The role of intergovernmental aid in defining fiscal sustainability at the sub-national level

Abstract

The concept of fiscal sustainability has become increasingly used over the last 20 years. However, much of the literature on fiscal sustainability at the sub-national level ignores the role of intergovernmental fiscal relations. To address this gap, this paper discusses a sufficient condition for sub-national fiscal sustainability and examines the importance of intergovernmental aid in determining that sustainability. Using panel data of counties and municipalities in the US, and using unit root and cointegration analytic techniques, this paper finds different levels of fiscal sustainability between counties and municipalities. We also find that intergovernmental aid plays an important role for fiscal sustainability for both counties and municipalities.

Keywords

cointegration federalism fiscal sustainability intergovernmental aid methods North America public administration unit root

Economic crises worldwide have affected local governments and their fiscal conditions (Cohen, 2011; Paulais, 2009). They have suffered from the sharp decline in revenues as the economic slowdown diminished real property values, other tax bases, and intergovernmental aid from higher level governments. Simultaneously, the demand for public services, in particular those related to unemployment and social welfare needs, increased, leading to a corresponding increase in government expenditures (Cohen, 2011; Paulais, 2009). However, with this fiscal stress, local governments may find it difficult to engage in new urban policies if they do not feel the policies can be fiscally sustained. As such, local governments have been trapped between decreasing revenues and increasing expenditure demands (Paulais, 2009), which brings the issue of fiscal sustainability into their current government agenda (GASB, 2011). Indeed, from Detroit, MI to Jefferson County, AL, insolvency is no longer a potential risk – rather, it has been realised.

This paper addresses the concept of fiscal sustainability at the sub-national level in the framework of a federal system that involves intergovernmental transfers. To be specific, we empirically examine fiscal sustainability of US counties and municipalities, in particular their ‘ability to balance revenues and expenditures over a long-term period’ (Ward and Dadayan, 2009: 465). Additionally, we disaggregate primary balances (total revenues less total expenditures) and explicitly introduce the context of intergovernmental fiscal dependency into our analysis. In so doing, this paper extends the previous fiscal sustainability analysis in three ways. First, most previous analyses have primarily focused on the aggregate primary balance and debt, with little disaggregation of this primary balance into its component parts. This paper extends these analyses by examining different types of budget balances. Second, the previous research was disaggregated only to the aggregate of sub-state level jurisdictions. This paper disaggregates the total of county and municipal jurisdictions into a separate analysis for counties and municipalities, which allows us to find different patterns of fiscal sustainability between these two types of sub-national governments. Finally, this paper examines not only if county and municipal governments are fiscally sustainable, but also addresses the question of how important is intergovernmental aid from the federal or state government to county or municipal fiscal sustainability. The decline or delay in intergovernmental aid has been asserted to negatively impact fiscal conditions of local governments, and the tendency becomes more visible during economic crisis (Hendrick, 2011). By explicitly taking intergovernmental aid into account, we highlight its importance to urban fiscal management and urban policy decisions.

The paper will be in five parts. The first will examine the history and definition of fiscal sustainability. The second section will then briefly discuss intergovernmental fiscal relations and how these can affect sub-national fiscal sustainability. The third section of the paper will describe formal statistical tests of sustainability, including stationarity, cointegration, and strong and weak sustainability, using four different budget measures. The fourth section provides empirical results and analysis, while the last section discusses the implications of these results.

History and concepts of fiscal sustainability

The concept of fiscal sustainability has a long history, although its nomenclature has often changed. In the late 1970s, immediately after the beginning of the tax and expenditure limitation movement, fiscal stress was the term used to explain the pressures that confronted state and local governments (Levine, 1980). The term fiscal sustainability did not appear in these earlier studies, although Gold (1995) did raise the possibility that states might face continuing long-term crises. After the Brundtland report (also known as Our Common Future) (World Commission on Environment and Development, 1987) on environmental sustainability, the term fiscal stress gradually changed into fiscal sustainability. Horne (1991), writing for the International Monetary Fund, identified indicators of fiscal sustainability, and by the mid-1990s, this change was nearly complete.

State and local governments are still being forced to confront the issues of fiscal sustainability. For example, in 2004, the Governmental Accounting Standards Board (GASB) issued statement no. 44 that focused on the reporting of the economic conditions of state and local governments – that is, the jurisdiction’s fiscal sustainability. This led to GASB’s fiscal sustainability project being moved from its research agenda to its current agenda (GASB, 2009). In 2011, GASB issued a preliminary views statement for comments on assessing economic conditions (replacing the term fiscal sustainability).

There are two distinct tracks in the analysis of fiscal sustainability concepts. One track emphasises the fiscal health of the jurisdiction in terms of whether or not the public sector is providing the appropriate level of public services financed with the appropriate level of revenue (Chapman, 2008; Ward and Dadayan, 2009; Zhao and Coyne, 2013). These are often based on median voter theory, on revenue and expenditure simultaneity, or on the role of the conflict between mandated expenditures and revenue limits. Fiscal analysis in this genre also often includes a discrete political dimension (Hendrick, 2011) as well as an explicit concern for fiscal capacity (Karuppusamy and Carr, 2012). This way of testing fiscal sustainability tends to analyse data in a very detailed and disaggregated manner – for example, it carefully analyses pension data as a contributor to fiscal sustainability (GASB, 2004). Concurrently, as summarised in Ward and Dadayan (2009), state structural deficits began to be seriously examined in the mid-1990s. Factors such as growing Medicaid costs, increasing school enrolments, pension problems, infrastructure deterioration, and dysfunctional tax systems were identified as some of the variables causing the imbalances (Chapman, 2008). Additionally, the World Bank published a series of papers, culminating with Burnside (2005) writing the seminal piece in developing a formal fiscal sustainability model. He formally developed a model that uses solvency conditions as a measure of sustainability and derives a series of results that indicate the necessary conditions for a country to be fiscally sustainable. However, this model is strictly focused on national level accounting, with one of its key variables being the primary balance – the difference between total revenues and total expenditures. Further, the principle balance is not disaggregated, and leaves out most of the discussion of federalism.

In a recent working paper from the Federal Reserve Bank of Boston (Zhao and Coyne, 2013), the definition of fiscal sustainability has been enriched by the introduction of a trend gap and an explicit inclusion of social insurance and income maintenance programmes (as well as the continued inclusion of other post-employment benefits and pension contributions). Since social welfare programmes often have an intergovernmental component this is the beginning of a discussion of intergovernmental effects. However, this paper still continues to aggregate state and local revenues and only concerns itself with federal revenue transfers to this aggregate.

The second track utilises formal statistical work that focuses on fiscal sustainability. Beginning with Hamilton and Flavin (1986), and continuing through Mahdavi and Westerlund (2011), this track employs the cointegration and stationarity components of sustainability.¹Hamilton and Flavin (1986) examined solvency conditions at the national level. Afonso and Jalles (2011) recently examined solvency conditions for 19 countries. Both of these papers use stationarity properties to determine if sustainability exists. They find that it generally does, although there are some qualifications attached. Using stationarity and cointegration measures, Mahdavi and Westerlund (2011) and Raju (2011) were the first to apply these concepts to the sub-national levels. Mahdavi and Westerlund (2011) first examined the aggregates of state and local governments and find that conditions for sustainability exist for broad measures, although there is some temporising for some narrower measures. Raju (2011), in examining states in India, found that there should be some concern about future sustainability. Of note, in both of these papers, there was an explicit recognition of intergovernmental transfers, with the fiscal sustainability results being much more fragile for measures that exclude intergovernmental aid.

While this track is open to criticism for both statistical techniques and narrowness of focus (Burnside, 2005), we believe that a more robust understanding of fiscal sustainability occurs when the statistical track is properly used as a complement to the fiscal health track. A one year balanced budget is not necessarily a sign of long run fiscal sustainability and thus long run statistical analysis can give insight into both necessary and sufficient conditions for sustainability.

It is in this sense that our paper is both consistent with and builds upon the Mahdavi and Westerlund (2011) work as it disaggregates into counties and municipalities and considers different impacts of intergovernmental aid on these different types of governments. Further, we have not found any published statistical work that disaggregates the analysis of fiscal sustainability to the county and municipal level, nor have we found any extended analysis of the importance of intergovernmental revenues to fiscal sustainability at this level of disaggregation.

Intergovernmental fiscal relationships

Over 40 years ago, Oates (1972) formalised, from an economist’s perspective, a federal system of governance. However, financing this federal system has always been complex. There are at least four different ways that intergovernmental fiscal interdependencies can affect local government finance. The first is tax structure interdependencies. A higher level of government can change a tax system and in so doing, affect the revenue receipts of lower levels of government in several ways: changing the tax structure itself; changing the definition of the tax base; and, indirectly, changing national tax rates that might change the national level revenue collected which would influence the amount of revenue available for revenue sharing and grants. If piggybacking occurs, the results of any change can be exacerbated. Esteller-Moré and Solé-Ollé (2002) found that a tax increase by one layer of government leads to a decline in taxes collected by other layers of government. Ultimately, this leads to an increase in the tax rates of the lower level governments.

The second instance of intergovernmental fiscal dependence comes from grants and other revenue flows. These are often given in a relatively uncoordinated manner and often audited only in an accounting sense rather than on an efficiency of use sense. Many are virtually impossible for a non-expert to track. There are times when grants from the same central government agency go to different state agencies and then to different local agencies. These grants may be accounted for under different names and acronyms.

Third, the existing quasi-economic-political relationship between governments can be described as evidencing either a hard or soft budget constraint. Under the soft budget constraint, the sub-national governments can confidently expect that the higher-level governments will come to its aid when it experiences fiscal stress. When budget constraints are soft, the sub-national jurisdictions can increase expenditures without facing the full costs of that increase. A hard budget constraint is exactly opposite; the sub-national governments are convinced that a bail out will not occur. The basic governance of a decentralised federal system might reflect a game of attempting to extract bailouts (Chapman, 2007). To the extent that the sub-national governments believe that the higher-level governments will come to their aid, they will act in a non-sustainable fashion. This is the example of a soft budget constraint. The way to minimise the non-sustainable influences of this game might follow Inman’s (2003) recommendations of constitutional regulations that limit the higher-level governments’ ability to bail out the lower-level government. In the United States, the hard budget constraint is usually dominant, in that municipalities that have been egregiously sloppy in their fiscal discipline have been forced into bankruptcy.

Fourth, there is some evidence that the existence of federalism adds a dimension to economic growth, although its existence alone is an insufficient explanation for growth (Okpanachi, 2010). Additionally, there is also some evidence that decentralisation of social protection expenditures is positively correlated with economic growth (Ezcurra and Rodríguez-Pose, 2009). Economic growth should influence both the supply of revenues and the demand for public services, and therefore have an influence on future fiscal sustainability concerns. However, Reingewertz (2014) noted that there are large disagreements regarding the influence of fiscal decentralisation on economic growth, and it may well be that intergovernmental grants are heavily influenced by political considerations. While we acknowledge that all four of these interdependencies are important, we focus on the second dimension of intergovernmental fiscal interdependency and examine how it affects local fiscal sustainability in the United States.

Local governments in the US

The US Census Bureau (2013) defines five types of local government in the United States. General purpose governments consist of counties, municipalities, and townships, and special purpose governments consist of school districts and special districts. Special purpose governments vary according to the number of functions, organisation characteristics and patterns of fiscal autonomy. Additionally, townships are regarded as minor civil divisions that exist to provide services or administer an area without regard to population (US Census Bureau, 2013). Townships exist in 20 states, and some of them overlap cities while others do not. Because of these variations in special purpose governments and townships, they will not be included in our database. Finally, there are no county governments (although counties exist) in Connecticut, Rhode Island and the District of Colombia, and counties are called ‘parish’ governments in Louisiana and ‘borough’ governments in Alaska. These are treated as county governments by the Census. We focus on counties and municipalities in our analysis.

At the sub-national level, counties and municipalities exist in most states in an overlapping relationship. Because of this relationship, they are often represented in econometric work as an aggregate sub-unit of state governments without any effort to differentiate between them. This tendency is also reflected in the studies of fiscal sustainability; counties and municipalities have been researched conjointly as local governments (Mahdavi and Westerlund, 2011; Sørensen et al., 2001) or municipalities are the focus of research on total local government fiscal adjustment (Buettner and Wildasin, 2002). However, counties and municipalities differ in terms of their state-local fiscal relationships and therefore have different revenue and expenditure structures as shown in Table 1. State constitutional and statutory restrictions influence local governments’ revenue structures by constraining the amount of intergovernmental aid allotted to them as well as their revenue-raising decisions (Pagano and Johnston, 2000). In many cases, intergovernmental aid accounts for a much greater portion of revenues to counties than municipalities. Although counties and municipalities both provide local services to their residents, counties deliver many state services (e.g. social services) as the administrative arm of state, and are also often responsible for such activities as election administration, the maintenance and construction of rural roads, urban zoning, and the implementation of enforcement codes. Municipalities spend a large amount on public utility operations (from which they also receive a large amount of revenue), as well as spending a greater percentage than counties on environment and housing. Table 1 further identifies the different budget categories from the broad (total revenue and expenditure) to the narrow (general revenue and expenditure). General revenue is defined as total revenue less other revenues while general expenditures are defined as total expenditures less other expenditures. Following Mahdavi and Westerlund (2011) who were concerned about budget category fungibility, we argue that an additional reason for these classifications is that general revenue and general expenditures are more directly controlled by the governing body as well as more visible to the public. To the extent that they are more visible, sustainability should be weaker when these measures are isolated: taxes and fees are harder to raise, and direct services are harder to cut. Additionally, we consider the role of intergovernmental aid in fiscal sustainability separately for counties and municipalities, in part because it is about twice as important as a revenue source for counties as municipalities. Since state-government aid to a county sometimes includes federal aid that is passed through to the county or municipalities in a county’s jurisdiction (Stein, 1984), we consider the aggregate intergovernmental aid from federal, state and local governments.

Table 1.

Revenue and expenditure summary for county and municipal governments (2006).

Revenue items (%)			County	Municipality
Total revenue	General revenue	Property tax	24.26%	16.74%
		Sales tax	7.57%	9.68%
		Income tax and other taxes	3.38%	8.46%
		User charges and miscellaneous	24.01%	21.63%
		IG revenue from state	28.41%	14.31%
		IG revenue from federal/local	4.72%	5.67%
	Other	Utility, liquor store, and insurance trust revenue	7.64%	23.51%
Total			100%	100%
Expenditure items (%)			County	Municipality
Total expenditure	General expenditure	Education	15.41%	10.11%
		Social service	27.98%	6.93%
		Transportation	7.24%	8.62%
		Public safety	15.05%	16.60%
		Environment and housing	7.51%	16.22%
		Government administration	11.08%	6.37%
		Debt interest	3.42%	4.05%
		Other(unallocable/ miscellaneous)	7.82%	8.59%
	Other	Utility, liquor store, and insurance trust expenditure	4.49%	22.50%
Total (not including capital outlay)			100%	100%

Note: IG = intergovernmental; the percentage data are calculated by using these original data of the US Census Bureau.

Source: US Census Bureau (2007).

With these variations in revenue and expenditure structures, which are derived from varying state-local fiscal relationships, it can be assumed that there could be different patterns of fiscal sustainability in counties and municipalities and that different approaches to current and future budget operations are required for counties and municipalities.

Formal tests of sustainability

As noted earlier, many writers have reached the conclusion that fiscal sustainability requires that the government budget be balanced over time (Afonso and Jalles, 2012; Mahdavi and Westerlund, 2011; Quintos, 1995).² To operationalise this concept of an intertemporally balanced government budget, the present value of the existing stock of government debt should be equal to the present value of future primary surpluses (Afonso and Jalles, 2012; Mahdavi and Westerlund, 2011; Quintos, 1995). Otherwise, in the long-run, the value of debt grows faster than the growth of the real interest payment, and the unbalanced government debt would result in unsustainable debt accounts (Quintos, 1995). This intertemporally balanced budget implies that the discounted value of debt converges to zero at the limit as time goes to infinity (Quintos, 1995).

Two representative frameworks that analyse the intertemporally balanced budget exist in the statistical studies of fiscal sustainability. The first framework (Hamilton and Flavin, 1986; Wilcox, 1989) uses a unit root test to determine if the time series of revenues as well as the time series of expenditures are stationary, I(0). If there is no unit root, then the fiscal system is sustainable. If the time series of a variable has a unit root, the series is not sustainable; there will be no return to the trend line of either revenues or expenditures. This first framework assumes that stationarity (that is, there is no unit root) is an indication of a sustainable deficit policy, which implies fiscal sustainability.

Murray (1994) used a metaphor of the drunkard’s random walk to explain a non-stationary process. The non-stationarity of a random walk is characterised by a growing variance (Murray, 1994). To forecast random walks of a time series, the most recently observed value of the variable is ‘the best forecaster of future values’ (Murray, 1994: 37). In his illustrative example, observers in a bar are most capable of guessing where the drunkard is by remembering where she was most recently. The longer time the observers stay in the bar, the more likely the drunk wanders far from where the observers last saw her. We can connect the unchanged variance and the consequent resilience of stationary processes to the concept of sustainability in that sustainability assumes the long-term recovery of a balanced budget after an economic shock or crisis despite a financial loss in the short-term.

The second framework used to examine the sustainability concept is the cointegration test on government revenue and expenditure with the discounted debt (Hakkio and Rush, 1991). According to this cointegration framework, even if government revenue and expenditure series follow a non-stationary process, fiscal sustainability still holds if they are cointegrated with each other (Hakkio and Rush, 1991; Haug, 1991; Mahdavi and Westerlund, 2011; Quintos, 1995; Smith and Zin, 1991; Trehan and Walsh, 1988, 1991). Formally speaking, cointegration is a necessary condition for the government deficit to be intertemporally balanced in present value terms because cointegration makes a linear combination of revenue and expenditure stationary (Hakkio and Rush, 1991).

To explain the concept of cointegration, Murray (1994) continued the analogy by referring to the relationship between the drunkard and her dog. The drunkard and her unleashed dog wander aimlessly. Even though the dog is seemingly not in the drunkard’s control, it can smell each new scent of the drunkard that crosses his nose and dictates a direction for its next step (Murray, 1994: 37). Therefore, the drunkard and the dog wander aimlessly (non-stationary), but they make sure that they have an eye on each other and do not separate by more than a certain distance (a linear combination). In this way, the drunkard and the dog are cointegrated. In another illustration, the relationship between intergovernmental revenue aid and local government expenditures could be explained in terms of cointegration. Both revenue transfers and government expenditures may drop dramatically in the aftermath of an economic crisis, and their deviations from the trend line may increase over time, meaning that they are non-stationary. However, if the revenue aid is earmarked to certain public services and the corresponding government expenditures, they do not seem to drift too far apart from each other because of a long-run structural relationship. This is even possible despite the presence of non-stationarity in revenue and expenditure streams, and this implies cointegration between the two. Therefore, the cointegration result can be used to decide if government revenues and expenditures are moving together within a long-run relationship.

Strong and weak sustainability

As noted earlier, fiscal sustainability requires an intertemporally balanced government budget. Following Quintos (1995), Mahdavi and Westerlund (2011) demonstrate that the structure between government revenue and expenditure for fiscal sustainability can be shown by running the following regression using panel data where R_it is government revenue for the ith local government in period t, G_it is government expenditure inclusive of interest payments, and $ε_{it}$ is a mean zero error term.

R_{it} = α_{i} + β \cdot G_{it} + ε_{it}

(1)

In her interpretation of β in equation (1), Quintos (1995) extended the empirical framework on deficit sustainability by introducing ‘strong’ and ‘weak’ sustainability conditions. Quintos (1995) argued that strong sustainability corresponds to the strict requirements of sustainability in previous literature: stationarity in debt (Hamilton and Flavin, 1986) and cointegration between revenue and expenditure (Trehan and Walsh, 1991). Weak sustainability, which is both an extension and a relaxation of the condition for sustainability, considers the cointegration condition as a sufficient condition, but not as a necessary condition. In this case, she argued that the debt balance in a strong sustainability situation goes to zero at a rate faster than the debt in a weak sustainability situation (Quintos, 1995).

The underlying assumption for an intertemporally balanced budget is that the government should run a ‘future surplus equal to its current market value of debt’ (Quintos, 1995). In econometric terms, when time goes to infinity, the first differenced government deficit (ΔB) in expected present value terminology (E_t) converges to zero, as below:

\begin{matrix} E_{t} \lim_{j \to \infty} \frac{Δ B_{it + j}}{{(1 + r)}^{j + 1}} = 0 where \\ Δ B_{it + j} = (1 - β) G_{it} - α_{i} - ε_{it} \end{matrix}

(2)

where β must be greater than zero and less than two, 0 < β < 2, to ensure that convergence approaches zero.

Strong sustainability holds when there is no unit root in the time series of revenue and expenditure variables. However, even if both variables have unit roots, meaning non-stationarity, $Δ B_{it}$ still can be stationary when the necessary and sufficient conditions for sustainability are satisfied: (a) $ε_{it}$ is stationary, meaning R and G are cointegrated; and (b) $β = 1$ (Quintos, 1995). As we can see in equation (2), since the expenditure variable that is non-stationary becomes zero in a condition of $β = 1$ , $Δ B_{it}$ can be stationary if $ε_{it}$ is stationary. Therefore, strong sustainability can be held either when both R and G follow stationary processes or R and G are cointegrated with $β = 1$ .

In defining ‘weak sustainability’, Quintos (1995) relaxed the condition of strong sustainability. She argued that $0 < β \leq 1$ in equation (1) is both a necessary and sufficient condition, and cointegration is just a sufficient condition for weak sustainability. As long as $β$ is between zero and one, $0 < β < 1$ , the condition of weak sustainability still holds regardless of the cointegration condition (Quintos, 1995). In this case, the process ΔB_it can follow a trend including a unit root, but it still goes to zero at a rate slower than strong sustainability. Therefore, according to Quintos (1995), when $β = 1$ , weak sustainability can be satisfied without cointegration, whereas strong sustainability can be satisfied with cointegration.

In this process of testing sustainability, Quintos (1995) assumed that revenue and expenditure are I(0) or I(1), which means that debt is integrated of at most second order, I(2). Because of this assumption, her framework has been criticised since it is only applicable to limited classes of stochastic processes (Bohn, 2007). Theoretically, debt can be integrated of higher order than second order, and Bohn (2007) showed that sustainability holds as long as debt is stationary in any finite number of orders of integration. Nevertheless, the Quintos’ framework is still useful for several reasons: it is empirically relevant to conceive cases where debt is I(1) or I(2) (Mahdavi and Westerlund, 2011); her cointegration test can provide support to the sustainability concept; and the strong and weak sustainability concepts explain departures from the benchmark case of β = 1 (Gabriel and Sangduan, 2011).

To examine whether it is strong or weak fiscal sustainability, the regression results from equation (1) provide various explanations of sustainability according to the values of the estimated β. To investigate these possibilities, we follow the Quintos’ (1995) tests of sustainability. The tests first begin with unit root tests of the revenue and expenditure variables, which indicate whether or not the variables are stationary. If the null hypothesis of the unit root test is rejected, then the tested variable is stationary. For fiscal sustainability, if both variables are stationary or I(0), then that strong sustainability holds. If the null is accepted, the tested variable is non-stationary or I(1), which is a necessary condition for the cointegration regression described in equation (1) above.

Next, an additional t-test must be conducted on the values of β from equation (2) to distinguish between strong and weak sustainability: β = 1 with cointegration for strong sustainability and 0 < β < 1, regardless of cointegration, for weak sustainability. In this situation, the first null hypothesis is β = 0 when the one-sided alternative is β > 0. If the null is accepted and β = 0, it means that deficit is growing and the conditions of fiscal sustainability are not satisfied. If the null hypothesis of β = 0 is rejected, then a second null hypothesis of β = 1 is tested with the two-sided alternative of β≠ 1. If the null is rejected to the left, which means 0 < β < 1, then sustainability is weak because expenditures are growing faster than revenues. If the null is rejected to the right, meaning that β > 1, then revenues are growing faster than expenditures. When the null of β = 1 is accepted, then the condition of cointegration becomes meaningful; if cointegration holds, this indicates strong sustainability; if not, this means weak sustainability. This implies that cointegration is a necessary condition of strong sustainability and a sufficient condition of weak sustainability.

To summarise, if β = 1 and the cointegration results imply stationarity, then there is strong sustainability. However, if the cointegration results imply non-stationarity, then there is weak sustainability. If 0 < β < 1 under both stationarity and non-stationarity, there exists weak sustainability. Finally, if 1 < β < 2 regardless of the cointegration results, we have a situation of sustainability which is unstable. Figure 1 summarises these relationships.

Figure 1.

Tests for fiscal sustainability.

Empirical approach

To test fiscal sustainability in counties and municipalities, we used the Annual Survey of Local Government Finances data from the US Census Bureau. The data set includes 610 counties and 671 municipalities for 37 years (1970–2006), and thus consists of 22,570 observations for counties and 22,354 observations for municipalities. The panel is unbalanced, meaning that the sample size of each local government varies, and the time series of each entity is not always 37 years.³

In order to investigate the importance of intergovernmental aid, we create four different measures of governmental balances that are calculated with and without intergovernmental aid in revenue side. The following Table 2, which is based on Table 1, defines these four different measures. Two measures relate to the primary balance analysis mentioned earlier, the jurisdiction’s total revenues and total expenditures (Measure 1) and general revenues and general expenditures (Measure 3). The other two measures subtract intergovernmental revenues from total revenues and general revenues, Measure 2 and Measure 4 respectively, thus attempting to determine the impacts of intergovernmental aid on the local jurisdictions’ fiscal sustainability. Intergovernmental aid is measured as the sum of intergovernmental revenues from federal, state, and local governments. As noted earlier, these four measures are used because: (a) they relate to the primary balance analysis used in the literature; (b) any further disaggregation would eliminate the possibility of fungibility among service functions; and (c) the attempt isolates the effects of intergovernmental transfers. Table 3 shows the descriptive statistics for these variables, calculated using the described data sets.

Table 2.

Four measures of governmental balances by county and municipality.

Measure	County and municipality
Measure 1	Total revenue (TR) and total expenditure (TE)
Measure 2	[Total revenue – total intergovernmental revenue (IGR)] and total expenditure
Measure 3	General revenue (GR) and general expenditure (GE)
Measure 4	[General revenue – total intergovernmental revenue (IGR)] and general expenditure

Table 3.

Descriptive statistics for revenue and expenditure variables by county.

	1970				2006
	Mean (SE)	Median	Min	Max	Mean (SE)	Median	Min	Max
County (N = 610)					County (N = 610)
TR	4273 (80,678)	1290	7.68	337,669	430,033 (1,131,625)	151,141	1070	2.16e+07
TE	4421 (83,945)	1324	7.68	351,998	413,107 (1,021,840)	148,273	2118	1.87e+07
GR	4196 (78,472)	1324	7.68	327,685	399,379 (948,427)	146,437	1070	1.70e+07
GE	4349 (82,434)	1300	7.68	345,649	392,704 (934,449)	144,898	2118	1.69e+07
TR-IGR	2570 (43,437)	768	7.49	173,050	280,636 (708,633)	96,584	1069	1.24e+07
GR-IGR	2493 (41,276)	763	7.49	163,066	249,982 (543,833)	91,428	1069	0.78e+07
IGR	1703 (38,167)	385	0	164,619	149,397 (457,247)	45,879	0	0.917e+07
Municipality (N = 634)					Municipality (N = 381)
TR	7858 (312,244)	2000	0	1,469,837	806,042 (4,430,654)	241,227	31,017	8.35e+07
TE	8237 (323,003)	2143	0	1,518,605	771,169 (4,346,130)	239,878	30,161	8.25e+07
GR	6596 (269,568)	1618	0	1,271,435	634,736 (3,719,231)	199,444	27,649	7.08e+07
GE	6833 (269,983)	1735	0	1,270,819	604,317 (3,475,376)	187,008	27,713	6.62e+07
TR-IGR	5776 (191,750)	1717	0	887,055	613,709 (3,105,720)	188,152	24,359	5.76e+07
GR-IGR	4514 (148,784)	1322	0	688,654	442,404 (2,381,281)	142,336	21,498	4.49e+07
IGR	2082 (122,073)	314	0	582,782	192,332 (1,352,708)	39,273	1655	2.60e+07

Note: All values are in thousands of dollars and 1970 values are adjusted for inflation.

Empirical results

The first step of empirical analysis is to examine if the times series of revenues and expenditures are stationary by using the unit root test. The presence of stationarity in both variables implies that strong fiscal sustainability exists. If they are not stationary, then their relationships need to be further analysed through cointegration analysis. In this paper, we apply the unit root test of Pesaran’s (2007) cross-sectionally augmented Dickey Fuller (CADF) panel unit root test that allows for cross-section dependence in panel data.

Table 4 reports the results of the panel unit root test and shows that the unit root null hypothesis can be rejected at the conventional significance levels (α = 1%, 5%) for the variables of TR, TE, and GE in counties and TR, TR-IGR, TE, and GE in municipalities. In the case of municipalities, TR is stationary even after subtracting intergovernmental revenue aid and, therefore, is sustainable regardless of external aid. However, counties become non-stationary when intergovernmental revenue aid is subtracted from TR, and thus the series TR-IGR is not sustainable. For the variables of GR and GR-IGR in both counties and municipalities, the null hypothesis cannot be rejected at any conventional significance level, and these variables are not stationary. Unit roots in these variables disappear when taking a form of the first differenced, which means that they all are integrated of order one, I(1).

Table 4.

Pesaran’s (2007) Dickey Fuller (CADF) panel unit root test results.

Variables		County (N = 610, T = 37)		Municipality (N = 671, T = 37)
		Test value (Z_tbar)	p-value	Test value (Z_tbar)	p-value
Measure 1	TR	−2.620	0.005	−3.852	0.000
Measure 1	TE	−6.624	0.000	−8.386	0.000
Measure 2	TR-IGR	4.543	1.000	−1.660	0.048
Measure 2	TE	−6.624	0.000	−8.386	0.000
Measure 3	GR	0.199	0.579	−1.439	0.075
Measure 3	GE	−5.519	0.000	−10.001	0.000
Measure 4	GR-IGR	0.996	0.160	1.656	0.951
Measure 4	GE	−5.519	0.000	−10.001	0.000

Note: The unit root test is implanted with a constant and trend in the test regressions and takes a unit root as the null hypothesis.

We next proceed by analysing the relationship between revenues and expenditures. As shown, TR and TE are both stationary, meaning fiscal sustainability, and thus additional tests for Measure 1 are not necessary. Since TR-IGR is non-stationary only for counties, further analysis will be conducted on Measures 2, 3, and 4 for counties while Measures 3 and 4 will be considered for further analysis for municipalities. Despite non-stationarity, strong sustainability can be held under the two conditions of: (a) cointegration; and (b) β = 1. Therefore, we can first test these non-stationary variables for cointegration by checking one of the conditions for strong sustainability and then test whether β = 1 given the cointegration condition. As the unit root test shows, the revenue and expenditure variables in Measures 2, 3, and 4 are integrated by different orders, with GE being integrated of order zero, I(0), and TR-IGR (counties only), GR and GR-IGR integrated of order one, I(1). For this reason we cannot use the traditional cointegration tests that assume all variables are integrated of a same order. We addressed this limitation by using the Autoregressive Distributed Lag (ARDL) bounds test developed by Pesaran et al. (2001), a widely used cointegration test for variables integrated with different orders. The bounds test is based on the joint F-statistic that shows asymptotic distribution under the null hypothesis of no cointegration. The null hypothesis is rejected when a value of the test statistic exceeds the upper critical bound value and is accepted if the test statistic falls below the lower bound value. Since the cointegration test raises a question of the relevant number of lags, we use the Akaike Information Criteria (AIC) which determines either one or two lag orders for variables. We report cointegration results in all cases of lag order of one, two, and three as shown in Table 5. The results imply cointegrated relationships between revenue and expenditure variables in all three lag order cases.

Table 5.

ARDL bounds test results.

Measures	County			Municipality
	Lags	F-statistic	Decision	Lags	F-statistic	Decision
Measure 2	1	3720.02	Cointegration	N/A
	2	2753.44	Cointegration
	3	1623.03	Cointegration
Measure 3	1	3308.20	Cointegration	1	3643.79	Cointegration
	2	1811.65	Cointegration	2	2394.98	Cointegration
	3	1190.68	Cointegration	3	1468.56	Cointegration
Measure 4	1	2639.11	Cointegration	1	3215.41	Cointegration
	2	1808.53	Cointegration	2	2297.84	Cointegration
	3	1231.34	Cointegration	3	1261.21	Cointegration

We next undertake t-tests for distinguishing strong sustainability (β = 1) and weak sustainability (0 < β <1). We estimate regression equation (1) and use a t-test for the null hypothesis that β = 0 versus the one-sided alternative that β > 0. If we do not reject the null hypothesis, then the relevant system is not sustainable. Table 6 indicates that all of the null hypotheses (β = 0) are rejected. The next step is to test β = 1 versus the two-sided alternative of β≠ 1. If we accept the null of β = 1, then there is strong sustainability because we can satisfy both conditions of the cointegration and β = 1. Even if we reject the null to the left, then β is between 0 and 1, thereby satisfying the condition of weak sustainability. According to Mahdavi and Westerlund (2011), 0 < β < 1 implies that the jurisdiction is sustainable now, but expenditures are growing more rapidly than revenues. If the null is rejected to the right, the jurisdiction is sustainable as long as β is between 1 and 2, 1 < β < 2. In this case, revenues are growing more rapidly than expenditures. However there is a strong possibility that this situation will not exist in the long run because of political pressures to either reduce revenues or increase expenditures to prevent a growing surplus. This is consistent with the Thompson and Gates (2007) argument that sustained revenue growth tends to encourage unsustainable spending increases. We call this situation fiscally balanced to differentiate it from the traditional fiscal sustainability concept.

Table 6.

A summary of strong and weak sustainability results.

Measure	County	Municipality
Measure 2 [(TR-IGR)_TE]	β = 0.5914 (0. 0015643) → Weak sustainability	N/A
Measure 3 [GR_GE]	β = 1.0187 (0.0007164) → Fiscally balanced	β = 1.0162 (0 .000712) → Fiscally balanced
Measure 4 [(GR-IGR)_GE]	β = 0.5165 (0. 0015589) → Weak sustainability	β = 0.64453 (0. 00069) → Weak sustainability
N	22,570 (610 counties for 37 years)	22,354 (671 municipalities for 37 years)

Note: Standard deviation in parenthesis.

As the results in Table 6 indicate, counties and municipalities are fiscally balanced for Measure 3; each $1 change in expenditures is associated with revenues going up by about $1.02 for counties and municipalities. This means that revenues are growing faster than expenditures, and therefore it is not the case of traditional fiscal sustainability. However, it can be fiscally balanced since the expenditure and revenue series are cointegrated with the revenue growth being less than two (1 < β < 2), meaning the revenue surplus does not increase indefinitely over time. None of them accept the null of β = 1, and therefore, strong sustainability does not hold in these measures for both counties and municipalities. However, both counties and municipalities become weakly sustainable for Measure 4; in the case of subtracting intergovernmental aid, revenues increase by about 52 cents for counties and 64 cents for municipalities for each $1 change in expenditures. Similarly, this is the case of weak sustainability since the change in revenues is greater than zero (0 < β < 1) and the expenditure and revenue series are cointegrated. In this analysis, without intergovernmental revenues, debt balances are always weakly sustainable at best.

The analysis results of county governments are roughly consistent with the result from Mahdavi and Westerlund (2011: 963). They found ‘weak sustainability’ in the narrowly defined balance, which is the balance of own-source general revenues without intergovernmental aid less current operation expenditures. According to them, the different level of sustainability in the fiscal balance reveals ‘a potential area of vulnerability masked by relatively broader balances including intergovernmental grants’.

Table 7 summarises overall findings of fiscal sustainability analysis for the four sets of revenue and expenditure measures by counties and municipalities.

Table 7.

Overall findings of fiscal sustainability analysis.

Measures		County	Municipality
TR; TE	Measure 1 TR;TE	No unit root → Fiscally sustainable	No unit root → Fiscally sustainable
TR; TE	Measure 2 TR-IGR;TE	Weakly sustainable	No unit root → Fiscally sustainable
GR; GE	Measure 3 GR;GE	Fiscally balanced	Fiscally balanced
GR; GE	Measure 4 GR-IGR;GE	Weakly sustainable	Weakly sustainable

Implications and future research

There are at least three implications of these results. First, there is a different level of fiscal sustainability between the four measures. As we suggested, using the narrower measures of GR and GE might make the local governments slightly less sustainable. This implication should lead to restrained budgetary behaviour on the part of counties and municipalities. If political leaders in either type of jurisdiction only examine total revenues and total expenditures, which include intergovernmental grants and other revenue sources, strong sustainability is easily shown. Therefore, they may believe that they are in a fiscally sustainable situation and may embark on polices, ranging from new expenditures on urban projects to cutting local taxes. However, when examining their fiscal positions from the general revenue–general expenditure perspective, they become politically less sustainable in that their revenues are growing faster than expenditures, which could lead to a political backlash against the revenue surplus and corresponding increases in spending. The appropriate behaviour for local government is to engage in cautious fiscal policies. Second, the intergovernmental aid has a significant impact on fiscal sustainability of local governments. This may occur because of the existence of soft budget constraints between the state and local governments which enable the local governments to ensure that their principle balance is never negative. It is interesting to observe that while the aggregate measures of total revenues and total expenditures are examined, strong fiscal sustainability still holds even after subtracting the intergovernmental aid (in case of municipalities). However, when the narrower definitions of revenues and expenditures are used, the jurisdictions become less sustainable (fiscally balanced at best) and run a surplus with the intergovernmental aid. In these cases, there may be some danger in their losing intergovernmental revenues because the state may believe that the local governments are too profligate, which may turn the jurisdiction’s fiscal balance into weak sustainability. The third implication is that when intergovernmental revenues are subtracted, fiscal sustainability becomes weaker for counties but does not change much for municipalities. A possible explanation for this result is that counties are often the administrative arm of the state and, therefore, if the state does not relax county responsibilities but does reduce intergovernmental aid, the fiscal sustainability of counties will suffer (Paulais, 2009).

To some extent, our findings that intergovernmental aid makes municipalities and counties sustainable make perfect sense. Local governments must take this aid into account when they make revenue and expenditure decisions. If this aid disappears, it should have precisely the effects that the statistical findings indicate although municipalities seem to be less affected than counties.⁴

Future research should examine several additional questions. First, what is the precise role of state mandates on counties and municipalities? This stream of research would include such questions as the differential effects of mandates versus across-the-board cuts when expenditures are reduced. Additional mandate questions could include investigating if mandated functions crowd-out other functions in order to maintain even weak sustainability or do mandates force sub-state governments to protect social services while other services become even more stressed? A second set of questions could address the variability of the estimated betas across geographic areas, across different population densities, or by geographic arenas. A third set of future questions comes from the findings of weak sustainability at the municipal level when intergovernmental revenues are eliminated. If intergovernmental aid is reduced, any tax and expenditure constraints on local spending become increasingly important. This could lead to a reduction in economic growth because of the reduction in infrastructure expenditures, or current service level expenditures. The endogeneity of revenues and expenditures could lead to a downward spiral, stimulated by the reduction in intergovernmental aid. This may be a strong downside risk to the budget constraints in our intergovernmental system. We will try to answer the questions raised in these potential extensions of the model when the data for such analysis become available.

Footnotes

Acknowledgements

This paper was presented at the Association for Public Policy Analysis and Management (APPAM) conference and the National Tax Association (NTA) conference. The authors would like to thank all the panel participants and the audience members for their helpful comments. We also wish to thank the anonymous referees for their suggestions.

Funding

This research received no specific grant from any funding agency in the public, commercial, or not-for-profit sectors.

Notes

References

Afonso

Jalles

(2011) A longer-run perspective on fiscal sustainability. ISEG working paper 2011/17. Department of Economics at the School of Economics and Management (ISEG), Technical University of Lisbon, Portugal.

Afonso

Jalles

(2012) Revisiting fiscal sustainability: Panel cointegration and structural breaks in OECD countries. European Central Bank working paper series. European Central Bank. Available at: http://www.ecb.europa.eu/pub/pdf/scpwps/ecbwp1465.pdf (accessed 27 August 2014).

Alexander

Barrow

(1994) Seasonality and cointegration of regional house prices in the UK. Urban Studies31(10): 1667–1689.

Baltagi

Song

(2006) Unbalanced panel data: A survey. Statistical Papers47(4): 493–523.

Bohn

(2007) Are stationarity and cointegration restrictions really necessary for the intertemporal budget constraint?Journal of Monetary Economics54(7): 1837–1847.

Buettner

Wildasin

(2002) The dynamics of municipal fiscal adjustment. CESifo working paper no. 649(1). CESifo Group, Munich, Germany.

Burnside

(2005) Fiscal Sustainability in Theory and Practice. Washington, DC: The World Bank.

Canarella

Miller

Pollard

(2012) Unit roots and structural change: An application to US house price indices. Urban Studies49(4): 757–776.

Chapman

(2007) Wealth and tax systems. In: Pagano

Leonardi

(eds) The Dynamics of Federalism in National and Supranational Political Systems. New York: Palgrave MacMillian, pp. 107–135.

10.

Chapman

(2008) State and local fiscal sustainability: The challenges. Public Administration Review68(S1): S115–S131.

11.

Chen

Guo

Zhu

(2011) The housing-led growth hypothesis revisited: Evidence from Chinese Provincial Panel Data. Urban Studies48(10): 2049–2067.

12.

Clark

Etile

(2002) Do health changes affect smoking? Evidence from British panel data. Journal of Health Economics21(4): 533–562.

13.

Cohen

(2011) Cities in times of crisis: The response of local governments in light of the global economic crisis: the role of the formation of human capital, urban innovation and strategic planning. Working paper 2011–01. Berkeley Institute of Urban and Regional Development, University of California, CA, USA. Available at: http://escholarship.org/uc/item/3432p4rb (accessed 22 June 2015).

14.

Esteller-Moré

Solé-Ollé

(2002) Tax setting in a federal system: The case of personal income taxation in Canada. International Tax and Public Finance9(3): 235–257.

15.

Ezcurra

Rodríguez-Pose

(2009) Decentralization of social protection expenditure and economic growth in the OECD. Publius41(1): 146–157.

16.

Gabriel

Sangduan

(2011) Assessing fiscal sustainability subject to policy changes: A Markov switching cointegration approach. Empirical Economics41(2): 371–385.

17.

Gold

(1995) The Fiscal Crisis of the States: Lessons for the Future (vol. 10). Washington, DC: Georgetown University Press.

18.

Government Accounting Standards Board (GASB) (2004) Economic condition reporting: The statistical section (statement no. 44). Available at: http://www.gasb.org/cs/ContentServer?c=GASBContent_C&pagename=GASB/GASBContent_C/ProjectPage&cid=1176156646154 (accessed 5 September 2014).

19.

Government Accounting Standards Board (GASB) (2009) GASB adds project addressing fiscal sustainability to current technical agenda: The user’s perspective. Available at: www.gasb.org/newsletter/fiscal_sustainability_dec2009.html (accessed 27 August 2014).

20.

Government Accounting Standards Board (GASB) (2011) Preliminary views of the governmental accounting standards board on economic condition reporting: Financial projections. Governmental Accounting Standard series no. 13–3. Available at: www.gasb.org/cs/ContentServer?site=GASB&c=Document_C&pagename=GASB%2FDocument_C%2FGASBDocumentPage&cid=1176159500453 (accessed 27 August 2014).

21.

Hakkio

Rush

(1991) Is the budget deficit too large?Economic Inquiry29(3): 429–445.

22.

Hamilton

Flavin

(1986) On the limitations of government borrowing: A framework for testing. American Economic Review76(4): 808–819.

23.

Haug

(1991) Cointegration and government borrowing constraints: Evidence for the United States. Journal of Business & Economic Statistics9(1): 97–101.

24.

Hendrick

(2011) Managing the Fiscal Metropolis. Washington, DC: Georgetown University Press.

25.

Horne

(1991) Indicators of fiscal sustainability. IMF working paper 91/5. International Monetary Fund, Washington, DC, USA.

26.

Inman

(2003) Transfers and bailouts: Enforcing local fiscal discipline with lessons from US federalism. In: Rodden

Eskeland

Litvack

(eds) Fiscal Decentralization and the Challenge of Hard Budget Constraints. Cambridge: The MIT Press, pp. 35–83.

27.

Karuppusamy

Carr

(2012) Interjurisdictional competition and local public finance: Assessing the modifying effects of institutional incentives and fiscal constraints. Urban Studies49(7): 1549–1569.

28.

Levine

(1980) Managing Fiscal Stress: The Crises in the Public Sector. Chatham: Chatham House Publishers.

29.

Mahdavi

Westerlund

(2011) Fiscal stringency and fiscal sustainability: Panel evidence from the American state and local governments. Journal of Policy Modeling33(6): 953–969.

30.

Murray

(1994) A drunk and her dog: An illustration of cointegration and error correction. The American Statistician48(1): 37–39.

31.

Oates

(1972) Fiscal Federalism. New York: Harcourt Brace Jovanovich.

32.

Okpanachi

(2010) Federalism and economic growth: The importance of context in Nigerian public finance Reform. Publius41(2): 311–335.

33.

Pagano

Johnston

(2000) Life at the bottom of the fiscal food chain: Examining city and county revenue decisions. Publius30(1): 159–170.

34.

Paulais

(2009) Local Governments and the Financial Crisis: An Analysis. Washington, DC: The Cities Alliance.

35.

Pesaran

(2007) A simple panel unit root test in the presence of cross-section dependence. Journal of Applied Econometrics22: 265–312.

36.

Pesaran

Shin

Smith

(2001) Bounds testing approaches to the analysis of level relationships. Journal of Applied Econometrics16: 289–326.

37.

Quintos

(1995) Sustainability of the deficit process with structural shifts. Journal of Business & Economic Statistics13(4): 409–417.

38.

Raju

(2011) Analyzing the fiscal health of state governments in India: Evidence from the 14 major states. Public Budgeting and Finance31(3): 49–72.

39.

Reingewertz

(2014) Fiscal decentralization: A survey of the empirical literature. Available at: http://papers.ssrn.com/sol3/papers.cfm?abstract_id=2523335 (accessed 12 January 2014).

40.

Smith

Zin

(1991) Persistent deficits and the market value of government debt. Journal of Applied Econometrics6(1): 31–44.

41.

Sørensen

Yosha

(2001) Output fluctuations and fiscal policy: US state and local governments, 1978–1994. European Economic Review45(7): 1271–1310.

42.

Stein

(1984) Municipal public employment: An examination of intergovernmental influences. American Journal of Political Science28(4): 636–653.

43.

Thompson

Gates

(2007) Betting on the future with a cloudy crystal ball? Revenue forecasting, financial theory, and budgets. Public Administration Review67(5): 48–66.

44.

Trehan

Walsh

(1988) Common trends, the government’s budget constraints, and revenue smoothing. Journal of Economic Dynamics and Control12(2–3): 425–444.

45.

Trehan

Walsh

(1991) Testing intertemporal budget constraints: Theory and applications to US federal budget and current account deficits. Journal of Money, Credit, and Banking23(2): 206–223.

46.

US Census Bureau (2007) US Census of governments: Local summary tables by type of government, 2007. Available at: https://www.census.gov//govs/local/historical_data_2007.html.

47.

US Census Bureau (2013) Individual state descriptions: 2012. Washington, DC: US Government Printing Office.

48.

Verbeek

Nijman

(1992) Testing for selectivity bias in panel data models. International Economic Review33: 681–703.

49.

Ward

Dadayan

(2009) State and local finance: Increasing focus on fiscal sustainability. Publius39(3): 455–475.

50.

Wilcox

(1989) The sustainability of government deficits: Implications of the present-value borrowing constraint. Journal of Money, Credit, and Banking21(3): 291–306.

51.

World Commission on Environment and Development (1987) Our Common Future (Brundtland Report). Oxford: Oxford University Press.

52.

Zhao

Coyne

(2013) Walking a tightrope: Are US state and local governments on a fiscally sustainable path? Working paper no. 13–18. Federal Reserve Bank of Boston, MA, USA.