State revenues tend to be procyclic, while the demand for state expenditures tends to be countercyclic. During recessions, particularly those of 1980-1982 and 1990-1991, states found themselves experiencing fiscal crises because revenues were insufficient to meet expenditures demands. In the short run, states faced with fiscal crises are pressured to reduce expenditures and increase taxes.1 In addition, states may use temporary measures such as accelerating tax collections and deferring expenditures, as well as bending accounting practices.2 Over the longer run, states can reduce their susceptibility to fiscal crises by relying on less cyclic tax bases or by smoothing expenditures over the business cycle by capturing and saving budget surpluses in boom years in order to finance expenditures during lean years. After the fiscal crises that states experienced during the 1980-1982 recession, many responded by creating rainy day, or budget stabilization, funds specifically for this purpose. Currently, over threequarters of the states have some type of rainy day fund. This article examines the degree to which these rainy day funds eased the fiscal stress experienced by states during the 1990-1991 recession.

Because most state governments face balanced budget constraints, budget stabilization funds and general fund surpluses are useful tools for smoothing state expenditures due to temporary fluctuations in tax revenues. The article examines states' actual reliance on tax increases, general fund surpluses, and rainy day funds to maintain expenditures during the declines in revenues suffered by states during the past three recessions. A measure of state fiscal stress during the 1990-1991 recession is then developed, and a regression analysis is performed to see whether the presence of an explicit state rainy day fund had a significant impact in lessening state fiscal stress, and whether the impact differed depending upon the specific provisions of the state rainy day fund.

STATE FISCAL POLICY AND THE ROLE OF BUDGET STABILIZATION FUNDS

In the analysis that follows, a state fiscal policy of neutrality over the business cycle is used as a benchmark for evaluating the use of budget stabilization funds.3 Under a policy of neutrality, states do not change their tax structures over the business cycle, so revenues move procyclically, and state expenditures grow at a constant rate through the business cycle. If states are constrained against deficit finance (as most are), a policy of neutrality would require building up surplus funds during the good years and using those funds during recessions.4 States desire rainy day funds to mitigate against the impacts of recessions that require states to reduce their expenditure growth or raise taxes, so the benchmark of neutrality makes sense as a measure of the effectiveness of rainy day funds. The use of fiscal neutrality as a benchmark does not constitute a policy recommendation, however. States may want to slow expenditure growth or increase taxes to some degree during a recession. While we take no stand on the desirability of fiscal neutrality as a policy goal, it is a good benchmark to measure the performance of a rainy day fund because the purpose of a rainy day fund is to reduce the need for tax increases and expenditure cuts.

Unlike the federal government, states are faced with constraints that limit their ability to finance temporary revenue fluctuations with budget deficits and surpluses. Only one state, Vermont, is completely free of balanced budget requirements.5 There are also ten states allowed to carry over a deficit, but of these ten, two are only allowed to carry this deficit over for one year, and a third can only carry over unanticipated deficits. Fortunately, the majority of these state balanced budget requirements are stock, rather than flow, in nature.6 This means that most states are allowed to spend more than they collect in tax revenues in a given year as long as they have surplus balances accumulated in prior years to cover the deficit. Because of these constraints, states have adopted the policy of building up budget surpluses during the boom years to help mitigate the fiscal problems associated with recessions.

From a practical standpoint, for most states it does not matter whether these surplus balances are kept in rainy day funds or just maintained as excess general fund balances. For the few states whose balanced budget requirements are flow in nature, such as Michigan, the setting up of a rainy day fund is the only way to circumvent the balanced budget constraint.7 As of fiscal year 1994, forty-four states had some type of rainy day fund, with almost three-quarters of these being established after the state fiscal crises experienced during the 1980-1982 recession.8 Table 1 shows the number of states with rainy day funds between fiscal years 1982 and 1994.

While many states could just rely on their surplus general fund balances, there are several reasons why establishing a rainy day fund may be preferable.9 Unlike general fund balances, rainy day funds often have specific provisions for saving and withdrawing funds, which prevents legislators from giving in to special interest demands to spend all of their revenues during boom years. Additionally, some tax and expenditure caps enacted during the Tax Revolt of the late 1970s and early 1980s contain provisions requiring state legislatures to, at least partially, refund tax revenues in excess of expenditures. As Gold (1983) points out, California's large surplus was an important factor in generating support for Proposition 13 reforms.10 A properly structured rainy day fund can provide a vehicle for saving during boom years in order to help cover revenue shortfalls during recessions.

The cyclical nature of revenues, however, is not the only reason why a state may wish to establish a rainy day fund. Some rainy day funds are structured so that they allow states to use them in cases of unanticipated revenue shortfalls, such as could happen with a bad revenue forecast, or in case of federal tax changes that may adversely affect state revenues.11 The use of these funds in case of adverse effects from federal tax law changes is specifically mentioned in the rules governing Virginia's rainy day fund. There has been substantial literature discussing the impact of the federal Tax Reform Act of 1986 on state tax revenues.12 While the 1986 reform created a windfall gain for most states, the possibility of a change that adversely affects state revenue is always present.

The majority of state rainy day funds require only regular legislative approval for the funds to be used, which allows states to cover unanticipated revenue shortfalls, but has the drawback of not providing very stringent controls to ensure that funds are left untouched until they are really needed. Some states have created rainy day funds without requiring that money be deposited in them, so the funds may have little or no money in them when a revenue downturn arrives. The current provisions and purposes of state rainy day funds differ substantially across states so specific provisions may influence how effective rainy day funds are at mitigating the effects of revenue downturns. The effectiveness of rainy day funds will be examined after first looking at the fiscal stress that rainy day funds are intended to reduce.

THE IMPACT OF RECESSIONS ON STATE FINANCES

Rainy day funds are set aside to lessen the degree to which states will have to reduce expenditure growth or raise taxes to cope with a recession. This section of the article looks at data from the last three recessions to see how recessions have affected the taxes and expenditures of states. The contractionary stages of these three recessions, from peak to trough, are November 1973 to March 1975, January 1980 to November 1982, and July 1990 to March 1991. Because the actual period of interest also includes the part of the expansionary phase when growth is still below average, these above listed periods do not perfectly correspond to periods of recessionary impact, but rather provide a basis for identifying years within which states might have been experiencing recession-induced fiscal stress.13

For expositional purposes, the recessions will be considered to be the fiscal years of 1974-1975, 1980-1983, 1990-1991. Table 2 contains the annual real percent growth in state general expenditures and tax revenues, discretionary tax increases (either base broadening or rate changes) as a percent of the previous fiscal year's tax revenues, and the adjusted growth in real tax revenues (taking out the revenue raised from discretionary tax increases) during the three recessionary periods listed above for all fifty states combined.14

From Table 2, it appears that state expenditure growth fared much better during the recession of 1990-1991 than it did during the other two recessions, falling below the 1980-1992 average only in 1990. The adjusted growth of tax revenues, which subtracts discretionary tax increases out of revenues, was impacted about as severely during the 1990-1991 recession as during the 1980-1982 recession, and worse in 1990-1991 than during the 1974-1975 recession.l5 The legislated tax increases, including both tax rate changes and base broadening, were more heavily relied upon during the 1990-1991 recession, however. During fiscal year 1992 alone, states legislated about $20 billion in new taxes, the largest single-year increase during any of the recessionary years considered here. Even though most states had saved relatively large surpluses during the 1980s, and most states had some type of rainy day fund by the 1990 downturn, tax increases were still relied upon to a great extent to compensate for revenue shortfalls. Additional data showing states' reliance on tax increases and stored surplus funds during, and in the years before and after, the 1990-1991 recession are given in Table 3.16

Table 3 shows that during fiscal years 1988 and 1989 states not only increased taxes, but also increased the size of their rainy day and surplus general revenue funds. The savings channeled into general revenue surpluses during these two fiscal years were almost three times larger than the savings to rainy day funds. The tax increases during fiscal years 1988 and 1989 were used to build savings, rather than expand current expenditures, as Table 3 shows. During both years, saving was greater than the amount of revenue derived from discretionary tax increases. Beginning in 1990, the behavior of states changed drastically with surplus balances being drawn down each year throughout the recession. During 1990, 1991, and 1992, states increased their reliance on tax increases to fund expenditures. During 1990 and 1991 alone, states raised almost $29 billion in new taxes. During 1990, states relied more heavily on previously saved general revenues, but with these balances running short, states turned to their rainy day funds in 1991.

The maximum reliance on rainy day funds was in 1991 when they were depleted by $725 million. There were thirty-eight states that had some type of rainy day fund in 1989 heading into the recession but only twenty-nine of them had positive balances. By the end of fiscal year 1991, eight of the states with positive balances in 1989 had completely depleted their rainy day funds, and another four had negligible balances. Of the forty-seven states with positive other balances in 1989, only thirty-six still had money remaining at the end of 1991. Five of the ten states that are allowed to carry over a deficit did so in either 1990 or 1991. In the latter half of fiscal 1991 and throughout fiscal 1992, with rainy day funds and other surpluses running low, states turned to tax increases to finance their expenditures.l7 When recessionary pressures eased in 1993, states still increased taxes but used this money to build back up their rainy day fund and other fund balances. Again, states placed more savings into other fund balances than into their rainy day fund balances.

The major tax increases instituted in fiscal years 1991 and 1992 show that state rainy day funds and other surplus balances were insufficient to maintain state expenditures during this recession. The following section of this article calculates a measure of the degree of fiscal stress experienced by each state during the 1990-1991 recession. These results will then be used to empirically investigate the impact of explicit state rainy day funds in easing state fiscal stress.

MEASURING STATE FISCAL STRESS DURING THE 199-1991 RECESSION

The purpose of a rainy day fund is to help a state maintain its expenditure growth while reducing its need to raise taxes during a recession. Thus, a natural measure of fiscal stress for purposes of analyzing rainy day funds is the amount of discretionary tax increases plus the amount that expenditures were reduced from their long run trend growth during a recession. Using this measure, a state that could maintain its long-run growth rate of expenditures without discretionary tax increases would have no fiscal stress. This measure of fiscal stress will be used as a benchmark for analyzing the effectiveness of rainy day funds, but it is not intended to be a policy recommendation. States might desire to cut back on expenditures to some degree, or raise taxes, during recessions, for example. This measure is natural for present purposes, however, because rainy day funds are intended to help states maintain their expenditures and reduce the pressures for increased taxes, which is exactly what is captured by this measure of fiscal stress.

Table 4 shows the amount of fiscal stress for each state between 1989 and 1992. The columns labeled "expenditure shortfall" show the dollar amount by which each state's expenditures were below its real trend growth rate in that fiscal year.ig A zero in this column means that a state's expenditure growth was either equal to or above its real trend growth. The columns labeled "tax increases" show the amount of legislated tax increases by fiscal year. These tax increase columns are cumulative, because an increase in the sales tax rate in 1990, for example, would not only be paid by citizens during the year in which it was implemented, but also during the subsequent years of 1991 and 1992. The final column of the table shows the total amount of tax increases and lower than average expenditure growth for each state as a percent of state general expenditures in 1988.

Table 4

Table 4 Continued

The numbers given in the final column of the table represent a measure of the degree of fiscal stress experienced by the state during the 1990-1991 recession, including fiscal years 1989 and 1992 when states also experienced below-average growth. Because the primary alternative to tax increases and expenditure reductions is to use previously accumulated surplus balances, these numbers also correspond to the amount by which surplus balances were insufficient to alleviate the fiscal stress of the recession. To see that accumulated balances can alleviate fiscal stress, look at Kansas, which has zeroes all the way across the table. This means that Kansas had no tax increases, nor did its rate of expenditure growth ever fall below its 1984-1992 trend growth rate. Kansas did, however, draw down $230 million previously accumulated fund balances in order to prevent fiscal stress, and enacted tax increases to take effect in 1993, a year beyond the data shown in the table.

In the Total row, the Total as a Percent of 1988 budget is the Total Fiscal Stress divided by the total of all state 1988 budgets. The 1988 budget is used because that is the year in which states would have had to have accumulated surplus funds in order to carry them through the next four lean years. If states had tax cuts or above-average expenditure increases, they received zeroes in the table. If these funds had been entered as negative numbers in the table instead of zeroes, the total would have been $57,359.5 million, which is shown as the Pooled Total. This is only 13.27 percent of the 1988 budget, or less than half the 29.31 percent total that results from not pooling states with positive and negative fiscal stress.

The 29.31 percent total might be thought of as gross fiscal stress, which is the total fiscal stress felt by those states that were negatively impacted. If this gross fiscal stress was offset by the above average expenditure growth that some states had in certain years, the resulting 13.27 percent total could be thought of as net fiscal stress. This suggests that states could benefit from pooling their financial risks from recessions, a concept that will be discussed further below. Given that many states created rainy day funds to reduce the impact of recessions after suffering through the 1980-1982 recession, the next section of the article examines whether states that had rainy day funds in 1989 were more likely to have less fiscal stress, and whether the specific deposit and withdrawal provisions of these funds made a difference.

THE IMPACT OF RAINY DAY FUNDS ON STATE FISCAL STRESS

Because some states have explicit rainy day funds while others do not, and because the provisions for depositing and withdrawing rainy day funds differ substantially across states, it is natural to ask whether rainy day funds helped to minimize the degree of fiscal stress experienced by states during the past recession. This section of the article uses regression models to look at this question empirically. For the first regression, the dependent variable will be the measure of fiscal stress as a percent of the state budget, as given in Table 4. A variable indicating whether the state had an explicit rainy day fund in 1989, RDF, is included as an independent variable. RDF is a binary variable taking on a value of 1 for states with rainy day funds and 0 otherwise. Also included in the regression are specific provisions of state rainy day funds.19 States that have explicit formulas or mandates for saving rainy day funds during the boom years are expected to have been better prepared for the recession than states whose fund saving is discretionary. The importance of these requirements was stressed by the National Conference of State Legislatures Fiscal Affairs and Oversight Committee.20 The dummy variable, REQUIRED SAVINGS, is included in the regression and is expected to have a negative impact on the degree of fiscal stress experienced by the state. Often state rainy day fund balances are capped at a certain percentage of general fund expenditures or revenues. If this cap is binding, it might prevent a state from saving sufficiently in its rainy day fund.21 A dummy variable, CAP, is included in the regression to capture the effect of these rainy day fund caps, and an additional variable, CAP%, denotes the percent of the state budget at which the fund is capped. Additionally, states differ in terms of the provisions for withdrawal of funds from their rainy day accounts. A variable indicating that the state's rainy day fund cannot be accessed by regular legislative action, WITHDRAWAL REQ, is included and is expected to have a negative impact on the degree of fiscal stress.

In addition to these variables describing the characteristics of rainy day funds, a measure of the degree to which states were impacted by the recession is needed. By including this variable, it is possible to determine the impact of the other variables on states who were hit similarly by the recession. This is important because states that experienced little fiscal stress could have been that way because they were either hit hard by the recession but well prepared, or because they were not hit very hard by the recession. Including a variable reflecting the severity of the recession on a particular state also controls for any selection bias arising from states that are hit harder by recessions being more likely to have rainy day funds.

SEVERITY is used in the regression model to calculate the impact on individual states due to reduced tax revenue growth. The rate of growth in state tax revenues during the recession, subtracting out any funds derived from discretionary tax increases, is used to calculate SEVERITY, under the assumption that states with slower revenue growth are the states that are more severely impacted by the recession. The sign is then reversed so that higher values of SEVERITY indicate a more severe impact. The result of this regression is given in the first column of Table 5.22

The regression, using fiscal stress as the dependent variable, shows that the mere existence of a rainy day fund does not alleviate fiscal stress, as shown by the insignificant coefficient on the Rainy Day Fund variable. If, however, the rainy day fund has a legal requirement that money be deposited into the rainy day fund, fiscal stress is significantly reduced. The coefficient on REQUIRED SAVINGS shows that the presence of a deposit requirement reduced the fiscal stress experienced by the state by approximately 17 percent of the state budget. Data on state Rainy Day Fund balances illustrates that states are not very disciplined with regard to saving money into their rainy day funds unless there is a legal requirement to do so, and the regression suggests that rainy day funds are likely to be ineffective without such a requirement. Of the thirty-eight states that had rainy day funds in 1989, twelve had REQUIRED SAVINGS.

The only other significant variable in the Fiscal Stress regression is SEVERITY, which measures the decline in tax revenues during the recession after removing any effects from rate increases. As expected, this variable indicates that greater tax revenue declines lead to greater fiscal stress. While the variable reflecting the presence of a cap on the size of the rainy day fund is the expected sign, it is insignificant. There are two possible explanations for this. State surpluses could have been so small that the cap was not binding, or alternatively, states were able to offset the impact of the cap by saving funds as other surpluses. The WITHDRAWAL REQ variable is also insignificant, meaning that the presence of a withdrawal requirement does not significantly affect the degree of state fiscal stress. The regression implies that a requirement that money be put into the Rainy Day Fund is much more important than a restriction on when funds can be withdrawn.

Another question of interest is how the division of fiscal stress between discretionary tax increases and expenditure reductions differed between states that had rainy day funds and those that did not. To answer this question, another regression was run with the same independent variables, but using the percent of the fiscal stress accounted for by expenditure shortfalls as the dependent variable. The only variable significant at the five percent level in that equation is the RAINY DAY FIND variable, meaning that states with rainy day funds are more likely to absorb fiscal stress through expenditure reductions than tax increases. The reason for this result is probably political. States that supported the creation of a rainy day fund are likely to be less supportive of tax increases, because rainy day funds were created, at least in part, in order to avoid tax increases.

States have implemented rainy day funds to try to reduce fiscal stress during recessionary times. This section finds that rainy day funds are successful in reducing fiscal stress only if they are accompanied by a provision that legally requires contributions into the rainy day fund. In addition, states with rainy day funds-whether or not they mandate contributions-are more likely to absorb fiscal stress through expenditure reductions than to react by increasing taxes.

STATE SURPLUSES AND FISCAL STRESS

The purpose of a rainy day fund is to reduce the need to legislate tax increases or reduce expenditure growth during recessions. This section calculates the level of fund balances that would have been needed prior to the 1990-1991 recession in order to completely eliminate the need for any expenditure reductions or tax increases. The elimination of all fiscal stress, measured this way, may or may not be an appropriate policy goal, but it is a good benchmark for examining the effectiveness of state rainy day funds and other surplus balances. States may want to have some tax increases or expenditure growth reductions during recessions, in which case this measure of fiscal stress would overstate the optimal size of a rainy day fund. On the other hand, states might want spending to be countercyclical because of the increased demand for social welfare expenditures during recessions, in which case this benchmark might understate the optimal size of a rainy day fund. We take no stand on optimal state fiscal policy, but use this measure of fiscal stress only as a benchmark. Because the benchmark incorporates reductions in spending growth and increases in taxes, it captures exactly the effects that rainy day funds are designed to reduce.

The National Conference of State Legislatures Fiscal Affairs and Oversight Committee suggests that states hold balances equal to 5 percent of the state budget.23 Gold, however, suggests that the volatility of the state's economy, as well as subjective views of the desirability of stability in tax rates, should be taken into account when deciding upon the optimal size of state rainy day surpluses.24 When looking at historical data, Crider (1978) suggests that it is important to consider the impact of a recession on how the actual budget differs from the potential budget (what the budget would have been in the absence of the recession).25 Thus, there is some reason to consider differences among states when examining the performance of rainy day funds.

Table 6 calculates the amount of funds each state would have to have had on hand in 1988 to completely eliminate any fiscal stress over the next four years of belowaverage economic growth. Because this measure is derived from an ex post standpoint, it would give the exact amount states would have needed to have had on hand to mitigate the effects of the 1991 recession. An alternative measure, however, is to add to this amount the actual surplus the state had coming out of the recession, under the assumption that those states wanted to carry those balances as a cushion. These two measures will differ only for states that did not use all of their surplus balances or states that ran a deficit. State balances can be referred to as fully funded if the balances would have been sufficient to carry the state through 1989-1992 without any fiscal stress. Table 6 shows, for each state, the amount of surplus balances the state had going into the recession, their usage of these balances by fiscal year, and the two measures of fully funded balances. "Fully funded size 1" represents the ex post amount just needed to eliminate all tax increases and reductions in expenditure growth, and "fully funded size 2" represents the amount in fully funded size 1 plus enough to leave balances the same at the end of the recession as they actually were.

It is apparent from Table 6 that most states had balances well below what would have been needed to maintain expenditure growth without tax increases through the 1990-1991 recession. The surplus needed to fund a more severe recession, such as 1980-1982 would be larger than those given here. These estimates show that for most states a 5 percent balance is insufficient to cover average expenditure growth without tax increases during even the relatively mild 1990-1991 recession. The Total row, second from the bottom on Table 6, shows that the average state would have to have had rainy day funds equal to 34.18 percent of its 1988 budget in order to be fully funded according to one estimate, and 32.79 percent according to the other, in order to avoid any tax increases or expenditure growth reductions.

Table 6 Continued.

This measure of full funding in the Total row adds all withdrawals from rainy day funds, but does not offset reductions of funds of some states by the increases of others. If the accumulated funds of those states with increases are used to offset decreases in others, the Pooled Total at the very bottom of Table 6 shows that states would have needed only between 16.76 percent to 17.47 percent of their 1988 budgets to be fully funded. If all states pooled their funds, they could have the same protection with a pooled rainy day fund of 17.47 percent of their budgets as would be produced with individual state funds averaging 34.18 percent of their budgets. For a multi-state fund to work, many problems would have to be worked out, including provisions to overcome the common pool problem.26 A comparison of the Total with the Pooled Total is most useful for showing that individual states can suffer considerably more fiscal stress than all of the states in the aggregate. From Table 6 it is evident that states differ substantially in the amount of surplus balances that would be needed to carry them through a recession with no fiscal stress.

RAINY DAY FUNDS AND AVAILABLE SURPLUS BALANCES

Rainy day funds can help states save to alleviate fiscal stress during recessions, but accumulated general fund balances would serve them just as well. There is always the temptation to spend surplus general fund balances, however, so explicit rainy day funds have the potential to help states accumulate funds to be used during recessions. Keeping in mind that rainy day funds are close substitutes for surplus general fund balances, do rainy day funds actually add to surplus balances available to states to smooth spending during recessions? To answer that question, a regression was run using actual surplus balances (SB) in 1988 as the dependent variable, to see how rainy day funds affected total funds available prior to the 1990-1991 recession.

The actual surplus balances should be a function of the amount states would like to have and the political institutional constraints that could keep actual balances from equaling desired balances. Desired balances should be an increasing function of fully funded balances, calculated in the previous section, so fully funded balances (FFB) is included as an independent variable. The other independent variables are the institutional features of rainy day funds that vary among states. The regression was run on

the forty-eight continental states for 1988, yielding:

SB = 224.2 + .03FFB - 42.5RDF + 169. lREQSAV - 63.5W1THDR (3.03) (4.58) (0.46) (1.88) (0.48)

- 383.7CAP + 50.7CAP% + . (2.57) (2.15)

R2 = .43 (t-values in parentheses)

The results of this regression lend additional insight into the regression results reported earlier, and show again that the most important feature for the effectiveness of rainy day funds is a provision requiring that money be put into them. The coefficient on FFB, fully funded balances, is significant and positive, providing the expected result that the larger a state's fully funded balance, as calculated in the previous section, the larger its surplus balances were going into the 19901991 recession. However, both SB and FFB are denominated in millions of dollars, so the coefficient shows that for each additional dollar of fully funded balances, actual surplus balances were larger by only three cents. The relationship is statistically significant, but small.

Both cap variables are significant at better than the .OS level. CAP is a binary variable showing that the existence of a cap on a rainy day fund lowers surplus balances by $384 million, while each percentage increase in CAP% increases surplus balances by $50 million. Thus, a cap of 5 percent using these figures would lower surplus balances by $384-$50*5=$134 million. A cap of 7.62 percent would be the point at which the two effects would cancel each other. Of the sixteen states with caps, only four have caps above this level, suggesting that in most cases, a rainy day fund with a cap is detrimental to a state's readiness for a recession.

The required savings variable, REQSAV, is significant at the .07 level, and reinforces the earlier finding that having an explicit requirement that the state deposit money into the rainy day fund is the most important factor in creating an effective rainy day fund. REQSAV is a binary variable, so states with a saving requirement, on average, entered the recession with $169 million more in surplus balances. In contrast, the lack of explicit withdrawal requirements did not seem to reduce the effectiveness of rainy day funds.

CONCLUSION

This article has focused on fiscal stress caused by recessions because states face more serious fiscal problems during recessions than during non-recessionary times. Rainy day funds provide a mechanism for easing the fiscal stress caused by economic fluctuations. States face other fiscal problems in addition to those brought on by business fluctuations, and those other problems may be more severe. The increased budgetary demands of Medicaid have strained state budgets, as have increased demands for expenditures in other areas such as prisons, law enforcement, and education. In addition, fiscal problems on the revenue side of the budget include uncertainties about federal assistance, and problems with tax bases such as general sales and motor fuels that have not kept up with income growth. This list is not meant to be exhaustive, but rather to illustrate that states face many fiscal problems that may be more serious than the cyclical variability of revenues.

Indeed, the cyclical variability of revenues is explicitly a short-term problem that appears only during recessions, whereas these other problems are more long-term in nature. Smoothing expenditures over the revenue cycle by using a rainy day fund will not directly solve many other major problems that states face, but it can help states cope with recessions. Because expenditures will be smoothed over the revenue cycle, it can also help states face up to their long-term budget constraints by preventing states from using temporary revenue gains during economic upswings to postpone dealing with longer-term fiscal problems. While rainy day funds are not the ultimate answer to most fiscal problems faced by states, they do deal directly with one clearly defined problem, and many states have established rainy day funds in order to try to ease the fiscal stress caused by recessions.

Rainy day funds are relatively recent as a nationwide phenomenon. Only twelve states had them in 1982, whereas forty-four states had them by 1994. Some form of saving over the business cycle is necessary for states if they do not want their expenditure growth rates to decline during recessions without raising taxes, because unlike the federal government, states typically face some type of balanced budget requirement. Only by saving money ahead of time can states avoid fiscal stress caused by reduced revenue growth during economic downturns.

Despite the fact that most states had rainy day funds during the 19901991 recession, states still exhibited a significant amount of fiscal stress. After depleting the majority of their surpluses by the middle of fiscal year 1991, states turned to major tax increases to maintain their expenditures. Even with these tax increases, most states were unable to maintain expenditure growth at previous levels. This article computed a measure of state fiscal stress during the 1990-1991 recession as the amount of discretionary tax increases plus the amount by which expenditure growth fell below average. Then, an empirical model was used to see whether the presence of an explicit rainy day fund had an effect on the degree of fiscal stress experienced by a state.

The mere existence of a rainy day fund in a state had no measurable effect in limiting fiscal stress. However, rainy day funds were effective in reducing fiscal stress if they had mandatory requirements for making deposits. One might be concerned that once established, the funds would provide a source of revenue that could be raided by the legislature rather than truly being saved for a rainy day. However, the empirical results indicate that rainy day funds are equally effective with or without a stringent requirement on the withdrawal of funds. The empirical results also show that for a given amount of fiscal stress, states that have rainy day funds are more likely to cope with that fiscal stress through spending reductions than through increases in taxes.

If states had wanted to maintain expenditure growth without tax increases during the 1991 recession, they would have needed rainy day (or other available) funds that were, on average, around 30 percent of their general expenditures. This is not intended as a recommendation that states have rainy day funds this large, but rather is an observation that if they do not, they are likely to have to rely on tax increases or declines in expenditure growth to cope with recessions. Given that many states created rainy day funds after the 1980-1982 recession to minimize future fiscal stress, the degree to which they actually helped is an empirical question which has not yet received much attention in the literature. The conclusions of this article suggest that properly structured rainy day funds did ease the fiscal crises during the past recession, but that they still were insufficient to prevent major tax increases and expenditure reductions.

ACKNOWLEDGMENTS

The authors wish to thank Stacy Mazer from the National Association of State Budget Officers and Corina Eckl from the National Conference of State Legislatures for their help in acquiring data for this article. The article was improved due to comments from seminar participants at Florida State University.