An experimental study of risk, ambiguity and market behaviour

Abstract

This thesis examines the consequences of Knight’s (1921) distinction between measurable risk and unmeasurable uncertainty or ambiguity for understanding the relationship between individual decision-making and market behaviour. Knight (1921) argued that investigating this distinction brings differences between theoretical predictions and observed market behaviour into focus. Ellsberg’s (1961) thought experiment challenged the descriptive validity of subjective expected utility theory. Knightian uncertainty and ambiguity challenge the ability of alternative decision theories to explain the interaction of individual traders within market institutions. Alternative predictions of the effects of Knightian uncertainty and ambiguity on market behaviour include increased price volatility due to Keynes’ (1921) animal spirits (Epstein and Wang, 1994) or more predictable behaviour governed by habit, rules of thumb and imitation (Heiner, 1983; Choi, 1993). This thesis investigates (1) the effect of Knightian uncertainty, ambiguity and ambiguity-aversion on the out-of-equilibrium behaviour of markets, (2) the effect of ambiguity and market interactions on individual decisions, and (3) the relative roles of risk and ambiguity preferences in understanding market behaviour. The key findings of this thesis include lower prices under ambiguity for equivalent probabilities under risk, no evidence that the effects of ambiguity on market prices are eliminated or significantly reduced over time, and an examination of the effect of ambiguity on the volume traded, bid-ask spreads and other aspects of trading behaviour. Most importantly, rather than increased price volatility, more predictable patterns of prices were observed under ambiguity compared with risk. This finding is compatible with traders who use heuristics to guide their behaviour, rather than discovering preferences over time as hypothesised by Plott (1996). Additionally, this thesis presents evidence that market interactions did not eliminate or significantly reduce the effects of ambiguity on individual decisions or over time. These findings emphasize the importance of rule-based behaviours in markets such as price following, momentum trading and similar phenomena that lead traders to pay increased attention to market prices under ambiguity, even when other traders do not have any more information than they do themselves. This suggests that the rules of thumb followed by individual decision-makers to overcome cognitive limitations interact in important ways in market institutions. Past research has suggested the interaction between risk preferences and the out-of-equilibrium behaviour of markets as an important direction for research (Duxbury, 1995; 2005). One of the few studies to relate measured risk preferences to experimental market behaviour, Ang and Schwarz (1985) presented evidence of a connection between less risk-averse behaviour, higher price volatility and faster rates of convergence to equilibrium. However, there are important psychological differences in the way people respond to the risk and the ambiguity that characterizes decision-making contexts in the field (Einhorn and Hogarth, 1986). This thesis compares measurements of risk and ambiguity preferences with other experimental decisions, finding differences between the contexts of risk and ambiguity. These contextual differences are particularly important for understanding the interaction of traders in markets if ambiguity-aversion is explained by the use of rules of thumb (Camerer, 1999), traders are more likely to apply rules of thumb under ambiguity (Heiner, 1983), or even imitate others under these conditions (Choi, 1993). Finally, this thesis presents comparisons of measured risk and ambiguity preferences with the direction of trade in double auction markets finding that ambiguity preferences are strongly related to the direction of trade under conditions of ambiguity. Combined with the observation of more predictable patterns of prices under ambiguity, this finding supports Camerer’s (1999) proposition that ambiguity preferences might be more important than risk preferences for understanding market behaviour. In contrast, less predictable prices observed under risk might reflect the behaviour of better informed traders and the potential for faster rates of convergence to equilibrium. Rather than a hindrance to the introduction of new markets, this form of price volatility might reflect a better functioning market. This thesis emphasizes the importance of the difference between risk and ambiguity for understanding market behaviour.