Is time inconsistency primarily a male problem?

Abstract

We conduct a simple experiment, using real money, that tests whether men and woman differ in time consistency. The experiment provides strong evidence of time inconsistency among males, but no evidence of such behaviour among females. Furthermore, the difference between males and females is statistically significant. This result could have important implications in marketing and in efforts to improve intertemporal decision-marking.

Acknowledgements

We gratefully acknowledge the financial support of the National Science Foundation (Award SES-0418450) and the Cornell University Institute for the Social Sciences. We thank Bill Schulze and Ted O'Donoghue for helpful comments. All errors are ours.

Notes

¹The example and definition are based on O'Donoghue and Rabin (1999).

²To our knowledge, only Harrison et al. (2005), McLeish and Oxoby (2007), Ashraf et al. (2006) and Tanaka et al. (2009) address gender differences in time consistency. Of these, only Ashraf et al. (2006) and Tanaka et al. (2009) directly test for a gender difference in time inconsistency and fail to find a statistically significant difference; however, the former uses hypothetical questions, and the latter involves a starkly different subject pool (Vietnamese households).

³Experiment materials are available at:http://www.people.cornell.edu/pages/jtp35/index_files/OnlineApp.pdf

⁴We removed five students who had participated in a pilot version of the same experiment.

⁵These statistics are available at http://www.people.cornell.edu/pages/jtp35/index_files/OnlineApp.pdf. In principle, we could have asked the subject to provide this information along with their gender. However, we did not do this for two reasons. First, we were concerned that a significant number of participants may choose not to respond to some of these extra questions, making them uninformative. Second, and more importantly, we wanted to focus entirely on the demographic measure for which we had a prior–gender. Collecting many demographics and than searching for differences in time consistency for all of them would raise concerns about the validity of our finding (Rubinstein, 2001).