The Economic and Budget Outlook€¦ · ISBN 0-16-045157-4. Preface This volume is one of a series...

CONGRESS OF THE UNITED STATESCONGRESSIONAL BUDGET OFFICE

A N

The Economicand Budget Outlook

U P D A T E

THE ECONOMIC AND BUDGET OUTLOOK:AN UPDATE

A Report to theSenate and House

Committees on the Budget

As Required by Public Law 93-344

The Congress of the United StatesCongressional Budget Office

NOTES

Unless otherwise indicated, all years referred to in Chapter 1 are calendar yearsand all years in Chapter 2 are fiscal years.

Some figures in this report indicate periods of recession using shaded verticalbars. The bars extend from the peak to the trough of the recession.

Unemployment rates throughout the report are calculated on the basis of thecivilian labor force.

Numbers in the text and tables of this report may not add to totals because ofrounding.

National income and product account data shown in the tables do not incorporatethe July 29, 1994, revisions, which did not significantly affect the economicoutlook. The discussion in Chapter 1 does, however, take account of therevisions.

For sale by the U.S. Government Printing OfficeSuperintendent of Documents, Mail Stop: SSOP, Washington, DC 20402-9328

ISBN 0-16-045157-4

Preface

T his volume is one of a series of reports on the state of the economy and the budgetthat the Congressional Budget Office (CBO) issues each year. It satisfies the re-quirement of section 202(f) of the Congressional Budget Act of 1974 for CBO to

submit periodic reports to the Committees on the Budget with respect to fiscal policy. Inaccordance with CBO's mandate to provide objective and impartial analysis, the reportcontains no recommendations.

The analysis of the economic outlook presented in Chapter 1 was prepared by theMacroeconomic Analysis Division under the direction of Robert Dennis and John F. Peter-son. Adrienne Kearney wrote the chapter, with contributions from Robert Arnold, LaurieBrown, Kim Kowalewski, Frank Russek, Jr., Matthew Salomon, and Christopher Williams.Matthew Salomon carried out the economic forecast and projections. Douglas Elmendorf,Victoria Farrell, Douglas Hamilton, Thomas Loo, Joyce Manchester, Angelo Mascaro, andJohn Sturrock provided comments and background analysis. Derek Briggs, Laurie Brown,Blake Mackey, John Romley, and Michael Simpson provided research assistance.

The baseline outlay projections were prepared by the staff of the Budget AnalysisDivision under the supervision of C.G. Nuckols, Paul N. Van de Water, Paul Cullinan,James Horney, Michael Miller, and Robert Sunshine. The revenue estimates were preparedby the staff of the Tax Analysis Division under the supervision of Rosemary Marcuss andRichard Kasten. Kathy A. Ruffing wrote Chapter 2 with contributions from Bryan Grote,Jeffrey Holland, and Richard Kasten. Matthew Salomon wrote Appendix A with LaurieBrown, and Robert Arnold wrote Appendix B. James Horney wrote the summary of thereport.

An early version of the economic forecast underlying this report was discussed at ameeting of CBO's Panel of Economic Advisers. Members of this panel are Michael Boskin,Barry Bosworth, Robert Dederick, Martin Feldstein, Benjamin Friedman, Lyle E. Gramley,Robert Hall, Lawrence Klein, Robert Lawrence, John Makin, Burton Malkiel, RudolphPenner, William Poole, Paul Samuelson, Charles Schultze, Robert Solow, James Tobin,Murray Weidenbaum, and Janet Yellen. Robert Gordon, Andrew Harless, James Medoff,Laurence Meyer, Michael Moran, and Edmund Phelps attended as guests. Although theseoutside advisers provided considerable assistance, this document does not necessarily reflecttheir views.

Paul L. Houts supervised the editing and production of the report, assisted by SherrySnyder. Major portions were edited by Paul L. Houts, Sherry Snyder, Sherwood Kohn, andChristian Spoor, with the assistance of Leah Mazade. Laurie Brown prepared the graphicsfor Chapter 1 and Appendix B. Marion Curry, Dorothy Kornegay, Linda Lewis, and L. RaeRoy assisted in the preparation and production of the report. With the assistance of MartinaWojak-Piotrow, Kathryn Quattrone prepared the report for final publication.

Robert D. ReischauerDirector

August 1994

Contents

ONE

TWO

APPENDIXES

A

B

C

SUMMARY

THE ECONOMIC OUTLOOK

CBO's Updated Economic Forecastfor 1994 and 1995 2

Factors Currently PropellingGrowth 2

Factors Currently RestrainingGrowth 7

A Neutral Factor: ResidentialConstruction 10

The Risk of Inflation 12Projections for the Years

Beyond 1995 18Risks to CBO's Economic Forecast 21

THE BUDGET OUTLOOK

The Deficit Outlook 28Changes in the Budget Outlook

Since Last Winter 29CBO Baseline Projections 34A Comparison with the

Administration's Projections 38The Federal Sector of the National

Income and Product Accounts 39

Evaluating CBO's Record ofEconomic Forecasts 45

Reestimating the NAIRU 59

Major Contributors to the Revenueand Spending Projections 65

IX

27

vi THE ECONOMIC AND BUDGET OUTLOOK: AN UPDATE August 1994

TABLES

S-l. Short-Term Economic Forecasts for1994 and 1995 x

S-2. Medium-Term Economic Projections xii

S-3. CBO Deficit Projections xiii

S-4. Changes in CBO Deficit Projections xv

1-1. The CBO Forecast for 1994 and 1995 3

1-2. The Fiscal Policy Outlook 8

1-3. Medium-Term Economic Projections forCalendar Years 1994 Through 1999 18

1-4. Medium-Term Economic Projections forFiscal Years 1994 Through 1999 19

1-5. Comparison of Congressional BudgetOffice, Administration, Federal Reserve,and Blue Chip Economic Forecasts 24

2-1. CBO Baseline Deficit Projections 29

2-2. Changes in CBO Baseline DeficitProjections 32

2-3. CBO Baseline Budget Projections,Assuming Compliance with DiscretionarySpending Caps 34

2-4. CBO Baseline Projections for MandatorySpending, Excluding Deposit Insurance 37

2-5. Comparison of CBO Baseline with OMBMidsession Review 38

2-6. Relationship of the Budget to theFederal Sector of the National Incomeand Product Accounts 40

2-7. Projections of Baseline Receipts andExpenditures Measured by the NationalIncome and Product Accounts 41

CONTENTS vii

A-l. Comparison of CBO, Administration, andBlue Chip Forecasts of Two-Year AverageGrowth Rates for Real Output 52

A-2. Comparison of CBO, Administration, andBlue Chip Forecasts of Two-Year AverageInflation Rates in the Consumer Price Index 53

A-3. Comparison of CBO, Administration, andBlue Chip Forecasts of Two-Year AverageInterest Rates on Three-Month Treasury Bills 54

A-4. Comparison of CBO, Administration, andBlue Chip Forecasts of Two-Year AverageLong-Term Interest Rates 55

A-5. Comparison of CBO, Administration, andBlue Chip Forecasts of Two-Year AverageInterest Rates on Three-Month TreasuryBills Adjusted for Inflation 56

A-6. Comparison of CBO and AdministrationForecasts of Four-Year Average GrowthRates for Real Output 57

B-l. Estimated Coefficients from Phillips CurveRegressions to Determine the NAIRU 63

FIGURES

S-l. The Federal Deficit xiv

1-1. The Federal Funds Rate Signals a Periodof Tighter Money 1

1-2. The Economic Forecast and Projections 4

1-3. Investment in Equipment Is StillSurging 5

1-4. The Burden of Household DebtRepayment Is Shrinking 7

1-5. Net Exports Move Accordingto World and U.S. Output 9

1-6. Mortgage Rates Have Turned Up 10

1-7. Housing Is Still Affordable 11

viii THE ECONOMIC AND BUDGET OUTLOOK: AN UPDATE August 1994

1-8. The Prices of Services Are Slowto Accelerate 15

1-9. Putting Recent Dollar Weaknessin Perspective 15

1-10. Real Short-Term Interest RatesAre Still Low 16

1-11. The Interest Rate Spread UsuallyNarrows Late in Expansions 17

1-12. Potential GDP Governs theProjection of GDP 20

1-13. The GDP Gap: GDP TypicallyOvershoots Its Mark 22

2-1. Changes in CBO Deficit Projections,March 1993 to August 1994 27

B-l. The Unemployment Rate, theReestimated NAIRU, and the Old NAIRU 59

BOXES

1-1. A New Measure of the Unemployment Rate 13

1-2. Using the NAIRU to Forecast Inflation 14

2-1. The Ten-Year Budget Outlook 30

Summary

T he Congressional Budget Office (CBO) pro-jects higher economic output and lowerdeficits in 1994 and 1995 than were antici-

pated last winter. In the longer run, however, thereis little change in the expected levels of gross do-mestic product (GDP) or deficits. Based on strongeconomic growth since the third quarter of 1993 andan expected continuation of high levels of businessinvestment and consumer spending, CBO haspushed its forecast of real growth in calendar year1994 up to 3.6 percent—almost 1 percentage pointhigher than the previous forecast. In addition toswelling the forecast levels of GDP in 1994 and1995, this unexpected growth has raised CBO's pro-jections of interest rates and consumer pricesthroughout the projection period. It has not, how-ever, significantly affected the estimate of potentialGDP or the projection of GDP for 1999.

CBO's baseline deficits will fall to $202 billionin fiscal year 1994 and to $162 billion in 1995-re-spectively $26 billion and $17 billion lower than thewinter baseline projections. Unfortunately, theshort-term improvement in the deficit outlook doesnot signal any brightening of long-term deficit pros-pects. CBO still projects that the deficit will beginto grow again—slowly in 1996 and more rapidly atthe end of the decade. By 1999, the deficit is ex-pected to reach $231 billion, somewhat higher thanthe $213 billion projected in the winter. CBO's lessdetailed 10-year budget projections indicate thatdeficits will continue to grow after 1999, and thereis no reason to believe that the trend will be re-versed after 2004 under current laws and budgetarypolicies.

The Economic Outlook

CBO's economic forecast reflects higher growth inthe short term than was envisioned in the winterforecast of The Economic and Budget Outlook:Fiscal Years 7995-7999, published in January 1994.This unexpectedly strong growth has pushed upanticipated interest rates and consumer prices buthas had little effect on CBO's projection of the sizeof the economy in 1999, the last year of the projec-tion period.

The Forecast for 1994 and 1995

CBO forecasts that real GDP will grow by 3.6 per-cent in 1994 and 2.7 percent in 1995, on a fourth-quarter-to-fourth-quarter basis (see Summary Table1). The growth rate for 1994 is almost 1 percentagepoint higher than was anticipated last winter. Theincrease is prompted by the strong growth recordedsince the third quarter of 1993 and CBO's assess-ment that the factors driving that surge-businessinvestment in plant and equipment and consumerspending for such durable goods as automobiles-will continue to spur growth through the middle of1995.

As a result of the rapid growth in 1994, CBOand most other economic forecasters agree that theGDP is near potential output (the level of real GDPthat is consistent with a stable rate of inflation).The other forecasters are divided, however, over the

x THE ECONOMIC AND BUDGET OUTLOOK: AN UPDATE August 1994

Summary Table 1.Short-Term Economic Forecasts for 1994 and 1995

Forecast

Nominal GDPCBO summerAdministrationBlue ChipCBO winter

Real GDPCBO summerAdministrationBlue ChipCBO winter

Consumer Price Index*CBO summerAdministrationBlue ChipCBO winter

Civilian Unemployment Rateb

CBO summerAdministrationBlue ChipCBO winter

Three-Month Treasury Bill RateCBO summerAdministrationBlue ChipCBO winter

Ten-Year Treasury Note RateCBO summerAdministrationBlue Chipc

CBO winter

1993

Fourth Quarter to Fourth Quarter(Percentage change)

5.45.45.44.9

3.13.13.12.3

2.72.72.72.7

Calendar Year Averages(Percent)

6.86.86.86.8

3.03.03.03.0

5.95.95.95.9

SOURCE: Congressional Budget Office; Office of Management and Budget, Mid-Session Review

1994

6.25.85.95.7

3.63.03.12.8

2.82.92.82.9

6.26.36.36.6

4.14.04.03.5

6.86.86.85.8

of the 1995 Budget

1995

5.35.65.85.4

2.72.72.62.7

3.23.23.43.0

5.86.26.06.4

5.54.74.84.3

6.87.07.26.0

(July 1994); EggertEconomic Enterprises, Inc., Blue Chip Economic Indicators (July 10, 1994).

a. The consumer price index for ail urban consumers.

b. The Bureau of Labor Statistics changed the unemployment survey in January 1994. The CBO summer forecast for 1994 and 1995 usesthe new survey method. The CBO winter forecast for those years has been adjusted upwardmake the forecasts comparable.

c. Blue Chip does not project a 10

Data for 1993 (shown in italics) use pre-1994 methodology.by one-quarter of a

-year note rate. The values shown here are based on the Blue Chip projection of

percentage point to

the Aaa bond rate,adjusted by CBO to reflect the estimated spread between Aaa bonds and 10-year Treasury notes.

SUMMARY XI

likely path of the economy from this point. Someanticipate that momentum will push actual GDPwell above potential, triggering a big increase ininflation and a spike in short-term interest rates sim-ilar to that experienced at the end of every postwarexpansion in the United States. Other forecastersbelieve that the growth of demand is already run-ning out of steam and that GDP will settle into apath just below potential output without much fur-ther action by the Federal Reserve to raise short-term interest rates.

CBO's forecast charts a middle course. RealGDP barely exceeds potential output and by the endof 1995 is beginning to fall back. Interest rates onthree-month Treasury bills are expected to increasefrom the unusually low rate of 3.0 percent in 1993to 4.1 percent in 1994 and 5.5 percent in 1995. Inthe absence of any unanticipated supply shocks, thefourth-quarter-to-fourth-quarter increase in the con-sumer price index will grow from 2.7 percent in1993 to 2.8 percent in 1994 and 3.2 percent in1995. Although greater than anticipated in the win-ter, these increases in inflation and short-term inter-est rates are smaller than those experienced at theend of previous postwar expansions.

The Congressional Budget Office believes thatthe turbulence associated with mature expansions inthe past is not likely to develop in 1994 and 1995because of the moderate progression of the currentexpansion, early action by the Federal Reserve torestrain inflationary pressures, and the low probabil-ity of a significant supply shock (such as the oilprice hikes that occurred as expansions were endingin the 1970s). The Blue Chip consensus of privateforecasters and the Administration's July 1994Mid-Session Review forecast also assume that theeconomy in 1994 and 1995 will not exhibit thespike in interest rates and in inflation associatedwith mature expansions in the past. In fact, theyassume significantly lower real growth in 1994 andlower short-term interest rates in 1995 than doesCBO-assumptions more in line with the smooth-landing scenario that predicts growth will gentlylevel off near potential output.

The unemployment rate is a key variable inassessing the state of the economy, but current dataon unemployment are difficult to interpret because

of a change in survey methodology introduced bythe Bureau of Labor Statistics last January. None-theless, it is clear that unemployment has fallenfaster than CBO anticipated last winter. The CBOforecast assumes that the unemployment rate willcontinue to fall-to 6.2 percent in 1994 and 5.8 per-cent in 1995 (based on the new methodology). At5.8 percent, unemployment will be below CBO'scurrent estimate of the nonaccelerating inflation rateof unemployment (NAIRU-the minimum rate ofunemployment consistent with a stable rate of infla-tion). A projected unemployment rate lower thanthe NAIRU is one sign of the slow buildup of infla-tionary pressures that is reflected in the Congres-sional Budget Office's forecast.

Projections for 1996 Through 1999

The Congressional Budget Office does not attemptto forecast cyclical fluctuations in the economymore than two years into the future. Beyond twoyears, CBO projects a course for the economy thatwill bring GDP for the last year of the projectionperiod (1999 in this case) to a level-consistent withthe average historical relationship between actualand potential GDP—slightly below estimated poten-tial output. The estimate of potential gross domes-tic product is based on an analysis of such funda-mental factors as growth in the labor force, produc-tivity, and national saving.

The increase in the growth forecast for 1994 didnot cause CBO to change its winter estimate thatpotential GDP would grow at an average rate of 2.4percent a year during the 1996-1999 projectionperiod. It did, however, push the level of real GDPforecast for 1995-the starting point for the 1996-1999 projections-above potential. Therefore, inorder to allow projected GDP to fall to its historicalrelationship with potential output in 1999, CBOreduced the projected average annual growth of realGDP between 1995 and 1999 from 2.6 percent to2.2 percent (see Summary Table 2).

The projection assumes that the interest rate onthree-month Treasury bills will fall from 5.5 percentin 1995 to 5.1 percent in 1996 and 4.9 percent forthe rest of the projection period. The projectedinterest rate for 10-year Treasury notes is 6.5 per-

xii THE ECONOMIC AND BUDGET OUTLOOK: AN UPDATE August 1994

cent throughout the 1996-1999 period, down from6.8 percent in 1995. Both the short- and long-termrates are somewhat higher than those projected inthe winter. The annual increase in the consumerprice index is expected to grow from 3.1 percent in1995 to 3.4 percent a year after 1996, up from thepeak rate of 3.1 percent in the winter projections.The unemployment rate is expected to increaseslightly after 1995, leveling off at 6.1 percent in1998 and 1999.

The Budget Outlook

The Congressional Budget Office projects that thedeficit will fall faster in fiscal years 1994 and 1995

than was anticipated in the winter baseline, pub-lished as Appendix A in CBO's April 1994 report,An Analysis of the President's Budgetary Proposals.There is little change, however, in the long-termbudget outlook; the deficit will still begin growingafter 1995, and by 1999 the projected deficit will beslightly higher than had been estimated in the winterbaseline.

The Outlook for the Deficit

CBO's baseline projections assume that the Con-gress makes no changes in current law and policiesthat affect tax revenues and mandatory spending.They also assume compliance with the discretionaryspending limits set in the Budget Enforcement Act

Summary Table 2.Medium-Term Economic Projections (By calendar year)

Actual1993

Forecast1994 1995 1996

Projected1997 1998 1999

Nominal GDP(Billions of dollars)

Real GDP (Billions of1987 dollars)

Real GDP(Percentage change)

Implicit GDP Deflator(Percentage change)

CPI-U (Percentage change)

Unemployment Rate(Percent)

Three-Month TreasuryBill Rate (Percent)

Ten-Year TreasuryNote Rate (Percent)

6,378

5,136

3.0

2.5

3.0

6.8

3.0

5.9

6,777

5,343

4.0

2.2

2.6

6.2

4.1

6.8

7,161 7,523 7,893 8,277 8,687

5,505 5,635 5,756 5,878 6,009

3.0

2.5

3.1

5.8

5.5

6.8

2.4 2.1

5.9 6.0

5.1 4.9

6.5 6.5

2.1 2.2

2.6 2.7 2.7 2.7

3.3 3.4 3.4 3.4

6.1 6.1

4.9 4.9

6.5 6.5

SOURCE: Congressional Budget Office.

NOTE: CPI-U is the consumer price index for all urban consumers.

SUMMARY xin

of 1990 (for 1995) and in the Omnibus Budget Rec-onciliation Act of 1993 (OBRA-93, for 1996, 1997,and 1998), which allow outlays to increase by lessthan $3 billion between 1994 and 1998-a cut ofmore than 10 percent in constant dollars. CBO as-sumes that discretionary spending will grow at thesame pace as inflation after 1998, when the discre-tionary caps have expired.

CBO projects that the baseline deficit will de-cline to $202 billion in 1994 and $162 billion in1995 (see Summary Table 3). The $162 billionrepresents the lowest deficit since 1989 and the low-est deficit as a percentage of GDP (2.3 percent)since 1979. It also represents a $128 billion reduc-tion below the record-high deficit of $290 billionposted in 1992. As recently as March 1993, CBOprojected that the 1995 deficit would be only a fewbillion dollars below that record-high level. Theimproved outlook for 1994 is the result of policychanges enacted in OBRA-93, economic perfor-

mance that has been better than anticipated, andreestimates of spending and revenues for a varietyof technical reasons. The policy changes accountfor a little less than half of the improvement in the1995 deficit.

The same factors that have pushed down theprojected level of the 1995 deficit since March 1993have also reduced the deficits for other years. Thecurrently projected deficit for 1998 is more than$160 billion lower than was anticipated in March1993. The policy changes enacted in OBRA-93account for almost 90 percent of the reduction, andthe rest is attributable to a strengthening of pro-jected economic performance. These factors havenot, however, altered the underlying trend of deficitsthat, after falling from the high levels of the early1990s, rise steadily as the decade draws to a close.CBO projects that the deficit will grow at an aver-age rate of more than 9 percent a year between1995 and 1999, reaching $231 billion in 1999.

Summary Table 3.CBO Deficit Projections (By fiscal year)

1993 1994 1995 1996 1997 1998 1999

Total Deficit

Standardized-Employment Deficit8

Total Deficit

Standardized-Employment Deficit5

Memorandum:Gross Domestic Product(Billions of dollars)

In Billions of Dollars

255 202 162 176 193 197 231

221 184 183 195 200 196 223

As a Percentage of GDP

4.0 3.0 2.3 2.4 2.5 2.4 2.7

3.4 2.7 2.6 2.6 2.6 2.4 2.6

6,295 6,677 7,070 7,431 7,800 8,179 8,581


a. Excludes the cyclical deficit and deposit insurance spending.

b. Expressed as a percentage of potential GDP.

xiv THE ECONOMIC AND BUDGET OUTLOOK: AN UPDATE August 1994

Since nominal GDP is expected to grow at an aver-age annual rate of about 5 percent during the sameperiod, the deficit will increase as a percentage ofGDP from 2.3 percent in 1995 to 2.7 percent in1999 (see Summary Figure 1).

Part of the fluctuation in deficits is the result ofcyclical changes in the economy. Revenues tend tobe lower and outlays higher when the economy isperforming below potential. The opposite occurswhen actual output exceeds potential. The stand-ardized-employment deficit is a measure of the dif-ference between revenues and outlays that wouldoccur if the economy were operating at potential (italso excludes the net expenditures of deposit insur-ance agencies and the Desert Storm contributionsreceived in the early 1990s). Taken as a percentageof potential GDP, this measure also declines fromhigh levels in the early 1990s, flattens out, and thenbegins rising at the end of the decade. Debt held bythe public, as a percentage of GDP, follows a simi-lar path.

CBO's abbreviated projections for the 2000-2004 period indicate that under current laws andbudgetary policies the growth in deficits experi-enced at the end of the 1990s will continue, withthe deficit equaling 3.6 percent of GDP in 2004.

Summary Figure 1.

The Federal Deficit (By fiscal year)

Percentage of GDP

5 -

4 -

3 -

1

Actual I Projected

Total

Standardized-Employment

1987 1989 1991 1993 1995 1997 1999


There is no reason to assume that the trend will bereversed after 2004 unless policies are changed. Infact, additional pressures on federal spending, whenbaby boomers start becoming eligible in large num-bers for Social Security and Medicare benefitsaround 2010, are likely to increase the growth ofdeficits.

The growth in the projected deficit after 1995will continue to be driven by the increasing costs ofthe two major federal health care programs: Medi-care and Medicaid. These two programs togetherwill grow at an average annual rate of approxi-mately 11 percent from 1995 to 1999, all other out-lays will increase at a rate of less than 4 percent,and revenues will increase at almost the same rateas GDP. If increases in spending for Medicare andMedicaid were more in line with the growth of oth-er outlays, the deficit would not grow as a percent-age of GDP. Comprehensive reform of the nation'shealth care system, currently being considered bythe Congress, could significantly affect spending forMedicare and Medicaid. Under any of the majorproposals considered thus far, however, any savingsin these two programs are more than offset by in-creased spending for other purposes, such as subsi-dies to allow low-income families to purchasehealth insurance. Increased revenues may keephealth care reform legislation from increasing defi-cits, but it is unlikely that the adoption of such aplan will significantly reduce them, at least throughthe next 10 years.

Changes in the Projections

Unlike the report that CBO issued in September1993, this update does not reflect fundamental shiftsin federal spending, revenues, or deficits. Lastyear's report showed the effects of the enactment ofOBRA-93, which reduced deficits by more than$400 billion over five years. Changes from lastwinter's projections that are reflected in this reportamount to a net reduction of just $18 billion overthe 1994-1999 period (see Summary Table 4). Leg-islation enacted since the winter baseline accountsfor less than $500 million of that net reduction;changes in CBO's economic forecast and technicalestimating assumptions are responsible for the rest.

SUMMARY xv

Changes in the economic forecast reduce pro-jected deficits in the short run (by as much as $8billion in 1995) and increase them in the long run(by $12 billion in 1999). The combination ofhigher-than-anticipated economic growth in 1994and 1995 and increased corporate profits push pro-jected revenues up in all years, but by less in 1998and 1999 than in the earlier years. The higherinterest rates that accompany faster growth drive upnet interest expenditures. The increases in projectednet interest also decline slightly after 1996, but notas rapidly as the reestimates of revenues decline.By 1998, the increases in net interest costs, togetherwith relatively small increases in other outlays,overwhelm the effect of higher revenues and pushdeficits up. In any case, compared with some of thelarge reestimates in previous reports, none of theincreases or decreases in any single year represent amajor shift.

Changes from the winter projections that cannotbe attributed to either legislation or the new eco-nomic forecast—so-called technical reestimates—alsotend to reduce the deficit in the early years of the

projection period (by $18 billion in 1994) and in-crease it later (by $6 billion in 1999). Technicalreestimates increase projected revenues in each yearthrough 1997 but decrease them in 1998 and 1999.Technical reestimates push down outlays projectedfor 1994 and 1995 but increase them after that.CBO estimates that Medicare and Medicaid expen-ditures in 1994 will be $4 billion lower than antici-pated in the winter baseline. Because it is not clearwhether these reductions represent a trend or anaberration in the patterns of Medicare and Medicaidspending, CBO has made no adjustment to the1995-1999 outlays projected for these two pro-grams. Reestimates of expenditures by deposit in-surance agencies are small compared with those inmany recent reports but still account for as much asa $5 billion shift in one year. Expenditures for theearned income tax credit (the refundable portion ofwhich is recorded as outlays) are consistently higherthrough the period than CBO's estimates in thewinter baseline. Reestimates in a number of otherprograms make up the remainder of the revisions inoutlays attributable to changes in technical estimat-ing assumptions.

Summary Table 4.Changes in CBO Deficit Projections (By fiscal year, in billions of dollars)

1994 1995 1996 1997 1998 1999

Winter Baseline Deficit

ChangesLegislative changesEconomic assumptionsTechnical reestimates

228

a-8

-18

180

a-8

-10

180

a-52


NOTE: The projections include Social Security and the Postal Service, which are off-budget.

a. Less than $500 million.

192

a-23

187

a64

213

a126

Total

Summer Baseline Deficit

-26

202

-17

162

-4

176

1

193

10

197

18

231

xvi THE ECONOMIC AND BUDGET OUTLOOK: AN UPDATE August 1994

Conclusion

Although a higher rate of economic growth and alower deficit in 1994 represent good news, the goodnews is tempered. The recent surge of growth doesnot herald an improvement in the rate at which po-tential output will grow. Because the economy isalready operating close to potential, an increase ingrowth above the rate of potential in the short runmust be offset by slower growth in the longer run.

Lower deficits in 1994 and 1995 also do notsignify any fundamental improvement in the long-term fiscal health of the nation. Under currentpolicies, deficits will begin to rise relentlessly as thedecade draws to a close. The current low levels ofthe deficit should be appreciated as the fruits of thespending reductions and tax increases enacted inOBRA-93. They should not, however, distractpolicymakers from the need for additional deficitreduction to increase national saving and investmentand improve the long-term prospects for theeconomy.

Chapter One

The Economic Outlook

S ince the third quarter of 1993, U.S. eco-nomic growth has been robust. Although asubstantial gap existed in mid-1993 between

the economy's output and what it could potentiallyproduce, the strong growth of the last three quartershas whittled away that gap: by mid-1994, both thecapacity utilization rate and the unemployment ratereached levels where inflationary pressures normallybegin to be a concern. The Congressional BudgetOffice (CBO) anticipates that the strong growth ofreal gross domestic product (GDP) will continueduring the second half of 1994 and early 1995.

The strength of the economy during the lastthree quarters was largely unanticipated. Both theBlue Chip consensus forecast and the CBO forecastprepared during the last quarter of 1993 indicatedabout 3 percent growth for the end of 1993 and thefirst half of this year.1 Instead, the economy grewat a vigorous 4.4 percent rate. Growth in employ-ment also accelerated. During the first 10 monthsof 1993, employment grew by an average of180,000 jobs a month, but between October 1993and July 1994, 270,000 jobs have been addedmonthly. Hours worked increased even morerapidly, and the unemployment rate fell from 6.7percent in January to 6.1 percent in July. Stronggrowth in investment and consumption of durablegoods spurred the pickup in economic activity. Ex-penditures by businesses for plant and equipment in-creased by about 14 percent over the three quarters,

See Congressional Budget Office, The Economic and Budget Out-look: Fiscal Years 1995-1999 (January 1994); and Eggert Eco-nomic Enterprises, Inc., Blue Chip Economic Indicators (January10, 1994).

and consumption of durable goods grew by morethan 8 percent. Given such a strong performance ofthe economy over the previous three quarters, thecurrent CBO forecast points toward higher GDP andinterest rates, and slightly more inflation, than lastwinter's forecast.

The Federal Reserve acted early this year at thefirst signs of the surprising economic strength re-ported in January. Between February 4 and May17, it increased the federal funds rate in four steps—from 3 percent to 4.25 percent. These moves sig-naled a major turning point in monetary policy,which had slowly lowered short-term interest ratesover the previous five years (see Figure 1-1). Other

Figure 1-1.The Federal Funds Rate Signalsa Period of Tighter Money

Percent20

15

10

1955 1965 1975 1985 1995

SOURCES: Congressional Budget Office; Federal ReserveBoard.

NOTE: Based on monthly data through July 1994.

2 THE ECONOMIC AND BUDGET OUTLOOK: AN UPDATE August 1994

short-term rates closely followed the hike in thefederal funds rate, but long-term rates climbed morethan short-term rates. Since January, three-monthTreasury bill rates have risen by about 135 basispoints, while 10-year Treasury rates have risen byabout 150 basis points. Policymakers hoped thatfinancial market participants would interpret theFederal Reserve's actions as prudent measures tohead off inflation. Instead, the financial communityanticipated further adjustments in the federal fundsrate and a rise in inflation. As a result, long-terminterest rates rose.

CBO's Updated EconomicForecast for 1994 and 1995

CBO expects the economy to continue to expand in1994 at an annual rate of about 3.6 percent on afourth-quarter-to-fourth-quarter basis (see Table 1-1and Figure 1-2). This forecast for growth in 1994is almost 1 percentage point higher than the rate of2.8 percent that was forecast last winter. The pri-mary reason for the increase is that business invest-ment in equipment and spending on consumer dur-ables, which accounted for the bulk of the unex-pected growth in demand recently, are likely to con-tinue to spur growth through mid-1995. Stronggrowth in disposable income, together with somepent-up demand for light vehicles (automobiles,small pickup trucks, sport/utility vehicles, and mini-vans), will stimulate consumption. Moreover, firmsnow operating at close to full capacity are likely tocontinue to expand their industrial base to meet theincreased demand for goods and services.

If the U.S. economy continues to grow abovepotential (the level of output consistent with stableinflation) during the last half of 1994 and early1995, the Federal Reserve is likely to increase short-term interest rates further in order to stem a rise ininflation. Three-month Treasury bill rates are ex-pected to escalate from their average of 4.0 percentin the second quarter to over 5l/2 percent during1995, approximately 1.2 percentage points higherthan the CBO forecast last winter. The forecast forlong-term rates (measured by the rate on 10-yearTreasury notes) has also been adjusted upward sincethe winter by almost a full percentage point to 6.8percent in 1994 and 1995.

Because the labor force is expected to growfaster in 1995, CBO forecasts the rate of unemploy-ment to fall only gradually, from an average of 6.2percent in 1994 to 5.8 percent in 1995. The currentforecast is about 0.5 percentage points lower thanlast winter's forecast.

Stimulated by activity in the household andbusiness sectors, the economy is forecast to grow ata 3l/2 percent rate through the second half of 1994,even though declining federal expenditures and adeterioration of net exports will restrain growth. In1995, however, GDP growth will slow as invest-ment and consumption subside, though the foreignsector is likely to stimulate growth by increasing thedemand for U.S. exports.

Factors Currently PropellingGrowth

Investment and consumption are the primary com-ponents of demand propelling growth in 1994. Theneed to meet the increased growth in final demandwill drive expenditures by firms on plant and equip-ment, while growth in disposable income will stim-ulate expenditures by households.

Business Investment

Capital investment by businesses-particularly in-vestment in equipment-has been the major sourceof the economy's strength for more than a year, andit should continue to be so for the rest of this yearbefore slowing down in 1995. A combination offactors has spurred spending on equipment sincemid-1992—strong growth in final demand, risingprofit margins, declining capital costs, and a contin-ued drive by corporations to strengthen their balancesheets. In addition, investment in construction,though not a strong source of growth, is apt todetract less from growth through the end of nextyear than it did during the last three years. By1995, with the economy running at capacity, risinginterest rates are likely to temper, though not re-verse, the growth in business investment.

CHAPTER ONE THE ECONOMIC OUTLOOK 3

Table 1-1.The CBO Forecast for 1994 and 1995

1993a 1994Forecast

1995

Nominal GDPCBO summerCBO winter

Real GDPb

CBO summerCBO winter

Implicit GDP DeflatorCBO summerCBO winter

Consumer Price Index0



5.44.9

3.12.3

2.22.5

2.72.7

6.25.7

3.62.8

2.52.8

2.82.9

5.35.4

2.72.7

2.52.6

3.23.0

Real GDP Growth6


Civilian Unemployment Rated


Three-Month Treasury Bill RateCBO summerCBO winter

Ten-Year Treasury Note RateCBO summerCBO winter


3.02.8

6.86.8

3.03.0

5.95.9

4.02.9

6.26.6

4.13.5

6.85.8

3.02.7

5.86.4

5.54.3

6.86.0

SOURCES: Congressional Budget Office; Department of Commerce, Bureau of Economic Analysis; Department of Labor, Bureau of LaborStatistics; Federal Reserve Board.

a. The data for 1993 are actual values for the summer forecast but are estimates for the winter forecast.

b. Based on constant 1987 dollars.

c. The consumer price index for all urban consumers (CPI-U).

d. The Bureau of Labor Statistics changed the unemployment survey in January 1994. The CBO summer forecast reported in this tableuses the new survey methodology. The CBO winter forecast has been adjusted upward by about one-quarter of a percentage point tomake the forecast comparable. Data for 1993, shown in italics, use pre-1994 methodology.


Over the last two years, the growth of invest-ment in equipment was spurred primarily by theeconomic expansion, which raised expectations forboth sustained growth in demand and the currentand future profitability of corporations. In a contin-uation of rapid growth that started in early 1992,

business fixed investment grew 13 percent over thepast four quarters. All of the growth was in equip-ment spending, with investment in business struc-tures virtually flat over the past two years. Outpac-ing the growth in real GDP, total spending onequipment grew by 17 percent over the past year,

Figure 1-2.The Economic Forecast and Projections

Real GDP GrowthPercent

1980 1985 1990 1995

Inflation3

12Percent

Actual Projected

1980 1985 1990 1995

Civilian Unemployment Rateb

10

8

6

4

2

Percent

Actual

I

Projected

CBOWinter(Adjusted)

1980 1985 1990 1995

Interest RatesPercent

1980 1985 1990 1995

SOURCES: Congressional Budget Office; Department of Labor, Bureau of Labor Statistics; Department of Commerce, Bureau of EconomicAnalysis; Federal Reserve Board.

NOTE: All data are presented on an annual basis; growth rates are year-over-year. For 1996 and subsequent years, the projections do notattempt to reflect cyclical patterns.

a. Consumer price index for all urban consumers (CPI-U). The treatment of home ownership in the official CPI-U changed in 1983. Theinflation series in the figure uses a consistent definition throughout.

b. From 1994 onward, the unemployment rate reported by the Bureau of Labor Statistics is not comparable with prior data. The discontinuityreflects an extensive revision of survey methods. CBO's summer forecast is made on the basis of the new methods. The winter forecastof the unemployment rate, originally reported on the old basis, has been adjusted upward by one-quarter of a percentage point.


Figure 1-3.Investment in Equipment Is Still Surging

10 Percentage of Real GDP

i t . . M\. .£. 18. i i i i i i i > i i i i i i i i I i 11 i i i i i i I :if i i i I i I

1950 1960 1970 1980 1990

SOURCES: Congressional Budget Office; Department of Com-merce, Bureau of Economic Analysis.

NOTE: The figure shows real investment in producers' durableequipment as a percentage of real GDP.

and spending for equipment other than computersgrew by 12 percent (see Figure 1-3).2

Will the equipment boom continue? For 1994,the prospects for continued strong growth are good.At midyear, new orders for nondefense capitalequipment (excluding aircraft and parts) were upnearly 23 percent over mid-1993 levels, consider-ably above the pace of shipments. Moreover, facto-ries are operating at historically high rates. Sincethe end of 1993, the Federal Reserve Board's indexof capacity utilization in manufacturing has re-mained above 82 percent, a sign of capacity con-straints. By pressing the capacities of current plantand equipment, growth in final demand should keepthe pace of investment spending from droppingmuch this year. Together with the relatively healthyprofit margins that currently prevail, the strength infinal demand is likely to overwhelm the adverseeffects of rising short-term interest rates on invest-ment plans for the remainder of 1994.

2. Computers should be considered separately from other investmentin equipment because the rapid growth in real investment in com-puters stems as much from the decline in the price of computersas it does from increases in nominal spending. Additional invest-ment in computers is probably worth much less to the productivecapacity of the economy~and its production requires much lessemployment and demand for the output of other industries-thanits size in the national accounts indicates.

In 1995, however, businesses are likely to pur-chase capital equipment at a slower rate. By then,the rapid growth of investment in the precedingyears will have boosted productive capacity to alevel sufficient to meet demand. With the economygrowing at its potential rate, the upward climb ininterest rates begun in 1994 should temper furthergrowth in spending on equipment by raising the costof borrowing and pressing profit margins.

Even so, investment in equipment is expected tohold at a fairly high level in 1995. Contributing tothis moderate response are the strengthened balancesheets of the nonfmancial corporate sector. Sincelate 1990, corporations have shifted their liabilitiestoward equity and long-term debt, which generallyimply more predictable financing costs than doesshort-term debt. Consequently, businesses are nowapt to be less vulnerable to short-term increases ininterest rates than they were in the 1980s.

Whereas spending on equipment generally fol-lows the business cycle, business expenditures forstructures appear to be governed by longer-termcycles. Throughout the boom in equipment invest-ment in 1992 and 1993, spending on constructionby businesses was sluggish. Recently, however,some signs of life have appeared. Construction ofcommercial space such as wholesale and retail out-lets has picked up, and industrial building activitygrew by more than 10 percent over the past year.Office construction continues to be weak, however.Though much lower than the astronomical peaks inthe 1980s, office vacancy rates remain high enoughin many areas to continue to cramp the growth ofspending on office buildings. Investment in petro-leum drilling also continues its secular decline, andboth hospital construction and utilities constructionhave remained flat. Overall, investment in businessstructures is expected to add little to GDP growthduring the forecast period, but at least it will not sapgrowth, as it did in the 1990-1993 period.

Inventory accumulation since the recession hasbeen lethargic compared with past recoveries andexpansions, and, although there was a spurt of in-ventory investment in the second quarter of thisyear, CBO does not expect strong inventory accu-mulation during the forecast. In fact, inventorieshave slipped relative to sales throughout this busi-ness cycle.


Inventories were reported to have jumped in thesecond quarter, however, prompting some concernthat economic activity would slow rapidly in thesecond half of this year. Analysts feared that thehigh level of inventories at retail and wholesaleoutlets would result in the cancellation of orders tomanufacturers and manufacturing output wouldtherefore slow. The increase in inventories wasindeed large for one quarter, but the ratio of inven-tory to sales is still quite low. Hence, the increaseis probably not a reliable signal of a slowdown inmanufacturing, and was largely the result of a tem-porary weakness in sales of durable goods andapparel. Even a moderate resurgence in sales ofthese goods will keep inventories in line.

Consumer Spending

The outlook for consumer spending continues to behealthy. Although the growth in consumer spendingeased in the second quarter of this year, the slow-down can be tied to special factors that are likely toreverse themselves in the coming months. Contin-ued gains in employment are expected to supportgrowth in disposable personal income and henceconsumer spending for the remainder of this year inspite of the recent rise in interest rates and theweakness in the stock market. The pace of con-sumer spending is likely to slow modestly nextyear, however, as higher interest rates begin to bite.

The reduction in the growth of consumer spend-ing in the second quarter of this year is not likely tobe sustained. The slower growth of spending onconsumer services, food, and other nondurablegoods came on the heels of strong increases in thefirst quarter, while the slowdown in spending onapparel may reflect its inherent volatility and prob-lems with the seasonal adjustment of apparel prices.A dip in purchases of motor vehicles stemmed inpart from the inability of manufacturers to satisfythe demand for a number of popular light vehicles.However, sales of vehicles are likely to reboundlater this year as manufacturers build more vehiclesto meet demand.

Indeed, the primary factor determining expendi-tures on consumption—namely, strong growth indisposable income—is expected to continue over the

forecast period. In the first seven months of 1994,about 2 million jobs were created. Moreover, theaverage hours each employee worked per week in-creased in the first half of 1994, and now exceedthe hours worked at the peak of the previous busi-ness cycle in 1990. Given the postwar downtrendin hours worked per week, this development isunusual and suggests that hours worked per weekwill not expand much more. Firms will probablyhire additional workers rather than try to increasehours per worker any further. This escalation indemand for labor will bolster the growth of totalcompensation and wages. In the CBO forecast, realcompensation per hour grows at a slightly briskerpace during 1994 and 1995 than it has in recentyears.

Strong household balance sheets will also con-tinue to support spending on consumption. House-hold net worth may have declined slightly in thefirst seven months of 1994, mainly as a result of adrop in stock and bond values. The Standard andPoor's 500, for example, dipped 3.9 percent be-tween January and July of 1994. But that drop instock prices is relatively small, and it is not likely tohave that much impact on consumer expenditures.

Developments on the liability side of householdbalance sheets also support consumer spending.The repayment burden has plunged in recent years(see Figure 1-4). In the first quarter of 1994, it hasstabilized, reflecting higher interest rates and stronggrowth in consumer installment credit. In fact, theincrease in installment credit suggests that consum-ers are comfortable with their financial situation andconfident about the growth in their future income.

The growth in consumer spending is likely toebb next year, however. Spending on consumerdurables should slow as more of the pent-up de-mand for durable goods is met. Moreover, stronggrowth in income should more than offset the effectof rising interest rates in the latter half of 1994.But higher interest rates will begin to take their tollin 1995 as the growth in disposable income flagssomewhat. As a result, the increased cost of bor-rowing as well as interest payments on existing con-sumer debt will dampen household expenditures onconsumption.


Figure 1-4.The Burden of Household DebtRepayment Is Shrinking

Percentage of Disposable Personal Income

1960 1970 1980 1990

SOURCES: Congressional Budget Office; Department of Com-merce, Bureau of Economic Analysis; Federal Re-serve Board.

NOTE: Debt repayments are shown as a percentage of dispos-able personal income. The latest data are for the firstquarter of 1994.

Factors CurrentlyRestraining Growth

Further fiscal tightening and net exports are ex-pected to restrain growth in 1994. However, in1995, the government sector should have a neutralimpact on growth, while recovering economies inthe rest of the world are likely to stimulate growththrough an increase in the demand for U.S. exports.

Fiscal Policy

Because of the underlying near-term strength inconsumption and investment, the fiscal restraint thisyear from the Omnibus Budget and ReconciliationAct of 1993 (OBRA-93) has not brought aboutmuch loss of output and employment. As it nowappears, that restraint has instead helped to checkinflationary pressures. Despite continued reductionsin real federal purchases, however, fiscal policy willassume a neutral stance in 1995, which will con-tinue throughout the projection period.

CBO estimates that under current tax and spend-ing policies the total budget deficit will declinefrom $255 billion in fiscal year 1993 to $202 billionin 1994 and to $162 billion by 1995 (see Table1-2). Thereafter, it escalates throughout the projec-tion period, reaching $231 billion by 1999. More-over, under current law, much heftier deficits arelikely in the next century, despite all of the deficitreduction undertaken in OBRA-93.

To measure the impact of fiscal policy on short-term growth, the total deficit is usually adjusted byremoving cyclical fluctuations in receipts and out-lays as well as outlays for deposit insurance. Acommonly used measure of fiscal policy that incor-porates these adjustments is the standardized-employment deficit. It calculates what the budgetdeficit would be without deposit insurance if theeconomy were operating at potential GDP. Changesin the standardized-employment deficit from oneyear to the next indicate the degree of fiscal stimu-lus or restraint. When this deficit rises, fiscal policyis considered stimulative; when it falls, fiscal policyis restrictive.

Based on the new economic and budget esti-mates in this report, the standardized-employmentdeficit will come down from 3.4 percent of potentialGDP in fiscal year 1993 to 2.7 percent in 1994 and2.6 percent in 1995. Thus, moderate restraint in1994 gives way to an essentially neutral policystance in 1995. In part, this movement reflects therecent rise in interest rates, which has resulted inhigher interest payments on new and refinancedfederal debt. But these additional interest paymentsmay have little impact on short-term growth. In-stead of spending this additional interest, debt-holders may choose to save most of it, especially ifthe higher interest earnings simply compensate themfor expected losses in their wealth as a result ofhigher inflation. Excluding interest payments, thestandardized-employment deficit shows about thesame amount of restraint in 1994 but somewhatmore in 1995-the first full year of higher interestrates in the CBO forecast.

With very little change after 1995 in the stan-dardized-employment deficit relative to potentialGDP, fiscal policy would no longer help to raisefuture living standards by stemming the federal


Table 1-2.The Fiscal Policy Outlook (By fiscal year on a budget basis)

1993 1994 1995 1996 1997 1998 1999

Total Budget DeficitStandardized-employment deficit*Cyclical deficit5

Memorandum:Deposit InsuranceNet Interest


25522161

-28199

20218422

-5202

162183-3

-17226

176195-7

-12245

193200-2

-5253

1971966

-5264

23122312

-4277

As a Percentage of Potential GDP

Total Budget DeficitStandardized-employment deficit8

Cyclical deficit5

3.93.41.0

3.02.70.3

2.32.6

0

2.42.6

-0.1

2.52.6

0

2.42.40.1

2.72.60.1


a. Excludes cyclical fluctuations and outlays for deposit insurance.

b. A negative value in this line indicates a surplus.

drain on national saving. Although making furtherdeficit reductions would change this pattern in fu-ture years, the longer such decisions are postponed,the more difficult they will be to make. CBO'sprojection does not explicitly include another reces-sion, but the chance that one will occur before theend of the decade is significant. When it does hit,many policymakers will want to avoid exacerbatingan economic downturn, thereby thwarting any at-tempts to reduce the federal deficit. If only for thisreason, it would be much easier to reduce deficitsnow rather than later.

Pending legislation to reform the health caresystem will probably have little effect on the size ofthe federal deficit, although it could significantlyalter the composition of receipts and outlays. It isalso unlikely to change private saving very much.Most proposals, however, would involve a redistri-bution of health care costs among different groupsof employers and employees, and thus could altertheir hiring practices and employment decisions.3

The extent of these distortions will depend on whatspecific provisions are ultimately enacted, but uncer-

tainty about this outcome could be affecting em-ployment decisions now.

State and Local Budgets

Although state and local governments are not ex-pected to add much to the overall economy in thenext two years, neither are they expected to restrainit, as they have in recent years. Actually, state andlocal governments are in their best fiscal positionsince the start of the recession in 1990.

Over the past few years, states have been moreconservative in their revenue estimates and haverestructured many of their programs to promote effi-ciency. As a result, 1994 revenue collectionsmatched or exceeded projections in almost all of thestates, and fewer states are having to make midyear

3. See Congressional Budget Office, An Analysis of the Administra-tion's Health Proposal (February 1994), and An Analysis of theManaged Competition Act (April 1994).


budget adjustments. Many states are actually con-sidering reducing taxes, especially for lower-incomefamilies. Taxes targeted for reduction include thesales tax, personal income tax, and corporate in-come tax in an effort to attract businesses. How-ever, taxes on cigarettes, motor fuels, and alcoholare still climbing.

Although that picture offers a rosier outlookthan states have had for the past several years, it isnot without potential pitfalls. Increasing costs ofgrants-in-aid programs such as Medicaid, federalmandates such as the Safe Drinking Water Act, andthe uncertain future of health care reform have thepotential to darken this outlook. In addition, themodest overall growth in state expenditures masksthe shifts that are occurring within state budgets.All major state functions except Medicaid and crim-inal justice declined as a percentage of state budgetsfrom 1987 to 1993. Although the rate of growth ofMedicaid spending slowed recently, the growth inMedicaid costs is still exceeding the rate of infla-tion. Spending on corrections is also growing, inpart because many states are under court orders torelieve overcrowding and improve prison conditions.

Figure 1-5.Net Exports Move Accordingto World and U.S. Output

Percentage Deviationfrom Trend

Percentage ofReal GDP

1070 1975 1980 1985 1990 1995

SOURCES: Congressional Budget Office; Department of Com-merce, Bureau of Economic Analysis; Federal Re-serve Board.

NOTE: U.S. real net exports rise when the rest of the world'sreal GDP rises relative to U.S. real GDP.

a. Relative output is the ratio of the rest of the world's real GDP,measured by a 28-country trade-weighted index, to real U.S.GDP. Data are shown with a nonlinear trend removed.

b. Real net exports as a percentage of real GDP.

Net Exports

The U.S. trade deficit is expected to widen through1994, but then begin to improve in early 1995.That change in trade performance reflects the differ-ent patterns of economic recovery in the majorindustrial countries. The U.S. economy, which wasone of the first to experience recession, recoveredearlier and is now growing more quickly than mostof its major trading partners. This faster growth inthe United States has produced correspondinglystrong growth in spending on imports by U.S. busi-nesses and consumers. Consequently, the growth inimports has exceeded that of exports, so it is notsurprising that the trade balance has deterioratedsince the U.S. recovery began (see Figure 1-5).

As the recoveries in the major trading partnersof the United States take hold in 1995 and the de-veloping countries continue to expand, the growthin U.S. exports is expected to outpace that of im-ports. In the CBO forecast, the world GDP index

of 28 countries, in which the GDPs are weighted byeach country's share of U.S. nonagricultural exports,is expected to grow more slowly than GDP in theUnited States in 1994, but to overtake it in 1995.

Two of the United States' major trading part-ners, Germany and Japan, have endured painfullyslow recoveries but are now showing signs of im-provement. In Germany, the first major drop inunemployment since late 1991 was posted in Juneof this year. Although the recent rise in long-terminterest rates, slow wage growth, and a restrictivefiscal policy are expected to weaken domestic de-mand in 1994, personal spending on consumption islikely to revive as the labor market stabilizes. As aresult, the saving rate could fall. According to asurvey of private forecasters, growth in output isexpected to register about 1.8 percent in 1994 (com-pared with a decline of 1.2 percent in 1993) and 2.6percent in 1995.4

4. See Consensus Economics, Inc., Consensus Forecasts (July 11,1994).


Although the Japanese economy remains weak,signs of recovery were evident in the first half of1994. In Japan's private sector, consumer spendinghas recently revived, and some gains in industrialproduction were reported in the first half of 1994.Excess capacity and falling prices, however, stillplague the Japanese economy. Consequently, ac-cording to the survey of forecasters, growth in out-put will slightly improve from 0.1 percent in 1993to 0.7 percent in 1994. In 1995, spending on con-sumption and an upturn in corporate investment areexpected to increase GDP growth to 1.9 percent.

Canada, Latin America, and the newly industri-alized countries of Asia, which together accountedfor approximately 50 percent of U.S. export demandlast year, should enjoy slightly faster growth in1994 and 1995, and strengthen export growth in theUnited States. According to the survey of forecast-ers, the Latin American countries are expected togrow at an average of 3.2 percent over the forecastperiod, while the newly industrialized countries ofHong Kong, South Korea, Singapore, and Taiwanwill continue to expand rapidly at nearly 7 percentover 1994 and 1995. At the same time Canada, theUnited States' major trading partner, is expected togrow at an average of 3.6 percent in 1994 and 1995.

and may not form a good basis for forecasts ofhousing. They also note that the decline in homeownership rates over the 1980s for many groupsmay imply a pent-up demand for nonrental units.

Interest Rates and Affordability

The rise in interest rates during the first half of thisyear will dampen residential investment, but thehigher rates are unlikely to reduce activity severely.Conventional mortgage interest rates have climbedfrom 7.1 percent in January to 8.6 percent for July,bringing rates back to mid-1992 levels. They arestill, however, below those of the 1980s, and aboutequal to the average for the mid-1970s (see Figure1-6). In addition, the availability of adjustable-ratemortgages provides home buyers with an opportu-nity to soften the effects of the increase in fixed-ratemortgages. The wide disparity between short-termadjustable-rate mortgages and long-term mortgageshas recently encouraged a resurgence of adjustable-rate financing.

Housing affordability declined recently, reflect-ing the increase in interest rates, but it still remainshigh (see Figure 1-7). The affordability index com-pares median family income with the income re-

A Neutral Factor:Residential Construction

Residential construction bolstered growth in demandover the last four quarters, but it is likely to waneduring the second half of this year and remain pallidduring 1995. The weak underlying demographics ofhousing demand during the 1990s, which tend toput a ceiling on housing starts, and the rise in mort-gage rates are the main reasons to expect slowergrowth. The CBO forecast assumes that housingstarts will be about the same on average over thenext year and a half as last quarter's rate of 1.44million units.

Although many analysts believe that the growthof residential construction will ebb, some forecastershave argued that residential investment will continueto contribute to GDP growth. These analysts arguethat demographic projections are highly uncertain

Figure 1-6.Mortgage Rates Have Turned Up

20

16

12

Percent

J_ _L1970 1975 1980 1985 1990 1995


NOTE: Contract interest rates on commitments for fixed-ratemortgages. Based on monthly data through July 1994.


Figure 1-7.Housing Is Still Affordable

160

140

120

100

80

Index

601970 1975 1980 1985 1990 1995

SOURCES: Congressional Budget Office; National Associationof Realtors.

NOTE: Monthly data are shown through June 1994. The indexequals 100 when median family income is just sufficientto qualify a family to purchase a median-priced home.

quired to purchase a home of median price. Af-fordability jumped sharply in the years since the1990 recession, and, although the recent drop is sig-nificant, affordability is still at high levels relativeto the last 15 years. Currently, according to theindex, the median family income is about 30 per-cent higher than necessary to qualify for the pur-chase of a typical home.

Expectations of future changes in housingprices, one of the important factors influencing thedecision to buy a home, are not reflected in the af-fordability index. If potential buyers anticipate rap-id appreciation in home values, they are more will-ing to buy, even when mortgage rates are high.This inclination was an important stimulant to de-mand in the late 1970s when housing prices wereappreciating rapidly. Since inflation in housingprices has been much more moderate recently, buy-ers are currently unlikely to see generous increasesin housing values. Therefore, even though afford-ability is better now than in the late 1970s, the fi-nancial incentives to buy a house may not be aspositive as the affordability index indicates.

Demographics

The underlying demographics imply general weak-ness in housing throughout the 1990s. Generallyslower growth in household formation and thechanges in the age distribution of the populationindicate that the demographic impetus to housingconstruction that was so strong in the 1970s and1980s will progressively weaken through this de-cade. The number of households, rather than popu-lation, is the main factor determining the need forhousing units. The Census Bureau reports that pop-ulation growth for the 1990s is likely to be aboutthe same as it was between 1965 and 1985, but nethousehold formation will probably be much slower.

The changing age distribution of householdsimplies that the recent weakness in rental marketsand first-time home buying will continue. Thenumber of households headed by a person age 25 to34 is declining, and this group constitutes the rentaland "starter" home market. Some of the weaknessin rental markets may be offset by the rapid growthin households headed by a person over 45 years old,but the net effect of these demographic shifts onrental markets is likely to be negative.

These demographic trends may, however, under-state future demand for housing. Unpredictablechanges in household formation, changes in employ-ment and income growth that affect the decision torent or buy, and a possible pent-up demand forowner-occupied housing may offset the generaldemographic pressures. The dip in home ownershiprates between 1980 and 1992 raises the possibilityof pent-up demand for owner-occupied housing.Although the decline in the overall home ownershiprate was not large-it fell from 65.6 percent in 1980to 64.1 percent in 1992-households in which thehead of household was 44 years old or younger(particularly those that have children) experiencedmuch sharper declines in home ownership rates,about 8 percentage points. The current relativelyhigh level of affordability of single-family housingmay permit home ownership rates for these groupsto increase.


The Risk of Inflation

Concerned about the unexpected pickup in growthin the U.S. economy and the low level of real short-term interest rates, the Federal Reserve increased thefederal funds rate by \1A percentage points in thefirst half of 1994. Policymakers at the FederalReserve noted that the rapid growth of late 1993and early 1994 and the speed at which excess ca-pacity in the economy was being eliminated in-creased the probability of higher inflation over thenext two years. Short-term interest rates needed torise, they argued, in order to curb future inflationarypressures.

Is the inflationary threat real? Although esti-mates of the highest level of output that can besustained without inducing higher inflation are un-certain, continued growth in excess of 2l/2 percent isindeed likely to exacerbate inflation eventually. Thepressures are not yet great, however, and inflationshould not pick up for a year or so, even if thepressures do become more intense.

The Economy Has LittleExcess Capacity

The major indicator of the economy's ability to ex-pand production rapidly without risking higher infla-tion is the unemployment rate. Right now, it indi-cates that the economy is at a high level of resourceuse (see Box 1-1). Therefore, in contrast to thethree years since the recession ended in early 1991,any further expansions of output cannot now rely onunderused resources. Instead, the rate of growth theeconomy can maintain at steady inflation rates islimited by the trends in the labor supply, capitalstock, and productivity.

The unemployment rate has dipped into a rangethat indicates a mildly inflationary situation. Onemeasure of labor market capacity-the nonaccelerat-ing inflation rate of unemployment (NAIRU)-isbased on the historical relationship between inflationand the unemployment rate (see Appendix B). Ifthe historical relationship continues to hold and theunemployment rate falls below the NAIRU, infla-

tionary pressures will slowly build; if unemploy-ment rises above the NAIRU, disinflationary forceswill prevail. How much the inflation rate changesdepends on the size and the duration of the differ-ence between actual unemployment and the NAIRU.Currently, the NAIRU is estimated to be about 6percent.

With the unemployment rate at 6.1 percent inJuly and the near-term outlook indicating (asidefrom the statistical uncertainties caused by the newemployment survey) that it will slip further, pres-sures on inflation are likely to increase. Economicexpansion along the lines of the CBO forecastwould presumably bring down the unemploymentrate to about 5% percent by the last half of 1995.The resulting increase in the underlying rate ofinflation is apt to be small—on the order of aboutone-quarter of a percentage point by the end of nextyear. CBO estimates that the current underlyingrate of inflation is between 2% percent and 3 per-cent, so given the forecast for unemployment, infla-tion would increase to a little over 3 percent by late1995.

The concept of the NAIRU has been useful forexplaining short-term movements in inflation in thepast, but it helps to clarify only the effects of excessdemand (see Box 1-2). Numerous events can haveshort-term effects on inflation, such as droughts, oilsupply shocks, changes in excise taxes, and importprices. These events could have a much largereffect on inflation during the forecast period thanchanges in excess demand. CBO has not, however,built any major supply shocks into its forecast forinflation through 1995.

The growth of wage rates and unit labor costshas not yet increased. Usually, however, increasesin wage rates and unit labor costs lag the tighteningof labor markets, so one should not be sanguineabout the lack of any acceleration to date. Gains inhourly labor compensation over the past four quar-ters have remained at 3 percent, and productivity,though slipping recently, grew 2.6 percent over thepast year. This combination of growth in wagesand productivity has kept the growth of unit laborcosts at 0.4 percent over this period. Inflationarypressures from labor costs, therefore, have not yetstarted to build.


The drop in the unemployment rate below theNAIRU suggests, however, that inflation in laborcosts will accelerate slightly during 1995. Wagegains are likely to pick up as labor markets tighten,and productivity growth is expected to fall off as itusually does in a late expansion. Thus, growth inunit labor costs will be higher than during the pastyear, though not so large as to cause a sharp in-crease in inflation.

Some analysts suggest that the standard NAIRUanalysis overstates the tightness of the labor market.Various other indicators of the degree of labor mar-ket tightness, such as the volume of help-wantedadvertising and the percentage of unemployed peo-ple who are not just on a temporary layoff, implymore slack than does the NAIRU analysis. Thehelp-wanted index has risen much less during thisexpansion than it did in previous expansions, and,

when adjusted for biases in the way the data arecollected, the help-wanted index appears quite low.

Although these conflicting signals about thelabor market introduce some uncertainty into theforecast, the relationship between inflation andunemployment has been closer in the recent pastthan between inflation and other measures of slackin the labor market. Thus, the CBO forecast antici-pates slightly higher inflation stemming from labormarket pressures.

Other ConsiderationsSignal Inflation

Other indicators of future inflation also imply thatinflationary pressures are building somewhat. Ca-pacity utilization and recent movements in some

Box 1-1.A New Measure of the Unemployment Rate

The Bureau of Labor Statistics (BLS) introducednew survey methods for the unemployment rate inJanuary 1994. These changes make it difficult tocompare the earlier data with that currently reported.Moreover, the change in the survey might havedistorted the picture of improvements in the labormarket over the first half of this year.

The unemployment rate is derived from theCurrent Population Survey (CPS). The Bureau ofLabor Statistics undertook a major redesign of theCPS in January 1994 to improve the accuracy andquality of the labor statistics drawn from this surveyof households. The redesigned survey obtains moreaccurate information about participation in the laborforce, most notably for women. The BLS also com-puterized the survey process, data collection, andtransmission to reduce errors.

Various background studies led the BLS toanticipate that the unemployment rate reported afterthe redesign would be higher than under the oldprocedures. The new survey was expected to resultin higher measured unemployment because it re-duced the likelihood that women who were lookingfor work would be misclassified as "keeping house."

Although most analysts thought that the unemploy-ment rate would rise by between 0.4 and 0.6 per-centage points with the shift to the new measure,many are less certain now that six months of dataare available. The new measure may be closer tothe old one than previously thought—only about one-quarter of a percentage point higher.

The new series may also not reflect recent trendsaccurately. The unemployment rate during the pe-riod of the new survey has fallen rapidly, from 6.7percent in January 1994 to 6.1 percent in July. Butthe measure reflects seasonal adjustments that try topurge typical seasonal changes in the unemploymentrate (such as a surge in hiring during Christmas) inorder to represent underlying trends better, and theseseasonal adjustments are based on the patterns ofpast years. The new survey has no history of sea-sonal patterns, however, so the seasonal adjustmentsof the old survey have been used to adjust the newdata. This new method may be causing an over- orunderstatement of the decline in the unemploymentrate since January. Therefore, the rapid drop in theunemployment rate this year may not be an accuratereflection of recent trends in the labor market.


producer prices are causes for concern, but fears ofrapid inflation stemming from the drop in the dollaror commodity price hikes appear to be overblown.

The reduction in excess capacity in manufactur-ing suggests that concerns about inflation are war-ranted, though inflation is unlikely to increase rap-idly. The capacity utilization rate for manufacturing

shows a high level of resource use, and some indus-trial prices have increased. The utilization rate iscurrently above 82 percent, a level that has beenassociated with an increase in goods inflation duringthe past 20 years. However, few industries havesignificantly raised prices to date. The overall pro-ducer price index for finished goods has risen onlyslightly over the last year, with only scattered in-

Box 1-2.Using the NAIRU to Forecast Inflation

The historical experience from 1987 to 1990 illus-trates the usefulness of the concept of the nonaccel-erating inflation rate of unemployment (NAIRU) inforecasting inflation. During that period, the NAIRUsignaled inflation, and inflation developed on cue.Even the magnitude of the increase in inflation wasclose to the expected amount.

By CBO's estimate, excess demand developednear the end of 1987—that is, the actual unemploy-ment rate fell below the estimated NAIRU level (6percent for that year, based on the old unemploy-

Higher Inflation FollowsLabor Market Tightening

Percentage Points Percent

-2

Inflation8

Labor Market Tightnessb

(Left scale)

\ -21984 1986 1988 1990 1992 1994

SOURCES: Congressional Budget Office; Department ofLabor, Bureau of Labor Statistics.

NOTE: Shading indicates the period when labor marketconditions are inflationary.

a. Consumer price index for all urban consumers (CPI-U),excluding food, energy, and used cars.

b. Labor market tightness is measured as the differencebetween CBO's estimate of the NAIRU and the actualunemployment rate.

ment survey). Aggregate demand continued to growfaster than the 2l/2 percent potential growth rate ofthe economy during 1988 and 1989. The unemploy-ment rate fell farther as a result, and ultimately wageand price inflation increased.

Yet inflation did not increase right away. Theunderlying rate of inflation (as measured by theconsumer price index excluding food and energy)remained virtually unchanged throughout the 1987-1989 period, even though the unemployment rateaveraged 5.3 percent from mid-1988 to the end of1989. It was not until late 1989, two years after theunemployment rate fell below the NAIRU, thatinflation picked up. Eventually, the underlying rateclimbed from 4.2 percent in 1987 to 5.4 percent inthe last half of 1990 (see Figure).

The increase in the underlying rate of inflationwas consistent with a commonly used rule of thumbfor forecasting the inflationary effect of maintainingan unemployment rate below the NAIRU. The ruleof thumb states that whenever the actual rate ofunemployment is 1 percentage point below theNAIRU for a year, the underlying rate of inflationwill increase by about one-half of a percentagepoint. That is, if the NAIRU is 6 percent and theactual unemployment rate remains at 5 percent for ayear, there will be one point-year of "excessive"employment for that year, and the underlying rate ofinflation would be about one-half of a percentagepoint higher than it would have been if the unem-ployment rate had been 6 percent.

According to the point-year rule, the underlyingrate of inflation should have increased about 1 per-centage point between late 1987 and the onset of therecession in mid-1990. The underlying inflation raterose by 1.2 percentage points, a close match.


creases in construction materials, motor vehicles,and some chemicals and metal products.

The recent increase in capacity constraints inmanufacturing may not have much effect on theconsumer price index (CPI), even if changes in theprices of goods did pick up. Changes in the CPIare not closely related to changes in the overallprices of goods, since the overlap between thegoods sector and the set of prices included in theCPI is relatively small. The price of goods accountsfor less than 44 percent of the items in the CPI,and, if food and energy—goods whose prices are notclosely related to capacity utilization-are excluded,goods account for only about 21 percent of the CPI.The prices of services in the CPI will ultimatelyaccelerate as excess resources are reduced, butprices for services will not increase rapidly. Infla-tion in services has fallen to about 3l/2 percent, andit will probably increase slowly, as it did in 1988and 1989 (see Figure 1-8).

Two indicators of potential inflation have beenwidely cited—the decline in the dollar and the in-crease in commodity prices. But both are unlikelyto be significant. Some analysts believe the drop inthe dollar against the German mark and the Japa-

Figure 1-8.The Prices of Services Are Slow to Accelerate

8Percentage Change from One Year Ago

j_ I I1984 1986 1988 1990 1992 1994

SOURCES: Congressional Budget Office; Department of Labor,Bureau of Labor Statistics.

NOTE: Growth year-over-year in the consumer price index forservices less energy services. Based on monthly datathrough July 1994.

Figure 1-9.Putting Recent Dollar Weakness in Perspective

160

140

120

100

80

Index: March 1973- 100

1970 1975 1980 1985 1990 1995SOURCES: Congressional Budget Office; Federal Reserve

Board.

NOTE: Trade-weighted index relative to the currencies of 10countries: Belgium, Canada, France, Germany, Italy,Japan, the Netherlands, Sweden, Switzerland, and theUnited Kingdom. Based on monthly data through July1994.

nese yen in the first half of this year will cause im-port prices and the prices of U.S.-produced goodsthat compete with imports to increase. Even thoughthe value of the dollar fell 7.8 percent between Jan-uary and July relative to the 10 major industrialcountries, the drop is unlikely to cause a significantincrease in the prices of imported goods. The re-cent weakness of the dollar is relatively small and,to date, only reverses an increase in the dollar thatoccurred during late 1992 and 1993 (see Figure1-9).

The increase in commodity prices during thefirst half of this year has also been mentioned as apossible indicator of an increase in inflation. Com-modity prices are not, however, a reliable signal ofhigher inflation. Commodity price indexes haverisen numerous times in the past for six-month pe-riods without any accompanying increase in CPI in-flation. In addition, commodity price indexes haveeased recently.

Prices of petroleum-one of the most importantcommodities-jumped between February and June,and they have increased only slightly since then;CBO does not anticipate that oil prices will experi-ence any further sharp increases. Although the rise


in petroleum prices will probably lead to a one-timeincrease in gasoline and fuel oil prices in late sum-mer, energy costs are unlikely to exacerbate infla-tion after that time.

Short-Term Interest RatesWill Increase Again

Early this year, signs of strong economic growthand the tightness of labor markets prompted theFederal Reserve to raise the interest rate on federalfunds. Its goal was to ease the nascent inflationarypressures and thereby increase the likelihood of aprolonged expansion. Although it became clear asthe year progressed that interest rates should rise—infact, they would have probably risen somewhatwithout Federal Reserve actions-how much highershort-term rates have to go to limit growth to a sus-tainable pace remains uncertain. The three-monthTreasury bill rate increased about 1.35 percentagepoints between January and July, and long-termrates jumped even more. Moreover, 10-year Trea-sury notes climbed from 5.8 percent to 7.3 percentduring the same period.

CBO anticipates that short-term rates will haveto increase further, by perhaps l*/2 to 2 percentagepoints before mid-1995, in order to slow economicgrowth to a sustainable pace (that is, GDP growthof less than 3 percent). Although real short-terminterest rates rose during the first half of this year,they are still quite low compared with similar pe-riods of late expansion. Moreover, real short-termrates are currently well below long-term rates, andare expected to rise given general pressures forhigher real rates in the mid-1990s.

Real Short-Term Interest Rates. Real short-terminterest rates were inordinately low during 1993,averaging 0.5 percent (see Figure 1-10).5 Such lowreal interest rates could not persist for extended pe-riods without ultimately overstimulating the econ-omy and risking higher inflation. By early August

Real interest rates cannot be measured directly, since they dependon expectations of inflation that are unknown. A commonly usedproxy, the nominal interest rate less actual (as opposed to ex-pected) inflation, is used here.

of this year, the real short-term rate had sprung upsharply, but will probably climb further given his-torical experience. The nominal three-month Trea-sury bill rate was 4.3 percent, and the real rate wasabout 1.2 percent in early August.

When the economy is in, or close to, a lateexpansion phase (that is, at or nearing full employ-ment), real short-term interest rates usually climbabove 2.5 percent. In two periods, 1973 to 1974and 1979 to 1980, the commonly used proxy forreal rates implies low real rates (see Figure 1-10).However, those periods were affected by major sup-ply shocks (primarily, sharp increases in oil prices),so that the increase in inflation was unexpected andtherefore the proxy for real rates was less useful.Although the behavior of real rates during thoseperiods is not clear, without a major inflationarysupply shock, real short-term rates on the order of2.5 percent are the historical norm.

Another pattern in previous expansions is thenarrowing of the difference between short- andlong-term interest rates late in the expansion, as the

Figure 1-10.Real Short-Term Interest Rates Are Still Low

Percent

-6 -

-101960

SOURCES:

1960 1970 1980 1990

Congressional Budget Office; Department of Com-merce, Bureau of Economic Analysis; Federal Re-serve Board.

NOTE: The real short-term interest rate is calculated by sub-tracting the growth of the consumer price index at anannual rate for all urban consumers over the subsequentthree-month period from the three-month Treasury billrate. Quarterly averages are shown.


Federal Reserve slows the growth of the moneysupply (see Figure 1-11). If this pattern holds upthrough this expansion, short-term rates will proba-bly have to rise further. The current spread betweenshort- and long-term rates is extremely large, partic-ularly given the stage of the business cycle. Nor-mally, short-term rates increase faster than long-term rates, and the difference between the two nar-rows as the expansion progresses.

The spread, however, has not narrowed thisyear. Long-term rates jumped by a surprisinglylarge amount early this year when short-term ratesstarted to increase, a response that appears to havestemmed from a reassessment of the risk of short-term rates increasing much more rapidly in thefuture than was previously thought. The perceptionof an increased risk of higher short-term rates, inturn, could have been affected by a reassessment ofthe risk of future inflation, particularly given thenews that the economy was much stronger thanforecast. CBO anticipates that long-term interestrates will ease over the next year and a half, asinflation remains relatively mild and growth is tem-pered during 1995, but the easing will be slow.Therefore, the narrowing of the spread is likely tooccur primarily as short-term rates rise.

Figure 1-11.The Interest Rate Spread UsuallyNarrows Late in Expansions

Percentage Points

1950 1960 1970 1980 1990


NOTE: The interest rate spread is the three-month Treasury billrate minus the yield on a 10-year Treasury note.

General Pressures for Higher Interest Rates. Anumber of trends indicate that real short-term inter-est rates will be greater than the current 1.2 per-cent—trends in the saving behavior of the govern-ment and private sector in the United States, therecovery of the industrialized countries of Europeand Japan, and the increasing capital demands ofnewly industrialized countries.

Real interest rates have varied widely over thepostwar period. The current real short-term interestrate is about the same as the rate that prevailed dur-ing the historical period from 1953 to 1967. Overthat period, large supply shocks were absent andinflation was quite low—averaging 1.5 percent.Consequently, this time frame can be used, tenta-tively, as a standard of comparison for the mid-1990s.

By that standard, real interest rates are likely torise from current levels. One factor pushing up realrates in the 1990s is the relatively low national sav-ing rate. Despite the progress in deficit reductionover the past four years, the projected federal bud-get deficit is a much larger share of GDP-2.5 per-cent-for the remainder of the 1990s than it was inthe 1953-1967 period, when the federal deficit aver-aged 1.1 percent of GDP. In addition, private sav-ing is much lower. Gross saving by businesses andhouseholds averaged 16.3 percent of GDP from1953 through 1967, but has recently averaged about12.6 percent of GDP. The continued high level ofgovernment borrowing, combined with low privatesaving, will put pressure on capital markets and onreal interest rates.

The growth of foreign economies is anotherreason to expect higher real interest rates. In themid-1990s, the European countries and Japan willhave recovered from their current economic difficul-ties. And CBO has long anticipated that worldcapital demands, driven by development needs inEastern Europe, the former Soviet Union, LatinAmerica, and the newly industrialized countries ofAsia, are likely to increase. The North AmericanFree Trade Agreement is also likely to encouragecapital investment in Mexico, although this demandwill be small in relation to the size of the worldcapital market.


Projections for the YearsBeyond 1995

CBO projects that real GDP will grow at an averageannual rate of 2.2 percent over the medium term-that is, between 1996 and 1999 (see Tables 1-3 and1-4). The projection for real growth assumes thatthe unemployment rate will average 6 percent dur-

ing that period (under the new definition of theunemployment rate). Inflation, measured by the an-nual rate of change in the CPI-U, is assumed toaverage 3.4 percent during that period. The three-month Treasury bill rate, which is forecast to in-crease between now and mid-1995, is expected todecline gradually thereafter. The 10-year Treasurynote rate averages 6.5 percent between 1996 and1999.

Table 1-3.Medium-Term Economic Projections for Calendar Years 1994 Through 1999

Nominal GDP(Billions of Dollars)

Nominal GDP(Percentage change)



Fixed-Weighted GDP PriceIndex (Percentage change)

CPI-U (Percentage change)8

1993

6,378

5.6

3.0

2.5

3.1

3.0

Forecast1994 1995

6,777 7,

6.3

4.0

2.2

2.7

2.6

161

5.7

3.0

2.5

3.0

3.1

1996

7,523

5.1

2.4

2.6

3.2

3.3

Projected1997 1998

7,893 8,

4.9

2.1

2.7

3.3

3.4

277

4.9

2.1

2.7

3.4

3.4

1999

8,687

5.0

2.2

2.7

3.4

3.4

Unemployment Rate(Percent)5


Ten-Year Treasury NoteRate (Percent)

Tax Bases (Percentage of GDP)Corporate profitsOther taxable incomeWage and salary

disbursements

Total

6.8

3.0

5.9

7.320.5

48.3

76.1

6.2

4.1

6.8

7.520.5

48.5

76.5

5.8

5.5

6.8

7.220.2

48.9

76.4

5.9

5.1

6.5

7.020.3

49.0

76.3

6.0

4.9

6.5

6.820.4

48.9

76.1

6.1

4.9

6.5

6.720.4

48.9

76.0

6.1

4.9

6.5

6.620.5

48.8

75.9


a. Consumer price index for all urban consumers (CPI-U).

b. The Bureau of Labor Statistics changed the unemployment survey in January 1994. The CBO projections reported in this table use thenew survey methodology. Data for 1993, shown in italics, use pre-1994 methodology.


CBO's medium-term projections do not reflectany attempt to estimate cyclical movements of theeconomy during the 1996-1999 period or the effectsof fiscal policy on the year-to-year changes in eco-nomic activity. Instead, CBO's projection for realgrowth is designed to approximate the level of eco-nomic activity on average, including the possibilityof above- or below-average growth. The projection

for potential GDP is based on an analysis of thefundamental factors underlying the economy, in-cluding growth of the labor force, the rate of na-tional saving, and growth of productivity. CBOprojects the path for real GDP by assuming that itwill grow smoothly to reach its average historicalrelationship with potential GDP by 1999 (see Figure1-12).

Table 1-4.Medium-Term Economic Projections for Fiscal Years 1994 Through 1999

Forecast

Nominal GDP(Billions of dollars)

Nominal GDP(Percentage change)



Fixed-Weighted GDP PriceIndex (Percentage change)

CPI-U (Percentage change)8

Unemployment Rate(Percent)6


Ten-Year Treasury NoteRate (Percent)

Tax Bases (Percentage of GDP)Corporate profitsOther taxable incomeWage and salary

disbursements

Total

1993

6,295

6.0

3.2

2.7

3.2

3.0

7.0

3.0

6.2

7.120.6

48.7

76.4

1994

6,677

6.1

3.9

2.1

2.6

2.6

6.3

3.7

6.4

7.620.5

48.4

76.5

1995

7,070

5.9

3.3

2.5

3.0

3.0

5.9

5.3

6.9

7.320.3

48.9

76.5

1996

7,431

5.1

2.5

2.6

3.1

3.2

5.8

5.2

6.6

7.020.3

49.0

76.4

Projected1997

7,800

5.0

2.2

2.7

3.3

3.4

6.0

4.9

6.5

6.920.4

48.9

76.2

1998

8,179

4.9

2.1

2.7

3.4

3.4

6.1

4.9

6.5

6.720.4

48.9

76.0

1999

8,581

4.9

2.2

2.7

3.4

3.4

6.1

4.9

6.5

6.620.5

48.8

75.9


a. Consumer price index for all urban consumers (CPI-U).

b. The Bureau of Labor Statistics changed the unemployment survey in January 1994. The CBO projections reported in this table use thenew survey methodology. Data for 1993, shown in italics, use pre-1994 methodology.


Growth

The projection for the years beyond 1995 is unusualin that the strong growth in the short-term forecastleaves the level of real GDP slightly above potentialoutput at the end of 1995. Therefore, real GDPmust grow more slowly than potential GDP in orderto restore the gap between the two to its historicalaverage. In the CBO projection, the growth of realGDP averages 2.2 percent between 1996 and 1999,slightly slower than the growth of potential GDP,which averages 2.4 percent during the same period.These rates of growth leave real GDP about 0.4percent below the level of potential GDP in 1999,roughly equal to the average historical gap betweenthese two variables.

CBO's projection for the growth of potentialoutput is little changed since last winter's report.However, several of the components that underliethe projection of potential output have been revised.In its winter report, CBO highlighted three sourcesof uncertainty about the projection for potentialoutput: the level of the NAIRU, slow growth of thelabor force, and rapid growth of total factor produc-tivity (TFP). Each has been revised since the winterforecast, but the revisions largely offset one another.

The Upward Revision to the NAIRU. CBO re-estimated the benchmark used to measure the stateof the business cycle, the nonaccelerating inflation

Figure 1-12.Potential GDP Governs the Projection of GDP

Billions of 1987 Dollars

1985 1990 1995


rate of unemployment (see Appendix B for a de-scription of how the NAIRU is estimated). Usingthe old definition of the unemployment rate, thevalue of the new estimate is about 5.8 percent in1993, or about 0.3 percentage points above the ear-lier estimate. The practical effect of this revision isto show that the economy was operating closer toits productive capacity in 1993 and early 1994 thanpreviously thought. For example, under the earlierestimate of the NAIRU, CBO estimated that the gapbetween output and potential output was about 2percent at the end of 1993. The new estimate, to-gether with the strong growth of real GDP in thefourth quarter of 1993, implies that the gap wasmuch smaller, perhaps 1.2 percent at the end of1993, and has been virtually eliminated during thefirst half of 1994. The higher estimate of theNAIRU also indicates that the trend growth of po-tential output is slightly slower than previouslythought.

CBO reestimated the NAIRU because of agrowing consensus that the economy is rapidly ap-proaching its productive capacity and could be indanger of overheating. The new estimate differsfrom the old for three reasons: first, it is calculatedusing a longer data sample; second, the data usedfor the calculation have been revised; and third, anew (and conceptually superior) measure of infla-tion is used for the calculation (see Appendix B).

The upward revision to the NAIRU is unrelatedto the revision to the Current Population Survey(CPS) undertaken by the Bureau of Labor Statisticsin early 1994 (see Box 1-1 for a discussion of theCPS revision). CBO estimates that the change inthe CPS adds another one-quarter of a percentagepoint to the NAIRU, starting in 1994. Therefore,on the new basis, the estimate of the NAIRU isabout 6 percent in early 1994. The adjustment forthe new survey technique, however, is subject to anextreme degree of uncertainty and will be monitoredin coming months to determine whether it is stilljustified.

Slower Growth of the Labor Force. As CBOpointed out in last winter's report, growth of thelabor force has been unusually slow since the lastbusiness cycle ended in 1990. This slowdownstems primarily from a decline in the growth of par-


ticipation rates in the labor force--the percentage ofthe working-age population that has or is activelyseeking a job—not in a slowing of populationgrowth. The major unresolved question is whetherthe tapering off of growth in the participation ofwomen in the labor force and the fall in the partici-pation rates of young people—both of which oc-curred about 1990—were short-run responses to acyclic slowdown in employment opportunities orwhether they represent a fundamental change in thepreferences toward work of some members of theworking-age population.

In the past, CBO's projections for growth in thelabor force embodied an assumption that the slow-down was exclusively a short-run phenomenon andthat participation rates would eventually regain theirearlier trend. However, the experience of the lastsix months has added weight to the argument thatparticipation rates will not rebound as much aspreviously thought. Currently, CBO projects thatthe labor force will grow at an average of 1.3 per-cent during the period from 1996 through 1999, arate that is about 0.1 percentage point below that as-sumed last winter.

The Rapid Growth of Total Factor Productivity.Although the years since the recession have beensomewhat bleak for employment growth, they havewitnessed impressive gains in productivity. Newlyreleased data indicate that total factor productivitygrew 2 percent in 1993, building on a hefty 2.7 per-cent gain in 1992.6 Both of these rates are abovethe trend rate observed since the early 1980s, andboth have raised CBO's estimate of trend growth inTFP. The improved outlook for productivity, com-bined with a stronger outlook for investment in cap-ital goods, has offset the downward pressure ongrowth of potential output caused by the higherNAIRU and slower growth of the labor force.

6. These estimates of total factor productivity were calculated byCBO. TFP is a measure of the productivity of both labor andcapital. A more comprehensive measure than labor productivity, itis defined as the growth in output above the growth of labor andcapital inputs.

Inflation

CBO projects that the gradual rise in the rate ofinflation during the short-term forecast will taper offin the years beyond 1995, and the growth of theCPI-U (consumer price index for all urban consum-ers) will average 3.4 percent a year during the medi-um term. The GDP deflator is projected to grow atan average rate of 2.7 percent between 1996 and1999. Both of these projections are higher than wasforeseen last winter because the economy is closerto potential than was previously thought.

Interest Rates

CBO projects that interest rates, both long- andshort-term, will decline slightly between 1996 and1999, though their ultimate levels are higher thanwhat was projected last winter. The projection forthe three-month Treasury bill rate drops from 5.6percent at the end of 1995 to 4.9 percent during1996 and thereafter. The 10-year Treasury note rateeases from 6.6 percent at the end of 1995 to 6.5percent during 1996 and remains steady at that ratethrough 1999.

Risks to CBO's EconomicForecast

The CBO economic forecast reflects a middle pathamong the uncertainties that affect any forecast.But it is certainly not the only possible outcome.Forecasters typically discuss two other views: atypical mature expansion through the end of 1995(which would carry a high risk of a recession some-time after the 1994-1995 forecast period), and a softlanding (in which the risk of recession soon after1995 is reduced by avoiding rapid growth in thenear term). Those are the relatively well-foreseenpossibilities. In addition, of course, the forecast issubject to risks that are not currently foreseen: inthe past, these unforeseen developments have ac-counted for a large part of forecast error.


A Typical Mature Expansion

The CBO forecast does not depict the typical behav-ior of an economy as its expansion matures andreaches a business cycle peak. Normally, when theeconomy approaches its potential, it continues togrow faster than its potential for some time, andactual GDP rises above potential GDP (see Figure1-13). That fast growth drives up inflation, short-term interest rates rise rapidly, and the economygoes into recession. Short-term interest rates usu-ally exceed long-term rates for a period of time-acondition that signals a considerable degree of mon-etary restraint. Every postwar expansion in theUnited States has ended with a sharp increase inshort-term interest rates (see Figure 1-1). The re-cord is similar in other countries. In the CBO fore-cast, however, the level of real GDP barely exceedsits potential and then quickly falls back. Althoughinflation and short-term interest rates increase, theyrise less than in preceding expansions.

If the economy did follow the typical end-of-expansion path, real GDP would be higher than inthe current CBO forecast. However, if such rapid

Figure 1-13.The GDP Gap: GDP TypicallyOvershoots Its Mark

Percentage of Potential GDP

-101950 1960 1970 1980 1990


NOTE: The GDP gap reported here represents GDP minus po-tential GDP expressed as a percentage of potential.Historically, expansions typically overshoot the mark sothat GDP comes to exceed potential GDP. The actionsof the Federal Reserve Board influence this outcome.

growth materialized, the Federal Reserve wouldprobably act more promptly to limit growth than ithas in the past, and as a result inflation need not besubstantially higher. In 1988 and 1989, and againthis year, the Federal Reserve demonstrated its will-ingness to raise rates to limit growth even beforesignificant inflation arose. In addition, the Chair-man of the Federal Reserve has many times empha-sized his determination to avoid letting inflationincrease again after enduring two recessions tolower it. Thus, the typical path would have some-what higher growth and significantly higher short-term interest rates than the CBO forecast. One ex-planation for the rise in long-term interest rates thatoccurred this spring would be that participants inthe financial markets put a lot of weight on the typi-cal path.

Such an outlook would raise the prospect of arecession to follow the end of the forecast period.Based on past experience, and the lags in operationof monetary policy, a recession would most likelyoccur beyond the end of CBO's short-term forecasthorizon-that is, in the 1996-1999 period when CBOdoes not attempt to predict cyclical movements inthe economy.

Why is this typical end-of-cycle scenario notCBO's forecast? First, the economy may not haveas much momentum as it did in previous businesscycles. This expansion has progressed at a moremoderate pace than average, stemming largely fromthe unusual degree of fiscal restraint and the need toabsorb structural problems such as overbuilding incommercial construction, changes in financial insti-tutions, and the restructuring of many large corpora-tions. Although these structural problems are lessof a problem now than during the last three years,fiscal policy continues to be much more restrictivethis year than is usually the case for periods whenthe economy approaches potential. Second, theFederal Reserve acted sooner than usual to restraininflationary pressures, and this action lessens thelikelihood that the economy will continue to growfaster than potential for a long period. Last, the twodeepest recessions of the postwar period were pre-ceded by major supply shocks-oil and other com-modity prices rose rapidly in 1973 and early 1974,and oil prices jumped again in 1979. Although asimilar supply shock is possible for the forecast pe-


riod, CBO does not consider it likely. Therefore,extremely rapid rises in inflation and interest rates,and the resulting deep recession, are less likely thansimple averages of past expansions and recessionsindicate.

A Soft Landing

Another view of the economy is that the growth ofdemand is already running out of steam. As aresult, growth will slow without much further actionby the Federal Reserve. Forecasters who maintainthis outlook point in particular to a slowdown infinal sales, especially the purchases of consumerdurables; the buildup of wholesale and retail inven-tories that occurred in the second quarter of 1994;and the continued weakness of net exports. Inaddition, they argue, the rise in long-term interestrates that occurred this spring will dramatically slowhousing starts and business fixed investment.

If this scenario developed, inflation would notincrease and the Federal Reserve would not need toraise short-term interest rates much further. Al-though such a soft landing indicates slightly slowergrowth in the near term, it is optimistic in the sensethat it would increase the likelihood that the onsetof the next recession would be later than that im-plied by the scenario for a mature expansion.

Why is this soft-landing scenario not in CBO'sforecast? The main reason is that CBO does notdetect any strong evidence that the factors drivingthe past year's growth-especially business invest-ment-have run their course. The evidence onwhich CBO's forecast is based was described earlierin this chapter. The economy is likely to continueto grow faster than its potential rate for a few morequarters. Further increases in rates combined withthe usual slowing of investment after a period ofrapid growth will be necessary to slow growth to itspotential.

Other Forecasts. The Administration and the JulyBlue Chip consensus forecasts call in large part forsoft landings. That is, the economy slows quicklyto a rate of growth close to that of potential byearly 1995 and, in real terms, short-term interest

rates do not rise significantly between now and1995 (see Table 1-5).

The current CBO forecast for real growth in1994 is higher by about 0.5 percentage points thanthe Administration or Blue Chip forecasts and some-what higher than the average of the upper and lowerbounds of the ranges anticipated by the FederalReserve. All forecasts, however, indicate GDPgrowth close to that of the economy's potential by1995. On the inflation front, the CBO, Administra-tion, and Blue Chip forecasts fall within the rangesanticipated by the Federal Reserve for 1994 and1995. Since February, the Federal Reserve hasrevised its inflation forecast for 1994 from about 3percent to the range of 2.75 percent to 3.0 percent,and expects inflation to increase moderately from1994 to 1995.7 In keeping with its outlook forstronger growth, the CBO outlook for civilian un-employment in 1995 is marginally lower than theother forecasts.

The forecasts differ most in their outlook forshort-term interest rates. The CBO forecast for thethree-month Treasury bill rate is almost 75 basispoints higher than that forecast by the Administra-tion and Blue Chip in 1995. Since all three fore-casts project approximately the same small increasein inflation of about 0.3 percent from 1994 to 1995,real interest rates rise by more in the CBO outlook.

Wider-Ranging Risks

Other risks to the forecast exist, but they are diffi-cult or impossible to evaluate. If it turns out thatpotential GDP is growing faster than CBO's currentestimate of 2.4 percent, the noninflationary growthof the economy could be substantially greater thanin the CBO forecast, both in the short run and overthe period to 1999. Although CBO believes rapidgrowth in productivity in the last three years is acyclical phenomenon and not a change in trend,some analysts feel trend productivity growth in the

7. Note that the Federal Reserve provided a forecast only for 1994 inits Monetary Policy Report to the Congress Pursuant to the FullEmployment and Balanced Growth Act of 1978 (February 1994).The range given-known as the central tendency-includes themajority of the forecasts of Federal Open Market Committeemembers and other Federal Reserve Bank presidents.


Table 1-5.Comparison of Congressional Budget Office, Administration, Federal Reserve,and Blue Chip Economic Forecasts

1993 1994Forecast

1995

Nominal GDPCBOAdministrationFederal Reserve8

Blue Chip

Real GDPb

CBOAdministrationFederal Reserve8

Blue Chip

Consumer Price Index0


Blue Chip

Civilian Unemployment Rated


Blue Chip


5.45.45.45.4

3.13.13.13.1

2.72.72.72.7

6.25.8

5.50 to 6.005.9

3.63.0

3.00 to 3.253.1

2.82.9

2.75 to 3.002.8

Average Level in the Fourth Quarter(Percent)

6.56.56.56.5


6.06.2

6.00 to 6.506.1

5.35.6

5.00 to 5.505.8

2.72.7

2.50 to 2.752.6

3.23.2

2.75 to 3.503.4

5.86.2

6.00 to 6.255.9

Three-Month Treasury Bill RateCBOAdministrationFederal Reserve8

Blue Chip

Ten-Year Treasury Note RateCBOAdministrationFederal Reserve8

Blue Chip9

3.03.03.03.0

5.95.95.95.9

4.14.0n.a.4.0

6.86.8n.a.6.8

5.54.7n.a.4.8

6.87.0n.a.7.2

SOURCES: Congressional Budget Office; Office of Management and Budget; Federal Reserve Board; Eggert Economic Enterprises, Inc.,Blue Chip Economic Indicators (July 10, 1994).

NOTE: n.a. = not applicable.

a. The Federal Reserve figures are the ranges-known as the central tendency-that include the majority of the forecasts of Federal OpenMarket Committee members and other Federal Reserve Bank presidents.

b. Based on constant 1987 dollars.

c. The consumer price index for all urban consumers (CPI-U).

d. The Bureau of Labor Statistics changed the unemployment survey in January 1994. The CBO forecast reported in this table uses thenew survey methodology. Data for 1993, shown in italics, use pre-1994 methodology.

e. Blue Chip does not project a 10-year note rate. The values shown here for the 10-year note rate are based on the Blue Chip projectionsof the Aaa bond rate, adjusted by CBO to reflect the estimated spread between Aaa bonds and 10-year Treasury notes.


1990s will indeed be higher. If so, the Federal Re-serve would have to recognize the potential forhigher growth, a difficult task. The main wayhigher potential growth would manifest itself isthrough lower inflation, but inflation lags behindgrowth by a substantial period. By contrast, thegoal of the Federal Reserve is to act in anticipationof inflation.

If other countries also expand at the same time,interest rates could be even higher than is typical atthe end of a business-cycle expansion. When busi-ness cycles are coordinated among countries, theycan have a powerful effect on world capital markets.In this case, the U.S. expansion, which started sev-eral years before the recoveries in Europe and Ja-pan, would have to be prolonged long enough forthe other economies to catch up.

Possibilities of a weaker outcome also exist.The slow growth of the labor force in the United

States could sharply reduce the growth of potentialoutput, implying that even moderate growth willcause the economy to be further above potentialthan CBO estimates. Increases in inflation wouldthen be larger and sooner than CBO anticipates, andthese increases would limit the Federal Reserve'sfreedom of movement. If that were to occur,sharply higher interest rates in 1995 and a greaterrisk of recession in 1996 could well be the result.

Finally, the role of political events cannot bediscounted. The Persian Gulf War complicatedpolicy management in 1990 and played an importantrole in the recession. Certainly, nothing suggeststhat the international scene is more stable now thanit was in 1990: the potential for destabilizing eventsexists in the Balkans, several countries of the formerSoviet Union, and Korea. How these events mightplay out in the world economy is impossible topredict.

Chapter Two

The Budget Outlook

A ccording to the latest projections of theCongressional Budget Office, the deficit inthe fiscal year ending on September 30,

1994, will barely top $200 billion. That will be thesmallest deficit chalked up in five years. Althoughthe intervening years witnessed several events thattemporarily ballooned the deficit—including a reces-sion and massive spending to resolve failed savingsand loan institutions-there is no denying the im-provement. CBO expects that the deficit in 1995will fall again in dollar terms (to $162 billion) andwill slip below 2l/2 percent of gross domestic prod-uct-the smallest deficit, in relation to GDP, in 16years. Much of the credit for the turnaround goesto the package of spending cuts and tax increasespassed a year ago, the Omnibus Budget Reconcilia-tion Act of 1993 (OBRA-93).

The year-to-year improvement in the deficitstalls after 1995. The deficit gets stuck at about 2Vipercent of GDP for several years and then startsclimbing at the end of the decade. In dollar terms,the deterioration starts sooner—indeed, immediatelyafter 1995.

The near-term outlook is somewhat sunnier thanin CBO's last report to the Congress. The deficit in1994 will likely be about $26 billion smaller thanCBO foresaw last winter; CBO has also trimmed$17 billion from its projected deficit for 1995.Although the revisions to revenues and spending areindividually modest-amounting to less than 2 per-cent of total revenues and outlays-they neverthelessjoin to reduce the deficit noticeably. They arealmost entirely attributable to new economic infor-mation and other developments; hardly any legisla-tion affecting the budget totals has been adopted sofar this year. These near-term revisions do not alterCBO's view of long-run budget trends. By the endof the decade, projected deficits are marginallybigger than those CBO published last winter.

The most striking aspect of CBO's semiannualbudget review is the absence of any big news.Seasoned budget-watchers are accustomed to signifi-cant revisions in the budget outlook every sixmonths or so, along with sharp disagreements be-tween the projections of CBO and the Administra-tion. Neither is present in this report. Indeed, inthe 12 months since the Congress passed OBRA-93,CBO has issued several new sets of projections, butnone has deviated very much from its predecessor(see Figure 2-1). All have conveyed the same un-derlying message: that OBRA-93 did not erase thedeficit but did arrest its upward march for aboutfive years. Nor are there significant differencesbetween the budget stories as told by CBO and bythe Administration in its midsession review, a com-parison that is presented later in this chapter.

Figure 2-1.Changes in CBO Deficit Projections, March 1993to August 1994 (By fiscal year)

Billions of Dollars400

300 -

200

100

March 1993(Pre-OBRA)

September 1993(Post-OBRA)

March 1994

1989 1991 1993 1995 1997 1999



The Deficit Outlook

At around $200 billion, the deficit in 1994 will bedown by more than $50 billion from last year'slevel and will be the smallest since 1989. Two con-ceited efforts at deficit reduction took place in themeantime, in 1990 and in 1993. For a few years inthe late 1980s, the deficit had seemed to settle atabout $150 billion but then shot up. To make mat-ters more awkward, statutory targets on the books atthe time-contained in the law popularly known asGramm-Rudman-Hollings—stipulated that the deficitwas supposed to be falling precipitously, not rising.The bleak outlook spurred a months-long budgetsummit between the Congress and the Bush Admin-istration, culminating in the Omnibus Budget Rec-onciliation Act of 1990. And when the budgetnumbers continued to worsen, despite the cutsadopted in 1990, a second big package (OBRA-93)followed less than three years later.

Both packages curbed spending, raised revenues,and downplayed fixed deficit targets, which hadproved to be elusive in practice, in favor of reformsin the budget process to ensure that the hard-wonsavings would not be undone. Both also achievedsignificant savings on the deficit front-about $500billion and $400 billion over their respective five-year horizons, according to past CBO estimates.1

Yet as was widely acknowledged when OBRA-93was enacted, the deficit is not about to fade away inthe absence of further policy changes.

Part of the deficit's jump in the early 1990sstemmed from temporary causes that were wellrecognized at the time. In fact, although economistsand financial market analysts use the deficit as arough-and-ready measure of the government's drainon national saving and on credit markets, they rou-tinely adjust it for ephemeral factors.

A prime example was the huge outlays fordeposit insurance in the 1989-1991 period as the

1. See Congressional Budget Office, The Economic and BudgetOutlook: Fiscal Years 1992-1996 (January 1991), Chapter 3, andThe Economic and Budget Outlook: An Update (September 1993),Chapter 2.

government moved to close or merge hundreds ofailing savings and loan institutions and, less promi-nently, commercial banks. Although such spendingswells the government's outlays, economists treat itas involving little or no fiscal stimulus. After all,most of the losses symbolized by deposit insuranceoutlays were incurred in the past-when institutionsmade bad loans and investments—even though thoselosses were invisible in the budget totals at the time.The government's subsequent actions to protect in-sured depositors, who had regarded their funds assafe, injected no extra stimulus into the economy.And some deposit insurance spending results in theacquisition of assets by the government that are soldlater, explaining why such spending swung belowzero beginning in 1993.

A second factor that is regularly segregatedwhen eyeing the deficit's trend is the effect of thebusiness cycle. When the economy is performingfar beneath its potential-as in a recession or theearly stages of a recovery-revenues falter, outlaysfor programs such as unemployment insurance rise,and the deficit is bigger than it would be otherwise.These effects are captured in the cyclical deficit.The deficit that remains, after subtracting the cycli-cal deficit and spending for deposit insurance, can-not be ascribed to a weak economy or to ephemeralfactors, and is labeled the standardized-employment,or structural, deficit. In 1991 and 1992, cyclicaleffects accounted for more than one-fourth of therecord-high deficits. But in today's buoyant econ-omy, as documented in Chapter 1, their contributionto the red ink is negligible.

A deficit of about $200 billion in 1994 marks adramatic improvement over the 1991-1993 periodtaken at face value. The improvement is less pro-nounced but still impressive when the deficit isconverted to a standardized-employment basis byadjusting it for the effects of deposit insurancespending and the business cycle (see Table 2-1 andSummary Figure 1). By either measure, the deficitbulged in the early 1990s and was reined in byOBRA-93.

Whatever measure is chosen, CBO's latest defi-cit projections tell the same story about the 1995-1999 period. The deficit falls from recent levelsthrough mid-decade, flattens, then begins to climb

CHAPTER TWO THE BUDGET OUTLOOK 29

again, picking up steam in 1999. This pattern is notan accident; pressures created by the growth inhealth care spending and federal interest costs andby the expiration of the stringent caps on discretion-ary spending in 1999 drive up spending faster thanrevenues. Indeed, CBO's broadbrush projections fora 10-year period suggest that such fundamentallong-run dynamics push the deficit up, not down(see Box 2-1).

Changes in the BudgetOutlook Since Last Winter

CBO has revised its budget projections relativelylittle since the beginning of the year. On balance,CBO has trimmed its projections of the deficit in1994 and 1995 but upped them in 1998 and 1999,with very small changes in the intervening years(see Table 2-2).

Table 2-1.CBO Baseline Deficit Projections (By fiscal year)

Total Deficit Assuming Discretionary Caps

Deficit Excluding Deposit Insurance

Standardized-Employment Deficit*

Actual1993

In Billions

255

283

221

1994

of Dollars

202

207

184

1995

162

180

183

1996

176

188

195

1997

193

198

200

1998

197

202

196

1999

231

235

223

On-Budget Deficit (Excluding SocialSecurity and Postal Service)

Memorandum:Deposit Insurance

Off-Budget Surplus

300

-28

258

-5

233

-17

254

-12


a. Excludes the cy^hal deficit and deposit insurance spending.

b. Less than $500 million.

c. Expressed as a percentage of potential GDP.

276

-5

285

-5

325

Social SecurityPostal Service

Total, Off-Budget Surplus

Hospital Insurance Surplus

Total Deficit Assuming Discretionary Caps


Standardized-Employment Deficit*0

47-1

45

4

As a Percentage

4.0

4.5

3.4

58-2

57

4

of GDP

3.0

3.1

2.7

71-1

71

7

2.3

2.5

2.6

771

78

6

2.4

2.5

2.6

83b

83

b

2.5

2.5

2.6

89-1

88

-5

2.4

2.5

2.4

94b

94

-14

2.7

2.7

2.6


Box 2-1.The Ten-Year Budget Outlook

The Congressional Budget Office's (CBO's) latestlook at trends in revenues and spending over thenext 10 years tells a familiar story—namely, that thefederal deficit is expected to head back up in the late1990s. The deficit is projected to double fromaround $200 billion in 1998 to nearly $400 billionby 2004 (see table at right). A more meaningfulfigure, the ratio of the deficit to gross domesticproduct (GDP), marches steadily upward from 2.4percent to 3.6 percent over the same period. Thistrend clearly indicates that the savings achieved in1990 and 1993 did not bring the deficit permanentlyunder control.

In deriving its 10-year projections, CBO doesnot assiduously estimate every component of revenueand spending as it does for the full-fledged baseline.Rather, it attempts to capture broad trends in reve-nues and spending based on reasonable assumptionsabout the path of the economy and other factors.Such projections aid policymakers in looking beyondthe usual five-year window that is mandated forbudget estimates and Congressional scorekeeping.

The current extrapolation points to a deficit of$397 billion, or 3.6 percent of GDP, in 2004. Thisestimate barely differs from CBO's last such esti-mate of $385 billion, or 3.5 percent of GDP, pre-sented last April in CBO's analysis of the Presi-dent's budgetary proposals.

As before, the long-run projections single outdiscretionary spending and health care programs asthe only broad areas likely to grow or shrink signif-icantly-in relation to GDP—over the next 10 years.The dollar caps on discretionary outlays push suchspending down from an estimated 8.2 percent ofGDP in 1994 to 6.7 percent in 1998. Although CBOassumes that discretionary spending will increasewith inflation beginning in 1999 (after the capsexpire), such outlays would still drift down as apercentage of GDP, reaching 6.1 percent by 2004.This continued, albeit slower, decline results simply

from the assumption that GDP will grow in realterms but discretionary spending will not.

CBO expects, however, that the two big healthcare programs, Medicare and Medicaid, will con-tinue to grow faster than GDP, as they have sincethey were created in the mid-1960s. Outlays forthese two programs combined are expected to climbfrom 3.6 percent of GDP in 1994 to 6.2 percent in2004. Through 1998, the drop in discretionaryspending (in relation to GDP) is enough to outweighthe steady rise in health care spending, but that is nolonger true once the caps expire. Thus, all otherthings being equal, the deficit would start to climbas a percentage of GDP. And other things areroughly equal: other major categories of spending, aswell as federal revenues, are expected to remainnearly constant as a share of GDP for the next de-cade. Social Security benefits, in particular, hoverjust under 5 percent of GDP through 2004—when theoldest members of the baby-boom generation arestill at least three years from eligibility for retire-ment benefits. As a consequence of the deficitoutlook, the debt-to-GDP ratio is basically flat (atabout 51 percent) through 1998 but then inches up,reaching 56 percent by 2004.

Any long-range projection inherently contains agreat deal of uncertainty. Over the 2000-2004 pe-riod, CBO assumes inflation (as measured by theconsumer price index) of 3.4 percent and real eco-nomic growth of 2.3 percent annually. Interest ratesare notoriously hard to forecast but are vitally im-portant to the budget outlook, given the growingfederal debt; CBO assumes a short-term interest rateof 4.9 percent on three-month Treasury bills and along-term interest rate of 6.5 percent on 10-yearTreasury notes. Of course, the economy will deviatefrom this path in ways that cannot be anticipated.Federal spending and revenues also may divergefrom their assumed paths for unpredictable reasons.But from today's vantage point, a prudent observershould expect the deficit to grow unless the tax andspending policies now in place are changed.


The Budget Outlook Through 2004 (By fiscal year)

1994 1995 1996

In Billions of

Revenues

OutlaysDiscretionaryMandatory

Social SecurityMedicareMedicaidCivil Service and Military RetirementOther

Subtotal

Deposit insuranceNet interestOffsetting receipts

Total

Deficit


Debt Held by the Public

1,265

545

3171588463

172794

-5202-68

1,467

202

207

3,440

1,363

546

3331779665

175847

-17226-77

1,525

162

180

3,611

1,433

550

35019510868

176898

-12245-72

1,609

176

188

3,801

1997

Dollars

1,492

547

36821612171

189965

-5253-75

1,684

193

198

4,011

1998

1,562

547

38823813574

1971,032

-5264-80

1,758

197

202

4,226

1999

1,632

566

40826315179

2051,107

-4277-83

1,863

231

235

4,476

2000

1,713

585

43029016883

2131,183

-2290-86

1,970

257

260

4,753

2001

1,799

605

45232018687

2201,265

-2307-90

2,086

287

288

5,059

2002

1,891

626

476354206

91227

1,354

-2325-94

2,209

319

321

5,397

2003

1,988

647

501391227

95235

1,451

-2346-98

2,343

355

357

5,771

2004

2,091

669

528434250100243

1,555

-2368

-103

2,488

397

399

6,188


Revenues

OutlaysDiscretionaryMandatory

Social SecurityMedicareMedicaidCivil Service and Military RetirementOther

Subtotal

Deposit insuranceNet interestOffsetting receipts

Total

Deficit




a. Less than 0.05 percent of GDP.

19.0

8.2

4.72.41.30.92.6

11.9

-0.13.0

-1.0

22.0

3.0

3.1

51.5

19.3

7.7

4.72.51.40.92.5

12.0

-0.23.2

-1.1

21.6

2.3

2.5

51.1

19.3

7.4

4.72.61.50.92.4

12.1

-0.23.3

-1.0

21.6

2.4

2.5

51.2

19.1

7.0

4.72.81.60.92.4

12.4

-0.13.2

-1.0

21.6

2.5

2.5

51.4

19.1

6.7

4.72.91.70.92.4

12.6

-0.13.2

-1.0

21.5

2.4

2.5

51.7

19.0

6.6

4.83.11.80.92.4

12.9

a3.2

-1.0

21.7

2.7

2.7

52.2

19.0

6.5

4.83.21.90.92.4

13.1

a3.2

-1.0

21.8

2.9

2.9

52.7

19.0

6.4

4.83.42.00.92.3

13.4

a3.2

-1.0

22.0

3.0

3.0

53.4

19.0

6.3

4.83.52.10.92.3

13.6

a3.3

-0.9

22.2

3.2

3.2

54.1

19.0

6.2

4.83.72.20.92.2

13.8

a3.3

-0.9

22.3

3.4

3.4

55.0

18.9

6.1

4.83.92.30.92.2

14.1

a3.3

-0.9

22.5

3.6

3.6

56.1


New legislation changes the outlook by lessthan $500 million in any year. Only two significantmeasures have been enacted. One made it easier forthe Federal Housing Administration to sell propertyit had obtained through foreclosure, resulting inlower holding costs over many years. Under the re-formed accounting rules adopted in 1990 for creditprograms, the budget recorded the expected savingsof about $400 million all at once when the legisla-tion was enacted.

The other, the Federal Workforce RestructuringAct of 1994, mandated reductions in federal civilianemployment of about 10 percent between now and1999 and permitted agencies to offer buyouts of upto $25,000 through March 1995 to induce employ-ees to leave. Because those changes would lead toadditional retirement benefits in the near term, theact also directed agencies temporarily to contributemore to the retirement funds. Only these narrowretirement-related effects of the legislation show up

Table 2-2.Changes in CBO Baseline Deficit Projections (By fiscal year, in billions of dollars)

Economic ChangesRevenuesOutlays

Net interestOther outlays

Subtotal, Outlays

Deficit

1994 1995 1996 1997

-9

3-12

-8

-20

15-312

-8

-20

17-314

-5

-12

13-211

-2


NOTE: Revenue increases are shown with a negative sign because they reduce the deficit.

a. Less than $500 million.

1998

-6

13-112

1999

Winter Baseline Deficit

Legislative Changes

228

a

180

a

180

a

192

a

187

a

213

a

-5

144

17

12

Technical ChangesRevenuesOutlays

Deposit insuranceStudent loansEarned income tax creditMedicaid and MedicareOffsetting receipts from agencies'

contributions to retirement fundsOther

Subtotal, Outlays

Deficit

Total Changes

Summer Baseline Deficit

-6

-1-21

-4

a-6

-12

-18

-26

202

-4

-5a20

1-3-5

-10

-17

162

-2

2120

1-34

2

-4

176

-1

1a20

1-14

3

1

193

1

aa20

2-13

4

10

197

3

aa20

2-13

6

18

231


on the official pay-as-you-go scorecard that theCongress uses to monitor legislation, and theyamount roughly to zero over the five-year period.The savings from the employment reductions willmaterialize in agencies' payroll accounts, which area significant part (approximately one-fifth) of thediscretionary spending cluster that is subject to thestatutory caps; that is, savings on the personnelfront were virtually inescapable under the belt-tight-ening already demanded by the caps.

A more important source of revision is changesin the economic outlook. Indeed, the underlyingpattern of changes in the deficit outlook-first down,then up—is explained by this source. As noted inChapter 1, the economy and, hence, revenues haveperked up faster than anticipated. But over time,revenues return nearly to the path that CBO previ-ously expected; after all, the economy cannot oper-ate above its potential indefinitely. At the sametime, less favorable trends take on added impor-tance. Chief among them is the outlook for interestrates. Partly as a consequence of the buoyant econ-omy, CBO now expects noticeably higher interestrates in the next two years than it did before,though by 1999 the increase tapers off to less thanhalf a percentage point. The federal debt held bythe public-now $3.4 trillion and expected to ap-proach $4.5 trillion by 1999-is big enough thateven such seemingly modest revisions in the as-sumptions about interest rates boost spending signif-icantly. And a similar twist affects the projectionsfor noninterest spending: reductions in outlays forunemployment insurance and similar programs oc-cur early but give way to the extra spending thatgradually results from higher inflation. In sum, re-visions attributable to the economic forecast shrinkthe deficit by $8 billion in both 1994 and 1995 butboost it by $12 billion in 1999.

Revisions that do not result from economicdevelopments or new legislation are dubbed techni-cal and, as this definition implies, can stem from avariety of reasons. On the revenue front, technicalrevisions are favorable for a few years but then fadeinto insignificance. Upward revisions to taxes oncorporate profits reflect the strong collections fromthis source (stronger even than suggested by data onprofits) and dominate the numbers at first, but theyare gradually overtaken by weaker trends in two

much bigger revenue sources, individual incometaxes and social insurance taxes.

On the spending side, deposit insurance~a vola-tile category of spending—is down by $5 billion in1995, chiefly from lower estimates of spending bythe Resolution Trust Corporation as it wraps up itsphase of the savings and loan cleanup before pass-ing the baton to the Savings Association InsuranceFund by July 1995. A downward revision in theoutlays for the government's student loan programis mostly a one-shot affair: the Student Loan Mar-keting Administration (nicknamed Sallie Mae) re-paid its entire $5 billion debt to the Treasury twoyears ahead of schedule. CBO had expected SallieMae to repay $2 billion of its debt in 1994 and theremaining $3 billion in 1996; Sallie Mae's decisionto pay the full amount this year leads to equal andoffsetting revisions in 1994 and 1996.

The earned income tax credit, a program aimedchiefly at supplementing the earnings of low-incomefamilies with children, is running $1 billion to $2billion a year higher than expected. Spending forMedicare and Medicaid in 1994 is down about $4billion from previous projections. Because the rea-sons for the slowdown are murky—and because $4billion represents less than 2 percent of those pro-grams' combined outlays—CBO has opted not torevise its 1995-1999 projections, which are servingas a base for estimating the budgetary effects of thecompeting health care proposals under considerationin the Congress. Estimated receipts of governmentretirement funds (such as Civil Service Retirementand Social Security) from federal agencies on behalfof their employees-an intrabudgetary collection-have been pruned by as much as $2 billion in 1999,partly because of the employment cuts that are nowincorporated in CBO's projections and partly toreflect other new information about coverage andcontribution rates.

Finally, other revisions are negligible in allyears except 1994, when they lead CBO to trim itsearlier projections by $6 billion. Lower-than-expected outlays in many departments this year-affecting programs as varied as international aid,space, justice, environmental spending, and creditprograms-are largely responsible. Such trends,though, never send clear-cut messages. Outlays can


emerge from the spending pipeline any time afteragencies commit or obligate the budget authoritythat they are granted in appropriation acts; sinceagencies do commit most of their allowable money,sluggish disbursements in one year presumably willbe offset in some future year. Furthermore, most ofthe programs that are spending less than anticipatedthis year are governed by the discretionary spendingcaps. The Congress decided in its budget resolutionadopted in May to spend almost the full amountallowable under the outlay caps in 1995-mimickingCBO's baseline assumption and hence obviating the

need for any revision in 1995 and beyond. Thatbudget resolution is guiding the 13 regular appropri-ation bills now wending their way through the Con-gress for the fiscal year beginning October 1.

CBO Baseline Projections

In 1994, federal revenues are expected to equal$1,265 billion and outlays $1,467 billion. Themajor components of these totals, and their pro-

Table 2-3.CBO Baseline Budget Projections, Assuming Compliancewith Discretionary Spending Caps (By fiscal year)

Actual1993 1994 1995 1996 1997 1998 1999


RevenuesIndividual incomeCorporate incomeSocial insuranceOther

TotalOn-budgetOff-budget

OutlaysDiscretionary*

DefenseInternationalDomesticUnspecified reductions

Subtotal

MandatoryDeposit insuranceNet interestOffsetting receipts


DeficitOn-budget deficitOff-budget surplus


510118428

98

1,154842312

29222

2280

542

762-28199-67

1,4081,142

267

25530045

3,247

549141463112

1,265929336

28020

2440

545

794-5

202-68

1,4671,188

279

20225857

3,440

600146498119

1,3631,003

360

27321

260-8

546

847-17226-77

1,5251,236

289

16223371

3,611

637149523124

1,4331,055

378

27721

267-15550

898-12245-72

1,6091,308

300

17625478

3,801

667153545127

1,4921,096

395

28322

273-31547

965-5

253-75

1,6841,372

312

193276

83

4,011

703157569131

1,5621,148

414

29122

282-48547

1,032-5

264-80

1,7581,432

326

197285

88

4,226

741162594135

1,6321,200

433

29823

290-46566

1,107-4

277-83

1,8631,525

339

23132594

4,476


a. Discretionary caps are set by law through fiscal year 1998. Discretionary outlays for 1999 are estimated as the 1998 amount adjusted forinflation. Estimates for particular subcategories-defense, international, and domestic-represent amounts that would be spent if 1994


jected growth over the next five years, are depictedin Table 2-3.

The Outlook for Revenues

Federal revenues are expected to equal 19.0 percentof GDP in 1994, rise to 19.3 percent in 1995 and1996, then slip back to 19.0 percent by 1999. Indi-vidual income taxes are the only source of revenuethat outpaces GDP, according to CBO's projections.From 8.2 percent of GDP this year, such taxes jump

to 8.5 percent in 1995 and essentially stay there-apattern largely traceable to the tax increases leviedby OBRA-93. That law approaches its full revenuepotential in fiscal year 1995. Its contributions infiscal year 1994 are smaller, for two reasons: higherwithholding from wages and salaries did not beginuntil January 1994; and on April 15, taxpayers wereallowed to pay only one-third of the extra liabilitiesOBRA imposed on high-income people for calendaryear 1993 (with the remaining two-thirds deferred tolater years). The government's other major sourcesof revenue-corporate income taxes, social insurance

Table 2-3.Continued

Actual1993 1994 1995 1996 1997 1998 1999

RevenuesIndividual incomeCorporate incomeSocial insuranceOther


OutlaysDiscretionary


8.11.96.81.6

18.313.45.0

8.22.16.91.7

19.013.95.0

8.52.17.01.7

19.314.25.1

8.62.07.01.7

19.314.25.1

8.52.07.01.6

19.114.15.1

8.61.97.01.6

19.114.05.1

8.61.96.91.6

19.014.05.0

DefenseInternationalDomesticUnspecified reductions

Subtotal

MandatoryDeposit insuranceNet interestOffsetting receipts


DeficitOn-budget deficitOff-budget surplus


4.60.33.6

08.6

12.1-0.43.2

-1.1

22.418.14.2

4.04.80.7

51.6

4.20.33.7

08.2

11.9-0.13.0

-1.0

22.017.84.2

3.03.90.8

51.5

3.90.33.7

-0.17.7

12.0-0.23.2

-1.1

21.617.54.1

2.33.31.0

51.1

3.70.33.6

-0.27.4

12.1-0.23.3

-1.0

21.617.64.0

2.43.41.0

51.2

3.60.33.5

-0.47.0

12.4-0.13.2

-1.0

21.617.64.0

2.53.51.1

51.4

3.60.33.4

-0.66.7

12.6-0.13.2

-1.0

21.517.54.0

2.43.51.1

51.7

3.50.33.4

-0.56.6

12.9b

3.2-1.0

21.717.83.9

2.73.81.1

52.2

appropriations were increased for inflation; unspecified reductions represent savings below those amounts required to satisfy the discre-tionary caps.

b. Less than 0.05 percent of GDP.


taxes, and other taxes such as excise taxes, estateand gift taxes, customs duties, and profits returnedby the Federal Reserve System—remain constant orfall slightly in relation to GDP.

The Outlook for Spending

The Budget Enforcement Act of 1990 (BEA) for-malized the use of several categories that CBO hadlong used to describe federal spending. Discretion-ary spending is funded anew each year through theappropriation process. Such spending encompassesnearly the entire budgets for defense and interna-tional affairs and a wide variety of domestic pro-grams—space and science, environmental protection,transportation, many education and social serviceprograms, veterans' medical care, and the operationof the Internal Revenue Service and the court sys-tem, to name a few.

Since fiscal year 1991, discretionary spendinghas been constrained by dollar caps. The BEA setcaps for 1991 through 1995, and OBRA-93 ex-tended their applicability through 1998. The capswill essentially freeze such spending at today's lev-els for four more years. Barring future emergen-cies, outlays will be held to the mid-$540 billionrange, and budget authority (the authority to commitfunds, the basic currency of the appropriation pro-cess) will be constrained accordingly. Within theselimits, policymakers must make trade-offs amongcompeting needs—defense, international, and domes-tic. Because the caps essentially hold discretionaryspending flat even as the economy grows, such out-lays shrink steadily in relation to the economy: from8.2 percent of GDP today-already far below thelevels of 10 percent to 12 percent that were typicalof the 1960-1989 period-they drift down to 6.6 per-cent in 1999.

All other categories of spending are controlledindirectly by the Congress. Mandatory spending isthe single biggest category of outlays, approaching$800 billion this year. It climbs steadily, outpacingthe growth of GDP, in CBO's projections. By1999, it amounts to 12.9 percent of GDP, up a per-centage point from today's levels. The familiar

benefit programs run by the government—led bySocial Security, Medicare, and Medicaid, a jointfederal/state program-dominate this huge categoryof spending (see Table 2-4). Lawmakers controlsuch spending indirectly, not by voting annual dol-lar amounts but by setting standards for eligibility,benefit formulas, and so forth. The big health careentitlements essentially explain why mandatoryspending, and eventually the deficit, are projected togrow in relation to GDP (see Box 2-1).

Deposit insurance outlays in CBO's newestprojections are subdued in comparison with thewildly fluctuating levels of recent years. Depositinsurance spending spiked to $58 billion in 1990and $66 billion in 1991, then registered negativeoutlays (that is, collections in excess of spending) of$28 billion in 1993. CBO expects that net outlayswill continue to be negative, but in shrinkingamounts after 1995, as proceeds from liquidationsbegin to dry up and as the government cuts theassessment that institutions must pay to the BankInsurance Fund on insured deposits.

Net interest lies outside policymakers' closecontrol; they influence such costs indirectly bymaking decisions that affect the size of the debt, butthey do not directly control the market interest ratesthat the Treasury must pay. Remarkably, net inter-est outlays in 1994 will be about $200 billion forthe fourth year in a row, even as the debt has grownby $1 trillion over the same period. This welcomestability-the result of refinancing large amounts ofmaturing debt at lower rates-will soon end. Netinterest is set to climb to $226 billion next year andto $277 billion in 1999, given CBO's outlook forinterest rates and continued federal borrowing.

The last category, offsetting receipts, is com-posed of various receipts and collections that arerecorded as negative outlays rather than as revenues.These receipts come either from the public (such asvoluntary Medicare premiums or licenses to useportions of the electromagnetic spectrum) or fromwithin the government (such as agencies' contribu-tions to retirement funds). They amount to a steady1 percent of GDP in CBO's projections.


Table 2-4.CBO Baseline Projections for Mandatory Spending, Excluding Deposit Insurance(By fiscal year, in billions of dollars)

Actual1993 1994 1995 1996 1997 1998 1999

Means-Tested Programs

MedicaidFood Stamps*Supplemental Security IncomeFamily SupportVeterans' PensionsChild NutritionEarned Income Tax CreditStudent LoansOther

Total, Means-Tested Programs

7625211647923

162

8426251737

1133

179

Non-Means-Tested

Social SecurityMedicare

Subtotal

Other Retirement and DisabilityFederal civilian"MilitaryOther

Subtotal

Unemployment Compensation

Other ProgramsVeterans' benefits0

Farm price supportsSocial servicesCredit reform liquidating accountsOther

Subtotal

Total, Non-Means-Tested Programs

All Mandatory Spending,Excluding Deposit Insurance

302143445

39264

69

35

171652

1151

600

762

317158475

4026

572

27

18116

-51241

615

Total

794

9626241837

1723

198

Programs

333177510

43275

75

22

1796

-11141

649

847

10827241838

2024

214

350195545

44295

78

24

1686

-31037

684

898

12128291938

2224

236

368216584

46305

82

25

1896

-41038

728

965

13529322039

2324

258

388238625

49325

86

27

1896

-51037

775

1,032

15131352039

2525

281

408263671

5135

591

28

1995

-68

36

826

1,107


NOTE: Spending for major benefit programs shown in this table includes benefits only. Outlays for administrative costs of most benefitprograms are classified as domestic discretionary spending; Medicare premium collections are classified as offsetting receipts.

a. Includes nutritir - assistance to Puerto Rico.

b. Includes Civil Service, Foreign Service, Coast Guard, and other retirement programs, and annuitants' health benefits.

c. Includes veterans' compensation, readjustment benefits, life insurance, and housing programs.


Table 2-5.Comparison of CBO Baseline with OMB Midsession Review (By fiscal year, in billions of dollars)

1994 1995 1996 1997 1998 1999

OMB Midsession Review Deficit* 220 167 179

SOURCES: Congressional Budget Office; Office of Management and Budget.

NOTE: Additional revenues are shown with a negative sign because they reduce the deficit.

a. Excluding the effects of the President's proposal for health care reform.

b. Less than $500 million.

190 192 207

DifferencesOutlays

Discretionary spendingDeposit insuranceMandatory spending and

offsetting receiptsNet interest

Total

Revenues

Deficit

CBO Baseline Deficit

-7-2

-2-2

-13

•6

-18

202

11

12

4

-9

-5

162

5-4

b3

4

-7

-3

176

2-2

3-2

1

2

3

193

-1-3

6-5

-3

7

5

197

12-1

6-5

11

12

24

231

A Comparison with theAdministration's Projections

On July 14, the Administration's Office of Manage-ment and Budget (OMB) issued its midsession re-view of the budget. Disagreements between theOMB and CBO projections are muted (see Table2-5). The only pronounced differences are in thefirst and last years of the projections, 1994 and1999.

In 1994, the Administration arguably did nottake full account of the sluggish spending that isdiscernible from the Monthly Treasury Statementand other data. CBO projects that the 1994 deficitwill be $202 billion, versus the Administration'sprojection of $220 billion; two-thirds of the differ-ence is attributable to CBO's lower projection ofspending and just one-third to greater revenues.

The two agencies' projections for the 1995-1999period are not strictly comparable. CBO's baselineestimates represent the implications of continuingcurrent taxing and spending policies, whereasOMB's projections incorporate the effects of a fewminor proposals that would alter current policies.The total effects of those proposals, however, arequite tiny, and most are concentrated in the discre-tionary spending area.2 In its baseline projections,CBO assumes that discretionary spending is equal tothe caps in 1995 through 1998; the Administration,in contrast, proposes spending that is slightly belowthe caps in 1996 and 1997. Furthermore, in 1999,

See Congressional Budget Office, An Analysis of the President'sBudgetary Proposals for Fiscal Year 1995 (April 1994), Table 1.One proposal that was identified as a departure from currentpolicy in that report-the effects of cutting federal employment onretirement contributions-is now substantially incorporated in theCBO baseline because of the enactment of limits on federal civil-ian employment in the Federal Workforce Restructuring Act of1994.


when the caps have expired, CBO returns to its tra-ditional baseline practice of estimating discretionaryspending at the previous year's level with an adjust-ment for inflation. OMB's projections, in contrast,incorporate hardly any rise in discretionary spendingin 1999, generating a $12 billion difference in thatfinal year.

Other differences between the two agencies'projections are similarly undramatic. By 1999,CBO expects $6 billion more in mandatory spend-ing than does OMB, chiefly in Medicaid, unemploy-ment insurance, and the earned income tax credit.As discussed in Chapter 1, CBO expects slightlyhigher rates on short-term Treasury bills than doesthe Administration, especially through 1995, but italso expects slightly lower rates on long-term Trea-sury notes and bonds for the entire five years ahead.Differences in the two agencies' projections of netinterest spending roughly track these contrastingassumptions about interest rates. Finally, CBO as-sumes a larger nominal GDP than does OMBthrough 1995 but chooses a more conservative as-sumption about the economy's long-run potential forgrowth thereafter. Largely as a consequence,CBO's revenues are smaller than the Administra-tion's in the last few years of the projection.

The Federal Sector of theNational Income and ProductAccounts

The projections summarized so far in this chapterdraw on the usual labels-revenues by source, out-lays by category—that are familiar to policymakers.Economists, though, often use another approach formeasuring the government's activities. The nationalincome and product accounts (NIPAs) divide thegovernment's spending and receipts into categoriesthat are conventionally used to analyze domesticproduction and income, helping economists to tracethe relationship between the government and othersectors of the economy.

Just a few major differences distinguish theNIPA versions of federal receipts and expendituresfrom their budget analogues. Netting and grossing

adjustments move some collections-most of whichare labeled offsetting receipts in the budget-fromthe spending to the receipts side of the NIPAs (seeTable 2-6). Most are recorded in the budget asnegative outlays because they are not deemed toresult from the government's taxing power. Shift-ing them to the receipts side of the NIPAs satisfiesthe desire of the accounts' users for a fuller pictureof government receipts regardless of source andclearly does not affect the total deficit. Such adjust-ments total $83 billion in 1994 and grow steadilythereafter.

Excluding lending and financial transactions, incontrast, does drive a wedge between the budgetand NIPA deficits. Macroeconomic analysis typi-cally disregards transactions that merely reflect thetransfer of existing assets and liabilities and there-fore do not contribute to current income or produc-tion. Prominent among such adjustments are thosefor deposit insurance outlays and direct loans madeby (or repaid to) the government. Other, relativelyminor factors that cause NIPA and budget totals todiverge are geographic adjustments (the exclusion ofPuerto Rico, the Virgin Islands, and a few otherareas from the domestic economic statistics) andtiming adjustments (such as the recording of corpo-rate taxes when accrued rather than paid, adjust-ments for irregular numbers of benefit checks orpaychecks because of calendar quirks, and so forth).

Tracing the relationship between the NIPA andthe budget data is complicated by the fact that theBureau of Economic Analysis regularly revises theNIPA data-sometimes by large amounts-but doesnot routinely publish a full-fledged bridge to thebudget totals. (Those totals, in contrast, seldomreceive more than negligible revisions.) Neverthe-less, when the dust finally settles, the NIPA deficitgenerally resembles the budget deficit excludingdeposit insurance—echoing CBO's frequent empha-sis on this measure in its regular budget reports.

The NIPA federal sector generally portraysreceipts according to their sources and expendituresaccording to their purpose and destination (seeTable 2-7). Receipts are split into four large cate-gories-personal tax and nontax receipts, corporateprofits tax accruals, indirect business tax and nontaxaccruals, and social insurance contributions—whoselabels summarize the nature of the collection and


Table 2-6.Relationship of the Budget to the Federal Sector of theNational Income and Product Accounts (By fiscal year, in billions of dollars)

Revenues (Budget basis)b

DifferencesNetting and grossing

Government contributionsfor employee retirement

Medicare premiumsDeposit insurance premiumsOther

Geographic exclusionsOther

Total

Receipts (NIPA basis)

Outlays (Budget basis)b

DifferencesNetting and grossing

Government contributionsfor employee retirement

Medicare premiumsDeposit insurance premiumsOther

Lending and financial transactionsDeposit insuranceOther

Defense timing adjustmentGeographic exclusionsOther

Total

Expenditures (NIPA basis)

Deficit (Budget basis)5

DifferencesLending and financial transactionsDefense timing adjustmentGeographic exclusionsOther

Total

Deficit (NIPA basis)

Actual1993a 1994

Receipts

1,153 1,265

55 5615 176 74 3

-2 -312 5

89 86

1,242 1,351

Expenditures

1,408 1,467

55 5615 176 74 3

23 -1•6 12 1

-8 -9-3 -16

88 59

1,497 1,526

Deficit

255 202

17 c2 1

-5 -6-14 -22

-1 -27

254 174

1995

1,363

582077

-3-2

86

1,449

1,525

582077

11-11

-9-4

89

1,614

162

101

-7-1

3

165

1996

1,433

6120

52

-33

87

1,520

1,609

612052

711

-10-2

83

1,692

176

81

-7-6

-4

172

1997

1,492

64223c

-32

89

1,580

1,684

64223c

221

-10-7

76

1,761

193

31

-7-9

12

181

1998

1,562

67263c

-3c

92

1,654

1,758

67263c

121

-11-7

83

1,841

197

41

-8-7

-10

187

1999

1,632

72273c

-3c

98

1,731

1,863

72273c

c21

-11-7

86

1,949

231

21

-8-7

-13

218

SOURCE: Congressional Budget Office.a. Estimate based on data from the Department of Commerce, Bureau of Economic Analysis, July 1994.b. Includes Social Security and the Postal Service.c. Less than $500 million.


Table 2-7.Projections of Baseline Receipts and Expenditures Measured by theNational Income and Product Accounts (By fiscal year, in billions of dollars)

Actual1993a 1994

Personal Tax and Nontax Receipts

Corporate Profits Tax Accruals

Indirect Business Tax and Nontax Accruals

Contributions for Social Insurance

Total

514

134

83

511

1,242

Receipts

558

157

91

545

1,351

1995

611

161

97

580

1,449

1996

650

164

97

610

1,520

1997

680

169

93

638

1,580

1998

717

174

96

667

1,654

1999

755

180

98

698

1,731

Expenditures

Purchases of Goods and ServicesDefenseNondefense

Subtotal

Transfer PaymentsDomesticForeign

Subtotal

Grants-in-Aid to State andLocal Governments

Net Interest

Subsidies Less Current Surplusof Government Enterprises

Required Reductions inDiscretionary Spending

307140447

63416

650

181

184

35

n.a.

292145437

66114

675

196

185

33

n.a.

292155447

70315

718

220

207

31

-11

297161458

74915

764

237

224

32

-23

304167471

79716

813

254

230

34

-41

313173486

84916

865

273

240

34

-58

324179502

90816

924

292

253

34

-57

Total

Deficit

1,497 1,526

Deficit

254 174

1,614 1,692

165 172

1,761

181


NOTE: n.a. = not applicable.

a. Estimate based on data from the Department of Commerce, Bureau of Economic Analysis, July 1994.

1,841 1,949

187 218


the identity of the payer. The term "nontax" is atip-off that NIPA receipts include some charges,such as fees and premiums, that are not generallytreated as revenues in the budget.

Federal spending can take the form of defenseand nondefense purchases (which enter directly intoGDP), transfers (most of which find their way intopersonal income and from there into consumption orsaving), grants to state and local governments(which may end up as state and local purchases ortransfers), net interest, and the subsidies less the

current surplus of government enterprises such asthe Postal Service and public housing authorities. Afinal category, required reductions in discretionaryspending, appears in Table 2-7 as a consequence ofthe discretionary spending caps that are mandatedby law. These caps will limit future spending forprograms funded through the appropriation process.Though no one can predict how particular programswill fare, the deepest effects of the required reduc-tions will almost certainly be felt in the NIPA cate-gories of defense and nondefense purchases andgrants.

Appendixes

Appendix A

Evaluating CBO's Recordof Economic Forecasts

S ince issuing its first forecast in 1976, theCongressional Budget Office (CBO) hascompiled a record for accurate economic

predictions that compares favorably with the trackrecords of five Administrations, as well as withconsensus forecasts of a sizable sample of private-sector economic forecasters. Although the margin isslight, CBO's forecasts have generally been closerthan the Administration's to the actual values ofseveral economic indicators that are important forprojecting the budget. Moreover, during the 11years for which comparisons are possible, CBO'sforecasts have been about as accurate as the averageof the 50 or so forecasts that make up the Blue Chipconsensus survey. Comparing CBO's forecasts withthat survey suggests that when CBO's economicpredictions missed the mark by a wide enoughmargin to contribute to sizable misestimates of thedeficit, those errors probably reflected limitationsthat confronted all forecasters.

These conclusions echo the findings of previousstudies published by the Congressional BudgetOffice and other government and academic review-ers. They emerge from an evaluation of the ac-curacy of short-term forecasts for four economicindicators: growth in real output, inflation in theconsumer price index (CPI), interest rates on three-month Treasury bills in both nominal and inflation-adjusted (real) terms, and interest rates on 10-yearTreasury notes and Aaa corporate bonds. In carry-ing out the evaluation, CBO compiled two-yearaverages of its f Beasts for the four indicators andcompared them with historical values, as well as

with the corresponding forecasts of the Administra-tion and the Blue Chip consensus.

Both CBO and the Administration have tendedto err toward optimism in their forecasts for a two-year horizon. In other words, the average forecasterror for real growth was an overestimate, and theaverage error for inflation was an underestimate.The Administration has also been more optimisticthan CBO in forecasting interest rates, with theaverage error being an underestimate. Overall, theaverage errors in the Administration's two-yearforecasts were slightly larger than in CBO's. More-over, an examination of longer-term projections ofgrowth in real output reaches similar conclusions:CBO's errors in projecting four-year average growthin real output were optimistic on average butsmaller than the Administration's. For the longer-term projections, both CBO and the Administrationrecorded larger errors on average than was the casefor their short-term forecasts. Finally, CBO's fore-casts appear to be about as accurate as the BlueChip consensus over the period for which compara-ble Blue Chip forecasts are available (1982-1992).

The differences among the three forecasts, how-ever, are not large enough to be statistically signifi-cant. The small number of forecasts available forthe analysis makes it difficult to distinguish mean-ingful differences in forecast performance fromdifferences that might arise randomly. Thus, thestatistics presented here are not reliable indicators ofthe future performance of any of the forecasters.


Sources of Data forthe Evaluation

Evaluating CBO's forecasting record requires com-piling the basic historical and forecast data forgrowth in real output, CPI inflation, and interestrates. Although each of these series has an impor-tant influence on budget projections, an accurateforecast of the two-year average growth in real out-put is the most critical economic factor in accuratelyestimating the deficit for the upcoming budget year.Two-year average forecasts published in early 1993and 1994 could not be included in the evaluationbecause historical values for 1994 and 1995 are, ofcourse, not yet available.1 The data were thereforecompiled using forecasts published early in theyears 1976 through 1992.

Selection of Historical Data

Which historical data to use for the evaluation wasdictated by the availability of actual data and thenature of the individual forecasts examined. Al-though CBO, the Administration, and Blue Chip allpublished the same measure for real output growth,selecting a historical series was difficult because ofperiodic benchmark revisions to the actual data.2

By comparison, not all of the forecasters publishedthe same measures for CPI inflation and interestrates, but the selection of historical data for theseseries was clear-cut.

Real Output Growth. Historical two-year averagesof growth in real output were developed from calen-dar year averages of the quarterly benchmark-years-weighted indices of real gross national product(GNP) and real gross domestic product (GDP) pub-lished by the Bureau of Economic Analysis (BEA).

1. The Clinton Administration adopted CBO's economic assumptionsas the basis for its budget in early 1993. As a result, once the1994 data are available, the errors for the early 1993 forecast willbe virtually the same for CBO and the Administration.

2. Before 1992, CBO, the Office of Management and Budget, andBlue Chip used gross national product to measure output. How-ever, beginning in early 1992, all three forecasters began to pub-lish forecasts and projections of gross domestic product instead.

The fact that several real GNP and GDP series werediscontinued because of periodic benchmark revi-sions meant that they were unsuitable historicalseries.

For example, during the 1976-1985 period, thethree forecasters published estimates for a measureof growth in real GNP that was based on 1972prices, the measure published by BEA at the time.In late 1985, however, BEA discontinued this 1972-dollar series and began to publish GNP on a 1982-dollar basis. As a result, an official series of valuesfor GNP growth in 1972 dollars is not available foryears after 1984; thus, actual two-year averagegrowth rates are not available to compare with theforecasts made in early 1984 and 1985. From 1986to 1991, forecasters published estimates of growthin real GNP based on 1982 prices. BEA revised thebenchmark again in the second half of 1991; itdiscontinued the 1982-dollar GNP and began topublish GNP on a 1987-dollar basis.3 Conse-quently, the historical annual series for 1982-dollarGNP is available only through 1990, and actualtwo-year average growth rates are not available forthe forecasts made in early 1990 and 1991.

By periodically updating the series to reflectmore recent prices, BEA's benchmark revisionsyield a measure of real output that is more relevantfor analyzing contemporary movements in realgrowth. But the process makes it difficult to evalu-ate forecasts of real growth produced over a periodof years for series that are subsequently discontin-ued. Recently, however, the difficulties presentedby periodic revisions of the data have been dimin-ished by the availability of new benchmark-years-weighted indices of real GNP and GDP. In 1992,BEA began regularly to publish and update thesealternative series for real growth.4

3. With the 1992 benchmark revision, GDP replaced GNP as thecentral measure of national output.

4. For details of the conceptual basis and empirical characteristics ofthis new series, see A.H. Young, "Alternative Measures of Changein Real Output and Prices," Survey of Current Business (April1992), pp. 32-48; I.E. Triplett, "Economic Theory and BEA'sAlternative Quantity and Price Indexes," Survey of Current Busi-ness (April 1992), pp. 49-52; and A.H. Young, "Alternative Mea-sures of Change in Real Output and Prices: Quarterly Estimatesfor 1959-92," Survey of Current Business (March 1993),pp. 31-41.

APPENDIX A EVALUATING CBO'S RECORD OF ECONOMIC FORECASTS 47

CPI Inflation. Two-year averages of inflation inthe consumer price index were calculated from cal-endar year averages of monthly data published bythe Bureau of Labor Statistics. Before 1978, thebureau published only one consumer price indexseries, known today as the CPI-W (the price indexfor urban wage earners and clerical workers). InJanuary 1978, however, it began to publish a sec-ond, broader consumer price index series, the CPI-U(the price index for all urban consumers). CBO'scomparison of forecasts used both series.

Until 1992, the Administration published itsforecasts for the CPI-W, the measure used to indexmost of the federal government's expenditures forentitlement programs. By contrast, for all but fourof its forecasts since 1979 (1986 through 1989),CBO based its inflation forecast on the CPI-U, amore widely cited measure of inflation and the onenow used to index federal income tax brackets. TheBlue Chip consensus has always published its fore-cast of the CPI-U. Although both the CPI-U andCPI-W may be forecast with the same relative ease,and annual fluctuations in the two series are virtu-ally indistinguishable, they do differ in some years;for that reason CBO used historical data for bothseries to evaluate the alternative forecast records.

Interest Rates. Two-year averages of nominal short-and long-term interest rates were developed from

calendar year averages of monthly data published bythe Board of Governors of the Federal Reserve Sys-tem.

The forecasts of short-term interest rates werecompared using historical values for two measuresof the interest rate on three-month Treasury bills:the new-issue rate and the secondary-market rate.The new-issue rate forecast by the Administrationcorresponds to the price of three-month bills auc-tioned by the Treasury Department-that is, it re-flects the interest actually paid on that debt. Thesecondary-market rate forecast by CBO, by contrast,corresponds to the price of the three-month billstraded outside the Treasury auctions. Because thesetransactions occur continually in markets that in-volve many more traders than do Treasury auctions,the secondary-market rate provides an updated eval-uation by the wider financial community of theshort-term federal debt. Blue Chip has alternated

between these two rates: publishing the new-issuerate from 1982 to 1985, switching to the secondary-market rate for the 1986-1991 period, and thenreturning to the new-issue rate in 1992. Clearly,there is no reason to expect the two rates to differpersistently; indeed, the differences between theircalendar year averages are minuscule.

The various forecasts of long-term interest rateswere likewise compared using historical values fortwo measures of long-term rates: the 10-year Trea-sury note rate and Moody's Aaa corporate bondrate. A comparison of forecasts is only possiblebeginning in 1984 because not all of the forecasterspublished projections of long-term interest ratesbefore that year. For forecasts made in early 1984and 1985, CBO projected the Aaa corporate bondrate. Beginning with its early 1986 forecast, how-ever, CBO switched to the 10-year Treasury noterate. The Administration has always published itsprojection for the 10-year Treasury note rate, butBlue Chip has published the Aaa corporate bondrate.

Separate historical values for real short-terminterest rates were calculated using the nominalshort-term interest rate and inflation rate appropriatefor each forecaster. In each case, the two-yearaverage nominal interest rate was discounted by thetwo-year average rate of inflation. The resultingreal short-term interest rates were very similar.Since there is no agreed-upon method for calculat-ing real long-term interest rates, they were not in-cluded in the evaluation.

Sources of Forecast Data

The evaluation used calendar year forecasts andprojections, which CBO has published early eachyear since 1976, timed to coincide with the publica-tion of the Administration's budget proposals. TheAdministration's forecasts were taken from theAdministration's budget in all but one case: theforecast made in early 1981 came from the ReaganAdministration's revisions to President Carter's lastbudget. The corresponding CBO forecast was takenfrom a projection published in its analysis of theReagan budget proposals. That forecast did not


include the economic effects of the new Administra-tion's fiscal policy proposals.

The average forecasts of the Blue Chip consen-sus survey were taken from those published in thesame month as CBO's forecasts. Because the BlueChip consensus did not begin publishing its two-year forecasts until the middle of 1981, the firstconsensus forecast available for use in this compari-son was published in early 1982.

whether individual forecasts are overestimates orunderestimates. The root mean square error—cal-culated by first squaring all the errors, then takingthe square root of the arithmetic average of thesquared errors-also shows the size of the errorwithout regard to sign, but it gives greater weight tolarger errors.

Measurement Issues

Measuring Bias and Accuracy

Following earlier studies of economic forecasts, thisevaluation of CBO's forecasts focused on two as-pects of forecast performance: statistical bias andaccuracy.

Bias

The statistical bias of a forecast is the extent towhich the forecast can be expected to differ fromwhat actually occurs. CBO's evaluation used themean error to measure statistical bias. That statis-tic—the arithmetic average of all the forecast er-rors—is the simplest and most widely used measureof forecast bias. Because the mean error is a simpleaverage, however, underestimates and overestimatesoffset each other in calculating it. As a result, themean error imperfectly measures the quality of aforecast—a small mean error would result either ifall the errors were small or if all the errors werelarge but the overestimates and underestimates hap-pened to balance out.

Accuracy

The accuracy of a forecast is the degree to whichforecast values are narrowly dispersed around actualoutcomes. Measures of accuracy more clearly re-flect the usual meaning of forecast performance thandoes the mean error. This evaluation used two mea-sures of accuracy. The mean absolute error~theaverage of the forecast errors without regard toarithmetic sign—indicates the average distance be-tween forecasts and actual values without regard to

These three statistics do not exhaust the availablesupply of measures of forecast performance. Forexample, to test for statistical bias in CBO's predic-tions, previous studies have used measures that areslightly more elaborate than the mean error. Thosestudies have generally concluded, as does this evalu-ation, that CBO's short-term economic forecasts donot contain a statistically significant bias.5

In addition, a number of methods have beendeveloped to evaluate a forecast's efficiency. Effi-ciency indicates the extent to which a particularforecast could have been improved by using addi-tional information that was at the forecaster's dis-posal when the forecast was made.6 To the extentthat the Blue Chip consensus forecasts represent awide variety of economic forecasters—reflecting abroader blend of sources and methods than can be

5. Another approach to testing a forecast for bias is based on linearregression analysis of actual and forecast values. For details ofthat method, see J. Mincer and V. Zarnowitz, "The Evaluation ofEconomic Forecasts," in J. Mincer, ed., Economic Forecasts andExpectations (New York: National Bureau of Economic Research,1969). That approach is not used here because of the small sam-ple size. However, previous studies that have used it to evaluatethe short-term forecasts of CBO and the Administration have notbeen able to reject the hypothesis that those forecasts are unbiased.See, for example, M.T. Belongia, "Are Economic Forecasts byGovernment Agencies Biased? Accurate?" Review, Federal Re-serve Bank of St. Louis, vol. 70, no. 6 (November/December1988), pp. 15-23.

6. For studies that have examined the relative efficiency of CBO'sforecasts, see Belongia, "Are Economic Forecasts by GovernmentAgencies Biased?"; and S.M. Miller, "Forecasting Federal BudgetDeficits: How Reliable Are U.S. Congressional Budget OfficeProjections?" Applied Economics, vol. 23 (December 1991), pp.1789-1799. Although both of the studies identify series that mighthave been used to make CBO's forecasts more accurate, they relyon statistics that assume a larger sample than is available. More-over, although statistical tests can identify sources of inefficiencyin a forecast after the fact, they generally do not indicate howsuch information can be used to improve forecasts when they aremade.


expected in any single forecaster—their use in thisevaluation can be interpreted as a proxy for an effi-cient forecast. The fact that CBO's forecasts areabout as accurate as Blue Chip's is a rough indica-tion of forecast efficiency.

More elaborate measures, however, are notnecessarily reliable indicators when the sample ofobservations is small, such as the 17 observationsthat make up the sample of CBO's two-year fore-casts. Small samples present three main types ofproblems for evaluating forecasts, including fore-casts based on the simple measures presented here.First, small samples reduce the reliability of statisti-cal tests that are based on the assumption that theunderlying population of forecast errors follows anormal distribution. The more elaborate tests offorecast performance all make such an assumptionabout the hypothetical ideal forecast with which theactual forecasts are compared. Second, in smallsamples, individual forecast errors have a relativelylarge weight in the calculation of summary mea-sures. The mean error, for example, can fluctuate inarithmetic sign when a single observation is addedto a small sample. Third, the small sample meansthat CBO's forecast history cannot be used in a sta-tistically reliable way to indicate either the directionor the size of future forecasting errors.

Apart from the general caution that should at-tend statistical conclusions based on small samples,there are several other reasons to view this evalua-tion of CBO's forecasts with particular caution.First, the procedures and purposes of CBO's and theAdministration's forecasts have changed over thepast 18 years and may change again in the future.For example, in the late 1970s, CBO characterizedits long-term projections as a goal for the economy,whereas it now considers its projections to be whatwill prevail on average if the economy continues toreflect historical trends. Second, an institution'sforecasting track record may not foretell its futureabilities because of changes in personnel or meth-ods. Finally, forecast errors increase when theeconomy is more volatile. All three forecastersmade exceptionally large errors when forecasting forperiods that included turning points in the businesscycle.

CBO's Forecasting Record

This analysis evaluated the Congressional BudgetOffice's forecasts over two-year and four-year pe-riods. The period of most interest for forecasters ofthe budget is two years. Because the Administra-tion's and CBO's winter budget publications focuson the budget projection for the fiscal year begin-ning in the following October, an economic forecastthat is accurate not only for the months leading upto the budget year but also for the budget year itselfwill provide the basis for a more accurate forecastof the deficit. A four-year horizon is used to exam-ine the accuracy of longer-term projections ofgrowth in real GNP.

Short-Term Forecasts

Historically, CBO's two-year forecasts are slightlymore accurate than the Administration's and sufferfrom slightly less statistical bias. In most cases,however, the differences are slim. Furthermore,CBO's forecasts are about as accurate as BlueChip's average forecasts.

An accurate forecast of two-year growth in realoutput is the most important factor in minimizingerrors in forecasting the deficit for the budget year.Accurate forecasts of nominal output, inflation, andnominal interest rates are less important for fore-casting deficits now than they were in the late1970s and early 1980s. The reason is that, givencurrent law and the level of the national debt, infla-tion increases both revenues and outlays by similaramounts. Revenues increase with inflation becausetaxes are levied on nominal incomes. Outlays in-crease because various entitlement programs areindexed to inflation and because nominal interestrates tend to increase with inflation, which in turnincreases the cost of servicing the federal debt.7

7. Rules of thumb for estimating the effect on the deficit of changesin various macroeconomic variables are given in CongressionalBudget Office, The Economic and Budget Outlook: Fiscal Years1994-1998 (January 1993), pp. 109-113.


Real Output Growth. For the two-year forecastsmade between 1976 and 1992, CBO had a slightlybetter record than the Administration in forecastingreal output growth (see Table A-l). On average,both CBO and the Administration tended to overes-timate growth of real output. For the 17 forecastsmade during the 1976-1992 period, the average er-rors were 0.4 percentage points for CBO and 0.5percentage points for the Administration. The rootmean square errors for this period were 1.0 percent-age point for CBO and 1.3 percentage points for theAdministration. CBO was closer to the true valuein nine of the 17 periods, the Administration wascloser in five periods, and the two forecasters hadidentical errors in three periods. In addition, CBO'sforecasts of two-year growth in real output made inthe 1982-1992 period were, on average, about asaccurate as the Blue Chip consensus.

Forecast errors tend to be larger when the econ-omy is more unstable. This tendency can be clearlyseen in the forecasts of real GNP growth by com-paring the large errors for 1979 through 1983-whenthe economy went through its most turbulent reces-sionary period of the postwar era-with the smallererrors recorded for later years. Similarly, the recentbusiness cycle accounts for the large errors in theforecasts made in 1989 through 1991; during thatperiod, CBO's errors were only slightly larger thanthose of the Blue Chip consensus.

CPI Inflation. The records for forecasting the av-erage annual growth in the consumer price indexover the two-year horizon were very similar (seeTable A-2). Both CBO and the Administration un-derestimated future inflation in their forecasts for1977 through 1980, and both tended to overestimateit in their forecasts for 1981 through 1986. Theaverage measures of bias and accuracy were virtu-ally the same for CBO and the Administration.CBO was closer to the true value in six of the 17periods, the Administration was closer in eight pe-riods, and the two forecasters had identical errors inthree periods. For the 1982-1992 forecasts, CBO'sinflation predictions appeared to be about as accu-rate as those of both the Administration and BlueChip.

Nominal Interest Rates. For the 1976-1992 fore-casts, CBO's record was slightly more accurate than

the Administration's for nominal short-term interestrates over the two-year horizon (see Table A-3).On average, the Administration tended to underesti-mate nominal short-term interest rates; CBO's meanerror was zero over this period. CBO was closer tothe true value in eight of the 17 periods, and theAdministration was closer in nine periods. How-ever, for the 1982-1992 period, the mean absoluteerror of CBO's forecasts was slightly above those ofthe Administration and Blue Chip.

For the 1984-1992 forecasts of long-term inter-est rates, CBO did significantly better than the Ad-ministration (see Table A-4). The Administrationtended to underestimate rates, and its mean errorwas larger than CBO's. In addition, the Adminis-tration's forecasts had a larger mean absolute errorand root mean square error. CBO was closer to thetrue value in six of the nine periods, and the Ad-ministration was closer in only three periods.

CBO's forecasts of long-term interest rates wereabout as accurate as those of the Blue Chip consen-sus. Both CBO and Blue Chip tended to overesti-mate long-term rates, each showing a mean error of0.3 percentage points.

Real Short-Term Interest Rates. For the forecastsmade in 1976 through 1992, CBO had a slight edgeover the Administration in estimating short-terminterest rates adjusted for inflation (see Table A-5).Again, the Administration was more likely thanCBO to underestimate interest rates, and its meanerror was greater. CBO and the Administrationrecorded similar mean absolute and root meansquare errors. CBO's forecasts were closer to theactual value in 10 of the 17 periods; the Administra-tion's were closer in seven. For forecasts madebetween 1982 and 1992, CBO's errors were gener-ally similar in both direction and magnitude to thoseof the Blue Chip consensus.

Longer-Term Projections

In forecasting real GNP growth for the more distantfuture, measured here as four years ahead, the Ad-ministration's errors were larger than CBO's. Al-though this comparative advantage for CBO doesnot directly affect the estimates of the deficit for the


budget year, accuracy in the longer term is obvi-ously important for budgetary planning over severalyears. Neither the Administration nor CBO, how-ever, considers its projections to be its best guessabout the year-to-year course of the economy. TheAdministration's projections each year are based onthe adoption of the President's budget as submitted,and in recent years CBO has considered its projec-tions an indication of the average future per-formance of the economy if major historical trendscontinue. Neither institution attempts to anticipatecyclical fluctuations in the projection period.

CBO's projections of medium-term growth inreal GNP for 1976 through 1990 were nearly alwayscloser to actual growth than were the Administra-tion's. The Administration's projections of theaverage annual rate of real GNP growth over fouryears showed an upward bias of 1.3 percentage

points, compared with an upward bias of 0.8 per-centage points for CBO (see Table A-6). Thosebiases resulted largely from the inability of theprojections made in early 1977 through 1980 toanticipate the recessions of 1980 and 1982.Through the subsequent years of expansion until themost recent recession, the upward bias was muchsmaller for the Administration's projections andsmaller yet for CBO's.

The size of the root mean square errors for theentire period for both CBO and, to a lesser extent,the Administration also results largely from errors inprojections made during the first five years. CBOhad a definite edge in the projections made in Janu-ary 1981 and 1982 and a lesser edge in later years.CBO's projections of four-year real GNP growthwere more accurate than the Administration's for 14of the 15 periods compared here.


Table A-1.Comparison of CBO, Administration, and Blue Chip Forecasts of Two-YearAverage Growth Rates for Real Output (By calendar year, errors in percentage points)

Actual

GNP1976-19771977-19781978-19791979-19801980-19811981-19821982-19831983-19841984-19851985-19861986-19871987-19881988-19891989-19901990-19911991-1992

GDPd

1992-1993

Statistics for1976-1992

Mean errorMean absolute

errorRoot mean

square error



errorRoot mean

square error

1972Dollars

6.75.23.91.31.10.20.75.2

bbbbbbbb

b

n.a.

n.a.

n.a.

n.a.

n.a.

n.a.

SOURCES: Congressional

1982Dollars

4.85.03.91.10.9

-0.30.55.25.13.03.13.93.51.7

C

C

c

n.a.

n.a.

n.a.

n.a.

n.a.

n.a.

Budget Office;

1987Dollars

4.84.73.81.10.5

-0.40.74.94.42.82.93.53.32.00.30.7

2.7

n.a.

n.a.

n.a.

n.a.

n.a.

n.a.

Office of

Benchmark-Years-

WeightedIndex

5.55.24.11.51.20.20.95.14.72.82.93.53.22.00.20.7

2.5

n.a.

n.a.

n.a.

n.a.

n.a.

n.a.

Management

CBOForecast

6.25.54.72.70.52.12.13.44.73.33.12.92.42.52.01.6

2.6

n.a.

n.a.

n.a.

n.a.

n.a.

n.a.

and Budget;

Error

0.70.30.61.2

-0.71.91.3

-1.70

0.50.3

-0.6-0.80.51.91.0

0.1

0.4

0.8

1.0

0.2

0.8

1.0

AdministrationForecast

5.95.14.72.90.52.62.72.64.73.93.73.33.03.22.81.4

2.2

n.a.

n.a.

n.a.

n.a.

n.a.

n.a.

Error

0.50

0.61.4

-0.72.41.8

-2.50

1.10.8

-0.2-0.21.22.60.7

-0.3

0.5

1.0

1.3

0.5

1.0

1.4

Eggert Economic Enterprises, Inc.

Blue ChipForecast

aaaaaa

2.03.54.33.23.02.82.12.21.91.2

2.3

n.a.

n.a.

n.a.

n.a.

n.a.

n.a.

, Blue Chip

Error

aaaaaa

1.2-1.6-0.40.30.1

-0.6-1.10.21.80.5

-0.2

n.a.

n.a.

n.a.

0

0.7

0.9

EconomicIndicator?, Department of Commerce, Bureau of Economic Analysis.

NOTES: Actual values are the two-year growth rates for real gross national product (GNP) and gross domestic product (GDP) last reportedby the Bureau of Economic Analysis, not the first reported values. The 1987-dollar series reflects the bureau's revisions publishedin July 1994. Revised estimates of the benchmark-years-weighted index, however, were not available at the time of publication.Forecast values are for the average annual growth of real GNP or GDP over the two-year period. The forecasts were issued in thefirst quarter of the initial year of the period or in December of the preceding year. Errors are forecast values minus actual values;thus, a positive error is an overestimate. The benchmark-years-weighted index of actual GNP or GDP was used in calculating theerrors.

n.a. = not applicable.

a. Two-year forecasts for the Blue Chip consensus were not available until 1982.

b. Data for 1972-dollar GNP and GDP are available only through the third quarter of 1985.

c. Data for 1982-doliar GNP and GDP are available only through the third quarter of 1991.

d. With the 1992 benchmark revision, GDP replaced GNP as the central measure of national output.


Table A-2.Comparison of CBO, Administration, and Blue Chip Forecasts of Two-Year AverageInflation Rates in the Consumer Price Index (By calendar year, errors in percentage points)

Actual

1976-19771977-19781978-19791979-19801980-19811981-19821982-19831983-19841984-19851985-19861986-19871987-19881988-19891989-19901990-19911991-19921992-1993



errorRoot mean

square error



errorRoot mean

square error

CPI-U

6.17.09.4

12.411.98.24.63.83.92.72.83.94.45.14.83.63.0

n.a.

n.a.

n.a.

n.a.

n.a.

n.a.

CPI-W

6.17.09.5

12.511.98.14.53.33.52.52.63.84.45.04.63.52.9

n.a.

n.a.

n.a.

n.a.

n.a.

n.a.

CBOForecast

7.14.95.88.1

10.110.47.24.74.94.13.83.94.74.94.14.23.4

n.a.

n.a.

n.a.

n.a.

n.a.

n.a.

Error

1.0-2.1-3.7-4.3-1.82.12.61.01.01.41.20.10.3

-0.1-0.70.60.5

-0.1

1.4

1.8

0.7

0.9

1.1


6.15.26.07.4

10.59.76.64.74.54.23.83.34.23.73.94.63.1

n.a.

n.a.

n.a.

n.a.

n.a.

n.a.

Error

0-1.8-3.5-5.0-1.41.62.11.51.01.71.2

-0.5-0.2-1.3-0.71.10.2

-0.2

1.5

1.9

0.6

1.0

1.2

Blue ChipForecast

aaaaaa

7.24.95.24.33.83.64.34.74.14.43.5

n.a.

n.a.

n.a.

n.a.

n.a.

n.a.

Error

aaaaaa

2.61.11.31.61.0

-0.2-0.1-0.4-0.70.80.5

n.a.

n.a.

n.a.

0.7

0.9

1.2

SOURCES: Congressional Budget Office; Office of Management and Budget; Eggert Economic Enterprises, Inc., Blue Chip EconomicIndicators', Department of Labor, Bureau of Labor Statistics.

NOTES: Values are for the average annual growth of the consumer price index (CPI) over the two-year period. Before 1978 the Bureau ofLabor Statistics published only one consumer price index series, known today as the CPI-W (the price index of wage earners andclerical workers). In January 1978, however, the bureau began to publish a second, broader consumer price index series, the CPI-U(the price index for all urban consumers). For most years since 1979, CBO forecast the CPI-U; from 1986 through 1989, CBOforecast the CPI-W. The Administration forecast the CPI-W until 1992, when it switched to the CPI-U. Blue Chip forecast the CPI-Ufor the entire period. The forecasts were issued in the first quarter of the initial year of the period or in December of the precedingyear. Errors are forecast values minus actual values; thus, a positive error is an overestimate.

n.a. = not applicable,



Table A-3.Comparison of CBO, Administration, and Blue Chip Forecasts of Two-Year AverageInterest Rates on Three-Month Treasury Bills (By calendar year, errors in percentage points)

Actual

1976-19771977-19781978-19791979-19801980-19811981-19821982-19831983-19841984-19851985-19861986-19871987-19881988-19891989-19901990-19911991-19921992-1993



errorRoot mean

square error



errorRoot mean

square error

NewIssue

5.16.28.6

10.812.812.49.79.18.56.75.96.27.47.86.54.43.2

n.a.

n.a.

n.a.

n.a.

n.a.

n.a.

SecondaryMarket

5.16.28.6

10.712.712.39.69.18.56.75.96.27.47.86.44.43.2

n.a.

n.a.

n.a.

n.a.

n.a.

n.a.

CBOForecast

6.26.46.08.39.5

13.212.67.18.78.56.75.66.47.57.06.84.7

n.a.

n.a.

n.a.

n.a.

n.a.

n.a.

Error

1.10.2

-2.6-2.4-3.20.93.0

-2.00.31.80.9

-0.6-0.9-0.30.62.41.5

0

1.5

1.7

0.6

1.3

1.6


5.54.46.18.29.7

10.011.17.98.18.06.95.55.25.96.06.24.5

n.a.

n.a.

n.a.

n.a.

n.a.

n.a.

Error

0.4-1.8-2.5-2.6-3.1-2.41.4

-1.1-0.41.31.0

-0.7-2.1-1.9-0.41-81.3

-0.7

1.6

1.7

0

1.2

1.3

Blue ChipForecast

aaaaaa

11.37.99.18.57.15.76.17.57.16.44.6

n.a.

n.a.

n.a.

n.a.

n.a.

n.a.

Error

aaaaaa

1.6-1.20.51.81.2

-0.5-1.2-0.30.72.01.4

n.a.

n.a.

n.a.

0.6

1.1

1.3

SOURCES: Congressional Budget Office; Office of Management and Budget; Eggert Economic Enterprises, Inc., Blue Chip EconomicIndicators] Federal Reserve Board.

NOTES: Values are for the geometric averages of the three-month Treasury bill rates for the two-year period. The actual values are pub-lished by the Federal Reserve Board as the rate on new issues, reported on a bank-discount basis, and the secondary-market rate.CBO forecast the secondary-market rate; the Administration forecast the new-issue rate. The Blue Chip alternated between the tworates, forecasting the new-issue rate from 1982 to 1985, the secondary-market rate from 1986 to 1991, and the new-issue rate againbeginning in 1992. The forecasts were issued in the first quarter of the initial year of the period or in December of the precedingyear. Errors are forecast values minus actual values; thus, a positive error is an overestimate.

n.a. = not applicable,



Table A-4.Comparison of CBO, Administration, and Blue Chip Forecasts of Two-Year AverageLong-Term Interest Rates (By calendar year, errors in percentage points)

Actual

1984-19851985-19861986-19871987-19881988-19891989-19901990-19911991-19921992-1993



errorRoot mean

square error

10-YearNote

11.59.18.08.68.78.58.27.46.4

n.a.

n.a.

n.a.

CorporateAaa Bond

12.010.29.29.59.59.39.08.57.7

n.a.

n.a.

n.a.

CBOForecast

11.911.58.97.29.49.17.77.87.1

n.a.

n.a.

n.a.

Error

-0.11.30.9

-1.40.70.6

-0.50.40.7

0.3

0.7

0.8


9.710.68.76.67.77.77.27.36.9

n.a.

n.a.

n.a.

Error

-1.81.50.7

-2.0-1.0-0.8-1.0-0.10.5

-0.4

1.0

1.2

Blue ChipForecast

12.211.89.98.79.89.58.78.78.4

n.a.

n.a.

n.a.

Error

0.21.70.8

-0.80.30.3

-0.30.30.7

0.3

0.6

0.7

SOURCES: Congressional Budget Office; Office of Management and Budget; Eggert Economic Enterprises, Inc., Blue Chip EconomicIndicators; Federal Reserve Board.

NOTES: Actual values are for the geometric averages of the 10-year Treasury note rates or Moody's corporate Aaa bond rates for the two-year period as reported by the Federal Reserve Board. CBO forecast the 10-year Treasury note rate in ail years except 1984 and1985. The Administration forecast the 10-year note rate, but Blue Chip forecast the corporate Aaa bond rate. Data are onlyavailable beginning in 1984 since not all of the forecasters published long-term rate projections before then. The forecasts wereissued in the first quarter of the initial year of the period or in December of the preceding year. Errors are forecast values minusactual values; thus, a positive error is an overestimate.



Table A-5.Comparison of CBO, Administration, and Blue Chip Forecasts of Two-Year Average InterestRates on Three-Month Treasury Bills Adjusted for Inflation (By calendar year, errors in percentage points)

ActualNewIssue

1976-19771977-19781978-19791979-19801980-19811981-19821982-19831983-19841984-19851985-19861986-19871987-19881988-19891989-19901990-19911991-19921992-1993



errorRoot mean

square error



errorRoot mean

square error

CPI-U

-0.9-0.8-0.7-1.40.83.84.85.14.43.93.02.32.82.61.60.80.2

n.a.

n.a.

n.a.

n.a.

n.a.

n.a.

CPI-W

-0.9-0.7-0.8-1.50.94.04.95.74.94.13.22.42.92.61.70.90.4

n.a.

n.a.

n.a.

n.a.

n.a.

n.a.

SecondaryMarket

CPI-U

-0.9-0.8-0.7-1.40.73.74.75.14.43.93.02.32.82.61.50.70.2

n.a.

n.a.

n.a.

n.a.

n.a.

n.a.

CPI-W

-0.9-0.7-0.8-1.50.83.94.95.64.84.13.22.32.92.61.70.90.3

n.a.

n.a.

n.a.

n.a.

n.a.

n.a.

CBOForecast

-0.81.50.20.2

-0.52.65.02.23.64.22.81.71.72.52.82.51.3

n.a.

n.a.

n.a.

n.a.

n.a.

n.a.

Error

0.12.21.01.7

-1.2-1.20.3

-2.9-0.80.3

-0.4-0.6-1.2-0.21.31.81.0

0.1

1.1

1.3

-0.1

1.0

1.2


-0.6-0.80.10.7

-0.70.34.23.13.43.63.02.11.02.12.01.51.3

n.a.

n.a.

n.a.

n.a.

n.a.

n.a.

Error

0.3-0.10.92.2

-1.6-3.7-0.8-2.6-1.4-0.4-0.3-0.2-1.9-0.60.30.61.1

-0.5

1.1

1.5

-0.6

0.9

1.2

Blue ChipForecast

aaaaaa

3.82.93.64.03.22.01.82.72.91.91.1

n.a.

n.a.

n.a.

n.a.

n.a.

n.a.

Error

aaaaaa

-1.0-2.3-0.80.10.2

-0.3-1.10.21.41.20.8

n.a.

n.a.

n.a.

-0.1

0.8

1.0

SOURCES: Congressional Budget Office; Office of Management and Budget; Eggert Economic Enterprises, Inc., Blue Chip EconomicIndicators', Department of Labor, Bureau of Labor Statistics; Federal Reserve Board.

NOTES: Values are for the appropriate three-month Treasury bill rate discounted by the respective forecast for inflation as measured by thechange in the consumer price index. The forecasts were issued in the first quarter of the initial year of the period or in December ofthe preceding year. Errors are forecast values minus actual values; thus, a positive error is an overestimate.

CPI-U = consumer price index for all urban consumers; CPI-W = consumer price index for wage earners and clerical workers;n.a. = not applicable.



Table A-6.Comparison of CBO and Administration Forecasts of Four-Year Average GrowthRates for Real Output (By calendar year, errors in percentage points)

Actual

1976-19791977-19801978-19811979-19821980-19831981-19841982-19851983-19861984-19871985-19881986-19891987-19901988-19911989-19921990-1993



errorRoot mean

square error

1972Dollars

5.33.22.50.70.92.7

aaaaaaaaa

n.a.

n.a.

n.a.

1982Dollars

4.33.02.40.40.72.42.74.14.13.53.32.8

bbb

n.a.

n.a.

n.a.

1987Dollars

4.32.92.10.40.62.22.53.83.63.23.12.71.81.31.4

n.a.

n.a.

n.a.

Benchmark-Years-

WeightedIndex

4.83.32.70.91.02.62.84.03.83.23.12.71.71.31.3

n.a.

n.a.

n.a.

CBOForecast

5.95.44.83.62.12.62.83.64.13.33.33.02.52.32.3

n.a.

n.a.

n.a.

Error

1.12.12.12.71.0

00.1

-0.30.30.20.20.30.81.01.0

0.8

0.9

1.2


6.15.44.83.72.63.73.83.34.34.03.83.43.23.23.0

n.a.

n.a.

n.a.

Error

1.32.12.22.81.51.01.0

-0.70.60.80.80.71.51.91.7

1.3

1.4

1.5

SOURCES: Congressional Budget Office; Office of Management and Budget; Department of Commerce, Bureau of Economic Analysis.

NOTES: Values are for the four-year growth rates for real gross national product (GNP) last reported by the Bureau of Economic Analysis,not the first reported values. The 1987-dollar series reflects the bureau's revisions published in July 1994. Revised estimates ofthe benchmark-years-weighted index, however, were not available at the time of publication. Forecast values are for the averagegrowth of real GNP over the four-year period. The forecasts were issued in the first quarter of the initial year of the period or inDecember of the preceding year. Errors are forecast values minus actual values; thus, a positive error is an overestimate. Thebenchmark-years-weighted index of actual GNP was used in calculating the errors.


a. Data for 1972-dollar GNP are available only through the third quarter of 1985.

b. Data for 1982-dollar GNP are available only through the third quarter of 1991.

Appendix B

Reestimating the NAIRU

T he Congressional Budget Office (CBO) hasrecently reestimated the nonacceleratinginflation rate of unemployment (NAIRU)

and has concluded that it is about three-tenths of apercentage point higher than was previously thought(see Figure B-l). The NAIRU is a summary mea-sure of capacity in the labor market that plays animportant role in CBO's projections of inflation andgrowth of real gross domestic product (GDP).1

Historically, the rate of inflation increases when therate of unemployment falls below the NAIRU anddecreases when the unemployment rate rises abovethe NAIRU. The upward revision implies, there-fore, that inflationary pressures are likely to occur ata slightly higher rate of unemployment than theywould under the old estimate. CBO's estimate ofthe NAIRU affects the projection for growth of realGDP because it is an important determinant (thoughnot the sole determinant) of CBO's estimate ofpotential output.

CBO reestimated the NAIRU because of agrowing consensus that the economy is approachingits productive capacity and could be in danger ofoverheating. The new estimate differs from the oldfor three reasons: first, it is calculated using alonger data sample; second, the data used for thecalculation have been revised; and third, a new (andbetter) measure of inflation is used.

For a discussion of the NAIRU, see Congressional Budget Office,The Economic and Budget Outlook: Fiscal Years 1995-1999 (Jan-uary 1994), p. 18. See also W.W. Lang, "Is There a Natural Rateof Unemployment?" Business Review, Federal Reserve Bank ofPhiladelphia (March/April 1990), pp. 13-22; or S.E. Weiner, "TheNatural Rate of Unemployment: Concepts and Issues," EconomicReview, Federal Reserve Bank of Kansas City (January 1986), pp.11-24.

The upward revision to the NAIRU is not re-lated to the revision to the unemployment portion ofthe Current Population Survey (CPS) made by theBureau of Labor Statistics in January (see Box 1-1in Chapter 1). For 1993 and earlier, CBO's esti-mated value of the NAIRU is based on the oldmethod for the unemployment survey. Beginning in1994, however, the NAIRU estimates should be ad-justed upward by another one-quarter of one per-centage point to account for the new survey meth-ods. The estimate of the 1994 level of the NAIRUis 6 percent using the new survey, a rate that iscomparable to the data on the unemployment ratenow being released.

Figure B-1.The Unemployment Rate, the ReestimatedNAIRU, and the Old NAIRU

Percent

1950 1960 1970 1980 1990

SOURCES: Congressional Budget Office; Department of Labor,Bureau of Labor Statistics.

NOTE: Data for the unemployment rate and the nonacceleratinginflation rate of unemployment (NAIRU) are on the oldbasis, as reported before the January 1994 revision ofthe Current Population Survey. Values of the NAIRU areestimated by CBO.


Procedure for Estimatingthe NAIRUComputing the NAIRU involves the statistical esti-mation of equations, known as Phillips curves, thatdescribe the primary influences on the short-termprocess of wage and price adjustment in the econ-omy.2 Phillips curves capture an important statisti-cal regularity: that higher rates of inflation have his-torically been associated with lower rates of unem-ployment. When total demand in the economypresses against the limits of what the economy cansupply, companies frequently raise the prices oftheir goods and services. They also typically hiremore workers to meet the larger demand, therebylowering the rate of unemployment and bidding upwages. When demand falls below the economy'scapacity, the opposite happens: companies cut pricesto move their goods while they reduce their payrollsto lower costs, thus raising unemployment andslowing the growth of wages.

The Phillips curve, in its simplest form, positsan inverse relationship between the rate of inflationand the rate of unemployment. Past research, how-ever, has found that the basic Phillips curve rela-tionship broke down during the 1970s, when theUnited States experienced high inflation and highunemployment at the same time. To cope with pe-riods like the 1970s, when fluctuations in total sup-ply dominated the economy, the simple theory mustbe augmented to include variables that capture theeffects of supply shocks (such as sharp increases inthe price of energy) and the process by which mar-kets form expectations of future inflation. A Phil-lips curve equation that allows for shifts in totaldemand and supply can successfully explain themovements of inflation and can be solved for thedesired estimate of the NAIRU.

For its Phillips curve, CBO estimates a regres-sion in which inflation—measured as the percentage

For other estimates of the NAIRU, see R.G. Gordon, "Inflation,Flexible Exchange Rates, and the Natural Rate of Unemployment,"in M.N. Baily, ed., Workers, Jobs and Inflation (Washington,D.C.: Brookings Institution, 1982); S.E. Weiner, "New Estimatesof the Natural Rate of Unemployment," Economic Review, FederalReserve Bank of Kansas City (Fourth Quarter 1993), pp. 53-69;and S.N. Braun, "Productivity and the NIIRU (and Other PhillipsCurve Issues)," Working Paper 34 (Federal Reserve Board of Gov-ernors, June 1984).

change in the overall price level-is the dependentvariable. The explanatory variables include lagged(that is, past) values of inflation (to represent ex-pected inflation); lagged values of the unemploy-ment rate of a reference group in the labor force (tomodel total demand); a variable measuring produc-tivity growth; a variable to control for food and en-ergy prices; and dummy variables to control for theimposition of wage and price controls during theearly 1970s. CBO uses the unemployment rate of areference group, married males, to measure the levelof demand in the economy because the overall un-employment rate is affected by changes in the com-position of the labor force. (For example, the over-all unemployment rate could rise despite unchangeddemand if there was an influx of youths, who tradi-tionally have a high rate of unemployment, into thelabor force.) The Phillips curve equation can besolved for the rate of unemployment that wouldkeep the rate of inflation constant, which is theNAIRU for the reference group.

The NAIRUs for individual demographic groupsin the labor force can be computed from the mar-ried-male NAIRU.3 CBO estimates regressions thatrelate the unemployment rate for each demographicgroup to the unemployment rate for married males.The NAIRU for each demographic group is calcu-lated by inserting the NAIRU for married males intoeach of these equations. The overall NAIRU is thencomputed as a weighted average of the NAIRUs ofthe demographic groups, with the groups' laborforce shares used as the weights. Note that theNAIRU for married males and for each of the otherdemographic groups is constant during the sample;the overall NAIRU varies over time only becauseshares of the labor force change over time.

Issues in Estimatingthe NAIRU

The most important estimation issue is what mea-sure of inflation to use. There are several choices:implicit deflators, fixed-weighted price indexes, and

CBO breaks the labor force down by sex, race (white/nonwhite),and age (16-19, 20-24, 25-34, 35-44, 45-54, 55-64, and 65 andover).

APPENDIX B REESTIMATING THE NAIRU 61

several alternative price indexes. Using an implicitdeflator, such as the implicit GDP deflator, is inap-propriate because it measures not only changes inprices but also changes in the mix of purchases.Thus, the series could show a decline in inflationnot because the growth of prices slowed, but be-cause consumers shifted their spending to goodswith prices that had increased less since 1987.

A fixed-weighted index, such as the fixed-weighted GDP price index, is a purer measure ofinflation because it computes the change in theprices of a fixed market basket of goods. It there-fore avoids the main problem associated with im-plicit deflators. However, it also has shortcomings,two in particular.

First, the index is likely to provide a misleadingpicture of inflation for years that are far removedfrom the base year, because the pattern of spendingis locked to that in the base year. The current baseyear, for example, is 1987; goods whose share ofconsumer and business expenditures has increasedduring the postwar period will be weighted tooheavily in the index early in the sample, because theindex assumes a mix of purchases from 1987. Theopposite holds true for goods whose expenditureshare has decreased through time—they will beunderrepresented early in the sample.

Second, the entire history of the series changeswhen the base year changes, which happens whenthe Bureau of Economic Analysis (BEA) rebench-marks the national income and product accounts(NIPAs). Since the rate of unemployment is notsubject to such revision, the entire relationshipbetween inflation and unemployment changes eachtime the NIPAs are rebenchmarked. The problem isparticularly acute for early years in the sample,because they are farther away from the base year.

Perhaps the best measure of inflation to use is aprice index whose weights change, but only infre-quently during the data sample. Year-to-yearchanges in such an index would reflect onlychanges in prices (not changes in the spending mix),but the weights ,r any given year would never betoo different from the actual pattern of spending forthat year. By this criterion, the CPI-U (the con-sumer price index for all urban consumers) seems to

be a good candidate.4 However, the CPFs weightsare somewhat out of date since they derive fromsurveys performed between 1982 and 1984. CBOtherefore uses one of BEA's alternative price in-dexes for gross domestic product, the benchmark-years-weighted price index.5 The advantage of thisindex is that its weights change infrequently duringthe postwar period, roughly every five years whenBEA rebenchmarks the NIPAs. Values of the indexin years since 1987 (the most recent base year) in-corporate weights computed from expenditure sharesin 1987 and in the last year of the sample, which isnow 1993. Thus, although post-1987 values of theindex are subject to revision until the next rebench-marking, more distant history is not.

Besides the benchmark-years-weighted priceindex for GDP, CBO estimated NAIRUs using fourother measures of inflation to gauge the sensitivityof the results. The other inflation measures are thebenchmark-years-weighted price index for personalconsumption expenditures (PCE), the CPI-U, thefixed-weighted PCE price index, and the fixed-weighted price index for PCE less food and energy.Fortunately, the estimates of the NAIRU computedusing these different measures of inflation all clus-tered in a small range.

The foregoing discussion ignores the CPS revi-sion introduced in January 1994 because all ofCBO's NAIRU estimates use the old definition ofthe unemployment rate. The Bureau of Labor Sta-tistics reckons that, during 1993 (when it conducteda trial survey on the new basis alongside the exist-ing CPS), the unemployment rate derived from thenew survey was, on average, about one-half of apercentage point higher than that derived from theold survey. However, recent movements in theunemployment rate suggest that the trial survey mayhave given a misleading impression of the actual

4. CBO's version of the CPI-U avoids the inconsistency (caused by achange in the treatment of home ownership in 1983) that distortsthe official series.

5. See A.H. Young, "Alternative Measures of Change in Real Outputand Prices: Quarterly Estimates for 1959-92," Survey of CurrentBusiness (March 1993), pp. 55-61; A.H. Young, "Alternative Mea-sures of Change in Real Output and Prices," Survey of CurrentBusiness (April 1992), pp. 32-48; and J.E. Triplett, "EconomicTheory and BEA's Alternative Quantity and Price Indexes," Sur-vey of Current Business (April 1992), pp. 49-52.


impact of the change in survey method on the un-employment rate. Recent evidence indicates that thedifference is smaller than previously thought-per-haps as small as one-tenth of one percentage point.CBO adjusts its estimate of the NAIRU by one-quarter of one percentage point (see Box 1-1 inChapter 1 for more details).

Results of the Estimation

CBO's Phillips curve estimates yield good results—the equations fit the data well and satisfy a range ofstandard diagnostic criteria (see Table B-l). Oneaspect of the estimates deserves specific mention.In order to solve such equations for the NAIRU, thecoefficients on lagged values of inflation must addup to one. In CBO's equations, the sum of the in-flation coefficients is constrained to equal one. Thisconstraint was tested statistically and was found to

be justified in each equation at conventional levelsof significance.

More important, the estimates of the NAIRUcalculated from the different equations are all closeto one another. CBO's preferred equation (usingthe benchmark-years-weighted price index for GDP)yields an estimate of the NAIRU for married malesof 3.55, which implies an overall NAIRU of about5.8 percent in 1993. Once it has been adjusted forthe new CPS, the estimate of the NAIRU is 6 per-cent in early 1994. The estimates computed fromthe equations that use alternative measures of infla-tion are all within two-tenths of a percentage pointof the first estimate. The estimate of the married-male NAIRU from the equation using thebenchmark-years-weighted PCE price index is 3.63,that from the CPI-U equation is 3.72, that from thefixed-weighted PCE price index is 3.69, and thatfrom the fixed-weighted price index for PCE lessfood and energy is 3.64.

APPENDIX B REESTIMATING THE NAIRU 63

Table B-1.Estimated Coefficients from Phillips Curve Regressions to Determine the NAIRU

Dependent Variable:

IndependentVariables

Constant

Lagged Inflation8

Lagged UnemploymentRate (Married males)6

Food and Energy Prices'

Productivity Deviation9

Wage and Price ControlsOnh

Off'

R-Bar Squared

Number of Observations

Benchmark-Years-Weighted

Price Index(GDP)

2.45(5.2)

1.0"

-0.69(5.4)

0.19(1.7)

-0.10(2.9)

-0.75(1.3)

3.19(6.6)

0.82

117

Benchmark-Years-Weighted

Price Index(PCE)

2.50(4.4)

1.0C

-0.69(4.5)

0.43(2.7)

-0.13(2.8)

-1.19(1.5)

1.17(1.7)

0.75

119

CPI-U

2.87(5.0)

1.0"

-0.77(4.9)

0.43(2.6)

-0.06(1.3)

-1.83(2.2)

0.99(1.4)

0.76

152

Inflation

Fixed-Weighted

Price Index(PCE)

2.67(5.5)

1.0"

-0.72(5.5)

0.34(2.7)

-0.13(3.4)

-1.25(1.8)

1.62(2.7)

0.80

152

Fixed-Weighted

Price Index(PCE less food

and energy)

1.92(5.3)

1.0"

-0.53(5.4)

n.a.

-0.10(3.0)

-1.37(2-3)

2.34(4.3)

0.83

152


NOTES: T statistics appear in parentheses below coefficients.

NAIRU = nonaccelerating inflation rate of unemployment; GDP = gross domestic product; PCE = personal consumption expendi-tures; CPI-U = consumer price index for all urban consumers; n.a. = not applicable.

a. In each equation, lagged values of inflation are assumed to follow a third-degree polynomial distributed lag, with the far end pointrestricted to zero.

b. Lag length is 20 quarters.

c. Lag length is 18 quarters.

d. Lag length is 12 quarters.

e. Four lagged values of the unemployment rate for married males.

f. One period lag of food and energy prices; defined as the difference between the rates of growth of the fixed-weighted price index for PCEand the fixed-weighted price index for PCE less food and energy.

g. The difference between the rates of growth of labor productivity in the nonfarm business sector and trend labor productivity. The trendvariable is segmented trend; its rate of growth is constant between business cycle peaks but differs between business cycles.

h. A dummy variable designed to control for the imposition of wage and price controls in 1971. (It equals 0.8 for the five quarters between1971:3 and 1972:3.)

i. A dummy variable designed to control for the termination of wage and price controls in 1974. (It equals 0.4 in 1974:2 and 1975:1 and 1.6in 1974:3 and 1974:4.)

Appendix C

Major Contributors to theRevenue and Spending Projections

The following analysts prepared the revenue and spending projections in this report:

Revenue Projections

Mark Booth Corporate income taxes, Federal Reserve System earnings, excise taxesDrew McMorrow Excise taxesPeter Ricoy Social insurance contributions, estate and gift taxesMelissa Sampson Customs duties, miscellaneous receiptsDavid Weiner Individual income taxes

Spending Projections

Defense, International Affairs, and Veterans' Affairs

Elizabeth Chambers Military retirement, defenseKent Christensen DefenseChristopher Duncan International affairsVictoria Fraider Veterans' benefits, defenseRaymond Hall DefenseWilliam Myers DefenseMary Helen Petrus Veterans' compensation and pensionsAmy Plapp DefenseKathleen Shepherd Veterans' benefitsLisa Siegel DefenseJoseph Whitehill International affairs


Human Resources

Wayne BoyingtonScott HarrisonJean HearneLori HousmanJulia IsaacsDeborah KalcevicLisa LaymanJeffrey LemieuxCory OltmanPat PurcellDorothy RosenbaumConnie TakataJohn Tapogna

Natural and Physical Resources

Kim CawleyPeter FontaineMark GrabowiczTheresa GulloJames HearnDavid HullMary MaginnissEileen ManfrediIan McCormickSusanne MehlmanDavid MooreJohn PattersonDeborah ReisRachel RobertsonJudith RuudBrent ShippJohn Webb

Other

Janet AirisEdward BlauJodi CappsKarin CarrBetty EmbreyKenneth FarrisGlen GoodnowLeslie GriffinBryan GroteVernon HammettSandra Hoffman

Civil Service Retirement, Social SecurityMedicareMedicaidMedicareFood stamps, foster care, child careEducationMedicareFederal employee health benefitsUnemployment insurance, training programsSupplemental Security Income, Social SecurityEducationPublic Health ServiceAid to Families with Dependent Children, child

support enforcement

Energy, pollution control and abatementEnergy, Outer Continental Shelf receiptsScience and space, justiceWater resources, conservation, land managementGeneral government, deposit insuranceAgricultureDeposit insurance, Postal ServiceAgricultureAgricultureJustice, Federal Housing AdministrationSpectrum auction receiptsTransportationRecreation, water transportationCommunity and regional development, natural resourcesDeposit insuranceHousing and mortgage creditCommerce, disaster relief

Appropriation billsAppropriation billsAppropriation billsBudget projections, other interestAppropriation billsComputer supportAuthorization billsBudget projections, civilian agency payCredit programsComputer supportComputer support

APPENDIX C MAJOR CONTRIBUTORS TO THE REVENUE AND SPENDING PROJECTIONS 67

Jeffrey Holland Net interest on the public debt, national incomeand product accounts

Deborah Keefe Computer supportFritz Maier Computer supportKathy Ruffing Treasury borrowing, interest, and debtRobert Sempsey Appropriation bills

CONGRESSIONALBUDGET OFFICESecond and D Streets, S.W.

Washington, D.C. 20515

THIRD-CLASS MAILPOSTAGE & FEES PAID

C.B.O.WASHINGTON, D.C.

PERMIT No. G-70

OFFICIAL BUSINESS

PENALTY FOR PRIVATE USE, $300

ISBN 0-16-045157-4

9 0 0 0 0

9M780160M451577I!

The Economic and Budget Outlook€¦ · ISBN 0-16-045157-4. Preface This volume is one of a series...

Documents

Transcript of The Economic and Budget Outlook€¦ · ISBN 0-16-045157-4. Preface This volume is one of a series...