(Archived document, may contain errors)
328 January 27, 1984 UNDERSTANDING THE FEDERAL DEFICIT PART 1: HOW-FORECASTERS GET IT WRONG INTRODUCTION As Congress convenes to discuss impending fe deral deficits gloomy forecasts are treated almost with the reverence once given to pronouncements from Mount Olympus. Lawmakers rushing to offer proposals-=usually tax increases-to bridge the budget gap seldom pause even for a moment to question the pred i ctions of the eco nomic soothsayers cates that such forecasters are no more reliable than fortune tellers or astrologers. Government predictions of deficits, in particular, are often wide.of the mark. At the beginning of each year, the Office of Managemen t and Budget (OMB) estimates the budget deficit and other economic aggregates for the fiscal year beginning nine months later Om's yearly budget deficit projec tions have on average erred by 254 percent of the estimated deficit between 1971 and 19 83. OMB has come within 50 percent of the actual deficit only five times in the last thirteen years. Moreover OMBfs errors are as likely to underestimate deficits as overestimate them Yet a review of the !lsciencetf of economic projections indi Example The Presid ent's FY 1983 Federal Budget, prepared by OMB, underestimated the active deficit by over $100 billion-a 100 percentorecasting error deficit in a forecast issued just two monmefore the end of the fiscal year--by $14.4 billion OMB later overestimated the sa m e This study is the first of a four-part series examining the nature and Part I1 will analyze the components of the effects of the federal deficit deficit; part I11 will explore the deficit's impact on the domestic economy and part IV will examine the tra d e impacts of the deficit. 2 According to one analyst, the FY 1984 budget deficit, based on the average OMB error of the recent past, could be anywhere in the range of 156 billion to $250 bi1lion.l And the FY 1987 defi- cit, applying OMB's average past err or rate; could end up anywhere between 60 billion and $244 billion. Given this dismal track record, how much should policy making be based on OMB's estimates of the deficit? Not much particularly when presumed deficits become the rationale for massive tax increases In fact, if history is any indication the current estimates of the 1984 budget deficit may be far too large OMB's past deficit errors have tended to follow 'the eco nomic cycle closely In recoveries actual deficits typically have been smaller th a n OMB predicted. also have tended to grow as the recoveries advanced. OMB forecasting errors Why does OMB miss the mark? The major reason is the inherent difficulty confronting any forecaster trying to predict the eco nomic cycle, particularly changes in such crucial variables as Gross National Product GNP unemployment, and inflation (CPI OMB's yearly GNP estimates erred an average of 36 percent between 1977 and 19 81. Its forecasts of percentage changes in CPI erred by an average of 130 percent between 1976 and 1982, and its annual estimates of percentage changes in unemployment erred by over 50 percent in the same period.2 Even a small change in these funda mental economic assumptions can have massive effects on budget deficits. A 2 percentage point sustained increase in the rate of economic growth, for example, would reduce the cumulative N 1984-1987 budget deficit by $440 billion, or one-half of its cur- rent pro jected level. The OMB is not the only forecaster to miss consistently, so errors cannot be blamed on politics or overoptimism among White House economists. A survey of leading private sector forecasters whose clients pay for accurate information, shows tha t they mis- calculated the timing of the economic recovery and missed the 1982 GNP growth by as much as 4 percentage points, or $120 billion in some cases. The bipartisan Congressional Budget Office (CBO also miscalculated the 1982 deficit by $80 billion- -a whopping 80 percent error. The problem is not with the forecasters, but with the inherent impossibility of predicting even a few months in advance the performance of a volatile $3 trillion economy. To generate accurate economic projections, forecasters would need perfect l Thomas S. McCaleb, Associate Professor of Economics, Florida State Uni versity, Tallahassee; formerly Senior Staff Economist, President's Council of Economic Advisors Federal Budget and Fiscal Policy in the 198O's unpublished paper, u ndated, p. 12. Randolph H. Boehm, Policy Analysis No. 25 (Washington, D.C The Cat0 Institute, June 30, 1983 pp. 14, 20, 24, and 27 3 foresight of the economic cycle, business and consumer psychology the impact of upcoming government taxes, the effects of b udget and monetary policy, foreign shocks such as the OPEC oil price runup, and such events as natural disasters and wars. Even then, the forecasters would face the sticky task of translating these factors into government revenue and expenditure figures f l imsy forecasts, every congressional initiative or Administration budget request based on an economic forecast should be accompanied by the track record of the forecasting agency In this way Ameri cans could judge how much credence to place in the predicti ons underpinning the legislation being passed in their.name In order that the public be protected from policies based on In short, no forecaster can be expected to model accurately the activities and daily decisions of millions of economic actors. Human be havior, unlike natural phenomena, cannot be fully quanti- fied.3 This is because human beings have free will and freedom of choice, and are never entirely predictable. Given these in- herent limitations, economic policy should not be guided by forecasts, e specially in turbulent times, but by fundamental principles designed to create a stable climate for long-term economic growth If Congress cannot read the future, it should not keep trying to make a mid-course correction whenever a new prediction is releas e d tory, and spending policies to encourage risk-taking and reward work--and stick to it It should chart a course of tax, regula THE FORECASTING RECORD Economic forecasting plays a fundamental role in economic policy making. While official government forec a sters claim that their long-range assumptions are not intended as precise forecasts of economic conditions, it is nonetheless true that these budget assumptions, as well as private sector forecasts, are used to jus tify economic policy initiatives of crit i cal importance. In par ticular, forecasts of enormous budget deficits "as far as the eye can see" have fueled the movement pressing President Reagan to back major tax increases OMB's record of predicting budget deficits and other economic aggregates is di s tressing. Each January OMB issues a forecast in the Budqet of the United States Government for the fiscal year beginning the following October OMB also updates this budget forecast in the Mid-Session Review, published around July each year amounts of the p rojected deficit, and .as a portion of GNP Chart 1 shows the errors in these OMB forecasts, in dollar The James B. Ramsey Economic Forecasting--Models or Markets (Washington D.C The Cat0 Institute, 1980 p. ix-xii. 4 CHART I Budget Fiscal Year 1971 1972 19 7 3 1974 1975 1976 1977 1978 1979 1980 1981 1982 1983 Federal Budget Deficit Projections 1971-1983 in billions of dollars Error Percentage of Projected Actual Error Pro j ect ion 1.3 11.6 25.5 12.7 9.4 51.9 43.0 47.0 60.6 29.0 15.8 29.7 91 -5 23.0 23.4 14.9 4.7 45.2 66.4 44.9 48.8 27.7 59.6 57.9 110.6 195.4 24.3 11.8 10.6 8.0 35.8 14.5 1.9 1.8 32.9 30.6 42.1 80.9 103.9 1869 102 42 63 38 1 28 4 4 54 106 266 272 114 Error Percentage of GNP 2.40 1.05 0.85 0.58 2.&0 0.88 0.10 0.09 1.40 1.20 1.50 2.70 3.30 Averag e Absolute Error 30.7 billion Average error projection 254 Average error GNP 1.42 Fiscal Year begins in October of the preceding calendar year 1983 began in October 1982 Hence FY Source: FY 1984 Budget of the United States and Cat0 Institute's "Forecasting the Economy Do Presidents Get It Right June 30, 1983 average error over the thirteen-year period is 254 percent of the projected deficit estimate. Om's projections have been between 50 percent above and 50 percent below the actual deficit only five times i n the last thirteen years ranged from a staggering 1,869 percent above to 63 percent below the yearly deficit forecast during the period has been over $30 billion. And deficit errors have averaged about 1.5 percent of GNP A number of important conclusions result from examining OMB's sorry record of deficit f~recasting Actual deficits have The average yearly absolute error This section closely follows the approach of the Economic Progress Report published by the Treasury of the United Kingdom, No. 160, Sept e mber 1983. 5 Bias in Errors Little or no systematic bias is evident in forecasting errors As Chart 1 shows, the OMB has overestimated and underes timated government deficits. In recent years the actual deficit has been much larger than OMB has forecast, b ut in three of the last thirteen years, actual deficits have been much smaller than the OMB originally predicted. Cyclical Nature of Errors Deficit forecasts show a cyclical pattern see Graph 1 If this When GNP has been on an upswing, actual deficits have tended'to be smaller than forecast; and when GNP has been falling, actual deficits have been larger than government forecasts pattern holds true for 1984, talk of $200 billion deficits will turn out to be well above the actual outcome Size of Errors Defic i t errors have been larger during periods of economic turbulence Deficit errors in the late 1970s, a period of relative economic stability were much smaller than they have been during the convulsive 1980s Timinq of Forecasts Not surprisingly, deficit forec asts made nine months ahead of the new fiscal year are more accurate than longer range fore casts but less reliable than those made during the fiscal year. But even this last group of projections often contains large errors. According to the U.S. Chamber o f Commerce, deficit pro jections made three months after the start of the fiscal year were subject to errors that ranged from 35 percent below forecasted deficits to 50 percent above.5 Errors are Growing Errors have increased in absolute terms, and as a p ercentage The FY 1983 deficit error as a portion of GNP of GNP since 1971 was the largest in the entire period greater fluctuation in the economic cycle. This probably reflects AN EXAMPLE OF FORECASTING INACCURACY The 'FY 1983 deficit forecast issued by OMB illuminates the difficulties in economic forecasting of the last thirteen years. In February 1982,.0MB forecast an FY 1983 budget deficit of $91.5 Richard W. Rahn, "Deficit Guessing Game Can Lead to Bad Economic Policies mimeograph, undated. 6 Errors as GNP 7 billion. The actual deficit turned out to be $195.4 billion--more than 100 percent over the original projection, or 3.3 percent of GNP. This astounding error resulted in large part from the poorer than expected performance of the economy. Then, hav ing underestimated the budget effects of the reces sion, OMB forecasters went on to miss the economic recovery. The Mid-Session Review, issued by OMB on July 25, 1983-just two months before the end of the fiscal year--projected a $209.8 bil lion federal d e ficit. The actual EY 1983 deficit came in $14.4 billion lower than OMB had predicted just 67 days before. OMB's forecasting error amounted to an annual error of $78.8 billion.6 The unemployment rate in the Mid-Session Review moreover was forecast at 9.6 p ercent by the end of FY 19 83. Yet the final jobless figure had dropped to 8.8 percent. As Dr. Richard Rahn the U.S. Chamber of Commerce's Chief Economist, asks With an error rate of this magnitude only two months in advance of the end of the fiscal year, how seriously should we take their projec tions for the next year, not to mention 1988?Il7 WHY FORECASTS GO WRONG The underlying problem with deficit forecasting is that economists cannot measure the deficit directly. The deficit is the difference between two very large numbers: estimated govern ment outlays and receipts. A small percentage change in either of these two massive numbers can have a large impact on the re sidual government deficit. Just a 2 percent increase in govern ment revenues, and a simi lar decrease in government outlays, for example, could change the budget deficit by $30 billion. There are a number of reasons why government and private forecasts alike have been wide of the mark in estimating federal revenues and outlays 1. Economic Cycl es No forecaster has been able to predict the economic cycle with consistency assumptions about GNP, inflation, and interest rates can have a dramatic effect on budget deficit forecasts. Economic growth-The economic growth rate has enormous consequences f or the budget deficit. During periods of rapid expansion, business profits, wages, interest, and dividend income surge while the Treasury gains new tax revenues generated from these sources. Greater economic growth, meanwhile, reduces Yet even a small cha nge in the underlying Ibid. Ibid 8 government.expenditures, since more jobs mean 1ess.unemployment compensation, welfare, and other social spending. As such, the deficit is .cut from two directions. The most powerful way to cut the deficit is by economic growth. According to CBO calculations, a 2 percentage point sus- tained increase in economic growth would reduce the cumulative deficit by $440 billion during 1984-1987.8 If the economic fore casts u n derestimated growth by such a degree, that alone would reduce upcoming budget deficit projections by as much as one half (see Chart 11 CHART 11 The Effect on Budget Deficit Projections of Selected Changes in Economic Assumptions Increases in billions of d o llars FY 1983 N 1984 FY 1985 FY 1986 FY 1987 Real Growth up 1 10 -28 -46 -62 -83 Unemployment down 1 25 -39 -39 -37 -37 Inflation down 1% 6 14 18 24 29 Interest Rates down 1 1 -5 -8 10 -12 Source: CBO The Economic and Budget Outlook: An Update," September 1982 Inflation--Inflation directly affects federal spending and revenues. Inflation boosts nominal corporate profits, personal incomes, payrolls, and sales and therefore increases tax collec tions-particularly as people are pushed into higher income tax b r ackets. (Beginning in 1985, however, indexation of brackets will reduce these inflation induced tax revenues spending also increases due to inflation, because many government programs, including Medicare, Social Security, and federal pay, increase with in f lation. Combining its effects on revenues and outlays, a 1 percent drop in the inflation rate actually would increase the government deficit by a total of over $85 billion between 1984-1987 Federal Interest rates--The 1.4 trillion government debt has made interest expense one of the largest and fastest growing items in the federal budget, skyrocketing to nearly 130 billion in FY 19 83. Unanticipated fluctuations in interest rates can have Congressional Budget Office The Economic and Budget Outlook: An Upda te Washington, D.C U.S. Government Printing Office, September 1982 9 significant effects on the accuracy of federal budget forecasts. According to the CBO, a one percentage point drop in interest rates for all government securities would cut the cumulativ e 1984-1987 budget deficit by $35 billion ment compensation expenditures and other social welfare spending. Increased employment expands the tax base and therefore sharply increases tax revenues. According to the CBO, a one percentage point drop in unemplo yment would reduce the budget deficit by 150 billion over the next four years Unemployment-A falling unemployment rate reduces unemploy 2. Complexities in Revenue and Outlay Estimates Even if the economic cycle could be gauged accurately, the forecaster m u st convert the resulting economic assumptions into revenue and outlay estimates To do this, he must try to unravel the complexities of spending patterns, the timing of income tax payments, the use of tax deductions .and shelters, the utilization of govern m ent programs, and many other factors. Errors made in these elements are compounded in the final revenue and outlay forecasts and then magnified in his deficit projection 3. Leqislative Chanqes Forecasts assume certain legislative changes.in tax and But ev e n the most seasoned political pundits budget policy rarely predict accurately what Congress and the President will do 4. Other Surprises There are many other unexpected events that may trigger a different deficit than that projected by economists disaster , such as a drought or a bad winter, can drastically alter government outlays and consumer spending patterns and have a significant effect on the deficit A natural OMB IS NOT ALONE OMB is not the only agency or organization to have trouble in developing ac c urate forecasts. Some critics say OMB's forecasts naturally are inaccurate because politics clouds its decision making. Others charge that administrations are inherently less competent or motivated than the private sector or the Congressional Budget Offic e forecasts, however, reveals that poor performance is not an OMB monopoly A recent review of both private sector and CBO The Congressional Budget Office Chart I11 shows the deficit estimates of the CBO for 1979 to In some years CBO has been noticeably mor e accurate in es 1983 timating the deficit than the OMB. In most years, however, the 10 CHART I11 Congressional Budget Office Deficit Estimates 1979-1983 in billions of dollars Fiscal Date of Year Estimate Estimate Ac tua 1 1979 12/77 $38 $27.7 1980 1/79 4 9 59.6 1981 2/80 21 57.9 1982 7/81 30 110.6 1983 2/82 157 195.6 Error Estimate 27 22 176 269 25 Average error projection: 104 percent. Source: Congressional Budget Office errors have been in exactly the same direction and of about the same magni.tude. CBO' s yearly deficit projections from 1979 to 1983 on average erred over 100 percent. The errors ranged from 27 percent below to 269 percent above projected deficits. Private Sector Forecasts of GNP and the inflation rate for 1981-1983, compared with the actu al results. These basic economic variables are among the most fundamental determinants of budget deficits that, although some forecasters came closer than others to the actual outcome in certain years, no forecaster was consistently accurate. These leadin g private sector forecasters erred by an average of 137 percent in their projections of GNP private sector analysts erred annually in their forecasts of the inflation rate by 21 percent Chart IV provides a sample of leading private sector forecasts It can b e seen The same IMPLICATIONS FOR ECONOMIC POLICY The poor record of all forecasters--private as well as government--holds important implications for current public policy 1) Congress should not place more confidence in forecasters than their record justif i es--especially in the case of projections of the economy four or five years hence extremely reluctant to use highly speculative forecasts to justify fundamental and far-reaching economic policy changes Policymakers should be 11 CHART IV Private Forecasts o f GNP Forecaster Chase Econometrics Data Resources Brookings Tomsend-Greenspan Wharton Average Prediction Actual Error Each Year Average Error Over 3 Years 137 percent INFLATION RATE percent change over previous fiscal year FY 1981 FY 1982 0.6 0 0.8 -0.5 0 .7 -2.0 1.5 -1.0 0.3 -0.3 0.8 -0.8 2.6 -1.9 225 138 FY 1983 2.1 1.6 3.3 2.0 2.4 2.3 3.4 48 Average prediction Ac tua 1 Average Error FY 1981 FY 1982 FY 1983 11.0 8.0 4.7 10.3 6.1 3.2 6 24 32 Average Error Over 3 Years 21 percent Estimate Source: The Washi n gton Post, January 8, 1984 2) Given the inherent limitations of forecasting, policymakers should rethink the case for tax increases based on the current gloomy budget deficit forecasts that these forecasts will be any more accurate than those in previous y ears-while it is certain that .tax increases could choke off the recovery There is no reason to suppose 3) liFine-tuningll the economy, national industrial plans, and efforts to smooth out the business cycle require accurate forecast ing of economic trend s and aggregates. Firm statistics simply do not exist, so such detailed policies are doomed to failure. In deed, they even can destabilize the economy, given that errors in 12 forecasting closely follow the business cycle. economic policies based on errone o us forecasts actually could magnify swings in the economy, providing excessive stimulus in recoveries and too much restraint in recessions. Rather than providing policymakers with information to counterbalance the economic cycle, forecasting actually tend s to intensify economic fluctuations Countercyclicali1 4 ally less than now forecast, if forecasting errors follow the typical cyclical pattern of the business cycle 5 disclaimer that economic assumptions are only careful guesses, not a scientifically reli a ble forecast of the likely outcome. Indeed, when policy initiatives and budgets are based on.officia1 forecasts, the Administration and congressional committees should be required to make.public the track record of the forecasting agency during the previo u s ten years so that the public might award the projections proper credence 6) Since forecasts are unreliable, solid economic principles should become the basis for economic policy. Such policy .should lay a stable foundation for long-term growth and not b e adjusted on a yearly basis to anticipate flimsy predictions of future eco nomic conditions. Congress and the Federal Reserve System should continue to move toward a tax system that rewards risk and effort, a'regulatory system that removes barriers to ent e rprise, a spend- ing reduction policy that shrinks the public sector, and a mone tary policy that provides for stable and low money growth The actual deficit for 1984-1985 probably will be substanti Accompanying every government budget should be a promine n t CONCLUSION Government and private sector 'forecasts generate more heat than light economists and policymakers alike is hardly warranted by the record. Official deficit forecasts for only one year have erred on average 254 percent a year over the last th irteen years private sector firms, which earn their living from forecasting are inaccurate in predicting the future. three, four, and five years ahead is rarely closer to reality than a 1ook.into a crystal ball. Would planning the economy, manipulating agg regate demand, and pursuing fashionable cure-alls such as industrial policy seem so desirable if the sorry forecast record were more widely known? Would economists and policymakers push for the large tax increases now being debated if they realized the ma s sive range of errors inherent in any deficit prediction? The flawed record of forecasting certainly should cause economic advisors and lawmakers to use forecasts in a far more cautious and tentative manner The naive confidence placed in economic forecasts by Even A forecast of two, 13 But what should economic policymakers use as a guide if the science of prediction is so crude and unreliable? Perhaps they should recognize that the first step to wisdom is to recognize the obvious limitations in any sort of p rediction and act accord ingly cit or any other key variable, it should not try to persuade the American people to accept frequent policy changes to fit the latest economic forecasts. It should move instead to establish a broad, long-term tax and monetary framework, based on the les sons of economic history, and leave the guessing game to the gamblers on Wall Street If Congress cannot predict the size of the federal defi Thomas M Humbert Senior Policy Analyst and Walker Fellow in Economics
328 January 27, 1984 UNDERSTANDING THE FEDERAL DEFICIT PART 1: HOW-FORECASTERS GET IT WRONG INTRODUCTION As Congress convenes to discuss impending fe deral deficits gloomy forecasts are treated almost with the reverence once given to pronouncements from Mount Olympus. Lawmakers rushing to offer proposals-=usually tax increases-to bridge the budget gap seldom pause even for a moment to question the pred i ctions of the eco nomic soothsayers cates that such forecasters are no more reliable than fortune tellers or astrologers. Government predictions of deficits, in particular, are often wide.of the mark. At the beginning of each year, the Office of Managemen t and Budget (OMB) estimates the budget deficit and other economic aggregates for the fiscal year beginning nine months later Om's yearly budget deficit projec tions have on average erred by 254 percent of the estimated deficit between 1971 and 19 83. OMB has come within 50 percent of the actual deficit only five times in the last thirteen years. Moreover OMBfs errors are as likely to underestimate deficits as overestimate them Yet a review of the !lsciencetf of economic projections indi Example The Presid ent's FY 1983 Federal Budget, prepared by OMB, underestimated the active deficit by over $100 billion-a 100 percentorecasting error deficit in a forecast issued just two monmefore the end of the fiscal year--by $14.4 billion OMB later overestimated the sa m e This study is the first of a four-part series examining the nature and Part I1 will analyze the components of the effects of the federal deficit deficit; part I11 will explore the deficit's impact on the domestic economy and part IV will examine the tra d e impacts of the deficit. 2 According to one analyst, the FY 1984 budget deficit, based on the average OMB error of the recent past, could be anywhere in the range of 156 billion to $250 bi1lion.l And the FY 1987 defi- cit, applying OMB's average past err or rate; could end up anywhere between 60 billion and $244 billion. Given this dismal track record, how much should policy making be based on OMB's estimates of the deficit? Not much particularly when presumed deficits become the rationale for massive tax increases In fact, if history is any indication the current estimates of the 1984 budget deficit may be far too large OMB's past deficit errors have tended to follow 'the eco nomic cycle closely In recoveries actual deficits typically have been smaller th a n OMB predicted. also have tended to grow as the recoveries advanced. OMB forecasting errors Why does OMB miss the mark? The major reason is the inherent difficulty confronting any forecaster trying to predict the eco nomic cycle, particularly changes in such crucial variables as Gross National Product GNP unemployment, and inflation (CPI OMB's yearly GNP estimates erred an average of 36 percent between 1977 and 19 81. Its forecasts of percentage changes in CPI erred by an average of 130 percent between 1976 and 1982, and its annual estimates of percentage changes in unemployment erred by over 50 percent in the same period.2 Even a small change in these funda mental economic assumptions can have massive effects on budget deficits. A 2 percentage point sustained increase in the rate of economic growth, for example, would reduce the cumulative N 1984-1987 budget deficit by $440 billion, or one-half of its cur- rent pro jected level. The OMB is not the only forecaster to miss consistently, so errors cannot be blamed on politics or overoptimism among White House economists. A survey of leading private sector forecasters whose clients pay for accurate information, shows tha t they mis- calculated the timing of the economic recovery and missed the 1982 GNP growth by as much as 4 percentage points, or $120 billion in some cases. The bipartisan Congressional Budget Office (CBO also miscalculated the 1982 deficit by $80 billion- -a whopping 80 percent error. The problem is not with the forecasters, but with the inherent impossibility of predicting even a few months in advance the performance of a volatile $3 trillion economy. To generate accurate economic projections, forecasters would need perfect l Thomas S. McCaleb, Associate Professor of Economics, Florida State Uni versity, Tallahassee; formerly Senior Staff Economist, President's Council of Economic Advisors Federal Budget and Fiscal Policy in the 198O's unpublished paper, u ndated, p. 12. Randolph H. Boehm, Policy Analysis No. 25 (Washington, D.C The Cat0 Institute, June 30, 1983 pp. 14, 20, 24, and 27 3 foresight of the economic cycle, business and consumer psychology the impact of upcoming government taxes, the effects of b udget and monetary policy, foreign shocks such as the OPEC oil price runup, and such events as natural disasters and wars. Even then, the forecasters would face the sticky task of translating these factors into government revenue and expenditure figures f l imsy forecasts, every congressional initiative or Administration budget request based on an economic forecast should be accompanied by the track record of the forecasting agency In this way Ameri cans could judge how much credence to place in the predicti ons underpinning the legislation being passed in their.name In order that the public be protected from policies based on In short, no forecaster can be expected to model accurately the activities and daily decisions of millions of economic actors. Human be havior, unlike natural phenomena, cannot be fully quanti- fied.3 This is because human beings have free will and freedom of choice, and are never entirely predictable. Given these in- herent limitations, economic policy should not be guided by forecasts, e specially in turbulent times, but by fundamental principles designed to create a stable climate for long-term economic growth If Congress cannot read the future, it should not keep trying to make a mid-course correction whenever a new prediction is releas e d tory, and spending policies to encourage risk-taking and reward work--and stick to it It should chart a course of tax, regula THE FORECASTING RECORD Economic forecasting plays a fundamental role in economic policy making. While official government forec a sters claim that their long-range assumptions are not intended as precise forecasts of economic conditions, it is nonetheless true that these budget assumptions, as well as private sector forecasts, are used to jus tify economic policy initiatives of crit i cal importance. In par ticular, forecasts of enormous budget deficits "as far as the eye can see" have fueled the movement pressing President Reagan to back major tax increases OMB's record of predicting budget deficits and other economic aggregates is di s tressing. Each January OMB issues a forecast in the Budqet of the United States Government for the fiscal year beginning the following October OMB also updates this budget forecast in the Mid-Session Review, published around July each year amounts of the p rojected deficit, and .as a portion of GNP Chart 1 shows the errors in these OMB forecasts, in dollar The James B. Ramsey Economic Forecasting--Models or Markets (Washington D.C The Cat0 Institute, 1980 p. ix-xii. 4 CHART I Budget Fiscal Year 1971 1972 19 7 3 1974 1975 1976 1977 1978 1979 1980 1981 1982 1983 Federal Budget Deficit Projections 1971-1983 in billions of dollars Error Percentage of Projected Actual Error Pro j ect ion 1.3 11.6 25.5 12.7 9.4 51.9 43.0 47.0 60.6 29.0 15.8 29.7 91 -5 23.0 23.4 14.9 4.7 45.2 66.4 44.9 48.8 27.7 59.6 57.9 110.6 195.4 24.3 11.8 10.6 8.0 35.8 14.5 1.9 1.8 32.9 30.6 42.1 80.9 103.9 1869 102 42 63 38 1 28 4 4 54 106 266 272 114 Error Percentage of GNP 2.40 1.05 0.85 0.58 2.&0 0.88 0.10 0.09 1.40 1.20 1.50 2.70 3.30 Averag e Absolute Error 30.7 billion Average error projection 254 Average error GNP 1.42 Fiscal Year begins in October of the preceding calendar year 1983 began in October 1982 Hence FY Source: FY 1984 Budget of the United States and Cat0 Institute's "Forecasting the Economy Do Presidents Get It Right June 30, 1983 average error over the thirteen-year period is 254 percent of the projected deficit estimate. Om's projections have been between 50 percent above and 50 percent below the actual deficit only five times i n the last thirteen years ranged from a staggering 1,869 percent above to 63 percent below the yearly deficit forecast during the period has been over $30 billion. And deficit errors have averaged about 1.5 percent of GNP A number of important conclusions result from examining OMB's sorry record of deficit f~recasting Actual deficits have The average yearly absolute error This section closely follows the approach of the Economic Progress Report published by the Treasury of the United Kingdom, No. 160, Sept e mber 1983. 5 Bias in Errors Little or no systematic bias is evident in forecasting errors As Chart 1 shows, the OMB has overestimated and underes timated government deficits. In recent years the actual deficit has been much larger than OMB has forecast, b ut in three of the last thirteen years, actual deficits have been much smaller than the OMB originally predicted. Cyclical Nature of Errors Deficit forecasts show a cyclical pattern see Graph 1 If this When GNP has been on an upswing, actual deficits have tended'to be smaller than forecast; and when GNP has been falling, actual deficits have been larger than government forecasts pattern holds true for 1984, talk of $200 billion deficits will turn out to be well above the actual outcome Size of Errors Defic i t errors have been larger during periods of economic turbulence Deficit errors in the late 1970s, a period of relative economic stability were much smaller than they have been during the convulsive 1980s Timinq of Forecasts Not surprisingly, deficit forec asts made nine months ahead of the new fiscal year are more accurate than longer range fore casts but less reliable than those made during the fiscal year. But even this last group of projections often contains large errors. According to the U.S. Chamber o f Commerce, deficit pro jections made three months after the start of the fiscal year were subject to errors that ranged from 35 percent below forecasted deficits to 50 percent above.5 Errors are Growing Errors have increased in absolute terms, and as a p ercentage The FY 1983 deficit error as a portion of GNP of GNP since 1971 was the largest in the entire period greater fluctuation in the economic cycle. This probably reflects AN EXAMPLE OF FORECASTING INACCURACY The 'FY 1983 deficit forecast issued by OMB illuminates the difficulties in economic forecasting of the last thirteen years. In February 1982,.0MB forecast an FY 1983 budget deficit of $91.5 Richard W. Rahn, "Deficit Guessing Game Can Lead to Bad Economic Policies mimeograph, undated. 6 Errors as GNP 7 billion. The actual deficit turned out to be $195.4 billion--more than 100 percent over the original projection, or 3.3 percent of GNP. This astounding error resulted in large part from the poorer than expected performance of the economy. Then, hav ing underestimated the budget effects of the reces sion, OMB forecasters went on to miss the economic recovery. The Mid-Session Review, issued by OMB on July 25, 1983-just two months before the end of the fiscal year--projected a $209.8 bil lion federal d e ficit. The actual EY 1983 deficit came in $14.4 billion lower than OMB had predicted just 67 days before. OMB's forecasting error amounted to an annual error of $78.8 billion.6 The unemployment rate in the Mid-Session Review moreover was forecast at 9.6 p ercent by the end of FY 19 83. Yet the final jobless figure had dropped to 8.8 percent. As Dr. Richard Rahn the U.S. Chamber of Commerce's Chief Economist, asks With an error rate of this magnitude only two months in advance of the end of the fiscal year, how seriously should we take their projec tions for the next year, not to mention 1988?Il7 WHY FORECASTS GO WRONG The underlying problem with deficit forecasting is that economists cannot measure the deficit directly. The deficit is the difference between two very large numbers: estimated govern ment outlays and receipts. A small percentage change in either of these two massive numbers can have a large impact on the re sidual government deficit. Just a 2 percent increase in govern ment revenues, and a simi lar decrease in government outlays, for example, could change the budget deficit by $30 billion. There are a number of reasons why government and private forecasts alike have been wide of the mark in estimating federal revenues and outlays 1. Economic Cycl es No forecaster has been able to predict the economic cycle with consistency assumptions about GNP, inflation, and interest rates can have a dramatic effect on budget deficit forecasts. Economic growth-The economic growth rate has enormous consequences f or the budget deficit. During periods of rapid expansion, business profits, wages, interest, and dividend income surge while the Treasury gains new tax revenues generated from these sources. Greater economic growth, meanwhile, reduces Yet even a small cha nge in the underlying Ibid. Ibid 8 government.expenditures, since more jobs mean 1ess.unemployment compensation, welfare, and other social spending. As such, the deficit is .cut from two directions. The most powerful way to cut the deficit is by economic growth. According to CBO calculations, a 2 percentage point sus- tained increase in economic growth would reduce the cumulative deficit by $440 billion during 1984-1987.8 If the economic fore casts u n derestimated growth by such a degree, that alone would reduce upcoming budget deficit projections by as much as one half (see Chart 11 CHART 11 The Effect on Budget Deficit Projections of Selected Changes in Economic Assumptions Increases in billions of d o llars FY 1983 N 1984 FY 1985 FY 1986 FY 1987 Real Growth up 1 10 -28 -46 -62 -83 Unemployment down 1 25 -39 -39 -37 -37 Inflation down 1% 6 14 18 24 29 Interest Rates down 1 1 -5 -8 10 -12 Source: CBO The Economic and Budget Outlook: An Update," September 1982 Inflation--Inflation directly affects federal spending and revenues. Inflation boosts nominal corporate profits, personal incomes, payrolls, and sales and therefore increases tax collec tions-particularly as people are pushed into higher income tax b r ackets. (Beginning in 1985, however, indexation of brackets will reduce these inflation induced tax revenues spending also increases due to inflation, because many government programs, including Medicare, Social Security, and federal pay, increase with in f lation. Combining its effects on revenues and outlays, a 1 percent drop in the inflation rate actually would increase the government deficit by a total of over $85 billion between 1984-1987 Federal Interest rates--The 1.4 trillion government debt has made interest expense one of the largest and fastest growing items in the federal budget, skyrocketing to nearly 130 billion in FY 19 83. Unanticipated fluctuations in interest rates can have Congressional Budget Office The Economic and Budget Outlook: An Upda te Washington, D.C U.S. Government Printing Office, September 1982 9 significant effects on the accuracy of federal budget forecasts. According to the CBO, a one percentage point drop in interest rates for all government securities would cut the cumulativ e 1984-1987 budget deficit by $35 billion ment compensation expenditures and other social welfare spending. Increased employment expands the tax base and therefore sharply increases tax revenues. According to the CBO, a one percentage point drop in unemplo yment would reduce the budget deficit by 150 billion over the next four years Unemployment-A falling unemployment rate reduces unemploy 2. Complexities in Revenue and Outlay Estimates Even if the economic cycle could be gauged accurately, the forecaster m u st convert the resulting economic assumptions into revenue and outlay estimates To do this, he must try to unravel the complexities of spending patterns, the timing of income tax payments, the use of tax deductions .and shelters, the utilization of govern m ent programs, and many other factors. Errors made in these elements are compounded in the final revenue and outlay forecasts and then magnified in his deficit projection 3. Leqislative Chanqes Forecasts assume certain legislative changes.in tax and But ev e n the most seasoned political pundits budget policy rarely predict accurately what Congress and the President will do 4. Other Surprises There are many other unexpected events that may trigger a different deficit than that projected by economists disaster , such as a drought or a bad winter, can drastically alter government outlays and consumer spending patterns and have a significant effect on the deficit A natural OMB IS NOT ALONE OMB is not the only agency or organization to have trouble in developing ac c urate forecasts. Some critics say OMB's forecasts naturally are inaccurate because politics clouds its decision making. Others charge that administrations are inherently less competent or motivated than the private sector or the Congressional Budget Offic e forecasts, however, reveals that poor performance is not an OMB monopoly A recent review of both private sector and CBO The Congressional Budget Office Chart I11 shows the deficit estimates of the CBO for 1979 to In some years CBO has been noticeably mor e accurate in es 1983 timating the deficit than the OMB. In most years, however, the 10 CHART I11 Congressional Budget Office Deficit Estimates 1979-1983 in billions of dollars Fiscal Date of Year Estimate Estimate Ac tua 1 1979 12/77 $38 $27.7 1980 1/79 4 9 59.6 1981 2/80 21 57.9 1982 7/81 30 110.6 1983 2/82 157 195.6 Error Estimate 27 22 176 269 25 Average error projection: 104 percent. Source: Congressional Budget Office errors have been in exactly the same direction and of about the same magni.tude. CBO' s yearly deficit projections from 1979 to 1983 on average erred over 100 percent. The errors ranged from 27 percent below to 269 percent above projected deficits. Private Sector Forecasts of GNP and the inflation rate for 1981-1983, compared with the actu al results. These basic economic variables are among the most fundamental determinants of budget deficits that, although some forecasters came closer than others to the actual outcome in certain years, no forecaster was consistently accurate. These leadin g private sector forecasters erred by an average of 137 percent in their projections of GNP private sector analysts erred annually in their forecasts of the inflation rate by 21 percent Chart IV provides a sample of leading private sector forecasts It can b e seen The same IMPLICATIONS FOR ECONOMIC POLICY The poor record of all forecasters--private as well as government--holds important implications for current public policy 1) Congress should not place more confidence in forecasters than their record justif i es--especially in the case of projections of the economy four or five years hence extremely reluctant to use highly speculative forecasts to justify fundamental and far-reaching economic policy changes Policymakers should be 11 CHART IV Private Forecasts o f GNP Forecaster Chase Econometrics Data Resources Brookings Tomsend-Greenspan Wharton Average Prediction Actual Error Each Year Average Error Over 3 Years 137 percent INFLATION RATE percent change over previous fiscal year FY 1981 FY 1982 0.6 0 0.8 -0.5 0 .7 -2.0 1.5 -1.0 0.3 -0.3 0.8 -0.8 2.6 -1.9 225 138 FY 1983 2.1 1.6 3.3 2.0 2.4 2.3 3.4 48 Average prediction Ac tua 1 Average Error FY 1981 FY 1982 FY 1983 11.0 8.0 4.7 10.3 6.1 3.2 6 24 32 Average Error Over 3 Years 21 percent Estimate Source: The Washi n gton Post, January 8, 1984 2) Given the inherent limitations of forecasting, policymakers should rethink the case for tax increases based on the current gloomy budget deficit forecasts that these forecasts will be any more accurate than those in previous y ears-while it is certain that .tax increases could choke off the recovery There is no reason to suppose 3) liFine-tuningll the economy, national industrial plans, and efforts to smooth out the business cycle require accurate forecast ing of economic trend s and aggregates. Firm statistics simply do not exist, so such detailed policies are doomed to failure. In deed, they even can destabilize the economy, given that errors in 12 forecasting closely follow the business cycle. economic policies based on errone o us forecasts actually could magnify swings in the economy, providing excessive stimulus in recoveries and too much restraint in recessions. Rather than providing policymakers with information to counterbalance the economic cycle, forecasting actually tend s to intensify economic fluctuations Countercyclicali1 4 ally less than now forecast, if forecasting errors follow the typical cyclical pattern of the business cycle 5 disclaimer that economic assumptions are only careful guesses, not a scientifically reli a ble forecast of the likely outcome. Indeed, when policy initiatives and budgets are based on.officia1 forecasts, the Administration and congressional committees should be required to make.public the track record of the forecasting agency during the previo u s ten years so that the public might award the projections proper credence 6) Since forecasts are unreliable, solid economic principles should become the basis for economic policy. Such policy .should lay a stable foundation for long-term growth and not b e adjusted on a yearly basis to anticipate flimsy predictions of future eco nomic conditions. Congress and the Federal Reserve System should continue to move toward a tax system that rewards risk and effort, a'regulatory system that removes barriers to ent e rprise, a spend- ing reduction policy that shrinks the public sector, and a mone tary policy that provides for stable and low money growth The actual deficit for 1984-1985 probably will be substanti Accompanying every government budget should be a promine n t CONCLUSION Government and private sector 'forecasts generate more heat than light economists and policymakers alike is hardly warranted by the record. Official deficit forecasts for only one year have erred on average 254 percent a year over the last th irteen years private sector firms, which earn their living from forecasting are inaccurate in predicting the future. three, four, and five years ahead is rarely closer to reality than a 1ook.into a crystal ball. Would planning the economy, manipulating agg regate demand, and pursuing fashionable cure-alls such as industrial policy seem so desirable if the sorry forecast record were more widely known? Would economists and policymakers push for the large tax increases now being debated if they realized the ma s sive range of errors inherent in any deficit prediction? The flawed record of forecasting certainly should cause economic advisors and lawmakers to use forecasts in a far more cautious and tentative manner The naive confidence placed in economic forecasts by Even A forecast of two, 13 But what should economic policymakers use as a guide if the science of prediction is so crude and unreliable? Perhaps they should recognize that the first step to wisdom is to recognize the obvious limitations in any sort of p rediction and act accord ingly cit or any other key variable, it should not try to persuade the American people to accept frequent policy changes to fit the latest economic forecasts. It should move instead to establish a broad, long-term tax and monetary framework, based on the les sons of economic history, and leave the guessing game to the gamblers on Wall Street If Congress cannot predict the size of the federal defi Thomas M Humbert Senior Policy Analyst and Walker Fellow in Economics
