Congressional Budget Office
Director’s Blog

Today CBO released an evaluation of the accuracy of its economic forecasts by comparing those forecasts with the economy’s actual performance and with the projections of other forecasters. The study examines the two-year ahead forecast accuracy and the five-year ahead forecast accuracy for a variety of macroeconomic variables, such as real GDP, inflation, and interest rates. Thirty-two CBO forecasts, those made early each year from 1976 to 2007, are included in the study.  Such evaluations help guide CBO’s efforts to improve the quality of its forecasts and also assist Members of Congress in their use of CBO’s estimates.

Since publishing its first macroeconomic forecast in 1976, CBO has compiled a forecasting track record that is comparable in quality with that of the Administration and that of the Blue Chip consensus. In particular, the accuracy of CBO’s two-year forecasts between 1982 and 2007 paralleled that of the forecasts published by the Blue Chip consensus and the Administration over the same period. The accuracy of CBO’s five-year projections also generally corresponded to that of other forecasters, although the Administration’s projections of types of income (such as wages and salaries, and profits) as a share of national output had slightly smaller errors. Comparing CBO’s forecasts with those of the Blue Chip consensus suggests that when the agency’s predictions of the economy’s performance missed by the largest margin, those errors probably reflected problems shared by other forecasters in predicting turning points in the business cycle.

CBO’s forecasting record provides a measure of the uncertainty underlying forecasts under normal circumstances. However, the current degree of economic dislocation exceeds that of any previous period in the past half-century, so the uncertainty inherent in current forecasts exceeds the historical average.

Today CBO released a brief that analyzes the effects of changes in the health insurance system on the U.S. labor market. In 2009, about three out of every five nonelderly American are expected to have health insurance that is provided through an employer or other job-related arrangement, such as a plan offered through a labor union. Changes to the health insurance system could affect labor markets by changing the cost of insurance offered through the workplace and by providing new options for obtaining coverage outside the workplace.

In the current system, employment-based plans are popular largely due to three reasons:

• They are subsidized through the tax code: Nearly all payments for employment-based insurance are excluded from taxable compensation and thus are not subject to income and payroll taxes.
• Employers offering coverage usually pay a large share of the premium – partly to encourage broad participation among their employees, so as to limit the potential for “adverse selection.”
• Larger group purchasers can spread administrative costs over a larger number of people, using these economies of scale to reduce costs imbedded in premiums.

Although employers directly pay most of the costs of their workers’ health insurance, the available evidence indicates that active workers—as a group—ultimately bear those costs.

Congress is currently considering proposals that would expand health insurance coverage. Those proposals would affect the labor market because of the close linkage between health insurance and employment. For example:

• Requiring employers to offer health insurance—or pay a fee if they do not—is likely to reduce employment, although the effect would probably be small. Those who would most likely be affected are currently paid close to or at the minimum wage. They would be more vulnerable to job loss because their wages could not be lowered sufficiently to absorb the cost of health insurance (if their firm decides to offer) or the fee (if their firm does not) without bumping into the minimum wage.
• Proposals that imposed surcharges on employers whose workers received subsidies directly from the government could have a larger impact on employment. (Such provisions are sometimes known as free-rider surcharges.) Many of the affected workers would be paid low wages that could not easily adjust to absorb the full cost.
• Providing new subsidies for health insurance that decline in value as a person’s income rises could discourage some people from working more hours.
• Subsidies could be targeted to small businesses, but employers or their workers might respond by taking action to qualify for the subsidies. For example, some firms might reorganize into smaller subsidiaries, and workers might move to smaller firms to take advantage of the new subsidies.
• Increasing the availability of health insurance that is not related to employment could lead more people to retire before age 65 or choose not to work at younger ages. It might also encourage other workers to take jobs that better match their skills—because they would not have to stay in less desirable jobs solely to maintain their health insurance.

The overall impact on labor markets, however, is difficult to predict. Although economic theory and experience provide some guidance as to the effect of specific provisions, large-scale changes to the health insurance system could have more extensive repercussions than had previously been observed and also may contain numerous pieces that would interact—affecting labor markets in significant but potentially offsetting ways.

Today CBO released an updated report on the treatment of federal receipts and expenditures in the National Income and Product Accounts (NIPAs). The Congress, executive branch agencies, and the press generally focus on the accounting of government finances presented in the Budget of the United States Government, which is prepared by the Office of Management and Budget (OMB). The budget is structured to provide information that can assist lawmakers in their policy deliberations, facilitate the management and control of federal activities, and help the Treasury manage its cash balances and determine its borrowing needs.

In contrast, the NIPAs—produced by the Bureau of Economic Analysis (BEA), an agency within the Department of Commerce—are intended to provide a comprehensive measure of activity in the U.S. economy, of which the federal sector is one part. Because the aims of the NIPAs differ from those of the federal budget, the two accounting systems treat some of the government’s transactions very differently, as described in this report. Despite the accounting differences, the government budget numbers reported by the BEA are not all that different from those reported under the OMB framework.

Today CBO released an updated report on the effects of the business cycle on the federal budget.

For example, during recessions, the budget deficit tends to increase because of the automatic stabilizers built into the budget: tax revenue tends to decline and certain forms of government spending, such as outlays for food stamps and unemployment benefits, tend to increase.

A budget measure that filters out cyclical factors is useful in several ways. For example, some analysts use such a measure to discern underlying trends in government saving or dissaving (that is, surpluses or deficits). Others use it to approximate whether the influence of the budget on aggregate demand and on the growth of real (inflation-adjusted) income in the short run is positive or negative. More generally, the measure helps analysts estimate the extent to which changes in the budget are caused by movements of the business cycle and thus are likely to prove temporary. The usefulness of the cyclically adjusted budget measure for such purposes is hampered by large but temporary federal measures, such as the Troubled Asset Relief Program (TARP).

Under CBO’s baseline assumptions, the cyclically adjusted budget deficit will rise sharply in 2009, to 9 percent of potential GDP (from 2.6 percent in 2008), but then decrease in 2010 and 2011 to 4.7 percent and 2.2 percent of potential GDP, respectively. Those deficits are smaller than the unadjusted deficit estimates because the latter include the automatic responses of revenues and outlays to the current recession. CBO expects that economic output will be much farther below its potential level in 2009, 2010, and 2011 than it was in 2008, which is to say that effects of the business cycle will substantially increase the federal budget deficit in those years. According to CBO’s baseline projections, in 2009 and 2010 the cyclical contribution to the budget deficit will climb to roughly 2.1 percent and 2.6 percent of potential GDP and in 2011 it will decrease to 2.2 percent.

Budget Deficits and Surpluses as Percentage of Potential GDP

In preparing its economic forecast published in September 2008, CBO estimated that 40 percent of the tax rebates issued in the spring and summer under the Economic Stimulus Act of 2008 would be spent within six months––raising the growth of consumption in the second and third quarters of 2008 by 2.3 percent and 0.2 percent, respectively, and reducing it by 1.0 percent in the fourth quarter, when the distribution of the rebates was expected to end. However, analysts disagree about the economic impact of tax rebates. One study of the 2001 rebates suggests that as much as two-thirds of those rebates was spent within six months. For the 2008 rebates, some analysts have put the figure as low as 10 percent to 20 percent––in contrast to CBO’s estimate of 40 percent. There is, however, disagreement among analysts about the economic impact of tax rebates. Today CBO released a brief that examines the issue in light of the evidence currently available.

Studies of tax rebates fall into three groups depending on the type of data employed: those based on detailed data about spending by individual households; those based on qualitative answers to surveys in which people were asked what they intended to do or had already done with their rebate check; and those based on national data on income and spending for the country as a whole. By itself, simple observation of aggregate  consumption over time may not detect the effect of rebates; no spike in spending corresponds to the spike in income. For that reason, CBO places more confidence in studies of the first two types, which rely on differences in spending by people who benefit from the tax rebates and those who do not.

The figure below illustrates why careful studies, not casual observation, are necessary.  It shows a counterfactual path for monthly consumption spending, constructed by subtracting from actual spending CBO’s estimate of the effect of the rebates. That estimate (based on an assumption that 40 percent of the rebates was spent) implies that the rebates raised the growth of consumption in the second and third quarters by 2.3 percent and 0.2 percent, respectively, but reduced it by 1.0 percent in the fourth quarter, when the distribution of the rebates ended. However, someone comparing the monthly course of consumption with that of income would be unlikely to detect any effect.

Note: The cumulative area between lines showing consumption with and without the effects of rebates is 40 percent of the area between the lines showing income with and without the rebates. In the figure, it is assumed that the 40 percent of rebates is spent over six months according to this pattern: 15 percentage points in the first month and 5 percentage points in each subsequent month. On the basis of those assumptions, CBO estimates that the rebates added 2.3 percent (at an annual rate) to the growth of consumption in the second quarter of 2008 and 0.2 percent in the third quarter but––because of those effects––reduced the growth of consumption by 1.0 percent in the fourth quarter.

I testified yesterday before the House Budget Committee regarding the state of the U.S. economy. I emphasized the following points:

• In CBO’s judgment, the economy will stop contracting and resume growing during the second half of this year, but the hardships caused by the recession will persist for some time. We now expect that the recovery will be more tepid than we had projected earlier. In particular, the growth in output later this year and next year is likely to be sufficiently weak that the unemployment rate will continue to rise into the second half of next year and peak above 10 percent. Economic growth over time will ultimately bring the unemployment rate back down to the neighborhood of 5 percent seen before this downturn began, but that process is likely to take several years.
• On the positive side, the fiscal stimulus provided by the federal government is now beginning to boost the economy, and financial markets show clear signs of improvement since the fall and winter. However, many factors will likely temper the strength of the recovery: the loss of household wealth, the fragility of financial institutions, persistently weak growth in the rest of the world, a surplus of housing units on the market, and low utilization of manufacturing capacity.
• Even if the economy returns to positive growth this year, the loss in output and income during this downturn will be huge. In CBO’s March forecast, the difference between the economy’s actual and potential output will average 7 percent of GDP (which is equivalent to about a trillion dollars) this year and next, and that gap in output will not close until 2013. CBO’s forecast in August is likely to show even larger shortfalls in output over the next few years. By this measure, the current recession and its aftermath will be the most severe economic downturn of the postwar period.

• The persistence of high unemployment in CBO’s forecast does not stem from a “failure” of fiscal stimulus. We expect that the stimulus legislation will boost GDP a little more than dollar-for-dollar of reduced tax collections and increased outlays. However, as large as the stimulus package is, the contraction in underlying demand is far larger, so the stimulus will offset only part of the contraction.
• Most experts believe that larger budget deficits are appropriate during recessions, because higher spending and lower taxes can bring the levels of resource use and output closer to the economy’s potential. Therefore, the extremely large deficit this year—roughly $1.7 trillion, or nearly 12 percent of GDP, in CBO’s March projection—serves a purpose. However, most experts also believe that persistent large deficits reduce capital accumulation and thereby slow the growth of output and incomes over time. Thus, the large deficits that CBO projects for the years after the economy has returned to full employment are more worrisome. Moreover, the sharp increase in debt this year and next raises the risk that investors might lose confidence in U.S. government debt as a safe haven. This risk heightens the importance of putting the budget on a sustainable path as the economy returns to full employment.

As I discussed yesterday, I participated in two panels at the Milken Institute’s Global Conference in Los Angeles on Monday.  The second panel was about “Infrastructure Projects as Economic Stimulus.” You can view the slides and webcast.

My main observations at this second panel were:

• Some of the infrastructure spending in the stimulus package would pass a cost-benefit test even apart from the recession.  For example, CBO’s analysis of infrastructure investment last year concluded that “additional spending of up to tens of billions of dollars each year on transportation infrastructure projects” could be justified as having benefits that exceed the costs.  We warned that economic returns on specific projects vary widely, so specific investments should be selected carefully.  In addition, we explained that some of the additional spending could be avoided by creating incentives to use existing infrastructure more efficiently—such as congestion pricing, which we analyzed more fully earlier this year.  Still, additional targeted infrastructure investment could be appropriate even in normal times. Moreover, these are not normal times, and it may be appropriate to undertake more immediate infrastructure spending than otherwise in order to put idle resources to use.

• CBO projected that infrastructure spending approved in the stimulus legislation would generate outlays—and thus economic stimulus—only gradually.  For example, we are looking for increases in federal highway spending to be 10 percent of the amount appropriated in the rest of fiscal year 2009, 25 percent in FY 2010, 20 percent in FY 2011, and a declining share thereafter.  Although the sluggishness of this projected spend-out surprised some observers, we explained that the need to draft plans, solicit bids, enter into contracts, and then to undertake the work (during appropriate weather) had led previous increases in budgetary resources for highways to be followed by increases in                                      

            Budgetary Resources and Outlays for Highways 

Source: Congressional Budget Office

outlays with a measurable lag. Lags in other areas of infrastructure spending can be even longer, especially where programs are new or receive significant boosts in funding relative to recent years—descriptions that fit provisions in the stimulus package focused on weatherization and broadband expansion among others.  CBO projects that total infrastructure outlays resulting from the stimulus package will peak in 2010 and 2011 but will remain significant for a number of years.  

Infrastructure Outlays as a Result of the American Recovery and Reinvestment Act



Source: Congressional Budget Office

• Very early data on the use of funds approved in the stimulus package are consistent with this perspective.  For example, the Department of Transportation has reported that $7 billion has been obligated for highway spending but only a few million federal dollars have been spent.

On Monday I participated in two panels at the Milken Institute’s Global Conference in Los Angeles.  I was honored to be invited, and I enjoyed the spirited and insightful discussions. You can view my slides and a webcast.

The first panel addressed the topic “U.S. Overview: When Will Growth Resume?”  My comments stressed four points:

• Even if economic growth resumes later this year, as a large majority of forecasters expects, the unemployment rate is likely to remain high for several years.  According to CBO’s March forecast, real GDP will begin to grow again in the second half of the year and will expand at a brisk 4 percent pace in both 2010 and 2011.  However, under our projections, the shortfall of actual GDP relative to its potential (that is, the “output gap”) will be so large by the end of this year—roughly 8 percent of GDP—that even this rapid growth rate will leave a measurable output gap through 2012.  The implication for economic policy is that actions to narrow that gap and bring the economy back to full employment more quickly are likely to receive serious consideration in the next few years.  (Read a longer discussion of our economic forecast.)

• The persistence of high unemployment in CBO’s forecast does not stem from a “failure” of fiscal stimulus.  We believe that last year’s tax rebates had a measurable impact on the economy, and that this year’s stimulus legislation will have a significant impact as well. CBO had expected that about 40 percent of last year’s tax rebates would be spent.  Although some economists have argued that the lack of a jump in consumption last year contradicts that expectation, we and other analysts think the data are quite consistent with it.  For example, see this illustration of actual consumption and disposable income (that is, income after taxes) compared with an estimate of what consumption and

  The Effect of Rebates on Disposable Income and Consumption (Monthly)

  

Source: Congressional Budget Office; Department of Commerce, Bureau of Economic Analysis.    (April 27, 2009)

spending from this year’s stimulus package has just begun. Once the spending and tax effects kick in, CBO expects that the package will boost GDP a little more than dollar-for-dollar of reduced tax collections and increased outlays—in other words, that the overall “multiplier” will be a little larger than 1.  Nevertheless, as large as the stimulus package is, the contraction in underlying demand is far larger, so the stimulus will offset only part of the contraction.

• The U.S. financial system is hardly out of the woods.  Despite the hundreds of billions of dollars of losses recognized by U.S. financial institutions in the past few years, most experts believe that hundreds of billions of dollars of further losses on mortgage-related assets and other assets will ultimately be incurred.  Unless the economy and real estate markets rebound more quickly than most forecasters predict, these further losses will be a drag on new lending for some time.  And international evidence on financial crises suggests that the crimp in lending caused by past losses can impede economic recovery for years.  Therefore, despite the vigorous and creative actions already taken by the Federal Reserve and Treasury, further actions are likely to be needed to return the economy to sustained growth. (See my testimony from January on options for dealing with the financial crisis.)

• The outlook for the federal budget over the next decade is grim.  To be sure, economic recovery and the waning of countercyclical tax, spending, and financial policies will cause the budget deficit to diminish sharply from its record-setting level this year if tax and spending policies set by current law are maintained.  In fact, CBO’s baseline projection, which follows current law, shows the deficit in the later part of the decade running just below 2 percent of GDP—a level that crowds out some private capital but allows debt to fall over time relative to GDP.  However, current law would lead to a significant increase in tax revenue relative to GDP (as the 2001-2003 tax cuts expire and as more households become subject to the Alternative Minimum Tax) and a significant decrease in defense and non-defense discretionary  spending relative to GDP (because the baseline assumes that these categories just keep pace with inflation).  If these components of the budget were instead maintained at their historical relationship to GDP, then the continued surge in spending on Medicare and Medicaid under current law would push the budget deficit to unsustainable levels.  (More discussion of this topic appeared in the blog last week.)

The CBO’s budget projections released last Friday are based in part on our economic forecast. In this blog entry, I want to discuss how our projection of real (that is, inflation-adjusted) GDP, one of the most important economic factors in the budget projections, compares with two other forecasts—the Blue Chip forecast (an average of about 50 private-sector forecasters), and the Administration’s forecast.

Often, it is useful to focus on growth rates, as we did in our report here. However, the effect of GDP projections on the budget—and especially on tax revenue—can be seen more clearly by comparing levels of projected GDP. As the figure below shows:

• CBO’s projection of real GDP is lower than that of the Administration throughout the next 10 years, and,
• CBO’s projection is stronger than the Blue Chip consensus of private forecasters in 2010 and beyond (although slightly weaker in 2009). A difference remains even though CBO’s projection of growth slows a bit after 2015 (the slope of the line flattens slightly), while the Blue Chip’s projection does not.

Comparison of CBO, Administration, and Blue Chip Medium-Term Projections: Levels of Real GDP, 2005 to 2019 (Billions of 2000 dollars)

Sources: Congressional Budget Office; Office of Management and Budget; Department of Commerce, Bureau of Economic Analysis; and Aspen Publishers, Inc., Blue Chip Economic Indicators (March 10, 2009).

For the differences in 2009, much of the story is behind us: real GDP fell at an annual rate of 6.2 percent in the fourth quarter of 2008 and now seems likely to be falling at a similar rate in the first quarter of this year. CBO’s forecast was completed a month and a half after the Administration’s forecast and a few weeks after the March Blue Chip survey. Economic news during that time—weaker employment, exports, and orders for manufacturers, and downward revisions to GDP growth in the fourth quarter of 2008 and to employment growth in December and January—caused many economists to sharply reduce their estimates for the level of economic activity in the first part of this year.

But, like the Blue Chip consensus, CBO projects the beginning of an upturn late this year (as shown in the figure below), reflecting in part the effects of the recent economic stimulus legislation (the American Recovery and Reinvestment Act) and very aggressive actions by the Fed and the Treasury.

The Gap Between Actual and Potential Output (Percentage of potential GDP)

Note: The gap is the difference between real (inflation-adjusted) gross domestic product and its estimated potential level (which corresponds to a high level of resource—labor and capital—use).

Sources: Congressional Budget Office; Department of Commerce, Bureau of Economic Analysis.

For the next few years, CBO projects faster growth than the Blue Chip, as the economy grows back toward CBO’s estimate of potential GDP (which corresponds to a high level of use of labor and capital resources). Still, the CBO forecast assumes that the gap between actual and potential output closes more slowly than in previous recoveries because of a persistent drag from financial markets, households’ loss of wealth, the overhang of vacant houses, and weak economic growth overseas. Therefore, CBO projects that the economy does not return to its potential level until 2014.

In the 2015-2019 period, the projected rate of real GDP growth averages 2.4 percent.  That rate is lower than during the period from 2010 to 2014, largely because there is no longer any gap to close between actual and potential GDP.

Projected growth from 2015 to 2019 is also below historical average growth rates, a difference that is more than accounted for by slower growth in the labor force because of the retirement of the baby boom generation.  Over the postwar period, the labor force grew at an average annual rate of 1.6 percent; by contrast, we project it to grow only 0.4 percent per year in the period from 2015 through 2019.  As a result, potential GDP grew 3.4 percent per year on average in the postwar period, but CBO expects that it will grow by only 2.4 percent annually (allowing for a tad more productivity growth) in the 2015-2019 period.  That demographic trend is reflected also in the Social Security Administration’s projections of the labor force, available here. CBO published its own analysis of demographic trends; while the numbers have changed a little with new information since then, the general story remains the same.

To be sure, all economic forecasts are subject to considerable uncertainty, as we emphasized in our report.  But I hope that this discussion of the logic behind the latest CBO forecast is helpful to readers of that report.

At the request of Senator Grassley and Congressman Camp, today CBO released a year-by-year estimate of the economic effects of the American Recovery and Reinvestment Act of 2009 (ARRA) as enacted on February 17, 2009. This is our first such analysis of the ARRA legislation as enacted. (In CBO’s letter to Senator Gregg, Senator Grassley, and Congressman Camp on February 11, 2009, we estimated an average of the effects of the House- and Senate-passed versions of H.R. 1 because the final language had yet been agreed upon; those numbers differ only very slightly from the estimates provided in today’s document).

We estimate that the legislation will raise gross domestic product (GDP) and increase employment in the short run—by adding to aggregate demand and boosting the utilization of labor and capital.  In contrast, we expect that the legislation will reduce output slightly in the long run because the resulting increase in government debt will tend to “crowd out” private investment and thereby reduce the stock of productive private capital.  That crowding-out effect will be diminished to the extent that some of the funding in the legislation will go for activities that could add to the nation’s long-term output. 

Today’s letter provides finer detail on how CBO estimated the short-run effects of the legislation. Different provisions in the law differ in both the magnitude and timing of their effects on aggregate demand. To simplify its analysis of the overall effects, CBO grouped the various provisions into a number of more general categories, and each category was assumed to have a range of effects on the economy that could by summarized by “multipliers”—the cumulative effect on output of a one-time increase in spending, or reduction in taxes, of one dollar. For example, a one-time increase in federal purchases of goods and services of $1.00 in the second quarter of this year would raise GDP by $1.00 to $2.50 in total over several quarters, with most of that effect in the first two quarters and little effect beyond a year. The multipliers are applied to outlays when they occur and to changes in taxes or transfer payments when they affect disposable income. Table 1 in today’s letter shows the categories to which CBO assigned the major provisions of ARRA. (In some cases, when different elements of a single provision were estimated to have different multipliers, the total cost of a provision was divided among more than one category. In those cases, the provision is shown in the table in the category to which most of its budgetary cost applied.)

Thanks to Ben Page and Robert Arnold of CBO’s Macroeconomic Analysis Division for their fine work on this analysis.