Tennis Serve Data May Elude Some as Serves Get Too Fast
by, , and
Read this article
- Access statistics
- 1,141 article downloads
- 2,251 complete issue downloads
- Total: 3,392
In our response to Krawczyk (2019), we emphasize the following points: (1) Our theoretical model incorporates a Tullock contest function which is the most commonly used tool in modelling any strategic contest, and controls for both the server’s and the receiver’s effort. (2) The panel nature of our data set allows us to control for unobserved heterogeneity at both the player and the match level and minimizes the omitted variable bias. (3) There is a difference between ‘risk’ and ‘effort.’ (4) There is a strong empirical pattern in our dataset which is robust to the use of different methodologies whether it is linear or semi-parametric. (5) Finally, given that Pope and Schweitzer (2011) was published in the top journal in the field and received a considerable number of citations, it should not come as a surprise that we apply their theoretical and empirical framework to tennis, which, like golf, has a well-defined reference point.
This article is a response to Unforced Errors: Tennis Serve Data Tells Us Little About Loss Aversion by Michał Krawczyk (EJW, March 2019).