Measurement Error, Luck and Risk-Taking by University Students in a Repeated Risky Investment Game

Abstract

We study whether luck in a repeated risky investment game can contribute to measurement errors and also help improve the predictive power of the game. We investigate how good and bad luck affect risk-taking behavior in the following rounds of the game. The luck outcome of the previous round is known when the subjects decide how risky their next choice in the game will be. A sample of 720 university students is used as subjects in the game in a recursive within-subject design. The results demonstrate a strong impact of luck on risk-taking behavior that lasts not only to the next round but also into another two follow-up rounds, with cumulative effects. A time delay of 1-2 months between Round 1 and Round 2 did not wipe out the luck effect and it was only marginally weaker than the luck effect from Round 2 to Rounds 3 and 4 that followed immediately after Round 2. Many recent studies have shown that risk preferences respond to shocks. This study indicates that random shocks such as luck in previous games (states of nature) influence risk-taking behavior. On average, luck stimulates more risk-taking and bad luck less risk-taking.

Keywords: Risky investment game, Measurement error, Repeated game of chance, Luck, Predictive power, Malawi