State revenue forecasting: An institutional framework.

The revenue forecasting environment can be as important as the projections themselves, since credibility ultimately influences utility. The independence and transparency that characterize Minnesota's forecasting framework help keep the state's budget process on track.

All revenue forecasters know a sad truth: the only thing certain about their forecasts is that they will be wrong. Whether measured individually or as a whole, state revenue forecasts are never accurate. Individuals who are unable to work in an environment in which they will never be right, and where changes to their projections are broadcast widely, quickly find other forms of employment. Those who remain, however, derive satisfaction from the knowledge that their work is an essential part of the budgetary process. Revenue projections provide the starting point for dealing with the challenges that accompany each new budget. By quantifying the size of the gap between spending and expected revenues, forecasts help focus the budget discussion, providing necessary discipline for negotiations between the executive and legislative branches of government.

Forecasts are never popular, nor are they ever fully accepted by policymakers. In both good times and bad, everyone would like to believe that revenues will exceed projections. Because forecasts are never accurate, they are easy targets for those whose back-of-the-envelope projections are more optimistic. Projected surpluses, particularly large surpluses like those that were widespread in the late 1990s, are an irritant to elected officials who could have used the unanticipated revenues for additional tax cuts or spending increases in popular programs. However, projected revenue shortfalls pose even worse problems for policymakers, since shortfalls create difficult decisions. Criticism of revenue forecasts is particularly acute during economic downturns. When revenues fall short of estimates, the forecasting methodology always receives extra scrutiny.

This article describes the institutional framework surrounding the production of the State of Minnesota's revenue forecast. This framework served the state well during the 1990-1991 recession and has worked reasonably well during the current downturn. Of course, there is no guarantee that it will continue to work in the future, or that it would work for other state governments. Although the forecasting framework has evolved over time, much of the initial structure was established following some unfortunate political meddling with state revenue projections that produced an overly optimistic revenue forecast just as the national economy was entering the twin recessions of the early 1980s. (1)

Minnesota's Revenue Forecast: An Overview

Minnesota's revenue forecast is prepared and issued by the Finance Department. This department is also responsible for preparing the governor's budget and for operating the statewide accounting system. State statutes require that a forecast be issued each November and February. November's forecast serves as the basis for the governor's budget recommendations, while February's forecast provides the governor and the legislature with an updated revenue outlook just prior to the legislature's deadline for setting its expenditure and tax targets.

Minnesota is a biennial budget state, and the constitution prohibits an operating deficit for the current biennium. As such, spending and/or revenues are expected to be adjusted in the off-year session should a deficit be forecast. Spending and revenue adjustments also have been made when the forecast showed a surplus, although the current governor's policy has been to rebate any surplus forecast after approval of the biennial budget.

Under Minnesota statutes, the Finance Department's forecast must extend through the period covered by the budget under consideration. It also must provide revenue and expenditure planning estimates for the biennium that follows. (2) This means that the November forecast in even numbered years covers three fiscal years, looking forward 32 months. Revenue planning estimates cover a fourth and fifth fiscal year, extending the period under consideration to a total of 56 months. Revenue and expenditure forecasts are based on current law, and neither reflect any of the governor's tax and spending initiatives nor anticipate any potential legislative action.

The Finance Department's forecast is the only forecast available to state policymakers. There is no competing forecast prepared by a legislative budget office or legislative research staff. Neither the revenue outlook nor the underlying economic forecast is the result of negotiations between the governor and legislative leadership. Instead, a small, full-time staff within the Finance Department prepares the forecast based on the United States economic outlook provided by a national forecasting firm. Every effort is made to ensure the forecast offers an independent assessment of the outlook for state revenues. Indeed, statutes prohibit the finance commissioner from informing the governor of forecast results prior to their official release.

The Goal: An Independent, Professional Forecast

Revenue forecasters quickly learn that forecasting proficiency alone does not ensure that their forecasts meet the objective of providing a mutually accepted starting point for budget negotiations. Even if by some miracle a particular forecast were to turn out to have been perfectly accurate, that forecast has no value unless it is the basis for setting the budget. If budget negotiations between the governor and the legislature are delayed by partisan arguments over the expected level of current law revenues or expenditures, the forecast has not effectively performed its role in the budgetary process. The same could be said if the legislature were to ignore the forecast and pass a budget in which spending exceeded reasonable revenue estimates.

Recognition of the fact that the forecast is not an end in itself, but rather a key early milestone in the larger, more important process of establishing and approving the state budget has helped shape the way Minnesota's revenue forecast is prepared. Although accuracy is the first priority, acceptance of the forecast is improved if it is believed to have been prepared independent of any particular political agenda. To maintain its reputation for independence, the Finance Department strives to make the forecasting process as transparent as possible.

Many of the changes in Minnesota's forecasting over the last 20 years have come from attempts to enhance the independence of the forecast or increase its transparency. Since 1984, for example, the revenue forecast has been prepared under the direct supervision of the state economist, who is a tenured University of Minnesota faculty member, and not a state employee. This helps insulate the forecast from political pressures since the state economist, who has final responsibility for the forecast, always has the option of returning to a full-time university position. The Finance Department maintains an annual contract with the university for half of this individual's time.

Starting Point: A Credible National Forecast

Minnesota's revenue forecast begins with a national outlook provided by a well-known national forecasting service. The state has contracted with the same firm for two decades, and the stability of this relationship helps reinforce the forecast's reputation for independence. Since many other private firms and most other states use the same national outlook, the possibility of the governor or the Finance Department "cooking the books" by shopping for or choosing a particularly rosy or gloomy scenario is eliminated. Although the national forecasting service has been accused of being excessively pessimistic, the revenue forecast is based on a set of economic assumptions that are clearly independent of the governor's agenda. State economists could certainly prepare their own baseline national economic forecast, but such a forecast would not carry the same credibility and would be subject to doubts about its independence.

The use of a national forecasting service does not mean that the results are accepted without question. Revenue forecasters know that forecasting services typically are slow to call the start of recessions and that forecasting models can produce projections that are decidedly either too optimistic or too pessimistic about the future course of the economy. As a check on the forecasting service's national economic outlook, the Finance Department assembles a Council of Economic Advisors to review the assumed rate of national economic growth prior to its use in the state forecast. Membership on the council is at the discretion of the commissioner of finance and is not subject to the gubernatorial appointment process.

Because Minnesota is a regional financial center, the state has the luxury of calling upon economists who work with national economic data on a daily basis. Several council members regularly prepare their own national forecasts. The seven council members are generally acknowledged to be the top macro and regional forecasters in the state. The council's membership includes the vice president for research of the Minneapolis Federal Reserve Bank and the chief economist of a major financial services firm.

The Council of Economic Advisors holds two working meetings prior to each forecast. These meetings, chaired by the state economist, are considered to be internal to the Finance Department and are closed to the public and the press to ensure that the discussions can be frank. A limited number of staff and members of the majority and minority caucuses of both the House and the Senate are allowed to observe in order to maintain legislative transparency. A report of the discussion is included in the document that accompanies the release of the forecast.

At the council's first meeting, which is held before the release of the national forecast expected to serve as the basis of the state's revenue forecast, members discuss their personal forecasts as well as the outlook for particular sectors or regions of Minnesota's economy with which they are familiar. At the second meeting, the focus is on whether the forecasting service's outlook is consistent with the broad consensus of economic forecasts. Although some council members are more optimistic and some less optimistic than the baseline forecast, all typically agree that the forecast is within reasonable bounds. Occasionally, however, there is a shared belief that the forecast is either too optimistic or too pessimistic. The procedure developed to handle those rare situations recognizes that the damage from an overly optimistic forecast is likely to be greater than that from an unduly pessimistic forecast, and that the credibility and reputation for independence that the forecasting service brings to the revenu e forecast is an important asset.

A decision to use something other than the forecasting service's baseline forecast requires a set of extraordinary actions. First, action only can be taken when a super majority rejects the baseline forecast. Finance Department economists and the Council of Economic Advisors agree that an asymmetric response is most appropriate in such instances. If a super majority of both the council and the Finance Department agree that the economic outlook provided by the national forecasting service is too optimistic, it will be substituted by a more pessimistic alternative. No substitution will occur if there is general agreement that the baseline forecast is too pessimistic. The revenue forecast still will be based on the forecasting service's projections, but the consensus belief that those projections are overly pessimistic will be noted in the materials accompanying the forecast's release. Under no circumstances would it be permissible for the underlying economic assumptions to be adjusted by the finance commissione r or a representative of the governor.

As might be expected, it is extremely difficult to get a super majority of economists to agree on anything. Consequently, there have been few situations in which the council considered requiring the use of an alternative national forecast. In November 1987, following the October stock market crash, the council believed the economic outlook sufficiently uncertain to urge the state to delay the forecast until January 1988, which it did. In October 1990, the baseline scenario from the forecasting service did not include a recession, even though the signs that a recession was well underway were obvious. The council and the Finance Department agreed that the November revenue forecast would have to be based on a scenario that included a national recession. Fortunately, the forecasting service's November outlook included a recession, so the state did not have to develop an alternative forecast. At the height of the uncertainty wrought by the Asian financial crisis, the council was sufficiently concerned about the sh ort-term economic outlook to seriously discuss the adoption of a more conservative forecast, but there was never sufficient agreement to warrant such action.

Monitoring the State Economy

Although Minnesota's economy is well diversified and its industrial mix closely mirrors the national mix, the revenue forecast is based on projections of state economic activity, not national average growth rates. Although maintaining and updating a fairly large model of the Minnesota economy is a significant undertaking that requires immersion in details, the Finance Department believes it improves the accuracy of the forecast. The department is a major consumer of local economic data collected by other state agencies, particularly on employment and income.

This data sometimes provides better insights about the direction of the economy than is available from the national forecasts. For example, information about the rapid increase in the use of performance-based competition in Minnesota influenced the November 2001 and February 2002 revenue forecasts. A quick analysis of a data set containing firm-level data on employment and wages revealed that bonus and stock options accounted for more than 5 percent of wages paid in the fourth quarter of 2000 and the first quarter of 2001. That measure of Minnesota's possible exposure to lower bonuses and a decline in the exercise of stock options was used in developing the forecast for Minnesota wage growth for the ensuing biennium.

In some instances, the Minnesota data is timelier than the national data, so using the state-specific numbers improves the forecast. For the fourth quarter of 2001, for example, Minnesota withholding tax receipts--which correlate closely with the level of wages, bonuses, and other performance-based compensation paid by employers in the state--were more than 1 percent below withholding collections from the prior year. This weakness appeared to continue into the first quarter of 2002. The official national income statistics, however, showed national wages growing at an average rate of more than 3 percent during fourth quarter of 2001. Although there is reason to believe that the growth rate for national wages will eventually be revised lower, basing Minnesota's revenue forecast on the national average growth rates would have almost certainly overstated future Minnesota revenues. Tying. Minnesota's revenue forecast to a state-specific outlook helps eliminate potential sources of error.

Transparency: The Key to Credibility

All revenue forecasters know that although the state and national economic performance is important, many factors affecting state revenues are only remotely connected to national economic aggregates. The growth (or decline) in capital gains realizations projected for the biennium is an obvious example of a major source of revenue that does not depend on GDP growth. Pension income, Social Security income, and IRA distributions also have no obvious tie to estimates of future real GDP growth rates. Changes in federal and state tax laws and federal administrative rules also can cause wide swings in revenues. Given the complexity of the interaction between the economic outlook and state tax law, revenue estimating models can easily seem like a gigantic black box to even the most sophisticated members of the legislature.

Finance Department forecasters recognize that not everyone needs (or wants) to fully understand the details of the models used to prepare the forecast, but they also recognize that a basic understanding of the way the revenue models work enhances the credibility of the forecast. As such, department economists always are available to explain the forecast and the forecasting models, and the published report contains a detailed discussion of the assumptions and the reasons for changing those assumptions from those in the previous forecast.

The most important institutional feature of the revenue forecast is the use of a shared income tax forecasting model. Legislative research staff and analysts from the finance and revenue departments jointly maintain the micro-simulation model used to generate the individual income tax forecast and fiscal notes for proposed tax changes. Legislative staff take responsibility for updating the micro-simulation model's code to reflect current state and federal law. The Revenue Department draws and prepares the nearly 20,000 tax filer sample used in the micro-simulation, and Finance Department economists provide the growth factors used to age the sample data into the future.

Finance, revenue, and legislative staff meet prior to the forecast to discuss concerns about the model and to ensure that the underlying parameter files reflect current law. As part of the forecasting process, the Finance Department calibrates the model so that the model liability matches that reported to the Revenue Department for the most recently completed tax year. Then, using the forecast for the Minnesota economy and other information, the Finance Department determines growth rates for each of the 36 types of income and deductions modeled. On release of the forecast, the Finance Department provides a file with the appropriate parameter growth rates to legislative staff and to the Revenue Department, and those parameters are used in all future fiscal notes on proposed income tax changes.

Because the model used to forecast the income tax is open, shared, and used by legislative staff for tax fiscal notes, the forecast appears less like a black box exercise. The shared model also reinforces the credibility of the forecast, since legislative staff could easily identify any unusual changes to variables driving the micro-simulation model. Equally important, sharing the simulation model for more than a decade has encouraged a spirit of cooperation between the Revenue Department and Finance Department analysts and their counterparts in the legislature. That cooperation has improved the quality of information available to decision makers, because no single individual can be an expert on all aspects of the tax code. This open collaboration among Finance Department staff, Revenue Department staff, and legislative staff is a major factor in supporting the credibility of the revenue forecast, particularly when times turn down.

Looking to the Future

State revenue forecasts will be examined closely over the next year as the national economy moves away from what is likely to be the mildest recession on record. Everyone from ordinary citizens to academic economists will seek explanations for the forecasts' inaccuracy. In some states, special blue-ribbon commissions will be established to evaluate the forecasting process and to suggest ways in which it could be improved. What most will find is that while more powerful computers allow forecasters to test more alternative scenarios than in the past, the underlying relationships between national economic variables and state revenues are still not well enough understood to greatly narrow the confidence band surrounding any revenue forecast.

For example, although the projected change in capital gains realizations accounted for more than one fifth of Minnesota's recently projected revenue shortfall, solid methods for forecasting changes in capital gains realizations simply do not exist. Capital gains taxes depend not only on what happens to the stock market over the next one or two years, but also on whether individuals decide to cash in their earnings or let them ride. Similarly, the increased importance of performance-based compensation has changed the way in which the future growth in wages should be estimated. So far, however, only ad hoc forecast adjustments have been possible, since data separating out bonus and stock option income from normal wages does not exist. When forecasts extend for multiple years, the already sizeable confidence bands widen significantly.

Although further research will gradually improve understanding of the interaction between state revenues and the economy, and forecasting models will improve over time, revenue forecasts will never be precise. State revenues are inherently volatile (and the more progressive the tax system, the more volatile the revenues), which means that on average there always will be sizeable differences between the forecast and actual receipts. But that does not mean forecasts can be ignored. Without a forecast, the alternative is making budget decisions with no guide as to how much spending can be afforded. Everyone chafes at the restraint and discipline applied by the forecast, but even critics agree that an unconstrained budget process without the discipline provided by the forecast would quickly lead to financial disaster.

NOTES

(1.) By the time those budget problems were finally resolved, there had been four special legislative sessions, Minnesota had been forced to borrow short term to overcome cash deficiencies, and its AAA bond rating had been downgraded to AA. Although most legislators who served during that period have now retired, memories of those difficulties still help shape attitudes about the revenue forecasting process.

(2.) Revenue planning estimates differ from the revenue forecast in the amount of detail used in the forecasting process. For example, while the revenue forecast is based on a Minnesota-specific economic outlook prepared by Finance Department economists, the revenue planning estimates for years four and five are based on estimated national economic growth and no attempt is made to fine tune them for Minnesota.

THOMAS F. STINSON, PH.D., is state economist for the Minnesota Department of Finance. Dr. Stinson is also an associate professor in the University of Minnesota's Department of Applied Economics.