Working Paper, Federal Reserve Bank of Atlanta 2002-26
This study reports the results of fifteen experimental asset markets designed to investigate the effects of forecasts on market prices, traders' abilities to assess asset value, and the link between the two. Across the fifteen markets, the authors investigate alternative forecast-generating processes. In some markets the process produces an unbiased estimate of asset value and in others a biased estimate. The processes generating the biased forecasts, though, are less variable than the process generating the unbiased forecast. The authors find that, in general, period-end asset price reflects private forecasts, regardless of the forecast-generating process. Subsequently, they investigate whether traders' abilities to use forecasts differ across the forecast-generating processes. The authors find that most are able to properly use unbiased forecasts. They refer to them as smart traders. By comparison, a significant proportion is unable to properly use biased forecasts (typically traders' adjustments for bias are insufficient). Linking market outcomes and traders' abilities, the authors find that asset price appears to properly reflect unbiased forecasts as long as the market includes at least two smart informed traders who have sufficient ability to influence market outcomes. To obtain a comparable result in markets with the biased forecast, at least three smart informed traders with sufficient ability to influence market outcomes are necessary.