The economy and the federal budget: guides to the automatic effects.

Holloway, Thomas M.

THE prospect of large federal budget deficits over the remainder of the decade has stimulated substantial interest in how the budget is affected by changes in economic conditions. economists and policymakers are especially interested in determining how much of the deficit will disappear "automatically" with economic expansion, declining unemployment, and inflation, and how much must be dealt with by specific policy changes.

That the budget is very sensitive to changes in economic conditions is dramatically illustrated by the fiscal year 1983 budget. When the 1983 budget was originally submitted to Congress in February 1982, the deficit was estimated to be $91.5 billion. By September 1983, the end of the fiscal year, the actual deficit was $195.4 billion. Overly optimistic assumptions about economic conditions accounted for $67.2 billion, or about two-thirds, of the $103.9 billion difference between the actual and estimated 1983 dificit.

The relationship between economic conditions and the budget is not a simple one. However, rough guides, or rules of thumb, can be developed to approximate some of the most important aspects of this relationship. This article presents one such set of rules for the federal budget on a national income and product accounts (NIPA) basis. In brief, after about 1 year.

* Each 1-percentage-point increase in the unemployment rate increases the deficit by about $25-$30 billion.

* Each $100 billion increase in current-dollar GNP decreases the deficit by about $34-$38 billion.

* Each 1-percentage-point increase in the inflation rate decreases the deficit by about $7-$9 billion.

All three rules are symmetrical; decreases in the unemployment rate, in the inflation rate, and in current-dollar GNP cause the deficit to change by the indicated amount, but in the opposite direction. To simplify the exposition, the remainder of this article discusses only increases or decreases in the indicators of economic conditions, not both increases and decreases.

These rules must be applied carefully. They are formulated to use in analyzing the budget in 1983-85. The first and third need to be scaled to the size of the economy if they are to be used in the analysis of a different time period.

In addition, each is designed to indicate the budgetary effects of a specific change in an economic condition while other relevant factors--such as other economic conditions--are unchanged. If these factors change simultaneously, the rules may not apply.

Moreover, the rules indicate average annual effects. Readers interested in the quarterly pattern of those effects are referred to the technical note at the end of this article. It presents simplified estimation procedures that closely approximate the results of the detailed model BEA uses to cyclically adjust the budget. These procedures may be used to simulate the quarterly pattern of budgetary effects.

The rules of thumb were derived from simulations of the models that BEA uses to cyclically adjust the budget and to estimate inflation-induced changes in the cyclically adjusted budget. The general procedure is the same for all three simulations.

First: Estimate the cyclically adjusted budget or inflation-induced changes in the cyclically adjusted budget with the appropriate model.

Second: Alter an indicator of economic conditions to represent the change that is to be studied (e.g., a 1-percentage-point increase in the unemployment rate).

Third: Reestimate the appropriate model based on the altered indicator of economic conditions.

Fourth: Compare the results of the third step with those of the first step to determine the effects on the budget of the change in economic conditions.

How these steps are implemented is discussed with each simulation. Unemployment rate increase

An increase in the unemployment rate directly increases unemployment-sensitive Federal expenditures such as unemployment insurance benefits and food stamp benefits. In addition, a 1-percentage-point increase in the unemployment rate is typically associated with a decrease in constant-dollar GNP of about 2 per cent. 1982. The full budgetary effects take about 1 year to develop. The simulation results for 1983 suggest that the increase in the budget deficit is equal to a little over one-third of the decrease in current-dollar GNP. More precisely, the simulation indicates that the increase in the deficit amounts to about 34 per cent of the decrease in current-dollar GNP when an adjustment of net interest is included in the budget measure and about 38 percent when the adjustment is excluded. In both cases, receipts decrease by an amount equal to about 29 percent of the decrease in GNP; expenditure increases account for the rest of the deficit change. Because the simulation results are expressed as percentages, they need not be scaled to the size of the economy. Thus, if current-dollar GNP had been $3,211 billion in 1983 instead of the $3,311 billion it actually was, the rule says that the deficit would have been $216 billion or $220 billion, respectively, instead of the actual $182 billion. Inflation rate increase

Inflation automatically increases receipts because higher prices mean higher dollar amounts subject to tax. Historically, in the case of the individual income tax, inflation has also resulted in higher marginal tax rates being applied to these higher dollar amounts. This "bracket creep" is scheduled to be eliminated in 1985; the economic Recovery Tax Act of 1981 provides for indexation of the individual income tax--automatic increases in personal exemptions and tax brackets in proportion to increases in a price index.

On the expenditure side, many programs are linked directly or indirectly to changes in some measure of the general price level. Because the effects on receipts have tended to be larger than the effects on expenditures, inflation has historically tended to move the budget toward surplus.

Table 2 shows the simulation results of the effects on the budget of a 1-percentage-point increase in the inflation rate without and with indexation of the individual income tax. Indexation was incorporated by assuming that the elasticity of personal income tax receipts with respect to inflation-induced changes in income subject to tax was 1.0 instead of the much higher values (1.63 to 1.66) that applied historically.

The simulation was done by increasing the rate of inflation 1 per centage point (at an annual rate) above the actual inflation rate starting in the first quarter of 1979. Because inflation is a measure of price change over time, the simulation shows the effects on the budget after 4 quarters, 8 quarters, and 12 quarters.

The effects on receipts without indexation of the individual income tax are larger than those on expenditures and develop more quickly. the complete budgetary effects on some indexed expenditures--those linked by legislation to changes in a price index--take more than 1 year to occur. The simulation provides a rule that suggests that at the end of four quarters, the cumulative effects of a 1-percentage-point increase in the inflation rate move the budget about $6 billion toward surplus. At the end of 8 and 12 quarters, the corresponding amounts are about $11 billion and about $15 billion, respectively. If, starting in the first quarter of 1979, the inflation rate had been 1 percentage point higher that it actually was the simulation result suggests that by the end of 1981, the deficit would have been about $81 billion instead of the actual $96 billion.

The effects on receipts are reduced with indexation, but the simulation results suggest that inflation still tends to move the budget toward surplus. At the end of four quarters, the cumulative effects move the budget about $4 billion toward surplus. At the end of 8 and 12 quarters, the corresponding amounts are about $8 billion and about $10 billion, respectively.

Because the budgetary effect of inflation is expected to increase with the economy, the rule of thumb for 1983-85 is derived by scaling the 1979-81 simulation results in the same way that the unemployment rate results were scaled. The 1983-85 rule is that the cumulative effects of a 1-percentage-point increase in the inflation rate starting in the first quarter of 1983 will move the budget about $7 to $9 billion toward surplus after 4 quarters (the end of 1983), $14 to $16 billion after 8 quarters (the end of 1984), and $16 to $19 billion after 12 quarters (the end of 1985).

Technical Note

The simulations discussed in the text provide general guides to the average effects on the budget of changes in economic conditions, but are not designed to generate quarterly estimates. The complete model that BEA uses to cyclically adjust the budget does provide quarterly estimates, but it is complex and requires many unpublished time series. The size and data requirements of the complte model preclude many applications where a rapid, even if approximate, solution is needed. For these applications, simplified estimation procedures have been developed that closely approximate the results of the complete model.

The simplified procedures consist of three equations; they estimate cyclical adjustments (i.e., differences between cyclically adjusted and actual levels) for total receipts, total expenditures, and the surplus or deficit: (1) TGU=1.266g (G-1 / G) TGF (2) EGU=(-0.00153UR-0.00049URTRIG -0.00010 4 sigma UR-.sub.1.+0.0047g i=1 +0.0182g-.sub.1.+0.016g-.sub.2.+0.0048g-.sub.3 +0.0027 7 sigma g-.sub.1.)G i=4 (3) SGU=TGU-EGU where: TGU=estimated cyclical adjustment of receipts; EGU=estimated cyclical adjustment of expenditures; SGU=estimated cyclical adjustment of the surplus or deficit; G=middle-expansion trend GNP; G=actual GNP; TGF=actual receipts; UR=actual less trend unemployment rate; URTRIG=UR when the national trigger was on, zero otherwise; g=GNP gap (1-(G/G)).

All dollar amounts are in current dollars. Estimates of the levels of the cyclically adjusted budget are obtained by adding the estimates from equations (1)-(3) to the appropriate components of the unadjusted budget.

As chart 8 shows, the simplified procedures capture almost all of the quarterly variation in the complte model's estimates for receipts and expenditures in the 1970-83 period, although estimates from the simplified procedures are generally below those of the complete model. For receipts, the correlation coefficient between the estimates of the simplified procedures and the complete model is 0.998, and the root mean square error is $1.9 billion; for expenditures, the correlation coefficient is 0.975, the root mean square error is $.9 billion; and for expenditures without the adjustment of net interest paid, the correlation coefficient is 0.999, the root meant square error is $.1 billion. (Expenditures without the adjustment for net interest paid are estimated by omitting all of the GNP gap (g) terms from equation (2)). Cyclical adjustments of the surplus or deficit from the two models may be compared by combining the top and middle panels of the chart or excluding the adjustment for net interest paid the top and bottom panels of the chart.