Do Local Governments Save and Spend Across Budget Cycles? Evidence From North Carolina

Abstract

The fiscal crisis encountered by state-local governments since 2008 has again made prominent the issue of how to better prepare for and stabilize expenditures during recessions. Does the stabilization function of government, and its theory, still hold? Previous studies focus on federal and state levels; only a few look at local governments. This article explores whether localities save and spend across the boom-bust cycle; we intend to identify the determinants of local government savings and estimate the impact of savings on stabilizing expenditures. Unlike some early evidence that shows countercyclical stabilization properties of local unreserved general fund balance, the empirical results of our study on North Carolina counties do not support the stabilization role by localities. This study carries timely and important implications for state/local policy making and financial operations; it also adds to the literature on the stabilization function of government.

Keywords

local financial management fund balance policy stabilization function of government

Introduction

Since the pioneering work of John Maynard Keynes in 1936 and especially the publication of Richard Musgrave’s Theory of Public Finance in 1959, the economic stabilization function of government has been gradually and widely accepted; but countercyclical spending was assumed to be the sole responsibility of the national government in the standard Musgrave theory (Musgrave, 1959; Oates, 1972). Later, Edward Gramlich (1987) advocated pursuing countercyclical policies for stabilizing expenditures at the subnational level (Hou & Moynihan, 2008). In the past two decades, public finance scholars have produced a rich literature that addresses key questions concerning the adoption, use, and consequences of budget stabilization funds, reserve funds, and other subnational fiscal stabilization tools, with almost an exclusive focus on the state level (Marlowe, 2005). Empirical research offers support for the positive impact of countercyclical spending on fiscal stress at the state level (Douglas & Gaddie, 2002; Hou, 2003a, 2005; Sobel & Holcombe, 1996; Wagner & Elder, 2007). Even though the political economy of the past several decades has implicitly ascribed both state and local governments an important role in the stabilization function (Marlowe, 2005), so far we know very little about the practice of countercyclical spending at the local level. Only a few studies in the literature looked at the role of local slack resources, either in the form of fund balances or stabilization funds, in local government finance and their impact on decisions about spending and revenues (Gianakis & Snow, 2007; Hendrick, 2006; Hou, 2008; Marlowe, 2005).The recent fiscal crisis has again brought the attention of academics and practitioners to the issue of how to better prepare for, and stabilize expenditures in, an economic downturn. Therefore, it becomes even more urgent than before to examine whether local governments possess the policy inclination, fiscal capacity, and policy tools to act countercyclically, because federal and state governments often cut their grants to localities in an economic downturn (or cannot provide as much as localities need, as in the most recent stimulus package).

The state of North Carolina offers a good case for us to examine the validity of the stabilization theory of local government. As a consequence of its rampant local government default on their municipal bonds during the Great Depression,¹ North Carolina developed a state oversight system that is considered a model by bond-rating agencies. The state exerts strong control over local fiscal practices; it is the only state that assumes the legal responsibility for issuing all local debts (Coe, 2007; Shelton & Tyer, 2000). Because of the financial discipline of its local governments and effective state oversight and technical assistance, North Carolina leads the nation in the number of local governments with the highest bond rating (Coe, 2007). Though 90% of states have adopted budget stabilization funds as a formal countercyclical fiscal policy instrument (National Association of State Budget Officers, 2007), budget stabilization funds remain rare among local governments; they rely mostly on year-end balances of the general fund.² This study attempts to examine the practices of such an exemplary system in managing local government fund balances and thereby adds to the small literature on countercyclical fiscal and budgetary policies at the local level. The empirical results of this study may carry timely and important implications for policy making and financial operation in the state and local sector.

To operationalize this study, we choose to focus on two research questions as follows: First, what factors affect the level of general fund balances in county governments (in North Carolina)?³ Second, what is the impact of general fund balances on local expenditures⁴ in general and their stabilizing effects during economic downturns in particular? The two research questions cover two sides of the same issue. No matter the balances of a locality’s general fund are large or small in level and ratio, presumably there is an implicit policy behind the balances. By identifying the determinants of general fund balances, we can estimate the separate effects of revenue sources, local government programs, and socioeconomic factors on the fund balances; then we estimate the effects of fund balances on local expenditures. Thus, we evaluate whether local governments exercise the stabilization role through fiscal policy.

This article is divided into seven sections. Following the introduction is a brief description of the budget stabilization theory and local fund balance policy. The third section introduces North Carolina counties’ finance structure and their fund balance policy. The fourth section reviews current literature on countercyclical expenditure stabilization at the local level, followed by a presentation of model specification, data, and estimation methodology. The sixth section presents empirical results with our interpretation, and the final section concludes with a discussion.

Budget Stabilization and Local Fund Balance Policy

The study of economic stabilization function of government emerged with the groundbreaking work of John Maynard Keynes in 1936. Scholars have since studied how central governments could adopt countercyclical monetary and fiscal policies to best deal with financial crisis to achieve economic stability over the business cycle (Hou, 2003b). The economic stabilization function of government has been gradually and widely accepted with the publication of Richard Musgrave’s Theory of Public Finance in 1959. Countercyclical fiscal policy has, however, been assumed to be the sole responsibility of the central government in the standard (or conventional) Musgrave theory (Musgrave, 1959; Oates, 1972). Edward Gramlich (1987) expanded the idea of practicing countercyclical fiscal policies to subnational governments (Hou & Moynihan, 2008). Gramlich observed that the conventional wisdom that only central governments should conduct fiscal stabilization policy might have been seriously misguided because empirical studies have shown that the migration of production factors is slow and limited in the short run; therefore, states should practice fiscal policies to stabilize their budgets. A series of studies, represented by Sobel and Holcombe (1996), Knight and Levinson (1999), Wagner (2003), and Hou (2002, 2003a, 2003b, 2005), provide evidence that the state level conducts stabilization practices. However, here the practices are specifically budget stabilization, not economic stabilization. A most recent research (Hou, 2010) draws a conclusion that with a whole set of restrictions (fiscal and budgetary institutions) the stabilization role of the states should be specified as the stabilization of their budgets, or public service provision, rather than the economy. Along the same line of argument, scholars began to investigate the stabilization role of local governments. This literature has been growing, but remains small and inconclusive. Given local governments’ limited fiscal capacity and constraints of fiscal and budgetary rules, we would not expect them to play an active role in stabilizing the economy. However, as frontliners in the provision of public services, local governments ideally should uphold the policy inclination to act countercyclically to stabilize their budgets, or service provision, because recessions are exactly the time when local residents with reduced income need the basic public services to be reliable so that they can get through the financial and employment hardships (Hou, 2010). The question, however, is whether they have the fiscal capacity, policy tools, and political support from voters to go against the tide in the larger economy.

In a recession, a subnational government could resort to capital markets or use intergovernmental grants to implement a countercyclical fiscal policy. However, intergovernmental grants are often unreliable when grantor governments are also in fiscal distress. Incurring public debt in economic downturns is technically sound; but drastic fiscal measures by state-local governments in bust years to balance their budget often adversely affect their credit rating, thus eroding their borrowing capacity on the capital market. Constitutional or statutory debt limitations also impose constraints on how much governments can borrow. Therefore, the only reliable means a subnational government could resort to in a recession is its fiscal reserves accumulated in boom years (Hou, 2003b).

The literature on fiscal federalism also touches on the issue of what role the central government should play in providing “insurance” against exogenous shocks of various sorts in the intergovernmental fiscal system. When state and local governments subjected to balanced-budget constraints are forced to cut spending in fiscal distress, it makes sense for the central government to provide some fiscal assistance so that they can maintain their level of spending. Indeed several scholars argued that the most effective countercyclical program would be one of federal assistance to state and localities in the context of recent recessions. However, centrally provided insurance to state and local governments comes at a price because of an obvious moral hazard problem. The availability of central insurance may create an adverse incentive for state and local governments not to make sensible provision for time of financial difficulties. They may always expect the central government to extend its assistance to relieve their fiscal distress from all sorts of causes. This could create a rationale for fiscal bailouts and a general softening of budget constraints (Oates, 2005). Here the fundamental challenge is in the design of fiscal institutions and budgetary procedures that address these incentive problems in ways that create hard budget constraints for state and local governments.

State and local governments accumulate reserves via either formal or informal vehicles. The formal means of saving is by budget stabilization funds, popularly called “rainy day funds,” which are created as a result of policy design. The finance law in most states usually includes stipulations regarding the deposits of previous-year budget surplus into the budget stabilization fund (Hou, 2004). States vary a lot in whether they allow local governments to create their budget stabilization funds. Some states allow their local governments to formally create budget stabilization funds, but others do not (Hou, 2008). The informal vehicle refers to resources accumulated in the general fund, other governmental funds, and enterprise funds that can be used for expenditure stabilization across business cycles (Marlowe, 2005). At the local level, formal saving devices are not as widely used as by the states (Hou, 2008). Local governments rarely establish a separate budget stabilization fund, relying more on the informal form of fiscal reserves, such as unreserved general fund balances (Marlowe, 2005). North Carolina counties, for instance, do not preserve rainy day funds. This study focuses on year-end balances of the general fund to test our research questions.

State governments and other stakeholders have a great interest in monitoring the fiscal conditions of their local governments. When local governments face severe fiscal distress, they may not be able to make timely payments to bondholders, vendors, and/or employees. They may also have to significantly cut back the provision of some essential local public services, creating a very visible and undesirable situation. Meanwhile, current and potential residents and businesses may face an increasing tax burden because local governments attempt to generate sufficient revenue to deal with the financial difficulties. In addition, the negative impact goes beyond the distressed community. Surrounding areas are sometimes forced to extend their coverage of service delivery to the distressed community without commensurate compensation. The financial difficulties of the distressed local government may hurt not only its own credit rating but also that of other localities and the state. Finally, all the state’s taxpayers may suffer from having to pay for grants and/or bailouts to the distressed community (Kloha, Weissert, & Kleine, 2005).

Therefore, probably no other single item in a local government’s financial statements attracts more attention than fund balance (Gauthier, 2009a), which is the difference between the assets and liabilities reported in the financial statements by the end of each fiscal year,⁵ as it is often used as an indicator to evaluate local fiscal health. State governments use fund balance as one of the major indicators to identify financially distressed local governments so state and local actions can be taken to restore financial stability (Kloha et al., 2005). Credit rating agencies often assess state and local governments’ creditworthiness based on the levels of their fund balance and unreserved fund balance in a government’s general fund. Similarly, laws and regulations also govern the appropriate levels of fund balance and unreserved fund balance for state and local governments (Government Finance Officers Association, 2002).

The Government Finance Officers Association (GFOA) recommends that state and local governments establish a formal policy on the level of unreserved fund balance that should be maintained in the general fund. The unreserved fund balance is expected to safeguard against the effects of unanticipated natural disasters and economic downturns and to help stabilize tax and fee rates and public service levels (Coe, 2007; Shelton & Tyer, 2000). Fund balance levels are also a crucial consideration in long-term financial planning (GFOA, 2002).

In practice, state and local governments often formulate a fund balance policy that establishes a minimum level at which fund balance is to be maintained. GFOA recommends that unreserved general fund balance should represent no less than 5% to 15% of general operating revenues, or no less than 1 to 2 months of general fund operating expenditures (Gauthier, 2002). A minimum of 5% of unreserved fund balance is recommended by GFOA for larger, in particular state, governments because larger governments often have more diversified and less volatile revenues and operating expenditures and normally find it easier to predict contingencies than do smaller ones (Joyce, 2001; Hou, 2004). Meanwhile, a number of reasons could justify a significantly higher minimum target level for local governments, including significant volatility in operating revenues or expenditures, exposure to natural disasters, rapidly growing budgets, or disparities in timing between revenue collections and expenditures (Gauthier, 2009b).

North Carolina Counties

This study uses counties as the unit of analysis because they are the frontliners in public service provision and the dominant form of local government in North Carolina. Benton (2002) suggests that counties have “become full-service governments” and “are also beginning to rival municipalities as major providers of services that traditionally have been dominated by cities” (p. 466). Counties in southern states have traditionally held all kinds of responsibilities as a “strong unit performing a large number of functions” (p. 19). They have fewer competing local government structures than in other regions of the country, thereby becoming the dominant form of local government (Marando, 1979, p. 414).

Several features distinguish counties in North Carolina from those in other states. First, in North Carolina almost all local governmental responsibilities rest with counties (and cities). The majority of local expenditures are also made through counties (and cities), rather than through special districts, school districts, and authorities, which are relatively much more important in many other states. Second, local responsibilities for a number of major “people” services, such as public health, education, and welfare, rest mainly with counties in North Carolina rather than with cities or special districts as in other states, and the county commissioners have limited discretion in whether or how much to fund these services. Third, county governments in North Carolina have more extensive authority than those in other states to provide urban type of services such as water and sewer services, solid waste collection and disposal, and fire protection and so on. Finally, North Carolina has a state-local revenue system that relies mainly on four major taxes—property, general sales, individual and corporate income, and gasoline. Rates for the sales and income taxes are relatively high in comparison with the same type of taxes in other states, whereas rates for the property tax are low compared with those in other jurisdictions. Since the property tax is relatively less important in North Carolina than it is in other states, whereas income and sales taxes have greater significance, the revenue structure of North Carolina tends to be more sensitive than that of most states to fluctuations in the economy (Bell, 2007).

North Carolina has a long tradition of using professional managers in local governments (Stenberg, 2007)⁶ and has developed a conservative budgeting and financial management system after the Great Depression. The legislature created a Local Government Commission (LGC) under the state treasurer’s office to impose financial controls and assist local governments in trouble. North Carolina is the only state that is legally responsible for issuing all local debt (Coe, 2007; Shelton & Tyer, 2000). The LGC recommends that localities maintain a minimum unreserved fund balance of 8% of annual expenditures, and retain a much higher amount of total fund balances (Coe, 2007). The LGC also proactively works with local governments to resolve their identified financial problems.

Literature Review

The existing literature on budget stabilization focuses on federal and state governments; only in recent years did some scholars begin to pay attention to the local level. We review this literature in two subareas that are related to this article: determinants of fund balance and effects of the saved balances.

Determinants of the Size of Unreserved Fund Balance

The GFOA recommends 5% to 15% as the benchmark of state-local savings. Previous studies, however, consistently found localities do not show the same propensity to limit their savings as states do and that localities maintain much higher levels. Minnesota cities (n = 103) kept an average 53% of their current expenditures as general fund balance in the 1990 to 2000 period. The fund balance was more than needed for cash flow purposes in more than half of the cities, and was equal to or more than a full year’s current expenditures in 20% of them (Marlowe, 2005). Chicago suburban municipalities kept a median 53% to 66% of unreserved fund balance from 1997 to 2003 (Hendrick, 2006). Most Mississippi counties maintained an unreserved fund balance ranging from 25% to 237% of their expenditures in 1995 and 1999 (Stewart, 2009).

Only a few studies attempted to explain the factors that determine the size of unreserved fund balance at the local level; and they reached no consensus. Hendrick (2006) found current fiscal performance, indicated as operating deficits or surplus, had a strong effect on municipal fund balances. She also found that debt levels had a negative impact: The higher the level of municipal debt was, the less reserves were used to fund capital expenditures. Size was another factor: Larger governments appeared to accumulate less reserves than smaller ones. Municipalities that were more politically conservative, reformed, and professional tended to accumulate more reserves. Not surprisingly, wealthier municipalities with lower spending needs had higher reserves than those that were poor and had higher spending needs. Hendrick’s analysis demonstrated that the main purpose of reserves was to cushion for unpredictable events and short-run instability. Unreserved fund balances became more important and interchangeable during fiscal downturns while maintaining reserves to compensate for risks became less important.

Stewart (2009) also found that current fiscal condition significantly influenced the level of Mississippi counties’ unreserved fund balances during two periods of relative resource abundance and scarcity. Consistent with the findings of Hendrick (2006), his study showed that the only variable of significance during both periods was debt level, which had a negative impact on unreserved fund balance. Other fiscal, demographic, and political variables included in the study explained little variation in fund balances, which may suggest that Mississippi counties were unique within themselves and it was hard to identify common factors to explain the behavior of unreserved fund balance.

Thus, it seems that applying a 5% to 15% savings benchmark across all jurisdictions does not provide a sufficient guide when localities assume different patterns in accumulating savings. Existing studies still cannot provide a good guideline for local governments to preserve an appropriate level of savings for particular conditions. Future research on this topic may need to explore other variables to explain the behavior of unreserved fund balance.

Effects of Unreserved Fund Balance on Government Expenditures

Studies of fiscal reserves at the state level find strong evidence that state savings help to maintain the level of government spending during downturn years (Hou, 2003a, 2005; Sobel & Holcombe, 1996). Few studies, however, provide direct evidence about the effects of unreserved fund balance on the stabilization of government expenditures at the local level. Marlowe (2005) examined whether local slack fiscal resources in the form of general fund balance had similar countercyclical effects; he found that municipal fiscal reserves in general had a marginal impact on the direction and size of current expenditure gaps, but unreserved general fund balance posed a countercyclical stabilizing effect during downturn years. However, he cautioned that the evidence was not conclusive.

Later studies, nevertheless, did not produce empirical results suggesting a stabilizing effect of unreserved general fund balance at the local level. Hendrick (2006) found some significant, positive impact of unreserved fund balance on the percentage change of noncapital expenditures with pooled data for Chicago suburban municipalities, but the results were not reliable for individual years. Based on Georgia county data, Hou (2008) investigated the determinants of the size of local fund balances and the impact of fund balances on local expenditures in downturn years. The empirical results of the study showed that local governments had few alternatives other than coping during economic downturns, given their less diversified revenue portfolio and strict balanced budget requirement.

Hendrick (2006) observes that municipal fiscal slack has “a fairly complex relationship with governments’ other fiscal structural features and environment,” which makes it difficult to detect the effects of fiscal slack on alleviating fiscal distress. Future research should attempt to develop a “theoretical model that incorporates relevant managerial, organizational, and political features . . . ” (p. 43). Pursuing this path of research, Hou (2008) suggested a number of parameters for the analysis of local governments’ fiscal behavior in economic fluctuations, including states’ permission for their localities to create their own rainy day fund, local governments’ adoption of a formal or informal countercyclical fiscal policy, and revenue portfolio and professionalism of local governments. This study continues on Hou’s (2008) exploration.

Model Specification, Data, and Methodology

Following the empirical framework adopted by previous research, we conduct empirical analysis to examine the factors that affect the level of general fund balances in North Carolina counties and the impact of these balances on county general fund expenditures in general and their stabilizing effects during economic downturns in particular, using a dataset of all the 100 counties in North Carolina for the period of 1990 to 2007. Since the sample period covers two most recent, full business cycles, point estimates thus obtained may bear more directly to current practices in local governments.

Dependent Variables

To identify the determinants of fund balances (answer our first research question), we use two dependent variable as follows: fund balance level and the balance as a share of overall expenditures. To estimate the effects of the balance (the second research question), our dependent variables are total expenditure, and year-to-year change (first difference) of total expenditure. To evaluate expenditure fluctuations, we construct an expenditure gap variable as the difference between actual expenditures and expenditures projected from a linear trend. Following previous studies, county expenditures are predicted as follows:

E_{i t}^{*} = a_{i} + b_{i} T

where E^*_it is the predicted expenditure for county i in year t, a_i the constant for county i, b_i the linear trend coefficient for county i, and T the value of the year in year T. Then the expenditure gap is expressed as the difference between the actual and the predicted expenditures as a percentage of the predicted expenditure:

Expenditure Gap = \frac{E - E *}{E *}

Positive expenditure gaps are expected in boom years, whereas negative expenditure gaps may arise during economic downturns.⁷

Explanatory Variables

To investigate the first research question, we choose five groups of explanatory variables: revenues, expenditures, economic indicators, social indicators, and county size. Revenue variables include the major revenue sources for county governments; these are property tax, sales tax, and intergovernmental revenue. Prior research indicates that the composition of governments’ revenue sources determines the stability of their revenue and in turn the level of their fund balance. Governments could plan on stable revenue sources more than on cyclical ones, for example, fines and fees (Stewart, 2009). In contrast, they may have to use unreserved fund balance to smooth the tax rate and offset erratic revenue fluctuations (Massey & Tyer, 1990; Stewart, 2009). Therefore, stable revenue sources would help governments to accumulate a higher level of fund balance (Marlowe, 2005; Stewart, 2009). Since property tax and local option sales taxes are the two most important revenue sources for North Carolina counties (though sales taxes are more prone to business cycle fluctuations), they are expected to facilitate maintaining fund balance. Intergovernmental grants to the county are more volatile because federal and state governments are likely to cut grants when they are under fiscal distress. Previous studies posited that local governments would accumulate more fiscal slack as a shield against revenue shocks if they have a heavier reliance on intergovernmental grants (Hendrick, 2006; Stewart, 2009). However, the negative relationship found between grants and fund balance suggests that local governments might not recognize the risk involved in relying on such a revenue source (Stewart, 2009) or that they might not have the capacity to cushion against such cuts that lead to unavoidably reduced local revenues.

County expenditures by function include expenditures for salaries and wages, capital outlays, and other operating purposes. Hendrick (2006) suggested that the levels of capital spending and total government spending, as measures of alternative sources of slack or fiscal flexibility, should have a negative impact on fund balance. In this study, we test the effects of different categories of spending on unreserved fund balance. Data for all the above-mentioned financial variables were collected from the Annual Financial Information Forms issued by the State Treasurer’s Office in various years.

We include county per capita income and unemployment rate as the third group of explanatory variables to capture the effects of economic factors on our dependent variables. As suggested by previous studies (Stewart, 2009; Wagner, 2003), we expect per capita income to be positively correlated with unreserved fund balance and unemployment rate negatively correlated. The income data are obtained from the U.S. Bureau of Economic Analysis; the unemployment rate data are from the U.S. Bureau of Labor Statistics.

Social indicators are the size of population, population squared, and shares of the school-age and the senior groups in the population, obtained from the North Carolina Office of State Budget and Management. Racial and ethnic diversity and the number of eligible people entitled to retirement may influence the behavior of the fund balance (Marlowe, 2004; Stewart, 2009); but such data are not available. Finally, we construct three dummy variables of county size by population to test the effects of the size of a jurisdiction on our dependent variables. The county size dummies were constructed according to the classification of the North Carolina treasurer’s office that roughly divides the 100 counties into four groups by their size of population; there are around 25 counties in each group.

We test the effects of fund balances on county expenditures and expenditure gaps. Fund balances are expected to have a positive impact on total expenditures. If fund balances have a positive impact on the expenditure gap during downturn years, they can be said to have a countercyclical effect. In contrast, they may have a procyclical effect if they increase a positive expenditure gap in boom years or enlarge a negative expenditure gap in lean years (Marlowe, 2005).

Revenue sources, economic indicators, social indicators, and county size variables are also included in the models. Higher tax revenues are expected to lead to higher expenditures. The effects on the expenditure gap depend on their stability over the business cycle. Grants are also an important variable because they may create cash flow problems that can be addressed only by accumulating more adequate fund balances by local governments (Marlowe, 2005). This study includes this variable to test its potential effects on county expenditures and expenditure gaps. Income and unemployment are included in the models to capture the variability in county economic conditions. The four social indicators are used to test their effects on the spending decisions of county governments. County size dummies are also included as controls. Table 1 provides the descriptive statistics of all variables included in this analysis.

Table 1.

Descriptive Statistics (NC County Data: 1990-2007)

Variable	Observation	Mean	Minimum	Maximum
Dependent/key variables
Fund balance available	1,799	95.98	−49.22	839.63
Fund balance available as % of total expenditure	1,799	22.11	−9.58	138.64
First difference of fund balance available	1,799	3.12	−206.10	794.64
Expenditure gap (%)	1,799	0.036	−45.11	89.15
Revenue sources
Property tax	1,799	204.94	65.19	622.19
Sales tax	1,799	79.01	28.95	258.06
Intergovernmental revenue	1,799	125.91	33.64	1,134.12
Total revenue	1,799	567.63	246.73	2,478.38
First difference of total revenue	1,699	15.92	−1,268.66	1,249.96
% of property tax in total revenue	1,799	37.22	10.51	58.76
% of sales tax in total revenue	1,799	14.88	4.48	27.46
% of intergovernmental revenue in total revenue	1,799	22.78	3.26	58.97
Expenditures
Salaries & wages	1,799	129.54	49.23	438.09
Other operating expenditures	1,799	191.41	54.93	944.50
Capital outlay	1,799	45.22	0.30	1,127.87
Total expenditure	1,799	549.28	247.98	2,291.70
First difference of total expenditure	1,799	17.05	−529.69	831.67
Debt service expenditure	1,799	38.34	0.00	329.40
Changes in debt service	1,790	23.13	−100.00	4,425.20
Economic indicators
Personal income (1,000)	1,799	12.93	7.86	22.02
Unemployment rate (%)	1,799	5.73	1.20	19.80
Social indicators
% of population under 18	1,799	23.74	14.95	30.99
% of population above 65	1,799	14.25	4.44	24.56
Population (1,000)	1,799	78.29	3.73	859.85
Population growth rate (%)	1,699	1.41	−3.34	8.22
County size by population (binaries)
Large (>100,000)	1,799	0.22	0.00	1.00
Medium (50,000-100,000)	1,799	0.25	0.00	1.00
Medium small (25,000-50,000)	1,799	0.24	0.00	1.00
Small (<25,000)	1,799	0.28	0.00	1.00

Note: All financial level figures are in per capita real terms.

To answer our first research question, we estimate two regression models as follows: per capita fund balance and fund balance share are regressed on the above-mentioned five groups of independent variables. To address the second research question, we estimate several empirical models to explore the effects of fund balances, both in level and ratio, on total county expenditures, the first difference of total expenditures, and expenditure gaps over the business cycle.

Estimation Methodology

We use the Newey-West estimator to deal with heteroskedasticity and autocorrelation in the dataset. OLS estimates are computed, and standard errors adjusted for heteroskedasticity and autocorrelation. We also add in county fixed effects in the model to control for time-invariant unobservable factors, and year fixed effects to control for yearly statewide changes.⁸

As certain county-specific characteristics may have differential effects on county expenditure gaps in various years, we estimate a Heckman two-stage selection model to address the potential selection bias. Population growth may create differential impact on individual county governments given the specific context of each county. Changes in debt service could affect the current revenues available depending on individual county debt service schedules and capital improvement plans (Marlowe, 2005). In addition, the differences in the composition of revenue portfolio may also affect a county’s capacity in adjusting to the changes in its economic conditions. Therefore, population growth, changes in debt service, and shares of property tax, sales tax, and intergovernmental revenue are included in the first stage as selection variables to estimate the likelihood for a county to experience a downturn year or upturn year. The likelihood of either a downturn year or upturn year, expressed as the inverse Mills ratio, is included in the second stage with county and year fixed effects as described above. The second stage estimates the effects of fund balances on county expenditures during the downturn and upturn years of the economic cycle.

Empirical Results

Determinants of General Fund Balances

Results on the determinants of fund balances, both in level and in ratio, are presented in Table 2. Among the major revenue sources, property tax exerts a statistically significant influence—each dollar of revenue from this source can raise balances by 35 cents or .05 percentage points. Given the average level of fund balances of US$96 or 22% of total expenditure, the 205 dollar average from this source can increase fund balances by US$72, a three-quarters increase in level, or 11 percentage points, a 50% increase in ratio. Also statistically significant is revenue from the sales tax, for which the magnitude is particularly large: Each dollar from this source can boost fund balance by as much as US$1.36 in level or .16 percentage points. In other words, the US$79 average sales tax revenue can raise fund balances by US$107, more than double the fund balances, or 12 percentage point, also an above-50% increase. Intergovernmental revenue does not pose a significant effect, probably because these grants all come with a specified purpose for use in the current fiscal year so that they do not add much to year-end balances.

Table 2.

Fund Balance Models

	Fund balance available		Fund balance available as % of total revenue
	Coefficient	z	Coefficient	z
Revenues Property tax	0.354***	4.080	0.053***	3.151
Sales tax	1.356***	6.299	0.157***	3.545
Intergovernmental	−0.017	−0.540	−0.006	−0.977
Expenditures
Salaries & wages	0.105	1.310	−0.009	−0.539
Other operating	0.022	0.663	−0.008	−1.264
Capital outlay	−0.028	−0.784	−0.014**	−2.030
Economic indicators
Personal income (1,000)	0.005*	1.694	0.001	1.556
Unemployment rate	−1.545779	−1.703	−0.438*	−1.962
Social indicators
% of population under 18	−0.841	−0.303	0.016	0.025
% of population above 65	3.572	1.060	0.153	0.219
Population (1,000)	−0.336**	−1.234	−0.144**	−2.065
Population squared	0.000**	0.774	0.0001*	1.822
County size
Large	13.926	0.894	5.602	1.625
Medium	2.794	0.233	2.312	0.947
Medium small	−12.118	−1.166	−2.236	−1.162
No. of observations	1,799		1,799
Prob > F	0.000		0.000
Centered R²	0.269		0.082

p < .1. **p < .05. ***p < .01.

Among the major expenditure categories, only capital outlay posts a statistically significant, negative effect in the ratios model: Each dollar of outlay reduces the fund balance by .014 percentage points; thus an average county with US$45 outlay in this category will have their fund balanced cut by .63 percentage points, which is only a 2.8% reduction off the average balance ratio, not a remarkable size. The two economic indicators are both significant (marginally significant in the level model), with the expected sign. Wealthier counties tend to have more in their year-end fund balances, both in the level and the ratio form. Per thousand dollar increase in per capita personal income can raise fund balance by US$4.6 or .91 percentage points. An average county with their personal income at the US$12,929 level can expect to have about extra US$60 in their balance or more than 11 percentage points above the average 22% level; thus, this wealth effect is quite substantial. Higher unemployment rates pull down fund balances, each percentage point by US$1.5 in level or .44 percentage points in ratio. An average county in an average year (with the unemployment rate at 5.73) may expect a fall of close to US$9 in level or 2.57 percentage points.

Of the social indicators, the shares of the young and the old in total population do not generate statistically significant results. Population size points to economies of scale—larger counties tend to maintain smaller fund balances in ratio; that is, the size of a locality matters in the negative direction. This effect, however, is not linear, as indicated by the positive sign of the quadratic term of this variable. The county size dummies are not significant; potentially the population variable may have picked up the effect. We tried removing these dummies but obtained no observable, statistically significant difference.

Effects of Fund Balances on Expenditures

We then test the effects of previous year fund balances (t-1) on total expenditures.⁹ The dependent variables are per capita real expenditure and the first difference (annual change) in total expenditure, thereby making the interpretation of the results straightforward. Results of this test are shown in Table 3.

Table 3.

Effects of Fund Balances on Total Expenditure

	Total expenditure				First difference of total expenditure
	Coefficient	z	Coefficient	z	Coefficient	z	Coefficient	z
Fund balance available_t−1	0.146**	2.063			0.080	0.883
% of fund balance available_t−1			0.322	1.336			0.829**	2.272
Revenue sources
Property tax	0.824***	4.718	0.841***	4.743	0.164	0.965	0.182	1.068
Sales tax	3.114***	4.558	3.277***	4.706	0.480	0.888	0.459	0.858
Intergovernmental	0.859***	6.350	0.858***	6.368	0.487***	3.772	0.495***	3.845
Economic indicators
Personal income (1,000)	3.875	0.795	4.427	0.905	2.947	0.690	2.454	0.574
Unemployment rate	−2.096	−1.045	−2.082	−1.032	0.589	0.274	0.792	0.370
Social indicators
% of population under 18	−3.553	−0.731	−3.995	−0.821	−1.253	−0.259	−1.419	−0.293
% of population above 65	6.244	0.919	6.627	0.974	7.208	1.100	7.056	1.085
Population (1,000)	1.541**	2.395	1.539**	2.410	1.172**	2.233	1.226**	2.328
Population squared	−1.085**	−2.288	−1.095**	−2.328	−0.902**	−2.371	−0.936**	−2.457
County size (binaries)
Large	−56.794	−1.475	−56.410	−1.450	13.805	0.423	7.221	0.221
Medium	−75.072**	−2.324	−75.234**	−2.298	23.716	0.913	19.475	0.751
Medium small	−69.458**	−2.420	−70.736**	−2.428	4.475	0.221	3.881	0.192
No. of observations	1,699		1,699		1,699		1,699
Prob > F	0.000		0.000		0.000		0.000
Centered R²	0.612		0.610		0.105		0.110

p < .1. **p < .05. ***p < .01.

Fund balances carried forward from the previous year exert a small but significant effect on total expenditures: Per dollar of balance can increase expenditure by 15 cents that is about US$14 increase for an average county (with a mean of US$96 of balance), which is a 2.6% push on the average of US$549 of total expenditure. Fund balance as a percent of total expenditure, however, is not statistically significant.

The three major revenue sources all present significant and large effects on total expenditures. Each additional dollar of property tax or intergovernmental grant is associated with an increase of more than 80 cents in total expenditures. Of particular interest is the large effect of sales tax revenue on total expenditure: each dollar from this source adds between US$3.11 and US$3.28 to the expenditure side; however, based on our limited data we cannot yet offer a reasonably satisfactory answer.

The economic indicators and the young and the old shares in the population do not produce significant results. The nonlinearity of population size keeps its significance. Total expenditure first rises with size, at US$1.54 for per thousand more residents. Then it drops, confirming again the economies of scale. The county size dummies correspond to this finding: All three are negative with two being significant, that is, all spend less than the small counties. Medium sized (50,000-100,000) and medium–small sized (25,000-50,000) counties are spending between US$70 and US$75 less per capita than the (default) small (<25,000) counties. This result points again to two issues: First, the county at the front line of service provision shoulders the task of supporting the daily life of residents; large or small, their functions are essential so that being small may in fact push up total expenditures. Second and related to the first, economies of scale is a factor we should not neglect.

Of the annual changes in total expenditure, fund balance as a percent of total expenditure is significant: Every percentage point can account for 83 cents in the marginal increase in total expenditure. The other significant factor is intergovernmental grants, each dollar accounting for about 50 cents of the annual outlay increase. Given the average amount from this source of US$126, intergovernmental grants can potentially account for US$62, 3.6 times the annual increase. Population and its squared term retain their significance and expected sign in this specification.

Effects of Fund Balances on Expenditures over the Economic Cycle

Our final test is to identify the effects of fund balances on expenditures during the downturn and upturn years of the economic cycle, which can be taken as the local government activity cycle. We do so by decomposing the expenditure gap variable into two groups: Those smaller than 0 are a proxy for the local downturn years and those larger than 0 are a proxy for the local upturn years.¹⁰ We use the James Heckman two-stage sample selection estimator for the test. The results are reported in Table 4 (results from the first stage are not presented).

Table 4.

Effects of Fund Balances on Total Expenditure in “Downturn Years” and “Upturn Years”

	Downturn years (expenditure gap < 0)				Upturn years (expenditure gap > 0)
	Coefficient	z	Coefficient	z	Coefficient	z	Coefficient	z
Fund balance available_t−1	0.003	0.995			0.021	1.631
% of fund balance available_t−1			−0.003	−0.171			0.103**	2.173
Revenue sources
Property tax	0.012	1.237	0.012	1.211	0.076***	3.577	0.079***	3.642
Sales tax	−0.003	−0.103	0.003	0.092	0.001	0.014	0.015	0.226
Intergovernmental	0.018**	2.065	0.018**	2.064	0.030***	3.130	0.030***	3.134
Economic indicators
Personal income (1,000)	0.023	0.064	0.051	0.145	−0.729	−0.822	−0.703	−0.793
Unemployment rate	−0.133	−0.821	−0.134	−0.830	0.253	0.390	0.292	0.452
Social Indicators
% of population under 18	0.066	0.185	0.042	0.118	0.596	0.571	0.553	0.533
% of population above 65	0.708*	1.711	0.709*	1.713	−0.372	−0.404	−0.296	−0.320
Population (1,000)	0.100***	2.604	0.099***	2.596	0.013	0.132	0.020	0.192
Population squared	−0.0001***	−2.732	−0.0001***	−2.743	0.00001	0.065	0.0000	−0.008
County size (binaries)
Large	−3.019	−1.219	−2.931	−1.186	−13.817**	−2.487	−14.750***	−2.598
Medium	0.236	0.122	0.347	0.181	−7.929*	−1.699	−8.556*	−1.799
Medium small	2.359*	1.951	2.407**	2.006	−6.823*	−1.787	−7.106*	−1.836
Constant	−30.042**	−2.468	−29.887**	−2.451	19.148	0.611	16.174	0.513
Wald χ² (p > χ²)	806.27 (.00)		810.85 (.00)		441.34 (.00)		439.42 (.00)
Log pseudolikelihood	−3,924.82		−3,925.22		−3,569.61		−3,568.79
Test of independent equations (p > χ²)	24.20 (.00)		24.40 (.00)		80.97 (.00)		81.50 (.00)
Censored observation	703		703		990		990
Uncensored observation	990		990		703		703

Note: Heckman two-stage sample selection models; Dependent variables are expenditure gaps.

p < .1. **p < .05. ***p < .01.

The models in general fit well. Given the sample period of this study, the number of observations of downturns (703) is smaller than that in upturns (990), which fits our intuition. We are interested in whether the fund balance variables exert positive impact during the downturn and how they affect the trend during the upturn. In downturns, the balances either in level or in ratio do not pose an effect that is, significantly different from zero. Among the revenue sources, only intergovernmental dollars show a significant estimate in both phases of the cycle; that is, during a downturn, other revenue sources do not seem to play a countercyclical role. The magnitude of the grants variable is .018 percentage points per dollar, which is about 2.2 percentage points for an average county (= 126 × .018), far from enough to raise expenditures up to the trend from the level of the constant (−30 percentage points). The share of seniors in total population shows a significant, positive effect in downturn years at the magnitude of .71 in both the level and the ratio models. At the mean share of seniors (14%), the senior population may reduce the expenditure gap by about 10 percentage points in a downturn. Population size has a similar effect at the magnitude of .1 in both specifications, which means the population variable may reduce the gap in downturns by 7.8 percentage points. Again, this effect is not linear; the quadratic term shows a negative effect.

In upturn years when the constant is already above the trend (by 16-19 percentage points), in contrast, the fund balance ratio is statistically significant (the level variable is marginally significant): The balance level can push up the trend further by close to 2 percentage points (96 × .021 = 1.97) and the balance ratio can send the trend up by about the same range (22 × .1 = 2.28). The magnitude of intergovernmental grants nearly doubles (from .018 to .03). Property tax revenue is now also significant at about .08 percentage points per dollar, the total effect of which can amount to 16 percentage points for an average county. The socioeconomic variables are not significant but all the three county size binaries are significant and negative, that is, the small counties now spend more than their larger counterparts. In sum, we see mainly procyclical forces instead of countercyclical ones.

Discussion and Conclusion

This article has examined the determinants of local government savings and estimated the effects of these accumulated savings on local expenditures. On the determinants of local savings, we found the property tax is a major contributor; intergovernmental grants are not, because they are for use in the current year. Having local option sales taxes is a strong plus at the local level: It may contribute dollar-for-dollar toward savings. Wealth is also a major savings contributor, indicating a strong positive relation between income and local savings capacity. Capital outlay reduces savings; so does unemployment. The population size of localities presents a nonlinear relationship with fund balances—within a certain range, economies of scale exist. Since local governments are at the forefront of public service provision, being small does not necessarily render any advantages but increases outlays.

Using county data in the state of North Carolina in the last two business cycles we have tested cyclical features of local expenditures and found their trend to be procyclical. Unlike some early evidence that suggests countercyclical stabilization properties of local unreserved general fund balance, the empirical results of our study do not show that local savings exert countercyclical impact on expenditure. Intergovernmental grants exert a small effect in the downturn but its effect in the upturn is much larger. This finding leads us to more questions for further exploration: If with a sizable ratio of savings localities still do not act countercyclically, they are simply following the economic and fiscal tide. The fact that large counties do come closer to the trend in upturn years than small ones do may suggest that being small loses the stand to go against the tide. If that is the case, how do we design fiscal institutions and budgetary procedures that allow the federal or state government to provide more effective fiscal assistance to localities without creating a problem of soft budget constraint? North Carolina has a conservative budgeting and financial management system, and the state government exerts unusually strong control over the fiscal practices of its local governments, such as their debt issuance. What can other states learn from North Carolina to improve their budgeting and financial management practices? Would local governments act differently in managing their savings and expenditures in states with a less effective oversight system? Are counties, in comparison with other local governments, more heavily influenced by their state governments when formulating their fund balance strategies, since they are generally agents of the state to a much greater extent than is true of other local jurisdictions? In addition, what are the effects of political factors on the practice of local savings and their capacity in stabilizing expenditures? Given the fact that final budget decisions are often made by elected politicians (county commissioners), to what extent do they affect recommendations made by professional county managers, following GFOA guidelines for fund balance targets? Are the decisions on local savings often affected by other political considerations for employee compensation, or commitments to interest groups and so on?

Our findings carry timely and important policy implications and contribute to the existing literature in several important ways. First, the mere size of savings at the local level is probably not enough to support a countercyclical fiscal policy. If this is true, then studies on the optimal size of savings will not shed light on the question. Second, the size of a jurisdiction does matter. Since being small does not render any advantage in this aspect, we should give up any unrealistic expectations for small localities to be fiscally “active” against the business cycle. Let them focus on providing services and not worry about finances. The current study is only among the first explorations, and more solid work is to be done with better data.

Footnotes

The authors would like to thank the co-editors in chief of this journal and anonymous reviewers for their valuable advice that helped them clarify some places of the text.

The author(s) declared no potential conflicts of interest with respect to the authorship and/or publication of this article.

The author(s) received no financial support for the research and/or authorship of this article.

Notes

References

Bell

A. F.

(2007). Article 1: An overview of local government. In Lawrence

D. M.

(Ed.), County and municipal government in North Carolina. Chapel Hill, NC: UNC-Chapel Hill. Retrieved from http://www.sog.unc.edu/pubs/cmg/

Benton

J. E.

(2002). Counties as service delivery agents: Changing expectations and roles. Westport, CT: Praeger.

Coe

C. K.

(2007). Preventing local government fiscal crises: The North Carolina approach. Public Budgeting & Finance, 27(3), 39-49.

Douglas

J. W.

Gaddie

R. K.

(2002). StateRainy day funds and fiscal crises: Rainy day funds and the 1990-1991 recession revisited. Public Budgeting & Finance, 22, 19-30.

Gauthier

S. J.

(2002). Reserve fund balance policy promulgated. American City & County, 117(4), 6.

Gauthier

S. J.

(2009a). Fund balance: New and improved. Government Finance Review. Retrieved from http://www.gfoa.org/downloads/NewFundBal_GFR_apr_09.pdf

Gauthier

S. J.

(2009b). What everyone needs to know about the new fund balance? Chicago, IL: Government Finance Officers Association.

Gianakis

Snow

(2007). Implementation of local government stabilization funds in Massachusetts. Public Budgeting & Finance, 27(1), 86-103.

Government Finance Officers Association. (2002). Appropriate level of unreserved fund balance in the general fund. Retrieved from http://www.gfoa.org/downloads/budget-appropriate.pdf

10.

Gramlich

E. M.

(1987). Subnational fiscal policy. In Quigle

John M.

(Ed.), Perspectives on local public finance and public policy (pp. 3-27). Greenwich, CT: JAI Press.

11.

Hendrick

(2006). The role of slack in local government finances. Public Budgeting & Finance, 26(1), 14-46.

12.

Hou

(2003a). What stabilizes state general fund expenditures in downturn years-budget stabilization fund or general fund unreserved undesignated balance? Public Budgeting & Finance, 23(2), 64-85.

13.

Hou

(2003b). Budget stabilization fund. In Rabin

Jack

(Ed.), Encyclopedia of public administration and public policy. New York, NY: Marcel Dekker Press.

14.

Hou

(2004). Budget stabilization fund: Structural features of the enabling legislation and balance levels. Public Budgeting & Finance, 24(3), 38-64.

15.

Hou

(2005). Fiscal reserves and state own-source expenditure in downturns. Public Finance Review, 33(1), 117-144.

16.

Hou

(2008, October 23-25). Parameters of local government options for a downturn: Evidence from Georgia counties. Paper presented at the 2008 Annual Conference of the Association for Budgeting and Financial Management (ABFM), Chicago, IL.

17.

Hou

(2010, February 25-27). Local fiscal behavior under state provisions: Do local governments smooth revenue from booms across busts? Paper presented at the seminar on “States as Facilitators or Obstructionists of Local Governments,” Florida State University, Tallahassee, FL.

18.

Hou

Moynihan

D. P. M.

(2008). The case for countercyclical capacity. Journal of Public Administration Research and Theory, 18, 139-159.

19.

Joyce

P. G.

(2001). What’s so magical about five percent? A nationwide look at factors that influence the optimal size of state rainy day funds. Public Budgeting & Finance, 21(3), 62-87.

20.

Keynes

J. M.

(1936).The General Theory of Employment, Interest and Money. New York: Harcourt, Brace & World.

21.

Kloha

Weissert

C. S.

Kleine

(2005). Someone to watch over me: State monitoring of local fiscal conditions. American Review of Public Administration, 35, 236-255.

22.

Knight

Levinson

(1999, September). Rainy day funds and state government savings. National Tax Journal, 52, 459-472.

23.

Marando

V. L.

(1979). City-county consolidation: Reform, regionalism, referenda, and requiem. Western Political Quarterly, 32, 409-421.

24.

Marlowe

(2004). The local fund balance: Explanations and implications (Unpublished doctoral dissertation). University of Wisconsin-Milwaukee, Milwaukee, WI.

25.

Marlowe

(2005). Fiscal slack and counter-cyclical expenditure stabilization: A first look at the local level. Public Budgeting & Finance, 25(3), 48-72.

26.

Massey

Tyer

C. B.

(1990). Local government fund balances: How much is enough? South Carolina forumml: A review of public affairs in South Carolina, 1, 40-46.

27.

Musgrave

R. A.

(1959). The theory of public finance. New York, NY: McGraw-Hill.

28.

National Association of State Budget Officers. (2007). The fiscal survey of states. Washington, DC: Author.

29.

Oates

W. E.

(2005). Toward a second-generation theory of fiscal federalism. International Tax and Public Finance, 12, 349-373.

30.

Oates

W. E.

(1972). Fiscal federalism. New York, NY: Harcourt Brace Jovanovich.

31.

Shelton

Tyer

(with Hembree

). (2000). Local government reserve funds and fund balance: Some applications of business concepts. Municipal Finance Journal, 21(1), 1-18.

32.

Sobel

R. S.

Holcombe

R. G.

(1996). The impact of state rainy day funds in easing state fiscal crises during the 1990-1991 recession. Public Budgeting & Finance, 16(3), 28-48.

33.

Stenberg

C. W.

III (2007). Article 5: Leading and governing in council-manager counties and cities. In Lawrence

D. M.

(Ed.), County and municipal government in North Carolina. Chapel Hill, NC: UNC-Chapel Hill. Retrieved from http://www.sog.unc.edu/pubs/cmg/

34.

Stewart

L. S. M.

(2009). Examining factors that impact mississippi counties’ unreserved fund balance during relative resource abundance and relative resource scarcity. Public Budgeting & Finance, 29(4), 45-73.

35.

Wagner

G. A.

(2003). Are state budget stabilization funds only the illusion of savings? Evidence from stationary panel data. Quarterly Review of Economics and Finance, 43, 213-238.

36.

Wagner

G. A.

Elder

E. M.

(2007). Revenue cycles and the distribution of shortfalls in U.S. states: Implications for an “Optimal” rainy-day fund. National Tax Journal, 60, 727-742.