In a real effort experiment with repeated competition we find striking differences in how the work effort of men and women responds to previous wins and losses. For women losing per se is detrimental to productivity, but for men a loss impacts negatively on productivity only when the prize at stake is big enough. Responses to luck are more persistent and explain more of the variation in behavior for women, and account for about half of the gender performance gap in our experiment. Our findings shed new light on why women may be less inclined to pursue competition-intensive careers. To the best of our knowledge, our paper is the first to report how the work effort of men and women responds to the outcome of previous competitions. In each of 10 rounds subjects are paired and informed of the value of the monetary prize that they are competing for. The prize, which can be interpreted as a relative-performance bonus, is awarded to one of the pair members depending on the relative work efforts of the pair members in the "slider task", which involves positioning a number of sliders on a screen, and some element of chance which we control. In our empirical analysis we explore how effort provision responds to the outcomes of previous rounds of competitive interaction, i.e., previous wins and losses. We use fixed effects dynamic panel data methods and control for permanent individual-level ability, time effects and prize effects. We exploit randomization induced by the experimental design to obtain a number of valid instruments for the variables measuring previous competitive outcomes. We note that the randomness present in the experimental design is critical to our identification strategy: it is this randomness that allows us to estimate the causal effect of previous competitive outcomes on current effort provision.
To download the PDF version of the working paper click here.