Academic journal article
By Swidler, Steve; Ketcher, David
Journal of Money, Credit & Banking , Vol. 22, No. 1
Economic Forecasts, Rationality, and the Processing of New Information over Time
ECONOMIC EXPECTATIONS PLAY A CENTRAL ROLE in explaining the dynamics of any equilibrium market. Because of their importance, the properties of economists' expectations of key price and income variables have been the focus of several earlier studies. Papers by Pesando (1975), Carlson (1977), and Pearce (1979) used the Livingston survey data and found that the consensus inflation forecasts were not rational. Figlewski and Wachtel (1981) and Pearce (1984) analyzed individual responses and concluded, respectively, that the inflation and stock price forecasts violated the rationality assumption. In contrast, Mullineaux (1978) found that the mean inflation forecasts appeared to be rational, and Brown and Maital (1981) concluded that the Livingston forecasts for several economic variables were unbiased, although information was underutilized. More recently, Schroeter and Smith (1986) made adjustments to the Livingston data and corrected for previous statistical errors only to find mixed results concerning the rationality of the Consumer Price Index and Producer Price Index.
This study probes the unbiasedness of economic forecasts which is a necessary condition for rationality. However, the present analysis uses a different data set than in previous studies, and examines, in detail, the process by which new information is incorporated into updated forecasts of an economic variable. Two important variables, real GNP growth and the inflation rate as measured by the Consumer Price Index, are the focus of this study.
The Blue Chip Economic Indicators provide the real GNP and inflation forecasts used in this paper. In contrast to previous studies of rationality, the analysis in this paper makes no assumptions about the relevant information set used by economists. Instead, the monthly revisions of the Blue Chip forecasts make it possible to directly analyze the value of new information. Using regression analysis, the empirical results suggest that revised forecasts provide more accurate estimates of the variable's actual value and that Blue Chip respondents process new information in a rational manner.
Since 1976, the Blue Chip Economic Indicators has surveyed prominent economists in universities and private industry, asking for their forecasts o several business variables. The number of economists has varied over the years; however, in recent surveys, approximately forty-five to fifty responses have been reported for each variable. While the forecasts are for annual values, the economists are polled monthly. Their forecasts are gathered by phone during the first three business days od each month, thus ensuring a timely reporting of the data in a newsletter published one week later.
Before 1980, the first forecast for each variable was given in June of the previous year, and a new forecast followed in the succeeding eleven months. Starting in 1980, the June figures became the last Blue Chip forecasts for a given year. Thus in more recent years, the first forecast for each variable was made in the beginning of July, eighteen months prior to the realization of the actual value on December 31st of the following year. The final forecast, produced in the beginning of June, was made seven months prior to the realization date. (1) This polling procedure from July to June makes it possible to infer the value of new information as it occurs over time and to see if economists use this information in a rational manner.
The two variables analyze in this study are the real GNP growth rate and the rate of inflation measured by the Consumer Price Index (CPI). The economic importance of examining these two key variables is clear, and not surprisingly, they have been the central focus of several of the papers cited above. Both rates are determined from annual index value as year over year percentage changes. …