County Fiscal Stress: Cause and Consequence in California After Proposition 13*

Robert W. Wassmer
Associate Professor

Charles Anders
Executive Fellow

Graduate Program in Public Policy and Administration
California State University, Sacramento
Sacramento, California 95819-6081
(916) 278-6304

rwassme@csus.edu
http://www.csus.edu/indiv/w/wassmerr

August 1999

* The paper was produced with the help of a 1997-98 Extramural Research Program Contract from the Public Policy Institute of California (PPIC). The opinions expressed here are only the authors. An extended version of this work is contained in our 1998 manuscript Causes of Fiscal Stress in California’s Counties.

County Fiscal Stress: Cause and Consequence in California After Proposition 13

Abstract

The legislative implementation of Proposition 13 requires that the state of California divide countywide property tax revenue among local governments. Until the early 1990s the allocations that existed in the three years prior to the passage of the proposition largely determined this division. In the early 1990s the state responded to the dual pressure of a recession and its constitutional obligation to fund a portion of K-14 education by shifting some of the allocation of property tax revenue away from county governments and toward public schools. Observers claim that the post-Proposition 13 method of property tax disbursement, and the changes to it in the early 1990s, exacted a toll on the fiscal well being of many of California’s counties. This empirical study seeks evidence for or against this claim through the analysis of fiscal data from California’s stand-alone counties. Due to Proposition 13 like property tax reforms having been adopted, or being considered by other states, interest in the results of this study should extend beyond California.

I. Background

Counties throughout the United States deliver services that are important determinants of quality of life. For instance, county governments in California administer a significant portion of the state’s criminal justice system, deliver public health and welfare service to the poor and uninsured, provide parks and recreation activities, and are the primary providers of local government services to individuals and business that reside in unincorporated areas. A county in poor fiscal health is less likely to deliver these services at the amount and level of quality desired by residents and business.

To assess the fiscal health of county and other forms of local governments, Bradbury (1982, 1983, and 1984) pioneered the development of measures of fiscal stress. She categorizes fiscal stress as either budgetary fiscal stress or citizen fiscal stress. Budgetary fiscal stress occurs when local government officials are unable to balance their current accounts. Citizen fiscal stress occurs when a local government imposes too high of taxes, fees, or charges for the services provided; or provides too low of services for the local revenue collected. The determination of too high taxes/fees/charges and too low services is done relative to comparative local governments in similar situations. As Bradbury points out, budgetary fiscal stress and citizen fiscal stress are interrelated and it is essential to consider both in any analysis of municipal fiscal stress. Though we briefly discuss other ways of measuring the fiscal health of local governments, we have chosen to adopt Bradbury’s specific classification of fiscal stress throughout this paper.

California counties in the first half of the 1990s offer a natural experiment regarding the response of local governments to exogenous fiscal shocks. In 1978, Proposition 13 froze the rate of property taxation throughout California at one percent and specified that the distribution of statewide property tax revenue among local governments be done according to law. At the time no such law existed and the state legislature specified that property tax revenue be gathered at the county level and distributed back to local governments within the county (special districts, schools, municipalities, and the county). Distribution was initially based upon the average percentage of total county property tax revenue received by a particular local government in 1975, 1976, and 1977. With only slight tinkering from the legislature, this method of distributing countywide property tax revenues remained in place throughout the 1980s.

In at least three ways, the passage of Proposition 13 and its implementation was a likely catalyst for fiscal stress in California counties. The first way was the one-percent cap on total property taxation in the state. Prior to Proposition 13, of all local governments in California, counties relied on property taxation the most. A second possible catalyst for fiscal stress in California counties was the freeze on the relative percentage of countywide property tax revenue that a local government gets to keep. Distributing current property tax revenue, based upon needs exhibited in the mid-to-late 1970s, is likely to leave some county governments with less property tax revenue than current service needs dictate.

A third possible catalyst for fiscal stress in California counties occurred in the early 1990s. The 1988 passage of California’s Proposition 98 established a minimum floor for state general fund revenue devoted to K-14 public schools. During the recession of the early 1990s, with increased social service demands and less sales and income tax revenue, the state found it difficult to meet the Proposition 98 funding floor. The state’s solution, based in Proposition 13’s allowance that the distribution of statewide property taxation be determined by the state, was the creation of the Educational Revenue Augmentation Fund (ERAF). In both 1991-92 and 1992-93, ERAF shifted property tax revenue away from other forms of local government in a county and toward school districts in the same county. Table 1 shows the average allocation of countywide property tax revenue in California from the fiscal year before Proposition 13 went into effect (1977–78) to 1994-95. As this table demonstrates, between 1993 and 1995 county government took the biggest fiscal hit due to the ERAF shift.

Table 1: Percentage Allocation of Countywide Property Tax Revenue to California’s Local Governments

Based largely upon the three catalysts just described, various observers claim that fiscal stress has occurred in all of California’s counties. In addition to these catalysts, the County Supervisors Association of California (1994) points out that in post-Proposition 13 California, county officials face highly constrained expenditure choices. In 1994-95, about 57 percent of expenditure in the average California county were devoted to categories in which local decisionmakers have no discretion. At the same time, under the constitutional restrictions imposed by Proposition 13, county decisionmakers have little ability to raise the local revenue necessary to meet the mandated expenditure obligations given to them by the state. The County Supervisors Association claims that fiscal stress is the natural byproduct.

One of Bradbury’s measures of fiscal stress is budgetary fiscal stress. Like other local governments in California, by law, in any given year a county cannot plan to have an operating budget that is in deficit. If an operating deficit occurs in one year, it must be corrected in the following year. Thus, a county government that exhibits an annual budget deficit, year-after-year, is likely feeling some degree of fiscal stress. In 1994-95, nearly half of California’s 57 stand-alone counties exhibited a six-year average pattern of annual budget deficits.

Given this background information, an analysis of the fiscal situation of California’s counties in the first half of the 1990s should be of interest to academics and policymakers for at least two reasons. The first is a better understanding of how local government responds to a loss in control of a primary revenue source. If the demand for services provided by the local government remain constant, such a loss is a classic example of Bradbury’s fiscal stress. In such a situation does the local government seek out other forms of discretionary revenue, cut discretionary expenditure, and/or try to run a long-term budget deficit? This question is addressed in this research.

A second reason to examine the fiscal situation of California counties in the early 1990s is that other states have adopted, or are considering the adoption of statewide property tax reforms like Proposition 13. A better understanding of the local fiscal stress that such reform can generate is thus important. In addition, there is continuing discussion in California about how to best undo the ERAF allocation, or even changing the way that property tax revenue in California has been distributed since Proposition 13. A better understanding of the effects that post-Proposition 13 property tax allocation has had on California county finances should offer valuable input into this discussion. For instance, should the property tax percentage taken away from a California county by ERAF be given fully back to it? Or should more be given back to counties that are currently experiencing greater relative fiscal stress and less given back to counties experiencing less relative fiscal stress? This issue is also addressed in this paper.

Our paper continues with a brief review of the previous literature on fiscal stress and empirical studies that have offered causal explanations for it. In Section III we offer a description of our regression model and the data used to estimate it. The results of the regression analysis are contained in Section IV. Section V describes how we use regression analysis to simulate the amounts of per-capita revenue and expenditure that each county in California would raise if it acted like the average California county. Section VI summarizes our results and the policy concerns that emerge from them.

II. PREVIOUS WORK ON MUNICIPAL FISCAL STRESS

There have been at least three methods used to quantify government fiscal stress. The first method (used by the federal government, academic analysts, and private municipal bond analysts) is to measure government fiscal stress with indices comprised of various demographic, economic, and social variables. A second method (employed by academic researchers such as Gramlich (1976), Martin (1982), and Pagano (1993)) is a direct account of the factors responsible for a government’s budget deficit. A third method, utilized by Ladd and Yinger (1989), involves a detailed comparison of the service demands and revenue capacity of local governments.

Some argue that local fiscal stress is an inevitable consequence of the political and economic environment in which a local government operates. Others contend that municipal fiscal stress is more the result of local officials making bad decisions given the environment in which they operate. Amidst these two schools of thought, researchers have identified particular factors that are expected to cause local fiscal stress. As an example, Bradbury (1983) identifies six important factors that affect a local government’s ability to cope with fiscal stress. These are the: (1) value of the local tax base, (2) amount of intergovernmental aid, (3) degree of overlapping government’s tax collections, (4) range of mandated service responsibilities, (5) local production costs, and (6) government’s service needs. Downing (1991) tests some of Bradbury’s propositions by asking officials in U.S. urban counties who self-report a very poor/poor financial condition to choose among a list of given factors as to what got them there. A decisive 97 percent chose increased expenditures for state required programs; 81 percent chose a decrease in federal assistance; 75 percent chose revenue constraints due to statewide tax limitations; while 69 percent chose a decrease in revenue sharing from the state.

There has also been research on the cause of isolated incidents of municipal fiscal stress. Gramlich (1976) found that New York City’s responsibility to provide welfare services, which are more appropriately provided by the state and federal government, led to its near bankruptcy during the mid-1970s. In his analysis of Philadelphia’s loan default in 1990, Inman (1995) found the city’s fiscal crisis to be the result of increased expenditures that were driven by an upward trend in public employee compensation and an increased demand for services coupled with a decline in intergovernmental aid. Previous empirical evaluations of fiscal stress in California counties suggest that it is a result of an insufficient local tax base combined with an increase in state mandated programs.

III. REGRESSION MODEL AND DATA DESCRIPTION

As Bradbury (1982, 1983, and 1984) has established, budgetary health and citizen-fiscal health can be used to appropriately gauge the fiscal health of a county government. We measure the budgetary health of a county in a given fiscal year as an average of the real budget deficits and/or surpluses in the current and past five fiscal years. To place this gauge of budgetary fiscal health in terms relevant to the fiscal stress experienced by the typical county resident it is divided by the county’s average population over the same period. This is designated as Per-Capita Budget Balance.

Poor fiscal health in a county can also manifest itself as county discretionary taxes, charges, and/or fees that are relatively too high for the level of county services provided; and/or county service levels that are relatively too low for the level of county taxes, charges, and/or fees collected. Thus, in addition to a measure of a county’s average budgetary situation, we need measures of the county’s reliance on discretionary own-source revenue and a proxy for services. As in most analyses of this sort, we use per-capita expenditure as a proxy for the level of services provided by a county. We are aware of the problems with doing this and include demographic and geographic factors that could alter the level of service quality delivered in two different counties spending the same amount per capita.

Counties in California can only raise own-source revenue through the use of special benefit assessments, license and permit fees, user charges, and an increment to the statewide rate of sales taxation. A special benefit assessment is a charge imposed on a landowner to pay for a public improvement or service that directly benefits the owner’s land. A license or permit fee is a charge for the right to do something. A user charge is a fee levied for the specific use of a county provided commodity or service. Finally, counties throughout California have the opportunity to propose up to an additional half-percent increment to the statewide rate of sales taxation. This increment must be passed by a majority of voters and can only be used to finance an authority designated to provide a specific service.

We account for a county’s Per-Capita Discretionary Revenue activity in a given year by the real per-resident dollar amount of revenue it raises in special benefit assessments, license and permit fees, user charges, and incremental sales taxation. Since license and permit fees are the fastest growing category of discretionary revenue for California counties, we also examine real dollar Per-Capita Fee Revenue separately. A county’s real dollar expenditure activity in a given year is designated Per-Capita Expenditure.

Bradbury (1982) argues that citizen fiscal stress on the revenue side is suitably gauged as relative per-resident over reliance on local revenue sources given the amount of local per-resident services provided (or per-capita expenditure). Also, citizen fiscal stress on the expenditure side is appropriately gauged as relative per-resident under expenditure, given the amount of local per-resident taxes, fees, and charges. In our analysis of a county’s use of local discretionary revenue, the statistical technique of regression analysis allows us to try and control for the amount of per-resident expenditure in a community by including exogenous explanatory variables that are expected to account for differences in per-resident expenditure across counties. Regression analysis also allows us to try and control for per-resident use of local revenue sources, in the analysis of a per-resident expenditure, by including exogenous explanatory variables in the analysis that are expected to account for differences in per-resident local discretionary revenue across counties.

Next we provide a simple model of the four basic factors that are expected to influence differences in per-capita fiscal activity across county governments. These factors are: (1) discretionary service choices made by citizens in the county, (2) service requirements imposed upon the county by the state, (3) sources of local revenue available to the county, and (4) efficiency in the production of county goods and services. If our three different measures of real per-capita county revenue and expenditure activity are lumped together and simply referred to as fiscal activity, the following general functional relationship accounts for all of this:

Fiscal Activity_i = f (Service Choices_i, Service Requirements_i, Local Revenue Sources_i, Production Efficiency_i), (1)

where,

i = 1, 2, 3, …57 for each of the stand-alone counties in California.

Equation 1 can accommodate the lumping of our six-year average measure of budgetary fiscal stress into the left-side characterization of fiscal activity, if for this case, each of the right-side causal factors is also measured as a six-year average.

In regard to the factors that are expected to influence the discretionary service choices made by a county’s board of supervisors, we offer this general relationship:

Service Choices_i = f (Median Income_i, Percentage Non-Adult_i, Percentage Senior Citizen_i, Poverty Rate_i, Unemployment Rate_i, Resident Political View_i, Employment Per Population_i, Percentage Population Unincorporated_i, Percentage Employment Retail_i, Percentage Employment Agriculture_i, Percentage Employment Manufacturing_i, Population_i). (2)

Service choices are an important and difficult relationship to specify. To appropriately gauge fiscal stress we need to control for county service (expenditure) levels that are low because that is what citizens and business desire. This control comes from the inclusion of variables that are expected to determine service choices. The extensive economic literature on factors that determine local service choices has been summarized by Fisher (Chapter 4, 1996); we use Fisher’s summary as the basis for a parsimonious choice of explanatory variables in equation 2.

Holding all else the same, a county with a higher median income is likely to have residents that desire a greater provision of services from their county. The service choices made by a county are also likely to be influenced by the percentage of county residents who are young, old, poor, or unemployed. The political view of residents, as proxied by the percentage of a county’s voters that are registered as Democrats, should also exert a positive influence on county service choices. Since counties are the first-provider of local government services to county residents in unincorporated areas, the greater the percentage of county residents in these areas, the greater should be the level of per-capita services in a county. Finally, county services are also provided to business. The greater the employment per resident in a county, the greater the business activity in a county and the greater should be the per-resident services provided by a county. The type of business in a county also influences county service choices. A county’s total population, in that it proxies for the urban, suburban, or rural nature of the county, may also influence service choices.

The factors that are expected to influence differences in the services that the state requires that a county provide are given below:

Service Requirements_i = f (Median Income_i, Percentage Non-Adult_i, Percentage Senior Citizen_i, Poverty Rate_i, Unemployment Rate_i, Percentage AFDC_i, Population_i). (3)

To appropriately gauge fiscal stress we also need to control for county service (or expenditure) levels that are high because of state requirements. The state of California requires its counties to provide services in the broad categories of criminal justice, health, and welfare. A theoretical argument can be made, and empirical evidence has been presented, that crime rates are higher in counties with more young adults, more poverty, lower median income, and greater unemployment rates. The eligibility of county residents for health and social support services is also directly related to the percentage of a county’s residents below the poverty line. Counties also administer social support services that primarily benefit children and the elderly. Resident demand for all health and welfare services also tends to move with the unemployment rate. The percentage of a county’s population that receives AFDC is a direct measure of a service that county government is in part responsible for. Finally, just as total population proxies for the urban, suburban, or rural character of a county, this characterization also influences the service requirements imposed upon a county.

The observable factors that are expected to influence the sources of local discretionary revenue (either actual or potential) in a county are:

Local Revenue Sources_i = f (Per-Capita Property Value_i, Percentage Property Tax Revenue_i, Percentage Property Redevelopment_i, Employment Per Population_i,Percentage Population Unincorporated_i, Percentage Employment Retail_i, Median Income_i). (4)

To understand citizen fiscal stress in the relative manner that Bradbury quantifies it, we also need to control for the revenue side of a county’s fiscal situation. We do this by including variables expected to influence a county government’s choice or constrained use of local revenue instruments. After Proposition 13, the property tax revenue collected by a county in California is positively related to the per-capita value of the county’s property tax base and the percentage of the countywide property tax revenue that is allocated to a county based upon the post-Proposition 13 distribution scheme and ERAF. In addition, the property tax revenue collected by a county is negatively related to the percentage of a county’s total assessed property value that is in redevelopment districts. All growth in property tax revenue within a redevelopment district, or the tax increment, is apportioned back to the redevelopment agency to finance the interest and principal on the debt issued to finance redevelopment. The county loses its portion of the tax increment collected within the redevelopment area for the duration of the project.

The sales tax component of a county’s potential discretionary revenue sources is measured by the relative size of a county’s retail sales industry. If a county has a greater percentage of its employment base in retail activity (holding employment per resident constant), it also has a greater amount of retail sales and sales tax base. The same can be said for a county with a greater percentage of the population in unincorporated areas. The higher this percentage, the greater the sales tax revenue retained by the county because in unincorporated areas there are no cities to split the revenue with. A county where residents have greater real disposable income should also generate more retail sales and hence more sales tax revenue.

Differences in the efficiency of producing local government services across California counties are also expected to exert an influence on the degree of fiscal activity observed in a county. In a simple fashion, relative production efficiency in California counties can be represented as:

Production Efficiency_i = f (Independent Charter_i, Population_i). (5)

Two measurable reasons why efficiency in the production of government services may vary across California counties are the organizational structure in which the county operates and the total population that it serves. The power to define a county’s organizational structure is constitutionally delegated to the state legislature by the California Constitution. However, counties have the authority to adopt charters through which they can create organizational structures entirely different from those prescribed by the state. Therefore, counties that have adopted a charter may be able create organizational structures that result in the more efficient or inefficient delivery of county services relative to counties subject to state general law. To control for any possible economies or diseconomies of scale that could result in production efficiency or inefficiency in a county, we also account for the total residential population in the county.

Equation 6 is the reduced functional form that represents all of the explanatory variables expected to influence the level of fiscal activity observed in California’s stand-alone counties in a given year:

Fiscal Activity_i = f (Median Income_i, Percentage Non-Adult_i, Percentage Senior Citizen_i, Poverty Rate_i, Unemployment Rate_i, Resident Political View_i, Employment Per Population_i, Percentage Population Unincorporated_i, Percentage Employment Retail_i, Percentage Employment Agriculture_i, Percentage Employment Manufacturing_i, Percentage AFDC_i, Per-Capita Property Value_i, Percentage Property Tax Revenue_i, Percentage Property Redevelopment_i, Independent Charter_i, Population_i, County Dummies), (6)

where,

Fiscal Activity_i = Per-Capita Budget Balance_i, Per-Capita Fee Revenue_i, Per-Capita Discretionary Revenue_i, Per-Capita Expenditure_i.

This reduced form regression specification should handle the functional performance non-comparability that occurs when comparing local government finances for relative fiscal health. The fact that all observations are taken from California counties reporting financial information in a standardized fashion to the state controller also handles fiscal management non-comparability. These fiscal-health comparability issues arise because local governments can be responsible for more or less services than others, and local governments can report financial statistics in different ways.

County dummies are a set of 56 variables (one for each stand-alone county with Los Angeles County excluded) that take on a value of one for a specific county and a zero otherwise. They are used in the regression analysis to control for causal factors that are constant in a county over time, but are not easily quantified and differ between counties.

The four casual relationships described in equation 6 are estimated using a pooled data set that combines county observations from one year with other years (specifically from the five fiscal years between 1990-91 and 1994-95). It is possible that the process generating the chosen measures of county fiscal activity is not the same in one year as it is in other years. Differences may be due to factors that are not measured by the set of right-side variables included in equation 6. If these excluded factors are effects that are constant across all counties and vary from year to year, then they can be picked up by the inclusion of a set of time dummies as explanatory variables.

Table 2 provides a summary of how each variable in equation 6 is measured and its source. This table also displays the mean, standard deviation, maximum value, and minimum value for all variables. The data contains 285 observations drawn from 57 counties over five fiscal years. With the exception of per-capita budget balance, the data in Table 2 is based on single year observations. All dollar values are in constant 1990 dollars based upon the California consumer-price-index deflator.

Table 2: Variable Names and Descriptions

IV. REGRESSION RESULTS

Before reviewing the regression results, we discuss how a few estimation issues were handled. The first issue is the functional relationship between the right-side explanatory variables and the left-side dependent variable. A linear specification restricts the regression to a fixed calculation of how a one-unit change in an explanatory variable affects the dependent variable across all values of the explanatory variable. A semi-log specification allows for an independent variable to exert a different per-unit influence at low values than at high values. There is not a theory that necessarily indicates which is the better specification for each of the four regressions. Were possible, we used the Box-Cox regression method in LIMDEP to decide the most appropriate functional form. For both per-capita discretionary revenue and per-capita discretionary expenditure the Box-Cox procedure converged at a pure semi-log specification. For per-capita fee revenue, the procedure converged at a Box-Cox l value of 0.05, or very close to a semi-log specification. For the per-capita budget balance regression we were forced to use a basic linear regression specification because the dependent variable exhibits negative values.

An additional regression issue is the possible presence of autocorrelation due to the pooling of cross-sectional values over time. Unfortunately, LIMDEP does not allow for a pooled Box-Cox regression that corrects for autocorrelation. As an alternative, we ran the per-capita budget balance regression in pure linear form and the other three regressions in semi-log form, and corrected for autocorrelation. We then tested for the significance of the estimated correlations. In no cases were the correlations statistically significant and this was taken as enough evidence to not correct for autocorrelation.

Another regression issue is the likely presence of heteroskedasticity. A statistical method has been developed by White to correct for this bias and we use it in all but the per-capita fee regression were it could not be used due to the Box-Cox technique. As an alternative we retrieved the residuals from the per-capita fee regression and checked for a relationship between them and county population. Finding a statistically significant relationship, the weighted-correction for heteroskedasticity available in LIMDEP was performed.

A final regression issue is the likely presence of multicollinearity in the original regression specification. In a first set of regression runs using equation 6, the explanatory variables poverty rate and the percentage AFDC never exhibited a statistically significant influence in any of the four regressions. The partial correlation coefficient between these two variables is 0.86. Thus we decided to drop poverty rate from our final regression specification. Poverty rate also exhibits a partial correlation coefficient of 0.72 with unemployment rate.

The regression results for each of the four dependent variables are recorded in Table 3. Regression coefficients derived for the dummy explanatory variables are listed at the top of this table. The appropriate F tests indicate that the group of fiscal year dummies as a whole, and the group of county dummies, exerted statistically significant influences in all regressions. The entry for each non-dummy explanatory variable in Table 3 first contains the elasticity if it was statistically significant. It then contains the regression coefficient, and below it in parenthesis, the coefficient’s standard error.

Since fiscal year 1990-91 is the dummy variable that was excluded from the regression analysis, the FY 1991-92 coefficient of 0.21 in the per-capita budget balance regression indicates that the six-year average budget deficit/surplus in the typical California county was $0.21 higher in fiscal year 1991-92 than in 1990-91. An easy way to interpret such a coefficient is to imagine a typical county in California that was identical in fiscal years 1990-91 and 1991-92 in terms of the explanatory variables used in the regressions. Just because a year passed, and nothing else changed, the six-year average budget surplus/deficit in this typical county increased by $0.21 (from an average value of $6.68). The remaining fiscal-year coefficients calculated for the budget balance regression indicate that the typical county’s budget surplus again rose in 1992-93, but fell in 1993-94, and even fell further below the 1990-91 level in 1994-95. California’s recession, that begin in mid-1990 and took some time to be reflected in the six-year average budget balances, is the obvious reason for this temporal pattern. Holding other explanatory variables constant, the per-capita amount of fee revenue collected in the typical California county continually rose from fiscal year 1990-91 to 1991-92, but fell from this level continuously until 1994-95. While the pure time trend in the typical county’s per-capita discretionary revenue fell continuously from 1990-91. Per-capita total expenditure, holding other factors constant, followed no discernable pattern over time.

Table 3: Determinants of Fiscal Activity in California Counties
(County Dummy Variables Included, Results Not Reported)

Statistical Significance in two-tailed test: *** = greater than 99%, ** = 90 to 99%, * = 85 to less than 90%. Per-Capita Budget Balance, Log of Per-Capita Fee Revenu, and Log of Per-Capita Expenditure corrected for heteroskedasticity using White's consistent covariance matrix estimator. Per-Capita Fee regression results corrected for heteroskedasticity using weighted least squares with county population as weight. Like the dependent variable, the non-dummy explanatory variables in the per-capita budget balance regression are six-year average values.

Our regression analysis also indicates that if two counties in California were identical based upon the explanatory variables included in the regression analysis, with the exception that one adopted an independent charter, the charted county would be spending 48 percent more per capita than the non-charted county. The chartered county would also be raising about 148 percent more in per-capita fees. These differences could be due to charter counties, ceteris paribus, being less efficient than non-chartered counties, or it could be due to residents in charter counties systematically exhibiting a greater taste for services.

Elasticity measures the percentage change in a dependent variable from its mean when an explanatory variable rises by one percent from it mean. For example, when the per-capita property value in the typical county rises by 10 percent (or $5.77 from a mean of $57.73), per-capita fee revenue falls by 3.9 percent (or $0.64 from a mean of $16.53). Or when the percentage employed in the retail sector rises by 10 percent (or 2.7 percent from a mean of 26.78 percent), per-capita fee revenue per person falls by 5.1 percent (or $0.84 from a mean of $16.53). Here are two clear indicators that the greater the local tax base in a county, the less the county relies on fee revenue. Since these regressions implicitly control for local expenditure (through the inclusion of explanatory variables that proxy for service choices and service requirements), a California county with greater per-capita property value and retail activity imposes less citizen fiscal stress on their residents in the form of fee payments required per resident.

As the percentage of a county’s population that is in unincorporated areas rises by 10 percent (or about 4.7 percent from a mean of 46.68 percent), per-capita total expenditure rise by 5.1 percent and per-capita fee revenue rise by 12.6 percent. These percentage increases result in an increase in per-capita expenditure of $54.67 from a mean of $1,072, and an increase in fee revenue per capita of $2.08 from a mean of $16.53. Such results are not surprising given that a county is the first provider of local government services to residents in its unincorporated areas. Greater local service provision necessarily requires greater local revenue. In California’s fiscal environment, increased fees are one of the few ways of getting this greater revenue. The regression results also indicate that a ten percent increase in the percentage population in unincorporated areas results in 21.6 percent decrease in the county’s long-term average budget surplus. This finding may point to a causal relationship between percentage population in unincorporated areas and county fiscal stress, that is the result of the constrained fiscal environment in California after Proposition 13.

A county’s real median income exerts a measurable influence on its fiscal activity. A ten percent rise in real median household income results in a whopping 1,130 percent increase in the average county’s per-capita budget balance and a slight 0.4 percent decrease in per-capita expenditure. On the citizen-choice side it is well established that an increase in median income should have a positive effect on per-capita county expenditure. The observed decrease in per-capita expenditure, due to an increase in median income, is likely due to higher income counties having to provide less state-required services. Criminal justice, health, and welfare services provided by the county are likely to be less on a per-capita basis in richer counties. Thus we can be fairly certain that a reduction in total expenditure per capita, following an increase in a county’s median income (holding other explanatory factors constant) is due to the negative service requirement effect being greater than the positive service choice effect.

Most pertinent to the important general question of whether local government seeks out other forms of discretionary revenue, cuts discretionary expenditure, or tries to run a long-term budget deficit after an exogenous fiscal shock, are the regression results related to the percentage countywide property tax revenue received by the county. Recall that this variable inversely measures the fiscal shock felt by California counties after the California Legislature imposed the post-Proposition 13 method of distributing property tax revenue and by the changes imposed by the state legislature in the early 1990s under ERAF. We find that a 10 percent increase in the percentage of countywide property tax revenue received by the average county, holding other factors constant, results in a 2.2 percent decrease in the use of fees per resident and a 2.4 percent decrease in all forms of local discretionary revenue per resident.

Based upon these results, there is little doubt that a California county is under greater fiscal stress if it receives a lower percentage of the total property tax revenue collected within its boundaries. Such counties ask their citizens to endure greater citizen fiscal stress in the form of larger amounts of local discretionary revenue payments per resident. Though it must also be made clear that percentage property tax revenue does not exert a significant non-zero influence on per-capita budget balance or per-capita expenditure. We can offer no proof that counties receiving a lower share of countywide property tax revenue, holding all else the same, are more likely to run long-term budget deficits or exhibit a lower level of per-capita expenditure.

Recall from the previous chapter that the percentage of a county’s residents that are young, old, receiving AFDC, or unemployed are expected to influence a county’s service choices and its service requirements. Thus it is difficult to decipher any of the elasticities derived for these explanatory variables. For instance, we find that a 10 percent increase in non-adults results in an 8.2 percent decrease in per-capita expenditure. We cannot be certain whether this is caused by less state-mandated services being provided to children or a decrease in demand for discretionary services by residents in counties with more children.

Though not an indicator of fiscal stress, it was informative to confirm that taste for county government services does vary by political affiliation. A 10 percent increase in the percentage of county voters registered as Democrats (from a mean of 47.5 percent to 52.25 percent) results in total expenditure per resident rising by 3.3 percent. To pay for the increased desire for county services, per-capita fee revenue rises by 16.1 percent (from a mean $16.53 to $19.19) and total per-capita discretionary revenue rises by 16.2 percent (from a mean $98.59 to $114.56). An increase in registered Democrats in a county does not impose greater fiscal stress in a county. Democrats appear to desire more services and pay for them in California’s constrained local fiscal environment with more fees.

Since the greater the employment per resident in a county the greater the business presence, it is not surprising that we find that a 10 percent increase in this variable results in a 1.1 percent increase in per capita expenditure. A greater non-residential presence requires additional service provision from the county. In addition we find the greater the manufacturing presence, as measured by percentage employed in manufacturing, the even greater the per-capita expenditure in a county. Finally as population increases in the typical California county, and other causes of county fiscal activity are held constant, per-capita fee decreases. This could be due to the efficiency of service delivery being greater in more populated counties (i.e., economies of scale).

V. REVENUE AND EXPENDITURE ACTIVITY RELATIVE TO CAPACITY

Since we have chosen to use Bradbury’s definition of fiscal stress, we need a method by which to quantify a relative over utilization of discretionary revenue and/or relative under provision of local services. In this section we offer a method by which to do this. California’s policymakers could use the results of this method to assist them in the equitable formulation of a new allocation scheme for property tax revenue collected in the state.

We begin by defining a county’s fiscal capacity as what it would be doing in regard to local fiscal activity (revenue or expenditure) if it acted like the typical California county that exhibited the same characteristics as included in a regression. A county’s fiscal effort corresponds to what a county is actually doing in regard to local fiscal activity. A relative measure of the degree that a county is utilizing its fiscal capacity therefore equals what a county is actually doing (its fiscal effort), divided by what it would be doing if it acted liked the typical California county with similar characteristics (its fiscal capacity). Specifically, we calculate this ratio (fiscal effort/fiscal capacity) for all 57 stand-alone counties, for all five fiscal years, for our two broad measures of county fiscal activity (per-capita discretionary revenue and per-capita expenditure).

The predicted value of a county’s per capita discretionary revenue and predicted value of a county’s per-capita expenditure is derived from regression analyses. The regression analyses used to predict fiscal capacity are conducted in the same manner as reported in Table 3, with the exception that county-specific dummies are excluded as explanatory variables. It was appropriate to include county dummies when determining the distinct influence of each of the explanatory variables thought to determine differences in fiscal activity across counties. But, if we desire to predict the amount of fiscal activity in a county if it acted like the average California county with the same determinant characteristics, it is not appropriate to control for county-specific influences that are constant over time. The control of such influences would necessarily pick up county-specific practices that are wasteful and inefficient. The only explanatory variables that should be accounted for are the ones that fall into the four causal categories first given in equation 1. We can debate about what specific factors to proxy for these four categories, but we should agree that county-specific dummies are not appropriate in a regression designed to predict a county’s fiscal capacity.

As an illustration of the just-described method, the per-capita discretionary revenue ratio is equal to the value of a county’s actual per-resident use of discretionary revenue, divided by its predicted use of per-resident discretionary revenue for a specific fiscal year. The predicted value of per-capita discretionary revenue is derived for a specific county in a specific fiscal year by plugging in values that the explanatory variables exhibit for that county in a given year. A ratio of 1.25 indicates that the county is using 25 percent more discretionary revenue per capita than the typical California county exhibiting the same characteristics. A ratio value of 0.75 indicates that the county is using 25 percent less discretionary revenue per capita than the typical California county in the same situation. For the ratio that measures actual revenue to predicted revenue, a larger number indicates a greater degree of relative citizen fiscal stress. For the ratio that measures actual expenditure to predicted expenditure, a larger number indicates a smaller degree of citizen fiscal stress. These ratios only represent a county’s fiscal stress relative to all other counties in California. They are not intended to be absolute measures of county fiscal stress.

To summarize the relative fiscal health of all of California’s 57 stand-alone counties in a single table, we calculate the average value of the five ratios calculated for discretionary revenue and total expenditure over the period 1990-91 to 1994-95. We also calculate the average long-term (six-year average) budget deficit/surplus recorded by a county over these same five fiscal years. These values are in Table 4.

Table 4: Ratios of Fiscal Activity to Predicted Values and Budgetary Health for California Counties in First Half of 1990s