Numerous psychological and economic experiments have shown that the exchange of promises greatly enhances cooperative behavior in experimental games. This paper seeks to test two theories to explain this effect. The first posits that individuals have a preference for keeping their word. The second assumes that people dislike letting down others' payoff expectations. According to the latter account, promises affect behavior only indirectly, because they lead to changes in the payoff expectations attributed to others. I conduct an experiment designed to distinguish between and test these alternative explanations. The results demonstrate that the effects of promises cannot be accounted for by changes in payoff expectations. This suggests that people have a preference for promise keeping per se.

2.3.1, 2.3.1

promisesv19.0.ztt: zTree program (zTree version 2.3.1) Screen.eps: screen shots (in german) Translation of screens.pdf english translation of the screens Translation of instructions.pdf english translation of instructions (German available upon request) readme_data.txt information on the other files .dat, *.tab, promises.do: see readme_data.txt LIST OF FILES: baseline.dat - Stata data file for "communication" and "no communication" treatments discussed in Appendix A switch.dat - Stata data file for "switching" treatments discussed in Section 4 promises.do - Stata code for tables, non-parametric tests, and regressions messages_baseline.tab - Tab delimited file containing text messages send in the baseline communication treatment, as well as the coded dummy "promise" messages_switch.tab - Analogous to previous file for the "partner switching" treatment NOTE: In order to run the .do file, you will need to adjust the paths to point to the appropriate directory.