CBO’s Economic Forecasting Record: 2017 Update
CBO’s economic forecasts have been comparable in quality to those of the Administration and the Blue Chip consensus. Large errors in CBO’s forecasts tend to reflect challenges faced by all forecasters.
For four decades, CBO has prepared economic forecasts to use in making its projections for the federal budget. Forecasts of output, inflation, interest rates, and wages and salaries, in particular, play a significant role in CBO’s budget analysis. For example, to project receipts from individual income taxes, CBO uses its forecasts of wages and salaries.
CBO regularly evaluates the quality of its economic forecasts for several reasons. One is to determine if it needs to change its forecasting methods. For example, partly in response to past forecast errors, CBO has changed the way it forecasts productivity growth and interest rates in recent years. Another reason for evaluating past forecasts is to calculate the errors in those forecasts, which in turn can be used to approximate the range of errors or uncertainty in the agency’s current forecasts. Finally, publishing such evaluations gives readers a tool to assess the usefulness of the agency’s projections and is thus one way in which CBO demonstrates its commitment to transparency.
To evaluate its economic forecasts, CBO compares them with the economy’s actual performance and with the Administration’s forecasts, which are published in the Office of Management and Budget’s annual budget documents, and the Blue Chip consensus—an average of about 50 private-sector forecasts published in Blue Chip Economic Indicators. Such comparisons can indicate the extent to which imperfect information and analysis may have caused CBO to miss patterns or turning points in the economy. They can also help the agency identify areas where it has tended to make larger errors than other analysts. This report evaluates CBO’s economic forecasts over two-year and five-year periods. The span of years that CBO examined for this evaluation differs by variable and by forecast period on the basis of data availability.
How Does CBO’s Forecasting Record Compare With Those of the Administration and the Blue Chip Consensus?
CBO’s forecasting record is comparable in quality to those of the Administration and the Blue Chip consensus. When CBO’s projections were inaccurate by large margins, the other two forecasters’ projections tended to have similar errors because all forecasters faced the same challenges. For example, all three sets of forecasts of inflation were relatively inaccurate during the late 1970s and early 1980s but generally became more accurate as inflation stabilized in more recent decades.
Do CBO’s Forecasts Exhibit Statistical Bias?
Statistical bias is the tendency of a forecaster’s projections to be too low or too high over a period of time. A simple and widely used indicator of bias is the mean error. By that measure, CBO’s forecasts of most economic indicators examined here have tended to be too high by small amounts, but the agency’s two-year forecasts of real (inflation-adjusted) output were slightly too low, on average.
After evaluating the mean errors of its forecasts, CBO reached two conclusions:
- CBO’s two-year forecasts of output growth and inflation have been less biased than its two-year forecasts of interest rates and the growth of wages and salaries, which exhibit a sizable upward bias—that is, they have tended to be higher than actual values by a larger amount (see figure below).
- For most economic indicators, the mean errors of CBO’s five-year forecasts (which are discussed in the second half of the report) have been slightly larger than those of the agency’s two-year forecasts. That pattern shows that CBO has a tendency to overestimate economic trends over the longer term.
Other forecasters’ projections generally exhibited bias of a similar magnitude and in the same direction. The mean errors of the Blue Chip consensus forecasts were very similar to those of CBO’s forecasts. The Administration’s forecasts of the growth of real output had larger mean errors than CBO’s forecasts and the Blue Chip consensus, but its forecasts of inflation and interest rates exhibited less upward bias than did the other two forecasters’.
How Accurate Are CBO’s Forecasts?
Accuracy is the degree to which forecast values are dispersed around actual outcomes. One widely used measure of accuracy is the root mean square error (RMSE). By that measure, CBO’s two-year forecasts are generally as accurate as those of the Blue Chip consensus and, for most economic indicators, slightly more accurate than the Administration’s two-year forecasts. The accuracy of all three sets of five-year forecasts is comparable.
Comparing the accuracy of its two-year and five-year forecasts, CBO observed the following:
- CBO’s five-year forecasts of output and inflation are more accurate than its two-year forecasts of those variables, in part because long-term forecasts rest more on underlying trends in the economy than on short-term cyclical movements, which are very difficult to predict.
- CBO’s five-year interest rate forecasts are less accurate than its two-year forecasts of those rates because of the large and unexpectedly persistent decline in longterm interest rates that began in the early 1980s.
- For its forecasts of wages and salaries, CBO’s findings are less clear-cut. The agency’s two-year and five-year forecasts of the growth of wages and salaries are equal in terms of accuracy, but its five-year forecasts of the change in wages and salaries measured as a percentage of output are more accurate than the corresponding two-year forecasts.
What Are Some Sources of Forecast Errors?
CBO’s and other forecasters’ largest forecast errors often stem from the difficulties of anticipating three key developments:
- Turning points in the business cycle—that is, the beginning and end of recessions;
- Changes in trends in productivity; and
- Changes in crude oil prices.
How Do Assumptions About Fiscal Policy Affect Forecast Errors?
Fiscal policy refers to the federal government’s policies on taxes and spending. Assumptions about fiscal policy are an important ingredient of an economic forecast because such policy affects output, inflation, interest rates, and wages and salaries. To provide lawmakers with a benchmark against which they can assess potential changes in the law, CBO constructs its economic forecasts under the assumption that federal fiscal policy will generally remain the same as under current law. By contrast, the Administration’s forecasts reflect the assumption that the policies in the President’s proposed budget will be adopted. Forecasters in the private sector (such as those who contribute to the Blue Chip consensus) form their own projections about the future of federal fiscal policy, so their forecasts reflect changes in law that they anticipate will be made.
Those different assumptions about fiscal policy account for some of the variation among forecasts and thus in forecast errors. Assumptions about fiscal policy can be particularly significant when policymakers are considering major changes to current law. For example, in 2009 and 2010, CBO’s two-year forecasts of real output growth diverged noticeably from the Administration’s forecasts and the Blue Chip consensus because of the different fiscal policy assumptions underlying the forecasts.
What Are the Limitations of This Evaluation?
This evaluation has three limitations. First, all forecasters change their procedures over time, which makes it hard to draw inferences about future errors. Second, because forecasters make different assumptions about future fiscal policy, it is difficult to compare the quality of forecasts without considering the role of expected changes in laws. Finally, the historical data (on output and income, for example) that forecasters use to make economic projections are often revised, which can complicate the task of interpreting forecast errors.
by Forecast Horizon and Forecaster
The root mean square error (RMSE) is a measure of forecasting accuracy. It will always be a positive value; the larger the value, the less accurate the forecasts.
Forecast errors are projected values minus actual values; the mean error is the average of the forecast errors. Mean error values greater than zero indicate that the forecaster overestimated the actual values, on average; negative values indicate that the forecaster underestimated the actual values, on average.
Date ranges refer to the years in which the forecasts were made.