Overconfidence in Currency Markets

Evidence from psychology shows that people tend to be overconfident in two dimensions: they underestimate uncertainty and overestimate their own abilities. This paper provides evidence that foreign exchange dealers are likewise overconfident in both dimensions.We study overconfidence because this specific departure from rationality has been the subject of substantial research showing that overconfidence could lead to severe distortions in asset prices. Further, the widespread human tendency towards overconfidence is one of the strongest findings in psychology. This evidence provides an objective basis for identifying a specific departure of potential economic significance, thus addressing a common concern: There is no objective basis for choice among the infinite possible alternatives to perfect rationality, so hypothetical departures from rationality might be designed to fit the data.We study dealers because they set the price in almost every currency transaction and they speculate with substantial resources at their disposal. This is relevant to a second key justification for assuming efficient markets: the market will achieve a rational equilibrium so long as the marginal agent is rational. If currency dealers are overconfident there may not be a rational marginal agent. Indeed overconfidence among dealers could bring price distortions even if they are not marginal agents (De Long et al. 1990).Our evidence concerns 400-plus North American traders with average trading experience of 12 years. Their modal rank is Senior Trader at a top dealing bank. These traders have had plenty of opportunities to learn their way out of biases and they face powerful financial incentives to do so. Our results indicate that currency dealers exhibit the same clear, strong tendencies to overconfidence documented among other groups.We also test Friedman's (1953) hypothesis that irrational dealers will be driven out of asset markets by trading losses. To the contrary, we find that experienced dealers are no less overconfident than inexperienced dealers. This result holds regardless of how overconfidence is measured, regardless of how experience is measured, and for different subcategories of dealers.Our evidence could help explain the high exchange-rate volatility of the floating rate period (Flood and Rose 1995), the profitability of trend-following trading strategies (Dueker and Neely 2007), and the apparent irrationality of exchange-rate forecasts (MacDonald 2000)