Assessing the Soothsayers: An Examination of the Track Record of Macroeconomic Forecasting
Greer, Mark R., Journal of Economic Issues
On the first business day of each year, The Wall Street Journal (hereafter WSJ) surveys the top macroeconomic forecasters in the United States in order to ascertain their predictions for the U.S. economy in the forthcoming year. Forecasts are made about the growth rate of real output, the inflation rate, and other macroeconomic aggregates. The WSJ derives a consensus forecast for each variable by taking the mean of the individual forecasters' predictions for that variable. Another forecast is published at the beginning of each July for the forthcoming 12-month period.
The WSJ currently polls its panel of economists on their predictions for short-term interest rates (as measured by the three-month Treasury bill rate), long-term interest rates (as measured by the 30-year Treasury bond rate), the growth rate of real output (as measured by real GDP), the rate of inflation (as measured by the CPI), and the exchange rate of the U.S. dollar against the Japanese yen. Forecasts of the unemployment rate were undertaken from 1986 until 1995. Each January, the economists are asked to forecast six and twelve months into the future. In forecasting interest rates and exchange rates, they prognosticate what their levels will be six and twelve months in the future. In the case of the economic growth and inflation variables, they forecast their rates of change for the first and second halves of the year.
The WSJ began publishing these forecasts at the beginning of 1981, and the macroeconomic aggregates covered in the survey have changed over time. While the survey has included the interest rate variables since 1982, only six-month predictions of these variables were undertaken until the 1984 survey, after which both six-month and one-year forecasting horizons were included. The survey has included a consensus forecast for the growth rate of real output since 1986, with real GNP used to measure output in the 1986-1991 surveys and real GDP used subsequently. The economists began forecasting the civilian unemployment rate in 1986 and finished doing so in 1995. Inflation predictions have been published regularly since 1986, and dollar-yen exchange rate forecasts have been undertaken since 1989.
The economists included in the WSJ's survey rank among the most highly credentialed and reputable in the profession. Some hold high-ranking positions in the nation's most powerful financial services corporations, others are partners in prestigious economic and management consulting companies, and yet others hold tenured professorships at top-tier economics departments. This group of individuals clearly possesses the capacity to bring the most advanced formal macroeconomic models and sophisticated econometric estimation techniques to the realm of macroeconomic forecasting. (Whether they actually find their expertise in economic theory and econometrics helpful in forecasting the economy will be examined later.)
This paper subjects the macroeconomic predictions of the WSJ's panel of economists to a straightforward directional accuracy test: how often did they correctly predict the direction of change in macroeconomic aggregates one year into the future?(1) For example, in 1996, the rate of inflation was 3.3 percent. In 1997, the rate of inflation fell to 1.8 percent. Therefore, 1997 witnessed a decrease in the rate of inflation compared to the previous year. In the WSJ survey published on the first business day of 1997, the consensus forecast called for a 2.9 percent inflation rate in 1997, which means that the panelists were predicting a decrease in the rate of inflation over the previous year. Since the forecasters correctly predicted the direction of change in this variable that year, this prediction was considered accurate for the purposes of this test. By contrast, at the beginning of 1993, the consensus forecast called for a 7.49 percent 30-year Treasury bond interest rate on December 31, 1993. Since the interest rate on this bond stood at 7. …