Allying techniques from experimental economics and experimental psychology, we relate market data to independent measures of the psychological characteristics of the actors involved. This enables us to test hypotheses about the consequences of psychological variables for market behaviour.
Our experimental approach relies on an asymmetric information trading game directly inspired by Plott and Sunder (1988). The value of the asset can be high (490), medium (240), or low (50). The traders observe different private signals. For example when the value of the asset is high, half the participants are privately informed that it is not low, while the others learn privately that it not medium. Traders can place limit and market orders in a call auction and an open outcry continuous market. There is a strong winner’s curse risk in this trading game. For example, if an agent with a bullish signal (not 50) offered to buy, say at 270, this bid would systematically be hit by traders with bearish signals (not 490), while traders with neutral signals (not 240) would be much more reluctant to engage in trading. Biais and Pouget (1999) show that in equilibrium in this trading game there should be no trade, except at fully revealing prices, and consequently no trading gains or losses.
While the experimental data suggests that a fair amount of information is revealed in the prices, we also observe significant deviations from equilibrium. Very high prices signal unambiguously that the asset value is 490 and very low prices signal that the value is low. However, in the experiment, transaction prices close to 240 convey a more ambiguous signal. For such prices the proportion of cases in our experiments where the true asset is 240 is only 52%. Consequently, in the 48% of cases where the price deviates from true value, some of the players must earn non-negligeable profits at the expense of others. In line with the behavioural game theory approach suggested by Camerer (1997), we study whether this phenomenon can be predicted by psychological factors.
A specific kind of overconfidence in one’s judgment, which we refer to as miscalibration, can offer an explanation for the failure of some participants to realize that their trades suffer from winner’s curse risk and are consequently loss making. Miscalibrated people tend to overestimate the precision of their information. We measure this bias using a confidence-interval task (Alpert and Raiffa, 1982).1 In an experimental asset market, Kirchler and Macejovsky (in press) used a confidence-interval technique and found evidence of overconfidence in predictions of price variations. In a financial market context with asymmetric information, Benos (1998), Odean (1998) and Daniel, Hirshleifer and Subrahmanyam (1998) show theoretically that this form of overconfidence leads to poor performance. Our experimental approach is particularly well suited to test this conclusion, since we can rely on direct measures of psychological variables, as well as of trading performance.
In these theoretical analyses, underperformance in the market will stem from overconfidence in the precision of one’s private signal. In the simple information structure of our game, participants cannot overestimate the precision of their private signal. Yet, we expect miscalibrated participants to overestimate the precision of their information set, which includes their signal as well as the observation of the market prices. When conditional uncertainty about the value of the asset is high, rational agents will recognize this. In contrast, miscalibrated traders will be less aware of this, and thus show excessive confidence in their assessment of the value of theasset. Hence, we expect them to be especially vulnerable to the winner’s curse. We identify market circumstances where this problem is likely to be particularly acute. As mentioned above, in our experimental data, when the opening price is close to 240, there is actually almost one chance in two (48%) that the true value of the asset is 490 or 50. In this context, miscalibrated participants whose signal does not rule out that the value is 240, will overconfidently believe the asset is worth 240. Thus they will be prone to fall into a winner’s curse trap whereby they will incur losses through trading with other participants who make gains at their expense.
In addition to studying a cognitive bias such as miscalibration, in this paper we also study how social dispositions can affect market performance. Self-monitoring is a disposition to attend to social cues and to adjust one’s behaviour to one’s social environment (Snyder, 1974). High self monitors are role players who habitually anticipate the effect of their behaviour on others, and in addition anticipate that the others will behave strategically. Anticipating that other market participants will be trying to manipulate the market as they themselves do, high self monitors will be less likely to take market prices at face value. Rather, they will reason about the signals and strategies that generated these prices. For example, when observing a price near to 240, they will not so readily jump to the conclusion that this indicates that the value is 240. Thus, they should be relatively unlikely to fall into winner’s curse traps and thus should avoid the corresponding trading losses.
