|M.Sc Student||Zel Dmitry|
|Subject||Repeated Decision Making and the House Money Effect|
|Department||Department of Industrial Engineering and Management||Supervisor||Professor Rann Smorodinsky|
In everyday life, people are often presented with virtually the same choices time after time, such as crossing a street on red light or betting in a casino. It seems natural to hypothesize that the outcomes of prior choices have certain influence on further decision making. Specifically, it is interesting to check whether subjects’ attitude to risk, when confronted with a choice, is affected by the prior outcomes of similar choices made in the past. One common measure of the attitude to risk is the money amount subjects are ready to put at stake. That is, in a lottery setting, the increase in betting amount is interpreted as an increase in risk taking, and vice versa, the decrease in the betting amount is interpreted as a decrease in risk taking, given that the lottery conditions remain unchanged.
Thaler and Johnson (1990) notice that prior gains and losses can dramatically influence subsequent choices in systematic ways. Specifically, they conclude that under certain circumstances, people tend to be more risk-taking following prior gains. This tendency is labeled as the House Money effect. However, decisions are rarely made in temporal isolation, hence it is necessary to define what "prior" gains or losses are. Having done that, the purpose of this work is to experimentally test the influence of prior outcomes on subsequent choices in different settings.
I employ two multiple-stage experiments, in which subjects are presented with identical lotteries. On each stage, subjects are required to decide on amount to bet. The probabilities of winning in each lottery are publicly known. I analyze the data from these experiments, along with data provided by Zak (2000), who ran a similar experiment, but didn’t disclose the winning probabilities to the subjects.
Under risk, subjects behave on the overall as predicted and exhibit increased risk taking the more they win throughout the experiment. However, in the lotteries immediately following a win, subjects depart from this pattern and decrease their bet, thus demonstrating the so-called gambler's fallacy. Under uncertainty, prior gains or losses had no effect whatsoever on further choices.