson, Sharon N. Kioko, and Samuel B. Stone, “Does Revenue Diversification Matter?” Working Paper presented at the 17th Annual Conference of the Association for Budgeting and Financial Management, 2005.

  12. 12. William M. Gentry and Helen F. Ladd, “State Tax Structure and Multiple Policy Objectives,” National Tax Journal 47, no. 4 (1994): 747772.

  13. 13. Oskar Ragnar Harmon and Rajiv Mallick, “The Optimal State Tax Portfolio Model: An Extension,” National Tax Journal 47, no. 2 (1994): 395401.

  14. 14. Jack P. Suyderhoud, “State-Local Revenue Diversification, Balance, and Fiscal Performance,” Public Finance Quarterly 22, no. 2 (1994): 168195.

  15. 15. Rebecca Hendrick, “Revenue Diversification: Fiscal Illusion or Flexible Financial Management,” Public Budgeting and Finance 22, no. 4 (2002): 5272.

  16. 16. Janine Berg, John Tepper Marlin, and Farid Heydarpour, “Local Government Tax Policy: Measuring the Efficiency of New York City's Tax Mix, FYs 1984–1998,” Public Budgeting and Finance 20, no. 2 (2000): 114.

  17. 17. Amilcare Puviani, Teoria della Illusions Finanziaria, reprint (Milano: Isedi, [1903] 1973): 5.

  18. 18. Domenico Da Empoli, “The Theory of Fiscal Illusion in a Constitutional Perspective,” Public Finance Review 30, no. 5 (2002): 377384.

  19. 19. Richard E. Wagner, “Revenue Structure, Fiscal Illusion and Budgetary Choice,” Public Choice 25 (1976): 4561.

  20. 20. Samuel H. Baker, “The Determinants of Median Voter Tax Liability: An Empirical Test of the Fiscal Illusion Hypothesis,” Public Finance Quarterly 11 (1983): 95108.

  21. 21. Erik Schokkaert, “Preferences and Demand for Local Public Spending,” Journal of Public Economics 34 (1987): 175188.

  22. 22. Walter S. Misiolek and Harold W. Elder, “Tax Structure and the Size of Government: An Empirical Analysis of the Fiscal Illusion and Fiscal Stress Arguments,” Public Choice 57, no. 3 (1988): 233245.

  23. 23. Geoffrey K. Turnbull, “The Overspending and Flypaper Effects of Fiscal Illusion: Theory and Empirical Evidence,” Journal of Urban Economics 44 (1998): 126.

  24. 24. R. D. Thomas and S. Boonyapratuang, “Local-Government Complexity—Consequences for County Property-Tax and Debt Policies,” Publius – The Journal of Federalism 23, no. 1 (1993): 118.

  25. 25. David Sjoquist, Mary Beth Walker, and Sally Wallace, “Estimating Differential Responses to Local Fiscal Conditions: A Mixture Model Analysis,” Public Finance Review 33, no. 1 (2005): 3661.

  26. 26. Geoffrey K. Turnbull, “Fiscal Illusion and the Output Expansion Hypothesis,” Public Finance Quarterly 21, no. 3 (1993): 305321.

  27. 27. U.S. Census Bureau website: http://www.census.gov/govs/www/cog2002.html

  28. 28. Town and township governments were excluded from the analysis.

  29. 29. Jeffrey M. Wooldridge, Introductory Econometrics (Mason, OH: Thomson South-Western, 2006).

  30. 30. Deborah A. Carroll and Keely Jones Stater, “Revenue Diversification in Nonprofit Organizations: Does It Lead to Financial Stability?”Journal of Public Administration Research and Theory Advance Access (2008). doi: 10.1093/jopart/mun025.

  31. 31. General revenue includes all amounts of money received by a government from external sources during its fiscal year other than for debt issuance, investment sales, agency or private trust transactions, and amounts transferred from other funds or agencies of the same government; it also excludes revenues classified as liquor store, utility, or insurance trust. General revenue can be further broken down into four main categories: taxes, intergovernmental revenue, current charges, and miscellaneous general revenue. U.S. Census Bureau, Government Finance and Employment Classification Manual, Chapter 7: Revenue.

  32. 32. Shannon (1987) suggested a diversified revenue structure should consist of property taxes, general sales taxes, and personal income taxes each contributing 25–43 percent of total revenues. Suyderhoud (1994) identified a diverse revenue portfolio as one in which property taxes, personal and corporate income taxes, general sales taxes, and all other revenues (including nontax revenues) contributed relatively equal amounts to total revenue. Carroll (2005) identified state tax revenue diversification as a relatively equal dependence on revenue from property taxes, personal and corporate income taxes, general sales taxes, motor fuel taxes, and all other taxes. At the local level in particular, Berg, Marlin, and Heydarpour (2000) examined diversification among New York City's property taxes, personal income taxes, sales taxes, and various business taxes.

  33. 33. Deborah A. Carroll, “Are State Governments Prepared for Fiscal Crises? A Look at Revenue Diversification during the 1990s,” Public Finance Review 33, no. 5 (2005): 603633.

  34. 34. Deborah A. Carroll, Robert J. Eger III, and Justin Marlowe, “Managing Local Intergovernmental Revenues: The Imperative of Diversification,” International Journal of Public Administration 26, no. 13 (2003): 14951519.

  35. 35. An HHI measure of revenue diversification is calculated as inline image, where Ri is the proportion of total revenue generated from each source and n represents the total number of revenue sources selected for measuring diversification.

  36. 36. Deborah A. Carroll and Benjamin J. Sharbel, “The Property Tax: Past, Present and Future,” in Public financial management, ed. Howard A. Frank (Boca Raton, FL: CRC Press, 2006): 151–178.

  37. 37. Taxes represent compulsory contributions imposed by a government for public purposes, other than for employee and employer assessments, contributions to retirement and social insurance trust systems, and special assessments to finance capital improvements. Tax revenue consists of gross amounts collected (including interest and penalties) minus amounts paid under protest and amounts refunded during the same period for all taxes imposed by a government, regardless of whether the government collects the taxes itself or relies on another government to act as its collection agent. U.S. Census Bureau, Government Finance and Employment Classification Manual, Chapter 7: Revenue.

  38. 38. The “other tax” revenue category consists of revenue generated from all other tax sources except property, sales and gross receipts, and income taxes.

  39. 39. Charges revenue consists of gross amounts received from charges imposed for providing current services or for the sale of products in connection with general government activities and excludes utility service charges. Miscellaneous general revenue consists of all other own-source general revenue not classified as taxes, intergovernmental revenue, or current charges; it also excludes liquor store, utility, and insurance trust revenue. U.S. Census Bureau, Government Finance and Employment Classification Manual, Chapter 7: Revenue.

  40. 40. Steven A. Finkler, Financial Management for Public, Health, and Not-for-Profit Organizations (Upper Saddle River, NJ: Pearson Prentice Hall, 2005).

  41. 41. John L. Mikesell, Fiscal Administration: Analysis and Applications for the Public Sector, 7th ed. (Belmont, CA: Wadsworth Publishers, 2007).

  42. 42. U.S. Census Bureau, Government Finance and Employment Classification Manual, Chapter 8: Expenditure.

  43. 43. An attempt was made to measure demand for municipal government services as the proportion of total expenditures for salaries and wages. The quality of these data was questionable, and so the variable was excluded from the analysis.

  44. 44. For all econometric models presented in this analysis, reduced form analyses were also conducted in which the statistically insignificant variables were removed and the regressions were reestimated. In all instances, the regression results were qualitatively unchanged. In addition, correlations among all variables were examined to ensure that the insignificant results in all models were not due to multicollinearity.

  45. 45. The relative effect of this dichotomous variable on the dependent variable was calculated using Halvorsen and Palmquist's (1980) approach. The percentage effect is equal to 100 × [exp(dichotomous variable)−1]. Using this approach, the number referred to in the text does not necessarily match the value of the coefficient shown in Table 3.