Skip Navigation
| Skip Navigation |
|
|
|
|
|
Reliable estimates of federal spending and revenues, and a clear understanding of associated issues, are important to the budget process. The Congressional Budget Office (CBO) has developed elaborate techniques to generate budgetary estimates and to analyze issues affecting spending and revenue totals. In order that these procedures and issues can be evaluated by those outside the agency, it is CBO's practice to provide published analyses of its estimating procedures and of current issues related to major spending items and revenue sources. This paper provides such analysis for individual income tax revenues, the largest single component of federal receipts.
The study was prepared by Frederick Ribe and Kathleen O'Connell of CBO's
Fiscal Analysis and Tax Analysis Divisions, respectively, under the direction
of Rosemary Marcuss. Mr. Ribe is the principal author. Ms. O'Connell is
responsible for parts of Chapters II and IV, and provided expert advice
during all other phases of the project. Others to whom the authors are
grateful for valuable comments are Timothy Considine, Edward Gramlich,
Robert Hartman, Frank de Leeuw, Robert Lucke, Steven Malin, Joseph Pechman,
and James Verdier. Susan Goeransson, Debra Holt, and Naif Khouri provided
research assistance, Patricia H. Johnson edited the manuscript, and Shirley
Hornbuckle and Linda Brockman typed it for publication.
Alice M. Rivlin
Director
August 1983
SUMMARY
CHAPTER I. INTRODUCTION
CHAPTER II. THE CBO INCOME TAX MODEL
CHAPTER III. FORECASTING ACCURACY OF THE CBO INDIVIDUAL INCOME TAX MODEL
CHAPTER IV. ISSUES IN INDIVIDUAL INCOME TAX REVENUES
APPENDIX: ANALYSIS OF ERRORS IN TREASURY REVENUE FORECASTS
| TABLES | |
| 1. | THREE SIMULATED FORECASTS OF REVENUES FROM INDIVIDUAL INCOME TAX USING CBO TAX MODEL |
| 2. | ANALYSIS OF ERRORS IN TREASURY DEPARTMENT FORECASTS OF INDIVIDUAL INCOME TAX REVENUES FOR FISCAL YEARS 1977-1981 |
| 3. | CHANGE IN PREDICTED INDIVIDUAL INCOME TAX REVENUES FROM REVISION IN CBO INFLATION FORECAST FOR 1983-1986 |
| FIGURES | |
| 1. | DISTRIBUTION OF TAXABLE INCOME FOR JOINT RETURNS OVER THE BRACKET STRUCTURE, FOR CALENDAR YEARS 1964 and 1979 |
| 2. | ACTUAL AND CBO PREDICTED DISTRIBUTIONS OF TAXABLE INCOME FOR JOINT RETURNS OVER THE BRACKET STRUCTURE FOR CALENDAR YEARS 1964 AND 1979 AND CBO PREDICTED DISTRIBUTION FOR CALENDAR YEAR 1985 |
| 3. | ACTUAL AND PREDICTED EFFECTIVE TAX RATES, CALENDAR YEARS 1964-1980 |
| 4. | IMPLICIT ELASTICITY OF TAX LIABILITIES WITH RESPECT TO TAXABLE PERSONAL INCOME UNDER INDEXED AND UNINDEXED TAXES, ASSUMING DIFFERENT RATIOS OF GROWTH IN THE GNP DEFLATOR TO TOTAL NOMINAL GNP GROWTH |
Developing budgetary and revenue forecasts to provide
assumptions underlying congressional concurrent resolutions on the budget
is an important function of the Congressional Budget Office (CBO). This
paper describes major aspects of the techniques used by CBO to forecast
revenues from the individual income tax: procedural methods, accuracy,
comparison with other techniques, and implications about the future behavior
of individual income tax revenues.
AVAILABLE REVENUE ESTIMATING TECHNIQUES
The most prominent available technique for estimating revenues from the individual income tax is the large-scale "microanalytic" model. This model consists mainly of a detailed computer-based miniature replica of the taxpaying population of the United States. The revenue implications of different economic conditions and statutory tax provisions are analyzed by calculating the effects that they would have on the liabilities of taxpayers in different categories, and then adding these estimates together to predict the economy-wide impact.
From a practical perspective, large-scale models are versatile and accurate,
but also cumbersome and expensive. These models are especially useful for
estimating the revenue implications of changes in tax statutes, which sometimes
require detailed analysis that only this technique offers. On the other
hand, the elaborate computations of large-scale models are sometimes unnecessary
to estimate how revenues change from year to year in response to changes
in economic conditions. In fact, if no significant changes in tax statutes
occur, such year-to-year revenue forecasts are sometimes developed using
extremely simple procedures consisting primarily of a rule of thumb for
predicting growth in taxes on the basis of the growth rate of aggregate
income. Of course, periods in which the tax structure is unchanged are
rare, especially since enactment of the Economic Recovery Tax Act of 1981
(ERTA), which mandated a series of important changes in the structure of
the income tax over five years.
THE CBO TAX MODEL
CBO's income tax model represents a compromise between the relatively complex and relatively simple alternatives just described. Although much smaller than large-scale models, it contains enough detail to account for the revenue implications of the significant current and prospective changes in the tax structure.
Essentially, the CBO model consists of four equations that determine how important constituent parts of the tax have behaved over long periods of time. When combined with current institutional information on statutory tax rates, brackets, and exemption levels, these equations permit forecasts of individual income tax liabilities and federal budget revenues from the individual income tax to be generated on the basis of a separate economic forecast of such variables as personal income, employment, and the price level. The model is designed specifically to account for the effects of several important recent and current developments on the income tax. These are:
The main advantages of the model are that it is of manageable size but still accurate and versatile enough to account for many different economic and legislative developments that affect individual income tax revenues. Unlike the other forecasting methods, the CBO model is well-documented and open to scrutiny and constructive criticism by tax professionals (or anyone else). Moreover, since the CBO procedure is different from most other techniques, it can provide a check on their accuracy.
The analysis of overall individual income tax revenues is divided into four parts. The first part produces an estimate of the tax base (gross taxpayer income minus all deductions and exemptions) on the basis of an economic forecast derived separately by CBO's Fiscal Analysis Division, together with assumptions concerning relevant provisions of the tax code. The economic forecast (covering gross national product (GNP), prices, employment, and other variables) is developed in the Fiscal Analysis Division separately from the revenue forecast developed in the Tax Analysis Division and the outlay forecast in the Budget Analysis Division. The three forecasts are, however, adjusted in light of one another to ensure consistency. Sometimes several different adjustments are made before a final set of forecasts is derived.
The second part of the analysis predicts the effective tax rate that applies to the tax base developed in Part 1. This computation involves predicting the pattern by which the tax base will be spread over the statutory tax bracket structure and how this pattern changes through bracket creep. By taking detailed account of the statutory characteristics of the bracket structure, this part of the model permits analyses of various changes in statutory bracket boundaries and rates, such as the changes mandated by ERTA.
The third part of the model develops a forecast of aggregate tax liabilities before credits using the results from Parts 1 and 2. Finally, the fourth part generates and incorporates estimates of tax credits and other tax items, and applies timing and other accounting adjustments to the tax liability forecast to derive a forecast of budget revenues from the individual income tax.
The Forecasting Record
The most important test of the reliability of any forecasting model is a comparison of the record of its forecasts, together with those of alternative procedures, to actual data after the forecasting period is completed. CBO's model is quite new, and therefore doesn't have an extensive record. Nevertheless, CBO has developed evidence on the model's reliability by developing "ex post" (or after the fact) forecasts of revenues in past years. This is done by putting oneself in the position of an imaginary forecaster several years ago who is using the CBO model to forecast revenues in later years that by now have been observed.
CBO has developed figures comparing the accuracy of ex post forecasts by its tax model with the Treasury Department's published individual income tax revenue forecasts for the same years. Attention is restricted to forecasting errors caused by technical inaccuracy in the procedures as opposed to errors resulting from inaccurate assumptions about the economy and about the passage of tax legislation affecting revenues. CBO's average errors for forecasts one, two, and three years into the future were all smaller than CBO's estimates of corresponding Treasury Department errors. Overall, these figures show that the CBO model is quite accurate, and compares favorably with alternative procedures.
Responsiveness of Individual Income Tax Revenues to Changes in Income
As this discussion has already noted, much interest centers on summary measures of the response of overall tax revenues to changes in the economy. Several recent developments, moreover, have altered the responsiveness of the tax. For example, the ERTA rate cuts that make the tax less progressive by reducing the top rates have likewise made it less responsive to economic changes, and the ERTA provision calling for indexation of certain tax provisions to the Consumer Price Index starting in 1985 will reduce the responsiveness even more.
The CBO income tax model can be used to calculate the response of revenues to different economic developments under various configurations of the tax laws. For example, under the bracket tax rates that will be in effect during 1984--following all the reductions that were mandated in ERTA--simulations with the model show that the responsiveness of tax liabilities to changes in GNP depends significantly on whether the GNP change primarily reflects inflation or represents a change in real output. Revenues change more sharply when GNP changes because of inflation rather than other factors. This occurs because increases in GNP resulting from inflation alone are taxed at higher rates as taxpayers are pushed into higher tax brackets. In quantitative terms, individual income tax revenues may grow as much as 70 percent faster than GNP if nominal GNP growth is entirely due to inflation.
By contrast, when GNP increases because of an increase in real output, some of the resulting income accrues to existing workers as a result of productivity increases, and this can cause bracket creep for these individuals. At the same time, however, much of the increase in aggregate income accrues to new workers who are drawn into employment by the upswing in the economy. New workers' incomes are taxed at relatively low rates, and this makes the responsiveness of income tax revenues to changes in real GNP relatively small. Estimates from the CBO model suggest that revenues increase roughly 30 percent faster than GNP when the GNP change is entirely real.
When the tax structure is assumed to be indexed to the price level, as it will be in 1985 and later years under the terms of ERTA, the tax will be more responsive to changes in aggregate income that are primarily "real" than to those that reflect inflation. Indexation prevents inflation from causing bracket creep, and this reduces the strength of the revenue response to inflation. Estimates from the CBO model suggest that revenues under the indexed income tax will grow at roughly the same rate as GNP when the GNP change is entirely due to inflation. This is a dramatic reduction in responsiveness from that in the unindexed tax, where revenues are estimated to grow as much as 70 percent faster than income under the same conditions. Income increases that result from strong growth in real output, by contrast, accrue in part to new workers and in part to existing workers as a consequence of growth in productivity. Productivity growth can cause these workers' incomes to move into higher tax brackets without protection from indexation, which is designed to eliminate bracket creep only when it results from inflation. Thus, some bracket creep may occur in response to real economic growth under the indexed income tax. As a result, CBO estimates that revenues from the indexed income tax will grow 30 percent faster than GNP if the GNP change is entirely real.
Economic Effects of Indexation
The shift in the behavior of the income tax that will be caused by indexation may have significant implications for the economy. Taxpayers will gain because indexation will protect them from unlegislated tax increases caused by inflation-induced bracket creep. Increases in economic efficiency may also result since marginal tax rates will be prevented from drifting upward. Indexation could also have adverse impacts, however. It could contribute to the tendency of structural budget deficits to increase over time (in the absence of legislation to control them). The expectation of large and rising deficits, in turn, may help explain the persistently high levels of interest rates, which may ultimately reduce living standards by reducing investment. Indexation may also reduce the value of the income tax as an automatic stabilizer of the economy.
This document is available in its entirety in PDF.