Fiscal Effects of GRS

correspondence address: David A. Latzko
     Dept. of Business & Economics
     Wilkes University
     Wilkes-Barre, PA 18766
     (717)-408-4718
     e-mail: dlatzko@wilkes.edu

* We are indebted to Bryan G. Harvey for research assistance. Louise Marshall provided helpful comments. The City of Wilkes-Barre budget data can be downloaded at http://wilkes1.wilkes.edu/~dlatzko/wbbudget.html

Municipal budgeting in Wilkes-Barre, Pennsylvania, before, during, and after general revenue sharing is examined. Only the size of the general fund balance was associated with the amount of revenue sharing dollars the city received. If taxpayers are loss averse and GRS was viewed as a nonrecurring revenue source, budgetary policy would be influenced by the political consequences of having to eliminate programs or raise taxes in the future. Grants would be accumulated for future use as a substitute to raising taxes or cutting programs. We find that the grants were partially "hidden" in the budget by underestimation of non-tax revenues.

State and local governments received general revenue sharing grants from the federal government from 1972 to 1987. These funds ought to have had the same impact on local government expenditures as an equivalent increase in the private incomes of city residents, with most of the grant going towards tax reductions (Bradford and Oates, 1971). However, empirical studies have established that the response of local government expenditures to revenue sharing was far greater than would be implied by the response of local spending to changes in income (Gramlich, 1977). Explanations for this flypaper effect have ranged from voter confusion between marginal and average tax prices (Courant, Gramlich, and Rubinfeld, 1979; Oates, 1979) to budgetary agenda manipulation (Romer and Rosenthal, 1980). Hines and Thaler (1995) recently proposed that loss aversion on the part of taxpayers might help explain the flypaper effect of revenue sharing grants. The purpose of this paper is to take advantage of the termination of federal revenue sharing a decade ago to examine the fiscal behavior of a third class city government in Pennsylvania before, during, and after federal revenue sharing in order to determine the effect of general revenue sharing on this municipality's budgetary policy. We also examine whether taxpayer loss aversion can provide a plausible explanation for the budgetary policy of city officials that is consistent with the city's experience.

This study examines the effects of revenue sharing on the budgetary behavior of the City of Wilkes-Barre, Pennsylvania over the years 1965 to 1994. Wilkes-Barre was incorporated in 1871. For much of its history, the city's economy was dominated by the anthracite coal industry. By the 1960's, the coal industry had declined and given way to a diversity of small commercial and industrial businesses. The city's population has been steadily declining since 1930 and is now around 45,000.

Wilkes-Barre is a third class, home rule city. The city council has always possessed the legislative responsibility to approve all budgets and tax levels. Expenditure and revenue data for the years 1965 to 1994 were taken from the city's adopted annual budgets and the year-end certified audits. The city budget for 1994 was about $23 million; the largest spending program is for public works, accounting for 25 percent of total city expenditures. The city's main sources of revenue are the earned income tax and property taxes, 37 and 20 percent of city revenues respectively. Between the years 1973 and 1986, the city budgeted an annual average of $1.2 million in federal general revenue sharing grants, an average of 12 percent of the city's budgeted revenues. The city also received general revenue sharing dollars in 1972 and 1987. Graph 1 shows the values of the level of budgeted city expenditures, tax revenues, and revenue sharing grants along with the size of the city's unappropriated general fund balance account on December 31 all expressed in real per capita terms.

We assume that budgetary decisions on city expenditures and tax revenues per capita are a function of variables reflecting the community's income and wealth (median income and total assessed property value, block grants received, and general revenue sharing grants per capita), the price of public expenditures (the property tax millage rate and the earned income tax rate), and the city's social and political characteristics (population, percentage of the population under 18 years of age, and dummy variables for form of government: weak mayor, city manager, or strong mayor). The values of these variables were taken from the city's annual budgets and various editions of the U.S. Census Bureau's County and City Data Book.

In order to correct for the possible endogeneity of the millage and earned income tax rates, two stage least squares was used. Table 1 presents regression results from the second stage. For the regression with total budgeted expenditures per capita as the dependent variable, the independent variables explain 96 percent of the variation in per capita budgeted expenditures. However, while the coefficient of the revenue sharing variable is positive, it is not significantly different from zero. The p-value for its t-statistic is 0.33. Revenue sharing did not have a meaningful impact on city spending. Revenue sharing had the expected negative effect on total budgeted tax revenues per capita, but this coefficient is not significantly different from zero either. The p-value of the t-statistic is 0.61. Revenue sharing did not have a statistically significant effect on either budgeted expenditures or tax revenues for the City of Wilkes-Barre. So what happened to the money?

The unappropriated general fund balance account is the city's contingency or rainy day fund. In addition to providing a source of revenue for current expenditures, the general fund balance account absorbs any unbudgeted surpluses and deficits. If the revenue sharing dollars were not converted into significant additional spending or tax reductions, the only place they could have ended up was in the city's unappropriated general fund balance account. This is confirmed by the estimates in the last column of Table 1 for which the dependent variable is the per capita level of the unappropriated general fund balance account. The equation explains nearly half the variation in the general fund balance, and the coefficient of the GRS variable is positive and significantly different from zero. The p-value of its t-statistic is 0.03.

The coefficients of the revenue sharing variable are estimates of the average elasticities over the study period. They tells us, for example, that a 1 percent increase in revenue sharing grants per capita caused the unappropriated general fund balance to rise by 0.76 percent. Based on these estimates, on a per capita basis, one dollar of budgeted revenue sharing grants resulted in about an additional 10 cents in budgeted city spending and 10 cents of budgeted tax reduction while the other 80 cents ended up in the city's fund balance account.

Revenue sharing grants did not result in significant additional spending or tax reductions. Rather, the flypaper effect manifested itself literally in Wilkes-Barre, Pennsylvania. The money stuck exactly where it hit: city hall. We argue that consistent with loss aversion on the part of taxpayers and the temporary nature of the program, electoral concerns led most of the GRS funds to be accumulated in the city's general fund balance account.
Taxpayers are loss averse when they are "much more sensitive to decreases in their welfare than to increases. This implies that the political cost of explicitly raising a tax is greater than the political benefit of an equivalent tax cut" (Hines and Thaler, 1995, p. 223). Because reductions in government service levels also decrease voter welfare, the political costs of reducing local government spending are greater than the political benefits of an equivalent increase as well. Local officials will be reluctant to raise taxes or cut expenditures because of the disproportionate political costs.

The Congressional legislation authorizing general revenue sharing was passed in 1972. The program was explicitly temporary in nature as it provided for state and local governments to receive financial grants only through 1976. General revenue sharing was renewed by Congressional action for three more years and continued to be renewed until 1986. The grants were disbursed with limited restrictions. Municipalities were required to report what they spent the funds on, but there were no formal applications for GRS funds. Recipients received entitlement checks regularly. The size of the grant was based on population, tax effort, and personal income. The formula utilized to calculate the level of revenue sharing funds a municipality was to receive favored those jurisdictions with high tax efforts.

We hypothesize that the temporary nature of revenue sharing along with loss aversion on the part of voters became important considerations to local elected officials in the development of budgetary policy. Significant income increases resulting from sources such as general revenue sharing ought to allow for significant reductions in tax levels or significant increases in spending. However, the reality of local politics is that the overriding objective of politicians is to get re-elected. With elections every four years, this is a long term objective. Local government officials would then view general revenue sharing as a short term and nonrecurring revenue source; city officials would be reluctant to integrate GRS funds into the operating budget because of the uncertainty generated by the temporary life of the program. If city officials increased expenditures or reduced taxes, they would be required to raise taxes or eliminate programs if revenue sharing was not renewed or extended. Either outcome will be viewed as undesirable by loss averse voters who desire that service levels are maintained and taxes are not raised and who, consequently, will punish the politicians at the ballot box. General revenue sharing funds would instead be placed in the unappropriated general fund balance account to serve as deferred taxation, keeping local tax and spending levels at the status quo and building up a general fund balance surplus for future use as a substitute to raising taxes or reducing service levels.

The previous analysis implies that general revenue sharing grants, if seen as a temporary revenue source as we suggest, should have had no significant effect on Wilkes-Barre's spending or tax levels. Revenue sharing dollars ought to have gone, instead, into the unappropriated general fund balance account. That is exactly what the evidence suggests happened.

The advent of revenue sharing posed a dilemma for city officials. The temporary nature of the program and electoral concerns seem to have dictated that revenue sharing grants result in neither more expenditures nor lower taxes, but the balanced budgeting practiced by municipalities forces elected officials to match anticipated revenues with anticipated expenditures. A solution is to deliberately underbudget city revenues so that realized revenues turn out to be greater than budgeted revenues.

We speculate that the introduction of general revenue sharing caused municipalities to underestimate non-tax revenue income items in order to substitute in the budget the federal revenue sharing dollars equal to the underestimated value. For example, if the city budgeted $500,000 in revenue sharing grants, non-tax revenue items would be deliberately underestimated by $500,000 in the budget. This would permit city officials to present a balanced budget and to avoid having to reduce taxes to balance the budget, thereby creating a revenue surplus when the actual collection of non-tax revenue items turned out to be above the underestimated level. This surplus would go into the city's unappropriated general fund balance account as a hedge against future politically damaging tax increases and spending cuts. This would account for the positive relationship between revenue sharing grants and the size of Wilkes-Barre's general fund balance account. Underestimating non-tax revenue items would be preferred to underestimating tax revenue items due to the structure of the GRS formula which rewarded recipients for a higher tax burden (high property tax millage and earned income tax rates).

Prior to revenue sharing, city officials had no incentive to inaccurately forecast revenues. Forecasts of non-tax revenue sources ought to be more accurate than those from the revenue sharing period. The termination of revenue sharing should bring a return to more accurate forecasts of non-tax revenue line items and a greater utilization of fund balances as a substitute for tax increases or program cuts.

The pattern of underestimation of non-tax revenue items during the years of general revenue sharing can be verified by the review of a series of consecutive prior years' audits to compare budgeted revenues with their actual levels. We calculated the percentage difference between budgeted revenue items and realized revenue items, which we call forecast errors, for the years 1969-1994. Forecast errors of non-tax revenue items averaged -24.75 percent (meaning an underestimation) during the years in which the city received general revenue sharing grants while the forecast errors in years in which revenue sharing dollars were not received averaged -7.2 percent. Underestimation of non-tax revenue items was significantly larger during the period of general revenue sharing than in the years before and after.

However, there is not a dollar for dollar correspondence between budgeted revenue sharing grants and the underestimation of non-tax revenues. Wilkes-Barre budgeted an annual average of $1,613,820 of revenue sharing grants but city officials underestimated non-tax revenue items during the GRS period by an annual average of just $674,074 in 1983-84 dollars. The correlation between budgeted revenue sharing grants and the size of the non-tax items forecast errors was -0.55 during the GRS period but just -0.21 over the entire study period. Our results indicate that an additional $1 of revenue sharing resulted in a $0.13 increase in the underestimation of non-tax items.

The forecast errors for tax revenue items (property taxes, earned income taxes, and the so-called nuisance taxes) averaged -1.88 percent during revenue sharing and -2.52 percent during non-revenue sharing years. As expected, this is not a significant difference.

Local municipalities received federal general revenue sharing grants from 1972 through 1987. This paper took advantage of the termination of federal revenue sharing to examine the budgetary behavior of a third class city government, Wilkes-Barre, Pennsylvania, before, during, and after revenue sharing and to investigate whether taxpayer loss aversion had an influence on municipal budgeting. Both the level of budgeted city expenditures and the level of budgeted taxation were insignificantly affected by revenue sharing grants. The size of the city's general fund balance was statistically associated with the amount of revenue sharing dollars the city received. An explanation consistent with the city's experience is that budgetary policy was influenced by the adverse political consequences of possibly having to eliminate popular programs or raise taxes in the future. So, local officials treated general revenue sharing, reflecting its temporary nature, as a short term and nonrecurring revenue source. As a result, general revenue sharing dollars were used to build up the fund balance for future use as a substitute to raising taxes or cutting programs. We find that the revenue sharing money was partially "hidden" in the budget through the underestimation of non-tax revenue items. This left the city's budget in balance and negated the need to increase spending levels or lower taxes.

Table 1
Two Stage Least Squares Estimates
-----------------------------------------------------------------------------

Independent                      Expenditures      Tax Revenues      Fund Balance
Variables
----------------                -------------     ----------------    ----------------
revenue sharing                      0.008               -0.0127                0.7594**
                                           ; (0.7445)            (0.5065)              (2.3942)

Adjusted R-squared                0.96                 0.79                      0.45
standard error                         0.05                 0.11                      0.76
sample size                               30                    30                         24
------------------------------------------------------------------------------

Notes: Absolute values of the t-statistics in parentheses. ** denotes significant at the 5 percent level.
* denotes significant at 10 percent level.

Bradford, David F. and Wallace E. Oates, 1971. "The Analysis of Revenue Sharing in a New Approach to Collective Fiscal Decisions." Quarterly Journal of Economics, vol. 85 (August), 416-439.

Courant, Paul N., Edward M. Gramlich, and Daniel L. Rubinfeld, 1979. "The Simulative Effects of Intergovernmental Grants: Or Why Money Sticks Where It Hits." In Peter Mieszkowski and William H. Oakland, eds. Fiscal Federalism and Grants-In-Aid. Washington, DC: Urban Institute, pp. 5-21.

Dommel, Paul R. and Keith P. Rasey, 1989. "Coping with the Loss of General Revenue Sharing in Ohio." Public Budgeting and Finance, vol. 9 (Autumn), 43-51.

Gramlich, Edward M., 1977. "Intergovernmental Grants: A Review of the Empirical Literature." In Wallace E. Oates, ed., The Political Economy of Fiscal Federalism. Lexington, MA: Lexington Books, pp. 219-239.

Hines, James R., Jr. and Richard T. Thaler, 1995. "Anomalies: The Flypaper Effect." Journal of Economic Perspectives, vol. 9 (Fall), 217-226.

Marando, Vincent L., 1990. "General Revenue Sharing: Termination and City Response." State and Local Government Review, vol. 22 (Fall), 98-107.

McKenna, Frank, Jr., Philip A. Russo, Jr., H. Kenneth Hilbbeln, and Douglas H. Shumavon, 1990. "National Fiscal Policy Changes and the Impact on Rural Governments: CDBG Cuts and the Loss of GRS." Public Administration Quarterly, vol. 14 (Fall), 324-352.

Oates, Wallace E., 1979. "Lump-Sum Intergovernmental Grants Have Price Effects." In Peter Mieszkowski and William H. Oakland, eds. Fiscal Federalism and Grants-In-Aid. Washington, DC: Urban Institute, pp. 23-30.

Romer, Thomas and Howard Rosenthal, 1980. "An Institutional Theory of the Effect of Intergovernmental Grants." National Tax Journal, vol. 33 (December), 451-458.

Stine, William F., 1994. "Is Local Government Response to Federal Aid Symmetrical? Evidence from Pennsylvania County Governments in an Era of Retrenchment." National Tax Journal, vol. 47 (December), 799-816.

David A. Latzko
Business and Economics Division
Pennsylvania State University, York Campus
office: 13 Main Classroom Building
phone: (717) 771-4115
fax: (717) 771-4062
DXL31@psu.edu
www.yk.psu.edu/~dxl31

THE FISCAL EFFECS OF GENERAL REVENUE SHARING: THE EXPERIENCE OF A PENNSYLVANIA THIRD CLASS CITY*