The Role of State Revenue Structure in the Occurrence and Magnitude of Negative Revenue Variance

Abstract

A negative revenue variance (also known as a revenue shortfall) is generated when the actual inflow of revenue falls short of the budgeted revenue. In an environment constrained by a balanced budget requirement, a negative revenue variance may result in a compensating cut in program expenditures. As such, it is imperative to explore the drivers of negative revenue variance. To answer these questions, we take a look at the states’ revenue mix, specifically, the diversification and elasticity of a state’s revenue structure. We establish a quantitative model to capture factors that affect the occurrence and magnitude of negative revenue variance. Our findings suggest that revenue diversification reduces both the occurrence and the size of a negative revenue variance. Elasticity, on the contrary, increases the occurrence but reduces the magnitude of the negative revenue variance. These findings provide additional evidence for the importance of fiscal planning and design of revenue structure that includes consideration of both diversification and elasticity of the revenue portfolio. Specifically, elasticity and diversification can be used in tandem to address an existing revenue shortfall.

Keywords

revenue structure revenue diversification revenue elasticity revenue variance

Introduction

The budget document is the culmination of both a technical and political process. On the technical side, costs are estimated, and revenues are forecasted. On the political side is the budget approval. In a political environment of scarce resources, not every program or organization will receive all of their requested funding. Therefore, the technical and political gymnastics of the budget process is ultimately about the government provision of programs and services.

Forecasting revenue flow is a critical part of the budget process. Thirty-one states are on an annual budget calendar, meaning the revenue forecasts may be 18 months out. For the remaining states on a biennial budget cycle, budget revenue forecasts can be 30 months out (Snell, 2011). There are months of negotiations and maneuvering within agencies, between agencies and the governor’s office or central budget office, between the governor’s office and the legislature, and within the legislative branch before adopting the budget. All of this takes place within the constraints of revenue projections, which are based on previous revenue flow and forecasted economic conditions. Throughout the execution phase of the budget process, the realities of the true economic conditions become clear as the economy progresses and revenues materialize. As revenues flow in, a variance usually occurs, which may lead to a mid-year adjustment to the original adopted budget. If the variance is due to lower than expected revenue inflow, then the expenditures may need to be adjusted downward. If the revenue inflow is higher than originally budgeted, then expenditures may be adjusted upward. These mid-year adjustments, also referred to as rebudgeting, can directly impact the provision of programs and services.

Although adjustments during a fiscal year are a regular and common part of the budget execution process (Cornia, Nelson, & Wilko, 2004; Dougherty, Klase, & Song, 2003; Forrester & Mullins, 1992), there are two primary reasons established in the literature. First, a common reason for mid-year budget adjustments is unintended misestimation of revenue flow. The level of misestimation can vary by revenue source (Rubin, 1987). Sometimes the misestimation is from unanticipated economic conditions (Dougherty et al., 2003). This type of adjustment is technical in nature. Second, some adjustments are intentional, managerial decisions. In particular, decision makers sometimes elect to underestimate anticipated revenue in a proactive attempt to avoid an imbalance or reserve additional funds for other uses (Dougherty et al., 2003).

Regardless of the reasons for a budget variance, implications are far reaching and often political. Forrester and Mullins (1992) explain how the rebudget process is not as visible as the official budget process. Reacting to necessary cuts due to revenue shortfalls appears technical and reactionary to revenue status and, therefore, can hide intentions to cut programs (Rubin, 1987). Cornia et al. (2004) explain how rebudgeting can serve to undermine the budget process. Because funds may ultimately be spent in ways not originally legislated or budgeted, legislative oversight and control are diminished, and management’s programmatic intentions may be undercut or shifted. Furthermore, the benefit of policy analyses performed as part of the creation of the original budget may be circumvented. Of course, all of these rebudgeting consequences are realized out of the limelight of media, general public, and interest group participation.

The purpose of this study is to examine the factors leading to the occurrence of negative revenue variances and their magnitude. We examine several factors that can influence revenue flow and subsequently lead to a negative variance by taking a look at the states’ revenue portfolio, specifically, the diversification and elasticity of state revenue systems. Although the consequence of a positive variance also has its own managerial and political implications and programmatic impact, we are particularly interested in the negative variance because of state balanced budget requirements and the depth and breadth of the Great Recession. We examine the influence of the revenue structure on the negative variance because of the potential for a substantial negative programmatic impact. For our analysis, we define revenue variance as the actual total revenue minus the original budgeted total revenue as reported in the Budgetary Comparison Schedule—General Fund of the state comprehensive annual financial reports (CAFRs).

The article is structured as the following: The next section offers a brief review of the literature on revenue structure with a special focus on revenue diversification and revenue elasticity. Then we outline the conceptual framework and methods and summarize the empirical findings. Lastly, we provide concluding remarks.

Revenue Structure

The Changing Revenue Structure

Ultimately, how a state’s revenue sources are combined is the result of a political process aimed at raising adequate revenue for planned programs and services. The type of revenue sources, the reliance on those sources, and the policy for each source result in the revenue mix or structure for the state.

State revenue structure is directly impacted by tax and expenditure limits (TELs). The late 1990s ushered in the latest wave of TELs, which were advocated by citizens and elected officials to restrain government spending. While findings of TELs impact on government spending have varied, the resulting effort, especially the effort to limit taxes, has impacted revenue structure. The limits force policy makers to find other sources of revenue that are not subject to the limitation (Johnston, Pagano, & Russo, 2000; Joyce & Mullins, 1991; Shadbegian, 1999). Focusing on state revenue structure, Kioko and Martell (2012) find that procedural limits, such as voter approval, limit the state’s ability to raise taxes. Wang (2012) finds that TELs lead to a regressive revenue structure due to increased reliance on sales taxes and user fees.

As evidence of the changing state revenue structure, a review of State Government Finances reveals that there has been less reliance on tax revenue and a greater reliance on charges since 2000. State tax revenue increased 40.9% from 2000 to 2011. As a percent of total revenue, taxes have gone from 42.8% in 2000 to 33.6% in 2011. Specifically, state general sales tax revenue has declined from 13.8% to 10.4% of total revenue. Selective sales taxes have gone from 6.16% to 5.8% of total revenue. Individual income taxes have dropped from 15.4% to 11.5% of total revenue. In the meantime, revenue from a nontax source, current charges, has increased 109.4%, which is an increase in reliance from 6.9% to 8.0% of total revenue (U.S. Census, 2012).

The Diversification of the Revenue Structure

Diversification is a characteristic of a jurisdiction’s revenue structure that is often analyzed. The fundamental idea is that a diversified revenue structure is more stable and predictable much like a diversified investment portfolio, which is drawn from the portfolio theory in corporate finance (Brealey & Myers, 1991). The portfolio of revenue sources is protected from the fluctuations of one particular revenue source. This assumes that the movement of one revenue source is not perfectly correlated with another revenue source; therefore, different revenues are not responding the same to changes in the economy (White, 1983). Conceptually, this portfolio stability makes planning and budgeting of government services more predictable.

This risk-reducing portfolio theory has not been greatly tested in public financial literature; therefore, the specific impact on revenue stability has not been clearly determined. White (1983) finds that the increased instability in a state’s tax revenue is due to less diversification marked by more reliance on unstable taxes such as personal income and corporate income taxes. Carroll (2009) finds that diversification’s impact on stability depends on how the structure is diversified. Yan (2012) finds that diversification’s impact on revenue stability depends on the stability of the economy. Chernick, Langley, and Reschovsky (2011) find that less diversification can lead to stability for large cities if it is a result of an increased reliance on the stable property tax.

The impact of diversification on other measures of fiscal performance has varied. Suyderhoud (1994) finds that diversification is associated with higher tax effort. Jordan and Wagner (2008) find this to be the case for small cities due to their limited revenue options. However, Hendrick (2002), Hannarong and Akoto (2004), and Pagano and Johnston (2000) find that diversification can lead to less of a tax effort if the method of diversification was to increase the use of nontax revenue sources.

Similarly, how a jurisdiction diversifies its revenue portfolio will determine the tax incidence (Ladd & Weist, 1987). Suyderhoud (1994) finds that diversification is associated with a less regressive system when analyzing state-local revenue. In analyzing small cities in Arkansas, Jordan and Wagner (2008) find that increased diversification results in increased use of the regressive sales tax.

Suyderhoud (1994) states that revenue adequacy is not directly measurable and “inferred” it by using Moody’s general obligation debt rating. He finds diversification is positively related to revenue adequacy. Jordan and Wagner (2008) use another measure of revenue adequacy, the percent difference between actual revenue and original budgeted revenue. They find that a greater diversification is associated with a smaller revenue shortfall. They also find that increased diversification is associated with greater than expected inflow of current year revenue over budgeted expenditures. Chernick et al. (2011) in a study of large central cities find that a diversified revenue structure is associated with generating more revenue.

The Elasticity of the Revenue Structure

A repeated theme in the diversification literature is that the method of diversification matters. In other words, the response of each revenue source to the economy matters. Elasticity is defined as the response of tax revenues to the business cycle (Sobel & Holcombe, 1996). In a study on state revenue elasticities, Felix (2008) points out that the ability of the states to balance their budgets through business cycles depends on the composition of states’ tax portfolios, which in turn determines the growth and stability of state tax revenues. The importance of each tax instrument’s sensitivity depends on its elasticity relative to tax revenue. Groves and Kahn (1952) estimate tax revenue elasticities for state and local governments and find that these elasticities are not constant over time. They also find that state and local tax systems are more stable than the federal government system.

Sobel and Holcombe (1996) argue that the elasticity is important because in the long run it is as an indicator of growth in tax revenue, and in the short run it is a measure of the cyclical variability of tax revenues. Felix (2008) defines long-run elasticity as the growth in tax revenue caused by the growth in personal income or other sources of tax revenue, and volatility of tax revenues is defined as the changes in those growth rates. Dye and McGuire (1991) state that these two characteristics of a revenue system are important because policy makers “must devise revenue systems that can both support expenditure programs over the long run and provide stable streams of revenue even as the underlying economy varies with the business cycle” (p. 55).

Earlier literature suggests that there is a trade-off between the two measures of elasticities (Groves & Kahn, 1952); however, more recent literature indicates that this may not be the case for some tax instruments (Felix, 2008). Dye and McGuire (1991) show that the trade-off between growth and volatility does not always hold as personal income taxes grow faster than sales tax but they are not more volatile. They also find that states with identical revenue shares for two taxes (personal income and sales) can experience different growth and variability in tax revenues. Sobel and Holcombe (1996) also report that the trade-off does not always hold, and corporate and sales tax revenues tend to grow at around the same rate; however, corporate tax revenues tend to have much more volatility.

Bruce, Fox, and Tuttle (2006) undertake an extensive study of the effect of the major state taxes (personal income and sales) in the long run and short run. They find that the average income elasticity of personal income tax is more than double that of sales tax in the long run. However, neither sales tax nor income tax is universally more volatile, and volatility is a function of tax policy regarding the tax base. The authors conclude that states have a number of options available to increase income elasticity of their tax structures. For instance, they can shift toward higher elasticity taxes (personal income tax) and shift away from low elasticity taxes (sales tax) or other combinations of their tax portfolio.

Felix (2008) analyzes the impact of factors affecting the growth and volatility of state tax revenue in the Tenth Federal Reserve District.¹ She concludes that the composition of a state’s tax portfolio has a major effect on the growth and stability of a state’s tax revenues. Felix finds that sales taxes tend to have the least volatility, whereas the personal income tax tends to be the most elastic over the past 40 years. Felix shows that in the Tenth District, personal income displays the most growth followed by severance tax, general sales tax, and corporate income tax—similar to national data.

Conceptual Model of Negative Revenue Variance

Given the significant negative programmatic impacts potentially associated with unexpected revenue shortfalls or negative revenue variance, especially when those shortfalls are large and caused by economic declines such as the Great Recession, we examine the influence of revenue structure on the occurrence of negative revenue variance and what determines the magnitude of the variance. We ask two questions. Why do negative revenue variances or revenue shortfalls occur? What contributes to their severity? We argue that the revenue structure impacts the inflow of revenue leading to negative revenue variances. In particular, building on the extant literature on revenue structure, we hypothesize that the answers depend on the diversification and elasticity of the state’s revenue portfolio.

The different mix of revenue sources, in accordance with portfolio theory in corporate finance, helps to lower the risk or volatility of revenue inflow as a whole as long as these revenue streams are not perfectly correlated (Brealey & Myers, 1991). The concept of revenue volatility can be defined as the degree of deviation or variation of actual revenue from its projected level. Negative revenue variance can be a manifestation of such revenue volatility. The revenue smoothing effect of diversification reduces volatility, subsequently reducing the chances of encountering a negative revenue variance (Hypothesis 1) as well as reducing the size of the gap (Hypothesis 3).

On the contrary, in terms of elasticity, a highly elastic revenue structure can deteriorate the state’s finances at a pace faster than the general declining economy. This may increase the likelihood of the occurrences of negative revenue variance and accentuate the magnitude of the existing shortfalls as a result. As such, we will expect that the high elasticity of a revenue structure contributes to the odds of experiencing a negative revenue variance (Hypothesis 2) as well as its size (Hypothesis 4).

The four hypotheses are summarized as follows:

Hypothesis 1: A more diversified revenue structure decreases the probability of the occurrence of negative revenue variance.

Hypothesis 2: A more elastic revenue structure increases the probability of the occurrence of negative revenue variance.

Hypothesis 3: A more diversified revenue structure decreases the magnitude of negative revenue variance.

Hypothesis 4: A more elastic revenue structure increases the magnitude of negative revenue variance.

An important contribution of this research is to disentangle the differential impact of the degree of diversification and elasticity of revenue portfolio on the occurrence and magnitude of negative variance. In the meantime, we recognize that other fiscal, institutional, political, and socio-economic factors can affect revenue variance, and they are captured by our model. A more elaborated discussion on the choice of variables and their measurements are provided in the next section.

Method

Data and Estimation Method

The following model specification is used to test the four hypotheses:

Revenue variance = f (\begin{array}{l} Revenue structure, Fiscal capacity, Fiscal institutions, Political factors, \\ Socio-economic factors \end{array}),

where

Revenue variance = general fund actual total revenue minus originally budgeted total revenue;

Revenue structure = level of diversification (modified Hirschman–Herfindahl Index [HHI] index), elasticity of structure (composite index);

Fiscal capacity = log of expenditures, tax effort as measured by total tax revenues as a proportion of state personal income;

Fiscal institutions = existence of balanced budget requirement as measured by existence of controls on supplemental appropriations and as measured by no deficit carryover, existence of tax limitations, existence of expenditure limitations, and whether there is a biennial budget;

Political factors = presence of Republican controlled governor, Republican controlled house, and Republican controlled senate;

Socio-economic factors = log of per capita state personal income, percent change in state personal income, and state population.

The revenue variance is calculated from the state CAFRs, and the revenue structure data for this analysis come from U.S. census state government finance data series. The Bureau of Economic Analysis provides the socio-economic data. Balanced budget data are obtained from the Smith and Hou (2013) study. The institutional and political data for states are obtained from National Conference of State Legislatures and the National Governors Association. All of the economic and financial data have been converted to 2009 constant dollars using the price deflator for state and local government consumption expenditures and gross investment. The unit of our observation is the U.S. state governments each year from 2007 to 2011. This study period includes the Great Recession, which ensures the presence of negative revenue variances and provides an ideal time frame to study the different impact of the structural characteristics of state revenue structure on their occurrences and size. After leaving out observations with missing data, the final data set consists of 232 observations from 47 states over the 5-year period.²

One challenge in our sample is that 127 of our 232 observations (54.7%) experienced a negative variance during our study period. This can raise a potential concern for selection bias as there may be some systematic differences or characteristics of these observations associated with their presence in the sample. To address this concern and examine our research hypotheses, we establish a Heckman selection model (Heckman, 1979) that captures factors that affect the occurrence and magnitude of negative revenue variance. The observations in the sample are first identified in terms of whether or not the state experienced a negative revenue variance. A probit model is then used in the first stage to estimate the probability that a state will experience a negative revenue variance. A Heckman selection procedure is then employed in the second stage to control for selection bias. The exclusion restriction or instruments used in the first-stage probit model include “Gubernatorial Election Year,” a dichotomous variable with one indicating it is an election year and zero otherwise, and 4-year dichotomous variables (Yr1-Yr4) with one indicating year 2007, 2008, 2009, and 2010 correspondingly, and zero otherwise. The second-stage model assumes that the magnitude of the negative revenue variance is a function of the degree of state revenue diversification, overall elasticity of state revenue portfolio as well as other variables indicating the fiscal capacity, fiscal institution, political and socio-economic influences, and is discussed in detail in a later section.

Revenue Variance

We calculate revenue variance as the difference between the actual total revenue and the original budgeted total revenue as reported in the Budgetary Comparison Schedule—General Fund found in the Required Supplementary Information section of the CAFR.³ The original budgeted revenue according to Governmental Accounting Standards Board (GASB) is based on the first complete appropriated budget (GASB, 1999). A negative variance occurs when the original budgeted revenue is greater than the actual revenue. In the first-stage probit model, the dependent variable is a dichotomous variable (Var_Neg) with one indicating a unit observation experiencing a negative revenue variance and zero otherwise, and we measure the magnitude of the negative revenue variance (Neg_Var_Pct) as the absolute percentage of its original budgeted revenue in the second-stage Heckman procedure.

Revenue Diversification Index

To measure the degree of diversification in a state’s revenue structure, we adopted the revised HHI developed by Suyderhoud (1994). This index is calculated based on the five general own-source revenue categories: general sales tax, personal income taxes, other taxes, general charges, and total other revenues.⁴ The index is defined as the following:

HHI = \frac{1 - \sum_{i = 1}^{5} {R_{i}}^{2}}{1 - \frac{1}{5}},

where R_i is the proportion of own-source revenue from each of the five main revenue categories.⁵ The HHI index ranges from zero to one with increasing values connoting more diversified state revenue structure.

Composite Index of Revenue Portfolio Elasticity

One major contribution of this study to the current literature is that we established a composite index that measures the long-run elasticity of a state revenue portfolio, which truly depicts how sensitive the entire revenue structure is relative to changes of statewide economy. The elasticity composite index is defined as the following equation:

ES = \sum_{i = 1}^{n} {ES}_{i} \times W_{i},

where ES = Revenue Portfolio Elasticity Composite Index, ES_i = the state personal income elasticity of R_i,⁶ and W_i = R_i as a percent of total general own-source revenue. The elasticity of R_i (ES_i) is defined as the ratio of the percentage change in revenue category i (R_i) to a given percentage change in state personal income (Y_t). To derive the numerical value of ES_i, we estimate the following regression equation:

\ln (R_{i t}) = α_{i} + β_{i} \ln (Y_{t}),

where R_it = revenue category i in year t, Y_t = state personal income in year t, and β_i = the coefficient beta, represents the income elasticity of revenue category i (ES_i).

Different from HHI, the value of ES is not bounded. Increasing value of the index indicates a more elastic revenue structure. It should be noted that both HHI and ES capture the general own-source revenues of a state, because general own-source revenues reflect the part of revenues that can be adjusted via a state’s fiscal autonomy or through the adoption of various kinds of revenue and tax policies.

Control Variables

Our analysis accounts for other factors than those of revenue structure that can also affect the occurrence and magnitude of revenue variance.

Fiscal capacity

Tax effort is a factor that indicates the extent to which the tax base is utilized for taxation. Drawn from Hendrick (2002), the variable is measured by the total tax revenues as a proportion of state personal income. To control for the needs of public expenditures, we include the log of the actual expenditures to reflect the size of the budget.

Fiscal institutions

Because state governments are generally not allowed to operate with deficit spending and they also have limited access to capital market, these fiscal disciplines can play a role in shaping the budgeting and financing decision. In particular, after the strict balanced budget requirement and a variety of TELs have been put into place, these fiscal institutions can have a strong influence on the budget outcomes. In our model, we incorporate two dichotomous variables, controls on supplemental appropriations and no deficit carryover, from the classification system of balanced budget requirement developed by Smith and Hou (2013) with value of one indicating the presence of these rules and zero otherwise. The balanced budget rules are chosen because they are proved to be most effective in affecting state expenditures. In addition, our study controls for the influence of TELs with dichotomous variables for each type of limitation: one indicating its presence and zero otherwise. In our second-stage model, we also include a dichotomous variable with one indicating the practice of biennial budget and zero otherwise to account for the influence of biennial budget practice on the magnitude of the revenue variance due to the extended lag between the revenue forecast and revenue realization.

Political factors

Many fiscal decisions are greatly affected by partisan interests. Following Smith and Hou (2013), we include the dichotomous controls for the election year as discussed earlier, the party of governor, house, and senate with one indicating Republican-dominated control and zero Democrat control.

Socio-economic factors

Last but not least, we control the influence of the wealth level, change of economic condition in each individual states, and its size by including the log of per capita personal income, percent change of state personal income, and population (in thousands).

Empirical Results

Table 1 provides the variable information and summary of statistics for the variables used in this study. As shown in Table 1, state revenue structures as a whole are quite diversified with an average HHI of 0.917. Also, the overall elasticity of their revenue portfolio is fairly high as well with an average of 1.367. It means for every dollar increase in state personal income, there is an increase of 1.367 dollars in state general own-source revenues. On average, a negative revenue variance occurred a little more than half of the time (0.547).

Table 1.

Variable Information and Summary of Statistics.

Variable	Definition	N	M	SD	Minimum	Maximum
Neg_Var	One indicating the occurrence of negative revenue variance and zero otherwise	232	0.547	0.499	0	1
Neg_Var_Pct	Negative revenue variance percent (absolute percent of its original budgeted revenue)	232	0.053	0.103	0	0.583
HHI	Hirschman–Herfindahl Index	232	0.917	0.076	0.626	0.999
ES	Elasticity composite index	232	1.367	0.448	0.463	3.273
Income Change	Percent change of state personal income	232	0.112	2.875	−10.327	9.606
Tax Effort	Total tax revenues as a proportion of state personal income	232	0.066	0.023	0.038	0.273
Ln Expend	Log of the actual expenditures	232	16.066	1.170	13.611	18.431
Gov Party	One indicating Republican Governor and zero otherwise	232	0.504	0.501	0	1
House Party	One indicating Republican controlled house and zero otherwise	232	0.422	0.495	0	1
Senate Party	One indicating Republican controlled senate and zero otherwise	232	0.448	0.498	0	1
Supp Controls	One indicating the presence of controls on supplemental appropriations and zero otherwise	232	0.332	0.472	0	1
No Deficit	One indicating no deficit carryover and zero otherwise	232	0.151	0.359	0	1
Tax Limit	One indicating the presence of tax limitation and zero otherwise	232	0.108	0.311	0	1
Spend Limit	One indicating the presence of expenditure limitation and zero otherwise	232	0.422	0.495	0	1
Biennial Budget	One indicating the practice of biennial budget and zero otherwise	232	0.349	0.478	0	1
Gubernatorial Election Year	One indicating the election year and zero otherwise	232	0.228	0.421	0	1
Population	Population (in thousands)	232	6,229.735	6,867.682	534.876	37,691.91
Ln Income	Log of per capita personal income	232	10.556	0.145	10.30939	10.9774
Yr1	One indicating year 2007 and zero otherwise	232	0.198	0.400	0	1
Yr2	One indicating year 2008 and zero otherwise	232	0.203	0.403	0	1
Yr3	One indicating year 2009 and zero otherwise	232	0.1982759	0	0	1
Yr4	One indicating year 2010 and zero otherwise	232	0.198	0.400	0	1

Table 2 provides the results of our two-stage Heckman selection model. The overall model is statistically significant, as indicated by the chi-square test. In the first-stage probit model (shown in the lower panel of Table 2), we examine a series of factors that predicts the likelihood that a state experiences a negative revenue variance. We find that the overall elasticity of the revenue portfolio is positively associated with the probability of experiencing a negative revenue variance, and it is statistically significant at the 1% level. This result suggests that the more elastic a state’s revenue portfolio, the more likely that a state will encounter a negative revenue variance. However, revenue diversification is negatively correlated with the occurrence of negative revenue variance, and it is statistically significant at the 10% level. This indicates that the more diversified a state’s revenue structure, the less likely that a state will experience a negative revenue variance. These two findings are consistent with our first two hypotheses.

Table 2.

Heckman Selection Estimation Results.

	Coefficient	SE	z	p > z
Neg_Var_Pct
HHI	−0.332	0.148	−2.24	.025
ES	−0.051	0.026	−1.91	.056
Income Change	0.002	0.004	0.55	.579
Tax Effort	1.189	0.547	2.17	.030
Ln Expend	0.000	0.013	−0.02	.986
Gov Party	−0.009	0.022	−0.44	.661
House Party	−0.034	0.027	−1.25	.210
Senate Party	0.020	0.028	0.70	.483
Supp Controls	−0.049	0.023	−2.13	.033
No Deficit	−0.056	0.031	−1.81	.071
Tax Limit	−0.037	0.041	−0.91	.361
Spend Limit	0.033	0.022	1.48	.139
Biennial Budget	0.032	0.024	1.37	.171
Ln Income	−0.084	0.083	−1.02	.309
Population	<0.000	0.000	−0.92	.356
Constant	1.343	0.925	1.45	.147
Neg_Var
Election Year	−0.304	0.288	−1.06	.291
HHI	−2.841	1.611	−1.76	.078
ES	0.727	0.253	2.87	.004
Income Change	0.007	0.055	0.13	.898
Tax Effort	−13.212	5.560	−2.38	.017
Ln Expend	0.063	0.131	0.48	.632
Gov Party	−0.311	0.221	−1.41	.160
House Party	0.371	0.279	1.33	.184
Senate Party	−0.695	0.269	−2.58	.010
Supp Controls	−0.354	0.227	−1.56	.118
No Deficit	0.470	0.302	1.56	.120
Tax Limit	−0.434	0.328	−1.32	.185
Spend Limit	−0.126	0.214	−0.59	.555
Ln Income	0.536	0.793	0.68	.499
Population	<0.000	0.000	−1.66	.098
Yr1	−0.010	0.301	−0.03	.973
Yr2	0.320	0.305	1.05	.294
Yr3	1.649	0.487	3.39	.001
Yr4	1.178	0.356	3.31	.001
Constant	−3.887	8.726	−0.45	.656
/athrho	−0.443	0.243	−1.83	.068
/lnsigma	−2.249	0.078	−28.96	.000
rho	−0.416	0.201
sigma	0.106	0.008
lambda	−0.044	0.023
Number of observations = 232
Censored observations = 105
Uncensored observations = 127
Wald χ²(15) = 58.25
Prob > χ² = 0.0000

Note. HHI = Hirschman–Herfindahl Index; ES = elasticity.

As the control variables in our first-stage model, year 2009 (Yr3) and year 2010 (Yr4) are statistically significant and positively correlated with a state’s probability of obtaining a negative revenue variance. This finding makes intuitive sense as the entire economy was in a deep recession in these 2 years. We also find that the variables of tax effort and Republican-controlled senate are negatively associated with a state’s probability of experiencing a negative revenue variance. In addition, population is negatively correlated with the probability of having a negative revenue variance.

The regression results of the second stage can be found in the upper panel of Table 2. Consistent with our expectation, revenue diversification is negatively correlated with the magnitude of the negative revenue variance, and it is statistically significant at the 5% level. One standard deviation increase of HHI (0.076) reduces the negative revenue variance by 2.5% of its original budgeted revenue (0.076 x -0.332) on average. Combining with the first-stage result, revenue diversification not only reduces the occurrence of a negative revenue variance but also reduces the size of the existing shortfall (Hypotheses 1 and 3). But different from what we expect, the increased elasticity of a revenue structure is negatively associated with the size of the revenue shortfall given the occurrence of negative revenue variance, and it is statistically significant at the 10% level. One standard deviation increase of ES (0.448) decreases the negative revenue variance by 2.3% of its original budget (0.448 x -0.051). The possible justification for this finding is that the increased elasticity helps state revenues rebound at a faster pace than the general economy following a deep recession because those more elastic sources are strong revenue generators for the states. Given the small coefficient, we exercise caution with this explanation.

Regarding the control variables, the tax effort variable is positively associated with the size of the negative variance with a 5% statistical significance. One standard deviation increase of tax effort (0.023) increases the negative revenue variance by 2.7% of its original budgeted revenue (0.023 x 1.189) on average. The result suggests greater reliance on the tax base tends to worsen the existing revenue shortfalls. However, the presence of controls on supplemental appropriations and no deficit carryover requirement on average reduces the negative revenue variance by 4.9% and 5.6% of its original budgeted revenue correspondingly. Both the controls on supplemental appropriations and no deficit carryover requirement are manifested as powerful tools to reduce the size of variance, and the effect is statistically significant at the 5% and the 10% levels, respectively.

Conclusion

Because the originally adopted budget is created prior to the inflow of actual revenues, the original budgeted revenue often varies from actual revenues. In practice, the detrimental impact from revenue volatility mainly manifests itself through the negative variance. Revenue stability and predictability depict the extent to which the actual level of revenue conforms to the budgeted revenue based on informed and professional forecasting. When it comes to the factors that influence such stability, scholarly debates more or less focus on either revenue diversification or the elasticity of revenue portfolio. Namely, the actual revenue falls short of its budgeted level. In this article, we contribute to the current literature by incorporating both revenue diversification and elasticity of revenue portfolio as determinants of the occurrences and magnitude of the negative revenue variance in an econometric model hoping to disentangle their independent effects.

A major finding of this research article indicates that a more diversified revenue structure decreases the probability of experiencing a negative revenue variance and it also helps to reduce the size of the negative variance. A more elastic revenue structure contributes to the odds of encountering a negative revenue variance but helps revenue to rebound or shrink the size of variance if the shortfall already exists.

Consistent with Jordan and Wagner (2008), our findings suggest that revenue diversification can be used as an active strategy to guard against the unexpected revenue shortage and to control the magnitude of a revenue shortage. Perhaps diversification is associated with revenue generation as found by Chernick et al. (2011). Furthermore, when the general decline of the economy and budget crisis are inevitable (i.e., negative revenue variances occur), the adoption of a diversified revenue portfolio can help the state turn the situation around faster than otherwise.

Our finding for elasticity’s association with a revenue shortage was not aligned with our hypothesis and suggests that higher elasticity is associated with a lower magnitude. Therefore, given the small coefficient for elasticity’s negative association with the magnitude of the shortfall, we cautiously suggest that a more elastic structure may also aid in the recovery. This, of course, is likely dependent upon how the state increases its elasticity. Dye and McGuire (1991), Bruce et al. (2006), and Felix (2008), all point to the higher revenue growth from a reliance on the elastic personal income tax. Bruce et al. conclude that decision makers can increase the elasticity of their portfolio by increasing reliance on more elastic taxes such as income taxes. Decision makers may also consider adjusting the structure of a revenue source such as the assortment of user fees to make user fees a more elastic source of revenue or adjusting the rate and base of sales taxes to increase the portfolio’s reliance on more elastic purchases.

The implication for budget decision makers is that the elasticity and diversification of the revenue portfolio can impact revenue adequacy and stability via different paths. While diversification helps to avert the potential shortfalls, it also contributes in getting the state out of its gloom. However, increased elasticity is associated with the occurrence of revenue shortfalls. However, the increased elasticity of state revenue portfolio caused by incorporating higher proportions of the cyclical revenue components or through adjusting the structure of individual revenue sources helps the fiscal recovery when the economy declines. Therefore, there is a trade-off with increased elasticity of revenue, which is increased probability of shortfall but a faster recovery. These two characteristics of a revenue portfolio work in tandem and operate as tools for decision makers. While the level of diversification can decrease occurrence and magnitude, holding diversification constant, increasing the elasticity of the portfolio could also decrease the magnitude of shortfalls.

Of course, smaller revenue shortfalls may allow for the opportunity to use other mechanisms, such as financial reserves to address the revenue gap. The larger the magnitude of the shortfall, the more likely programmatic expenditure cuts will have to take place as a mid-year budget adjustment, which has its own efficiency, accountability, and oversight implications.

It should be noted that our study examines a time period that overlaps one of the greatest recessions in history. Accordingly, it provides ample observations for studying revenue diversification and elasticity during revenue shortfalls. These findings may have general implications for revenue windfalls, a positive variance. As expressed by Dougherty et al. (2003), a positive variance could be the result of an intentional underestimation of revenue with the purpose of creating a surplus, which allows for organizational slack and avoids the risk of violating the balanced budget requirement. Future research calls for extending the data set to a different time frame to determine the consistency of patterns.

Considering the work of Felix (2008) and Bruce et al. (2006), future research may look at a longer time period at each revenue source to examine the impact of elasticity and diversification given tax policy. This would allow the isolation of various policies that could impact the elasticity of each revenue source, such as income bracket levels, income exemptions, rate structure, and goods and services exemptions. With tax policies as a focus, this would also suggest another look at revenue shortages, especially relatively large revenue shortages, which most assuredly impact the provision of goods and services.

Footnotes

Declaration of Conflicting Interests

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

Funding

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

Notes

Author Biographies

Meagan M. Jordan is an associate professor at the School of Public Service at Old Dominion University. Her research includes revenue structure and management, popular financial reporting, fiscal transparency, and state and local budgeting and budget theory.

Wenli Yan is an assistant professor in the Wilder School of Government and Public Affairs at Virginia Commonwealth University. Her research focuses on the impact of revenue structure on revenue volatility of public and nonprofit organizations, and also the subsequent impact on financial sustainability of these organizations.

Somayeh Hooshmand is a PhD student in the School of Public Service at Old Dominion University. Her research focuses on community public health, health policy and health outcomes.

References

Brealey

R. A.

Myers

S. C.

(1991). Principles of corporate finance (4th ed.). New York, NY: McGraw-Hill.

Bruce

Fox

W. F.

Tuttle

M. H.

(2006). Tax base elasticities: A multi-state analysis of long-run and short-run dynamics. Southern Economic Journal, 73, 315-341.

Carroll

D. A.

(2009). Diversifying municipal government revenue structures: Fiscal illusion or instability? Public Budgeting & Finance, 29, 27-48.

Chernick

Langley

Reschovsky

(2011). Revenue diversification and the financing of large American central cities. Public Finance and Management, 11, 138-159.

Cornia

G. C.

Nelson

R. D.

Wilko

(2004). Fiscal planning, budgeting, and rebudgeting: Using revenue semaphores. Public Administration Review, 64, 164-179.

Dougherty

J. D.

Klase

K. A.

Song

S. G.

(2003). Managerial necessity and the art of creating surpluses: The budget execution process in West Virginia cities. Public Administration Review, 63, 484-497.

Dye

R. F.

McGuire

T. J.

(1991). Growth and variability of state individual income and general sales tax. National Tax Journal, 44, 55-66.

Felix

R. A.

(2008). The growth and volatility of state tax revenue sources in the tenth district. Economic Review, 3, 63-88.

Forrester

J. P.

Mullins

D. R.

(1992). Rebudgeting: The serial nature of the municipal budgetary process. Public Administration Review, 52, 467-473.

10.

Governmental Accounting Standards Board. (1999). Statement No. 34: Basic financial statements and management’s discussion and analysis for state and local governments (No. 171-A). Norwalk, CT: Governmental Accounting Standards Board.

11.

Groves

H. M.

Kahn

H. C.

(1952). The stability of state and local tax yields. The American Economic Review, 42, 87-102.

12.

Hannarong

Akoto

J. B.

(2004). State and local fiscal structures and fiscal stress. Journal of Public Budgeting, Accounting & Financial Management, 16, 40-61.

13.

Heckman

(1979). Sample selection bias as a specification error. Econometrica, 47, 153-161.

14.

Hendrick

(2002). Revenue diversification: Fiscal illusion or flexible financial management. Public Budgeting & Finance, 22, 52-72.

15.

Johnston

J. M.

Pagano

M. A.

Russo

R. P.

(2000). State limits and state aid: An exploratory analysis of county revenue structure. State and Local Government Review, 32, 86-97.

16.

Jordan

M. M.

Wagner

G. A.

(2008). Revenue diversification in Arkansas cities: The budgetary and tax effort impacts. Public Budgeting & Finance, 28(3), 68-82.

17.

Joyce

P. G.

Mullins

D. R.

(1991). The changing fiscal structure of the state and local public sector: The impact of tax and expenditure limitations. Public Administration Review, 51, 240-255.

18.

Kioko

S. N.

Martell

C. R.

(2012). Impact of state-level tax and expenditure limits (TELs) on state government revenues and aid to local governments. Public Finance Review, 40, 736-766.

19.

Ladd

H. F.

Weist

D. R.

(1987). State and local tax systems: Balance among taxes vs. balance among policy goals. In Stocker

F. D.

(Ed.), The quest for balance in state-local revenue structures (pp. 39-69). Cambridge, MA: Lincoln Institute of Land Policy.

20.

Pagano

M. A.

Johnston

J. M.

(2000). Life at the bottom of the fiscal food chain: Examining city and county revenue decisions. Publius: The Journal of Federalism, 30, 159-170.

21.

Rubin

I. S.

(1987). Estimated and actual urban revenues: Exploring the gap. Public Budgeting & Finance, 7(4), 83-94.

22.

Shadbegian

R. J.

(1999). The effect of tax and expenditure limitations on the revenue structure of local government, 1962-87. National Tax Journal, 52, 231-237.

23.

Smith

D. L.

Hou

(2013). Balanced budget requirements and state spending: A long-panel study. Public Budgeting & Finance, 33(2), 1-18.

24.

Snell

R. K.

(2011). State experiences with annual and biennial budgeting. Washington, DC: National Conference of State Legislatures. Retrieved from http://www.ncsl.org/documents/fiscal/BiennialBudgeting_May2011.pdf

25.

Sobel

R. S.

Holcombe

R. G.

(1996). Measuring the growth and variability of tax bases over the business cycle. National Tax Journal, 49, 535-552.

26.

Suyderhoud

J. P.

(1994). State-local revenue diversification, balance, and fiscal performance. Public Finance Quarterly, 22, 168-194.

27.

U.S. Census. (2012). State government finances. Washington, DC: U.S. Government Printing Office.

28.

Wang

J. Q.

(2012). Who pays and who benefits? The impact of state tax and expenditure limits on tax progressivity and redistributive spending. Journal of Public Budgeting, Accounting & Financial Management, 24, 660-682.

29.

White

F. C.

(1983). Trade-off in growth and stability in state taxes. National Tax Journal, 36, 103-114.

30.

Yan

(2012). The impact of revenue diversification and economic base on state revenue stability. Journal of Public Budgeting, Accounting & Financial Management, 24, 58-81.