Home > Centre for Philosophy of Natural and Social Science > Events > Abstracts > Choice Group Seminars - Abstracts


Choice Group Seminars - Abstracts

Luck, Insurance and Fairness

Social decisions under risk are a theoretical nightmare. The problems created by risk are numerous: 1) how to combine ex ante and ex post considerations which all appear relevant? 2) is it fair to rely on hypothetical choices made under a veil of ignorance in order to devise redistributive policies? 3) should we distinguish option luck from brute luck and compensate the effects of the latter only? 4) if prevention policies appear generally better both ex ante and ex post, why do we succomb to the temptation of implementing rescue policies? I propose to address these problems with the help of a simple theory based on three conceptual tools: 1) a comprehensive notion of individual well-being; 2) the principle that individuals are never responsible for their luck; 3) solutions from the theory of fair compensation.

Individual Characteristics, Risk Taking and Anomalies

  • Wednesday, 18 January 6:00 - 7:30 pm
  • Shepley Orr
  • University College London

The paper addresses three main questions. Are there types of individuals whose risk attitudes (and their correlates) have cross-situational stability and external validity? Second, are there types of individuals who are particularly liable to commit anomalies i.e., do anomalies correlate? Third, do certain variables predict loss-aversion, and are risk- and loss-aversion correlated? To examine these questions, we conducted a study where respondents were presented with a number of different types of question/task. For individual characteristics, we used attitudinal measures of both self-reported risk behaviour and a new instrument we devised to measure the variable we call "sentimentality", where we hypothesized that the latter would predict loss aversion. We measured four common anomalies in the literature: the common ratio effect, preference reversal, the Ellsberg paradox, and willingness-to-accept (WTA) vs willingness to-pay (WTP) disparities. Other individual characteristics we measured were uncertainty over preferences, risky choice behaviour, ambiguity aversion, and loss aversion. By examining the correlations among the variables, we were able to answer our questions as follows. First, while self reported risk behaviours do in fact seem to correlate across situations (i.e., the alpha on the reliability analysis of the self-reported risk behaviour scale is high), this scale did not correlate with risk-taking using standard gambles. Second, we did not find that anomalies correlated, leading us to reject the hypothesis that some types of agents are "anomaly-prone". The answer to the third question above is more complicated. We found that many of our variables concerning risk seeking and loss aversion did in fact correlate, but others did not. When looking at the patterns in the data as a whole, including the anomalies, we found two relatively distinct clusters of factors. However, we found that where predicted relationships were not found between the two clusters, it was often the case that the method of elicitation for the variable itself varied. That is, we found that when the same construct (e.g., risk aversion, loss aversion) was measured by both choice and valuation tasks, that correlations were lower than expected. This is consistent with the failure of invariance, raising questions about the transferability of data generated by any specific type of task.

Preferences Over Monetary Time Sequences: Theory and Evidence

Lowenstein and Sichermann (1991) and several others have argued that there is evidence of decision makers preferring improving sequences of outcomes (e.g. rising wage profiles or lowering discomfort sequences): "To most persons, a deteriorating series of utility levels is a rather close approximation to the least attractive of all possible patterns". This fact, especially for monetary sequences, runs contrary to any currently accepted economic model of decision over time, as it implies negative discounting. Most of this evidence comes from survey data, in which respendents were answering hypothetical questions. We ran an experiment in which subjects were provided with solid financial incentives when choosing between monetary sequences. We find that a majority of people choose 'rationally', so that the psychological heuristics proposed by Lowenstein and collaborators do not explain the data at all. On the other hand, a significant proportion of the data is inconsistent with any positive discounting model. In addition, an analysis of irrational choices uncovers some striking and non-obvious patterns of association (between different types of irrational choices and between rational and irrational choices). We propose a two stage elimination decision model. Decision makers use sequentially an incomplete version of standard exponential discounting in the first stage and the 'increasingness' heuristics in the second stage. This model explains the data far better than any pure discounting model (exponential, hyperbolic or otherwise), and is consistent with the observed association patterns in irrational choices.