Investment decisions and time horizon: risk perception and risk behavior in repeated gambles

To investigate the effect of time horizon on investment behavior, this paper reports the results of an experiment in which business graduate students provided certainty equivalents and judged various dimensions of the outcome distribution of simple gambles that were played either once or repeatedly for 5 or 50 times. Systematic mistakes in the ex-ante estimations of the distributions of outcomes after (independent) repeated plays were observed. Despite correctly realizing that outcome standard deviation increases with the number of plays, respondents showed evidence of Samuelson's fallacy of large numbers. Perceived risk judgments showed only low correlations with standard deviation estimates, but were instead related to the anticipated probability of a loss (which was overestimated), mean excess loss, and the coefficient of variation. Implications for future research and practical implications for financial advisors are discussed.