What Do Economists Know? An Empirical Study of Experts' Expectations

For more than three decades, economic columnist Joseph A. Livingston has canvassed a panel of economists twice a year, eliciting their six-month and twelve-month forecasts for more than a dozen key variables. This study analyzes whether the experts' predictions are unbiased, and whether complete use was made of all relevant, known information (unbiasedness and completeness being necessary conditions for fully rational expectations). Little bias was found in either the half-year or full-year predictions, but extensive underutilization of information-particularly data on monetary growth-occurred. "To prophecy is extremely difficult-especially with respect to the future." Chinese proverb Do ECONOMISTS' EXPECTATIONS regarding key price and nonprice variables utilize all known, relevant information, in an unbiased, efficient manner? This is a worthy subject for research, for several reasons. Properties of experts' predictions likely form an upper bound for those of laymen. Further, as John Muth [14] has noted, "the character of dynamic processes is typically very sensitive to the way expectations are influenced by the actual course of events" (p. 316); hence, we need to know precisely how events do affect expectations. Finally, the common practice of replacing a variable's (generally unobserved) expectation with a proxy based on its past values will be unbiased (and will not cause bias in other