West Virginia's Tax Structure and Rainy Day Fund

Russell S. Sobel

On March 11, 1994, the West Virginia state legislature adopted H.B. 4018, an amendment to the state code that created the revenue shortfall reserve fund. This fund is often called the rainy day fund. It sets aside revenue during healthy fiscal years to meet unanticipated revenue shortfalls during poor fiscal years. West Virginia is the 45th state to create a fund of this type. For the vast majority of states, the main reason for establishing a rainy day fund was to help cope with the predictable losses of tax revenue associated with economic recessions. In fact, most states created their funds soon after a recession ended (Gold 1983). Because revenue fluctuates systematically with the ups and downs of the state economy, a rainy day fund can smooth the revenue available over the business cycle, In this manner, states ease the fiscal stress associated with recessionary times.

The degree to which a state's tax revenue fluctuates with the economy is one of its most important budgetary considerations. If a state's revenue is very unstable over the business cycle, it must take precautions to prevent major fiscal crises from happening during recessions. One of its options is to set aside a fairly large rainy day fund balance. Another option is to reexamine its tax structure to see if any changes can be made to promote a more stable flow of revenue.

This article analyzes the revenue stability provided by West Virginia's current tax structure and how the rainy day fund can help to avoid fiscal crises during future recessions. It also discusses the principles of stability-oriented budgeting, and how these principles can be used to improve West Virginia's revenue structure.

West Virginia's Revenue Structure: Budgeting for Stability

West Virginia's revenue comes from many different sources such as federal grants, roadway tolls, higher education tuition payments, lottery proceeds, and, of course, taxes. Each revenue source is dedicated to a state fund, and these funds differ by the types of expenditure programs they can finance. There are four important state funds: the general revenue fund, the state road fund, federal funds and special revenue funds.

Over one-third of the state's $5.9 billion in revenue in FY 1995 was in the state general revenue fund (approximately $2.3 billion). This is the only state fund not earmarked for a specific purpose. It represents the resources of the state available for appropriation by the legislature. The fund contains twenty-four revenue sources, including approximately $743 million from the consumers'sales tax and $709 million from the personal income tax. The business and occupation tax ($196 million), severance tax ($176 million) and corporate net income tax ($145 million) are also important revenue sources in the state general revenue fund (Petry 1995).

About 14 percent of the state's revenue in FY 1995 was in the state road fund (approximately $850 million). The fund's revenue can be used only for expenditures related to the construction and repair of public highways. This fund receives revenue from transportation-related taxes such as the excise tax on gasoline ($212 million), the privilege tax on automobiles ($112 million), licenses and registration fees ($70 million), the consumers'sales tax on fuel ($65 million), and federal highway and safety money ($395 million) (Governor's Office 1995; Petry 1995).

Together, the general fund and the state road fund encompass the vast majority of state tax collections, and the revenue in these funds are concentrated in a few major tax sources. Understanding the susceptibility of these tax sources to recessions is important in analyzing the stability of West Virginia's tax revenue.

Federal grants (approximately $2.1 billion) accounted for about 35 percent of the state's revenue in FY 1995 (Governor's Office 1995). They are given to the state to be spent on specific expenditure items. Federal aid for roads is deposited in the state road fund and the remainder of federal money (approximately $1.7 billion) goes into the federal funds account. While federal money represents a very important source of revenue for West Virginia, its variability is out of the state's control. This is particularly apparent during the current federal budget debate in Washington.

About 17 percent of the state's revenue in FY 1995 was in the special revenue fund (approximately $1 billion). It is an aggregation of many smaller funds. The largest single category of special revenue funds is the portion raised by colleges and universities from tuition payments and other charges and fees, such as athletic ticket and bookstore sales. Revenues from the soft drink tax and health care provider tax also go into special revenue funds (Governor's Office 1995).

Because these funds have different revenue sources, they have differing degrees of variability over the business cycle. West Virginia can use this knowledge to its advantage by matching expenditure programs that are relatively easy to cut during recessions with revenue sources that are relatively unstable over the business cycle. The most important expenditure programs, on the other hand, the ones that are very difficult to cut during recessions, should be funded with relatively stable revenue sources.

The State General Revenue Fund's Stability

The key to the state general revenue fund's stability is diversification. Because its revenue is not earmarked for specific purposes, its revenue sources can include a mixture of both stable and unstable tax sources. During a recession, the unstable revenue source will provide less revenue, but this shortfall will be at least partially offset by other, more stable, general fund revenue sources.

This study focuses on the fluctuations in revenue due to the business cycle, "cyclical variability," and not the year-to-year fluctuations due to other factors such as federal tax law changes, labor disputes, weather conditions, or court settlements. The major difference between these two types of variability is that random variance is much easier to deal with than systematic cyclical variability. For example, if a state has five sources of revenue, some will be randomly higher than usual and others will be randomly lower than usual in any given year. The combined revenue from all sources is more stable than the revenue from any one individual tax. This is the principle of diversification that most investors use when designing a portfolio of stocks. Cyclical variability, on the other hand, cannot be so easily avoided. Even the portfolio of a diversified investor will fluctuate with the overall market. Thus, states with many revenue sources are not totally protected when a recession hits. Their revenue will fall because all major sources of tax revenue fluctuate with the business cycle. Fortunately, states can take advantage of the fact that different taxes have differing degrees of cyclical variability.'

An important thing to remember is that changes in tax laws and tax rates also influence tax revenue. Special attention must be paid to distinguish fluctuations related to the economy from fluctuations related to legislative action. Legislative tax changes are not a central consideration in this analysis because they do not systematically induce recessionary budgetary problems. If anything, tax law changes are induced by cyclically fluctuating revenue. A more stable source of revenue will demand less legislative interference. If stability is a desirable feature in a tax system, however, legislative changes should be kept to a minimum. The following sections of this article explore the susceptibility of West Virginia's major tax sources to economic fluctuations.

The Consumers' Sales and Service Tax

The consumers' sales and service tax is the state's largest source of tax revenue. Because it is based upon consumer purchases, its responsiveness to economic fluctuations depends on how consumers' purchases vary with changes in their income. Because consumers' purchases of some items are much more sensitive to income changes than others, an important feature of the sales tax is which goods are exempt from the tax. When comparing the 46 states that currently have a sales tax, the most commonly exempted items are prescription drugs, food, gasoline, liquor, and clothing. These items are usually exempted from taxation because the burden falls more heavily on families with lower incomes. However, because purchases of these types of items do not vary much with income levels, exempting them from the sales tax makes the revenue from the tax less stable over the business cycle. Figure 1 shows how retail sales fluctuate with the economy at a national level. These data have been adjusted for inflation and represent the annual percent change in total retail sales, retail sales excluding food, and the U.S. economy as measured by gross domestic product (GDP). The figure shows that when food purchases are not included, the retail sales tax base is more severely influenced by economic conditions, with more revenue being generated when the national economy is growing and less when it is in recession.

West Virginia has removed several items from exemption; most notably, gasoline in 1983, soft drinks in 1988, and food purchases in 1989. Additionally, the sales tax rate was increased from 3 percent to 5 percent in 1981, from 5 percent to 6 percent temporarily in 1988, and this final increase was made permanent in 1989. Thus, the 1980s was a period of increased reliance on the consumers'sales tax, through both the removal of exemptions and increasing tax rates. These changes were primarily enacted in a search for additional revenue, but the removal of the food and gas exemptions, most notably, will create additional revenue that is more stable than the prior revenue from the consumers' sales tax.

The additional revenue generated by these changes was often dedicated to specific programs outside the state general revenue fund, however, limiting the stabilizing influence that these changes will have on the state general fund. The additional revenue from the 1983 revision of the liquor and wine exemption, for example, goes into the drunk driving prevention fund, while the revenue from the removal of the gasoline and special fuel exemption flows into the state road fund.

Perhaps no characteristic of the sales tax varies more across states than the taxation of services. A recent study by the Federation of Tax Administrators (1994) indicated that West Virginia taxed 152 of the 164 major services included in their study, the third highest number in the nation. Only New Mexico and Hawaii ranked higher, both taxing 155 of the 164 services. Although the appropriateness of taxing so many services is subject to debate, it may provide additional stability to the state's sales tax revenue.

West Virginia also has a use tax that supplements its consumers' sales tax. It applies to out-of-state transactions where the consumers'sales tax does not apply, such as mail-order purchases and purchases made by West Virginia businesses from companies outside the state. The use tax's cyclical variability is probably higher than the consumers'sales tax because a much larger proportion of the use tax is accounted for by business purchases, which are more sensitive to the economy.

The West Virginia Public Affairs Reporter, a quarterly refereed journal of the Institute for Public Affairs, is published four times a year. Each issue attempts to bring responsible and intelligent information to bear on West Virginia's governmental problems. The Reporter's content is written by public officials, administrators, faculty, research associates, or others familiar with West Virginia affairs. In all cases, however, the views of the authors are not necessarily those of the Institute for Public Affairs nor of West Virginia University.

Selective Excise Taxes

Selective excise taxes are basically sales taxes on certain items. The most common type of selective excise tax is a sin tax, such as one levied on alcohol, tobacco, or gambling. Taxes on privilege or luxury items are also common. Consumers' purchases of items such as tobacco and liquor are very stable over the business cycle because they do not depend highly on income levels. However, because consumers are becoming more conscious about their health, revenue from these taxes is declining. They are not a good source of long-term revenue growth for the state, but they are a relatively stable source of revenue during recessions.

Selective excise taxes on luxury goods place a larger burden on individuals with higher incomes. However, purchases of luxury items are very responsive to income and employment levels, and fluctuate highly over the business cycle. Revenue from the state's privilege tax on motor vehicles, for example, varies widely from year to year. For comparison purposes, the annual percentage change in tax revenue from the cigarette tax and motor vehicle privilege tax are given in Figure 2 along with data on the state economy. In this figure, the state economy is measured by total personal income. All data have been adjusted for inflation.

Figure 2 shows that revenue from the motor vehicle privilege tax has more variation through time than revenue from the cigarette tax, In fact, the only noticeable change in cigarette tax revenue is the spike in FY 197879 when the tax rate was raised from $0.12 to $0.17 per pack. The recession in the early 1980s, and again the recession in the early 1990s caused large declines in the revenue from the motor vehicle privilege tax, while revenue from the cigarette tax remained fairly constant. This highlights the fact that some taxes have much more variation than others, and that this variation is due to the sensitivity of consumers' purchases to changes in income. Purchases of new automobiles are much more sensitive to income changes than are purchases of tobacco products. The motor vehicle privilege tax's variability is also magnified by the way it is imposed. The entire privilege tax is due at the time of purchase, even if a consumer finances the automobile. This means that the revenue from the tax is based upon purchases within the year, not on the number of automobiles owned, which would be much more stable. Local revenue from the personal property tax on automobiles, for example, is much more stable because it is based on automobile ownership, rather than just purchases during the year.

The soft drink tax is another example of a selective excise tax created to provide funding for a specific purpose; in this case, the state's health sciences school programs. Like "sin" type taxes, the tax on soft drinks is fairly stable over the business cycle. Consumers' purchases of soft drinks are not very sensitive to the economy. Thus, this tax provides a fairly stable level of funding over the business cycle. However, like other per unit taxes, the revenue does not automatically increase with inflation.

Business and Occupation PrivilegeTax

The business and occupation privilege (B&O) tax has undergone major changes in its history. In 1985, the state legislature repealed the tax for all businesses except public service and utility businesses, and replaced it with a system of new taxes. The new taxes include the severance tax, business franchise tax, and corporate income tax, which will be discussed later. In 1995, the state legislature adopted H.B. 2267, again drastically changing the B&O tax. Instead of taxing actual production, the new tax on electric utility companies is based on the average annual units of generation produced from 1991 to 1994. The stream of revenue from this revised tax will be almost constant each year because of the way it is calculated. That is good in the sense of revenue stability, but it will also reduce the tax's long-run growth potential. Revenue from the tax will not automatically increase with inflation, nor increase with growth in electricity generation. Also, under the new provisions, power produced outside the state, but sold within West Virginia, will be taxed based upon actual yearly electricity sales within the state. This portion of the tax will fluctuate through time with changing amounts of imported electricity. Nonetheless, the overall impact of this tax change is going to be a very large increase in the stability of the revenue from the B&O tax.

The Severance Tax

While the severance tax applies to many different natural resources, including timber and limestone, the major component is coal. Coal severance accounts for almost 90 percent of all severance tax revenue. Because the tax is levied on the value of extracted material, changes in the revenue from this tax can come either from changes in the prices of natural resources or changes in production levels. For example, between 1973 and 1974 the average price of coal nearly doubled, going from $8.53 to $15.75 per short ton (Energy Information Administration 1991). This caused the severance tax revenue from coal to increase substantially. Compared with a tax based upon the tonnage of coal, the tax on the value is only advantageous during periods when the price of coal is rising. Since 1982, however, the price of coal has been failing, reducing the state's revenue from this tax.

The remainder of fluctuation in coal severance revenue comes from changes in coal production, which is subject to very high variation from random events, especially a major coal strike. However, severance revenue may actually be a stabilizing influence on state revenue during recessions because almost 90 percent of all coal production is used by electric power companies. Thus, if a recession causes the economy to fall, the only major influence it will have on the coal industry is through changes in the demand for electric power and the demand for electric power is relatively stable during recessions. Additionally, since 80 percent of West Virginia coal is exported outside the state, revenue depends more upon national economic conditions than it does on the state economy.

While this is true for more general economic recessions, it is not true for all recessions. Severance taxes can actually be a very unstable source of revenue during a state recession that begins in the coal industry. In this case, severance revenues will decline by a much larger amount than if the recession had been caused by a more national event, such as a stock market crash. Thus, as the state economy continues to become less dependent on the coal industry, severance revenue should become more stable over the state business cycle. Also, fluctuations due to strikes should become lower in the future as the industry becomes less unionized. Thus, the long range forecast for state severance taxes is good. Predictions about the state economy generally forecast that employment in the coal industry will decline but that the total production of coal will remain fairly constant (Hammond and Thompson 1995). It appears that severance tax revenue will be a stabilizing influence on state revenue in the future, even during recessions.

Corporation Net Income Tax

The corporate net income tax is one of the most variable of all sources of tax revenue. Corporate net income declines quite drastically during recessions and expands rapidly during economic booms (see Figure 3). For example, in FY 1991, revenue from the corporate net income tax was down 25 percent from the previous fiscal year, Recessions have a major impact on the revenue from this tax, and it is currently the most unstable source of revenue in the state's general revenue fund.

The Personal Income Tax

The personal income tax is the state's second largest source of tax revenue. At a national level, personal income taxes are more stable than corporate net income taxes and also more stable than sales taxes that exempt food. They are, however, less stable than sales taxes that include food sales (Sobel and Holcombe 1994). Nonetheless, the personal income tax is a strong source of revenue, both for its long-run growth and stability over the business cycle.

Figure 3 shows the percent change in the inflation-adjusted tax revenue from the personal income tax in West Virginia compared with the state economy and the revenue from the corporate net income tax. The stars in the figure represent changes in the personal income tax rates that took effect in fiscal years 1984 and 1988. After taking out the rate changes' impact, the revenue from the tax itself remained fairly stable through both the recession in the early 1980s, and the one in the early 1990s.

Gasoline Excise Tax and Special Fuel Tax

Taxes on gasoline and other fuels have grown in importance in recent years. While revenue from the gasoline iue iri reuerit yearb. vvilme revenue irurn Life qa6oiine

excise tax and special fuel tax historically supports the state road fund, additional revenue generated from a recent five cents per gallon increase is being directed into the federal aid highway matching fund. This 1993 rate increase is temporary, and is scheduled to expire in 2001.

Gasoline has been subject to the consumers' sales tax since 1983. The revenue generated from the taxation of gasoline under the consumers'sales tax also goes into the state road fund. A 1992 U.S. Census Bureau study showed that West Virginia ranked 14th highest of all states in per-person gasoline taxes. The reason for the continual increase in gasoline tax rates through time is that they are levied on a per-gallon basis, which means that revenues do not automatically increase with inflation. This requires the state to increase the per-gallon rate frequently to keep revenue growth at an even pace with inflation. The consumers' sales tax that applies to gasoline at a 5 percent rate is also subject to this problem because of the way it is calculated. Instead of being a percent of the retail price, as it is calculated for other items, the tax is based upon the average wholesale price. Like the severance tax, this tax has a minimum floor that prevents adverse fluctuations in the price of gasoline from having a large impact on revenue. Currently, the wholesale price is below the floor amount of 97 cents per gallon so even if gasoline prices rise, revenue will not increase unless the wholesale price rises above the floor amount. Taxes on gasoline are, however, fairly stable over the business cycle, as consumers' purchases are not very sensitive to income fluctuations.

A Final Comparison

While the logic behind the analysis of revenue variability is straightforward, making direct comparisons of each source of tax revenue is difficult. Using yearly data from 1963 to 1994, however, it is possible to decompose the variation in each source of tax revenue into its two major components, cyclical variability and random variance. Cyclical variability is measured as the change in revenue historically associated with a 1 percent change in the state economy. The movement in tax revenue not accounted for by movements in the state economy is then considered the random variance. This methodology for determining these estimates is fairly standard in the field of tax analysis (see Sobel and Holcombe 1994). It is important to remember that these estimates are based upon historical relationships between the taxes and the state economy. They are not intended for use in revenue forecasting, but as a guide to the relative variability of the different taxes.

Table 1 lists these estimates for each of West Virginia's major taxes. The first column shows the percent of the revenue in the fund that comes from the specific tax, measuring its relative importance for the state. The column labeled cyclical variability shows the change in the revenue from the tax that has historically been associated with a 1 percent change in the state economy, as measured by state personal income. Thus, for example, a 1 percent change in the state economy is usually associated with about a two and one-half percent change in the revenue from the corporate net income tax. The next column gives the ranking of the taxes from highest to lowest cyclical variability. The next column gives an index of the tax's random variability. A value of one would suggest that all of the movement in the revenue from the tax are due to random fluctuations, while a very low number would suggest only a small degree of random variance in the tax. This random variance measure excludes any movements in revenue associated with the business cycle or tax law changes. It can be viewed as the predictability of the tax revenue from that source. The final column gives the ranking, from highest to lowest, of the tax's random variance. A rank of one is the least desirable tax from the standpoint of stable revenue in both rankings.

The state road fund is a well-designed fund for revenue stability. It includes the most cyclically variable tax, the motor vehicle privilege tax, but also includes one of the most stable taxes, the gasoline excise tax. This is a good balance because fluctuations in one tax are partially offset by the stability of the other. Additionally, the motor vehicle privilege tax has much less random variance than the gasoline excise tax. Again, this is a good match between two taxes that reduces overall variability. The soft drink tax, which goes to fund health sciences school programs, is also one of the more stable tax sources over the business cycle. It does, however, have a fairly high degree of random variance.

The taxes that compose the state general revenue fund are perhaps the most important. Two estimates are given in the table for West Virginia's largest revenue source, the consumers'sales tax. The first is with food purchases included as they were before 1979 and are today. The second is the estimate for the sales tax when food purchases are exempt, as they were from 1979 to 1989. There is a substantial difference in the cyclical variability of the revenue from the sales tax with food exempted than with it included. With food exempt, the consumers' sales tax is the second most variable of all taxes, averaging a 2.86 percent response to a 1 percent change in the state economy. The sales tax's variability is significantly reduced to 0.915 when food purchases are included, making it the second most stable tax source of the ones listed. It is also one of the most predictable taxes, with random revenue fluctuations small compared with the other taxes.

The personal income tax is also one of the more stable revenue sources, averaging a 1.1 percent response to a 1 percent change in the state economy. These results confirm that the sales tax is more stable than the personal income tax when food is included, but is more variable when food is exempt. Additionally, it is confirmed that the corporate net income tax is the most variable revenue source in the state general revenue fund.

Historically, the B&O tax on electric utilities is one of the least cyclically variable revenue sources for West Virginia. It ranks highest, however, in random variation among all of the tax sources. Thus, the tax is fairly recession proof, but is unpredictable. This estimate is, of course, for the old B&O tax, and is not representative of the future variability of the tax under the new provisions. The results for the severance tax on coal are given in two ways. The first is just the regular estimate of the relationship between the state economy and the revenue from the tax. This result shows the tax having a high degree of cyclical variability, changing by 2.143 percent for every 1 percent change in the state economy. The second estimate is obtained by seeing what would have happened if the state economy had perfectly followed the national economy over the period. This more accurately reflects the future variability of this tax as West Virginia's economy becomes less dependent upon the coal industry and more representative of the U.S. economy. These results show that the severance tax on coal is one of the most unstable taxes when the coal industry is hit hard by a recession, but that it is more stable during normal national economic slowdowns.

The cigarette excise tax is also one of the most stable general revenue sources, moving approximately 1 percent in response to a 1 percent change in the state economy. It also is very predictable, with little random variance. The last two general fund revenue sources in the table are the insurance tax and the use tax. Both taxes have a high degree of cyclical variability. The insurance tax, however, is subject to more year-to-year random fluctuation than the use tax.

In summary, these results suggest the existence of substantial differences in the cyclical variability of West Virginia's major tax sources. Tax policy aimed at increasing reliance on the more stable sources, and lowering reliance on the most variable sources, will result in a more stable stream of revenue over the business cycle. Just for comparison purposes, the total cyclical variability of the state general revenue fund was also estimated over the same period and is given in the first line of the table. Overall, its revenue changed by 1.413 percent for every 1 percent change in the state economy. The 1980-1982 recession, for example, caused about a 7 percent decline in the state economy. This suggests that West Virginia can expect a cumulative loss of approximately 10 percent of its general fund revenue (inflation-adjusted) during a severe recession. Making cutbacks in state expenditures of this size can be a major budgetary problem for a state like West Virginia. To prevent a similar budgetary situation from happening during the next major recession, the state can attempt to reduce the variability of its tax structure and it can also begin saving money to offset the revenue losses. This is precisely the opportunity presented by the state's newly created rainy day fund.

West Virginia's Rainy Day Fund

As mentioned in the introduction, West Virginia created a rainy day fund in 1994 to save money during healthy fiscal years to help finance expenditures during recessions. The optimal size of a state's rainy day fund is mainly determined by how much revenue losses it can expect during a recession. A state with a relatively stable tax structure will need less money in its rainy day fund than a state with a relatively unstable tax structure.

There is no particular reason that a state must have an explicit rainy day fund to save for recessions. The vast majority of states are allowed to carry around general fund surpluses for many years. Thus, a state may just keep running up its general fund surplus for the next recession. There are, however, some reasons why an explicit rainy day fund may be preferable. Some researchers have pointed out that large general fund surpluses were partially responsible for the voters' motivation for

enacting tax limitation laws in California and Alaska (Gold 1983; Oakland 1979; and Mattoon and Testa 1992). Also, some states have laws requiring them to refund some of their surplus tax revenue. For these states, the importance of establishing an explicit rainy day fund is clear. For other states the advantage of a fund is to put in place explicit requirements for saving money and then to put explicit restrictions on when the money can be used. Creating a rainy day fund through a constitutional amendment is thus preferable to creating one statutorily. A constitutional rainy day fund creates a stronger guarantee that rainy day funds will be saved and also that they will be used only when necessary. This can help to increase the state's bond rating further, as it is a stronger assurance that rainy day funds will be present when needed. West Virginia's current fund is statutory, not constitutional.

State rainy day funds can be designed to serve several different purposes. West Virginia's fund allows it to be used for any revenue shortfall that would otherwise require the governor to reduce appropriations. This will happen whenever actual revenues are less than were forecasted. A federal tax law change, for example, may cause a revenue shortfall that would allow the state to use this fund. Alternatively if the state revenues turn out to be randomly below what had been forecast, this would also allow access to the fund. The fund is set up in a way that will reduce the number of times that the governor must make midyear appropriation cuts. This can be very advantageous to avoiding declines in the state's bond rating (Mooney 1994). Unfortunately, the procedures governing West Virginia's revenue shortfall reserve fund will prohibit the state from using the money during a recession if the falling revenues were accurately forecasted. This defeats the fund's purpose of providing revenue stability during recessions. Another problem with the current structure of the fund is that during a prosperous fiscal year this fund could be drawn down if the revenue forecast was too high, even though the state economy was not in a recession. Thus, the fund may be depleted during healthy fiscal years, and may not be available during poor fiscal years.

To give a literal example, the fund cannot be spent on all rainy days. On a rainy day, the fund can only be used if it were predicted to be sunny. When it rains, and it was predicted to rain, the fund cannot be used. If a goal of the fund is to help prevent fiscal crises, this single policy is perhaps the major barrier to achieving that goal. Unless the trigger mechanism is changed, the legislature may have to amend the state code during the next recession to use the money in the fund. Without changing the code, the only deliberate way to use the fund during a recession is to over-forecast revenues. While revenues are frequently less than were forecast during recessions, this cannot be expected, especially with the increasing accuracy of state revenue forecasts.

Many states with rainy day funds have procedures in place to ensure that their rainy day funds can be accessed during recessions and that they are not abused at other times for political purposes. Several states have an explicit formula for withdrawals (ACIR 1994). Indiana, for example, can only use its "counter-cyclical revenue and economic stabilization fund" if adjusted personal income declines by more than 2 percent in a given year. Maryland can use its "revenue stabilization account" only if the state unemployment rate is greater than 6.5 percent, and greater than the rate 12 months earlier. Michigan's "countercyclical budget and economic stabilization fund" is perhaps the most rigidly designed fund, it can only be used if the annual growth rate in real personal income is negative, and then the amount that can be withdrawn is set by a formula. Other states are not so stringent, but have provisions directly regarding economic recessions. Pennsylvania, for example, can use its "tax stabilization reserve fund" if the governor declares an emergency or a downturn in the economy, and the withdrawal is approved by a two-thirds vote of each house. Up to half of Oklahoma's "constitutional reserve fund" can be used if the coming fiscal year's general revenue fund is less than the current year's amount. Virginia's "revenue reserve fund" can be used for revenue losses caused by economic conditions or by changes in federal tax legislation. On top of these provisions, most states also require legislative approval, some even requiring supra majority approvals. The point is that many states have an explicit provision allowing them to use their fund during a recession.

Many states also have explicit provisions governing the depositing of money into their rainy day fund. West Virginia's fund has a provision that 50 percent of the year-end surplus must be deposited in the fund. Indiana must annually deposit an amount equal to the annual growth rate in personal income in excess of 2 percent times the previous year's general fund revenue. Michigan must deposit an amount equal to the annual growth rate in real personal income in excess of 2 percent times general fund revenue of the fiscal year. Washington's "budget stabilization fund" requires deposits in the amount of the projected growth in real personal income minus 3 percent times the previous fiscal year general state revenue. Many other states have provisions for the annual deposit of a certain percentage of general revenue, while others have provisions for depositing a percentage, or all, of the year-end surplus, Some states rely only on legislative appropriations for deposits. Research on the effectiveness of state rainy day funds usually concludes that they are under-funded without some kind of explicit formula for depositing money. It is often too difficult for the state legislature to appropriate money to the fund when there are so many expenditure programs in need of extra money. This could be a substantial problem for West Virginia in the future. The way the fund is currently set up, deposits are made only when the state runs a general fund surplus. This can arise either when the leg

islature does not appropriate all of the forecasted revenue or when actual revenue is higher than forecasted. Thus, most of the deposits will be unintentional, due to revenue forecast errors. Additionally, when large forecast errors are made, the governor sometimes adjusts the revenue estimates, allowing the legislature to enact supplementary appropriations. Currently, it is very difficult for the legislature to make a specific dollar deposit into the fund.

Most states also place a cap, or maximum size, on their rainy day fund. Generally the caps range from 5 percent to 10 percent of general fund revenue. West Virginia's cap is 5 percent of the previous fiscal year's general fund appropriations. The presence of a cap is more important when a strict formula is used to set deposits into the fund. The cap, however, also has a drawback of constraining the amount of funds a state may save for a recession.

Several factors go into determining the optimal size of a state's rainy day fund. First, and most important, is the cyclical variability of its revenue. After the state knows how much revenue losses to expect during a typical recession, it must then decide how to handle the losses. The state could allow expenditures to fall by the total amount of the revenue losses. In this case, there is no need for a rainy day fund. More than likely the state will wish to maintain expenditures above what the total revenue losses would dictate. In this situation, the state could increase taxes to maintain expenditures, by either raising existing tax rates or by establishing new taxes. This is precisely what West Virginia did in the late 1980s and early 1990s when it increased the sales tax rate and removed the food exemption. These recessionary tax increases, however, are not only politically costly, but they can drive the state economy further into the recession. The alternative to raising taxes is to draw on previously stored surplus balances from the rainy day fund. Thus, the optimal size of the fund depends upon the cyclical variability of current taxes, the ability of the legislature to cut current expenditure programs, and finally, the willingness of the legislature to increase taxes.

During the poor fiscal times of the late 1980s and early 1990s, West Virginia was among the states with the largest tax increases in the nation. In FY 1990 alone, West Virginia had a larger percent of its revenue generated from tax changes than any other state in the nation (National Conference of State Legislatures 1991). Almost 19 percent of FY 1990 tax revenue was generated from tax increases. Again in FY 1993 and FY 1994 the state received a substantial portion of its revenue from tax increases. Almost 9 percent of FY 1994 revenue was the result of tax increases, making West Virginia the 4th highest in the nation in that year. Between 1989 and 1994, West Virginia enacted approximately $800 million in new taxes. The establishment of a properly structured rainy day fund could prevent the need for such large tax increases during future recessions.

The funding necessary to ease West Virginia's fiscal pressure is a very complicated issue. As mentioned earlier, a balance of about 10 percent would have been needed to cope with the recession of 1980-1982, which was mildly severe. The precise figure is something that deserves a detailed analysis of its own, but a rough estimate would be somewhere between 4 percent and 8 percent of general fund revenue. The National Conference of State Legislatures has recommended that states carry a 5 percent balance, which is probably sufficient for minor economic fluctuations. West Virginia's revenue shortfall reserve fund has a cap at 5 percent, effectively limiting the amount that it can save. Another important consideration is whether additional money should be saved to help support expenditures from funds other than the general revenue fund. If the state would like to maintain expenditure levels in the state road fund or programs financed partially by federal aid, a higher balance would be needed, perhaps ranging from 10 percent to 15 percent of the general fund budget.

It is very important, however, that the fund is built up to at least the 5 percent level. Going into the 1990-1991 recession, the average state had only about a 2 percent surplus reserve. Most states ran out of rainy day funds very quickly and had to turn to tax increases during the second fiscal year of the recession.

At the end of FY 1993-94, West Virginia made its first deposit into its rainy day fund: $20,564,000, or approximately 1 percent of the state's general revenue fund. The state's surplus at the end of FY 1994-95 was $86.3 million (Smith 1995) resulting in another deposit of $43.1 million. The forecast for FY 1995-96 shows a potential deposit that is much less than the 1994 and 1995 amounts, perhaps as low as $3 to $6 million (WV Senate Journal 1995). Thus, after the deposit at the end of FY 1995-96, the rainy day fund will have approximately $67 million, or about 2.9 percent of the estimated state general fund appropriations at that time. A 5 percent rainy day fund would contain approximately $116 million. Assuming no withdrawals from the fund, and an average deposit rate of $10 million per year, it will take until FY 2000-01 to amass enough of a surplus to reach 5 percent of the FY 1995-96 budget. Of course, the amount that will be needed to equal 5 percent of that year's general fund appropriations will be larger. If the rate of deposit into the fund were raised to $20 million per year, the fund would reach the 5 percent level in FY 1998-99. The real question is how much time is left until the next recession. If the next recession was not until 2005, then a slow rate of deposit would be fine. If a recession happens in the next few years, the fund may be depleted without helping much, and the savings process must start over.

Conclusions

The most important and immediately needed change to the code governing the rainy day fund is a revision of the provisions governing its usage during recessions. As the fund is currently set up, it can only be used to cover the difference between actual revenues and forecasted revenues. Thus, another great depression could hit, and if the governor's office accurately forecasted the decline in revenue, the state could not use the fund. With the increasing accuracy of state revenue forecasts, this problem might even become worse in the future. During the next recession it will be hard for many voters to see their expenditure programs cut while the state has a large balance in its rainy day fund that cannot be used.

The state should also adopt a more specific formula for the depositing of funds into the rainy day fund. Under the current provisions, deposits are only made if there is a surplus generated by the legislature not appropriating all forecasted revenue, if the governor underestimates revenue, or if the legislature makes a direct appropriation to the fund. The first two options are accidental in nature, while the third option is very unlikely given the demands placed upon the legislature for other spending programs. Explicit provisions for deposits and withdrawals into West Virginia's rainy day fund should be created, allowing for a more deliberate control of the fund's size. Changing the fund to a constitutional, rather than a statutory provision, would also aid in making the fund more stable. The fund's statutory nature does not prevent the legislature from accessing the fund for any purpose through a supplemental appropriation. This uncertainty can damage the state's bond rating, raising the cost of borrowing money. Combined, the changes mentioned above would greatly enhance the state's ability to stabilize expenditures over the business cycle. It is worth noting that these changes need not come at the expense of other rainy day fund purposes, such as accommodating poor revenue forecasts and state-wide emergencies. Many states have created constitutional rainy day funds which serve all these purposes. Additionally, changing the state tax structure to reduce revenue variability would allow the state to maintain a smaller rainy day fund without compromising stability.

References

Department of Tax and Revenue. January 1995. West Virginia Tax Expenditure Study: Consumers
    Sales Tax and Use Tax Expenditures. Charleston, WV: Research and Development Division,
    Department of Tax and Revenue.

Energy Information Administration. November 1991. Quarterly Coal Report, April-June 1991.
    Washington, D.C.: Energy Information Administration.

Federation of Tax Administrators. 1994. Sales Taxation of Service: An Update. Washington, D.C.:
    Federation of Tax Administrators.

Gold, Steven D. 1983. Preparing for the Next Recession: Rainy Day Funds and Other Tools for the
    States. Legislative Finance Paper No. 41. Denver: National Conference of State Legislatures.

Governor's Office, 1995. The Executive Budget, Fiscal Year 1996. Charleston, WV: The
    Governor's Office.

Hammond, George and Eric Thompson. 1995, West Virginia Economic Outlook. Morgantown,
    WV: Center for Economic Research, West Virginia University.

Legislative Auditor's Office. 1994. Digest of Revenue Sources in West Virginia. Charleston, WV:
    Legislative Auditor's Office.

Mooney, Christopher Z. 1994. "The West Virginia State Budget Process." The West Virginia
    Public Affairs Reporter 11:1 (Winter): 2-16.

National Conference of State Legislatures. Various years. State Budget and Tax Actions. Denver:
    National Conference of State Legislatures.

Oakland, William H. 1979. "Proposition 13, Genesis and Consequences." Review of the Federal
    Reserve Bank of San Francisco, pp. 5-14.

Petry, Kerri. 1995. Tax Analyst, West Virginia Department of Tax and Revenue. Memorandum
    concerning state revenue for FY 1995. September 26.

Smith, Roger L. May 19, 1995. Director, West Virginia Department of Administration, Finance
    Division, Budget Section. Interviewed by Russell S. Sobel.

Sobel, Russell S. and Randall G. Holcombe. 1994. "Measuring the Income Elasticity of Tax Bases
    Over the Business Cycle." West Virginia University, Department of Economics, Working Paper
    No. 94-11.

U.S. Census Bureau. 1994. Statistical Abstract of the United States. Washington, D.C.: U.S.
    Government Printing Office.

U.S. Advisory Commission on Intergovernmental Relations. 1994. Significant Features of Fiscal
    Federalism. Washington, D.C.: U.S. Advisory Commission on Intergovernmental Relations.

West Virginia Blue Book. 1995. Ed. Darrell E. Holmes. Charleston, WV: R.R. Donnelley and Sons.

West Virginia Senate Journal. March 11, 1995. Senate Executive Message No. 19 (As Corrected).

__________

Russell S. Sobel is Assistant Professor of Economics and Research Associate, Center for Economic Research, at West Virginia University.