Dying to Win? Olympic Gold Medals and Longevity

This paper investigates how status affects health by comparing mortality between Gold medalists in Olympic Track and Field and other finalists. Due to the nature of Olympic competition, analyzing performance on a single day provides a way to cut through potential endogeneity between status and health. I first document that an athlete’s longevity is affected by whether he wins or loses and then detail mechanisms driving the results. Winning on a team confers a survival advantage, with suggestive evidence that higher mortality among losers may be due to poor performance relative to one’s teammates: slower team members die earlier than faster members within losing teams, but mortality does not differ between faster members of losing teams and all members of winning teams. Since I show that ability is likely not driving longevity differentials, slower members on losing teams perhaps feel guilty for “letting their teammates down.” Consistent with the importance of performance relative to other team members, I document there is less within-team variation in longevity among losing teams when relative performance cannot be objectively measured using results from the Tug-of-War.

However, winning an individual event is associated with an earlier death. By analyzing the best performances of each athlete before the Olympics, I demonstrate that an athlete’s performance relative to his expectations partly explains the earlier death of winners in individual events: on average, Olympic Gold medalists expected to win, but losers exceeded their expectations. Additionally, being considered a “favorite” but failing to win is associated with a higher mortality hazard than athletes initially ranked outside the top five who also lost.

My results are robust to estimating a range of parametric and semi-parametric survival models that make different assumptions about unobserved heterogeneity. A number of sensitivity tests verify that no country- or time-specific subset of the data drives the results. My central estimates imply lifespan differentials of a year or more between winners and losers. Status among these athletes is likely not correlated with income due to the prevailing system of amateurism during early Olympic Games. I also show that differences in ability between Olympic finalists—which may positively correlate with latent health—do not predict mortality by comparing athletes who ever held a World Record (the highest ability group) and those who did not.

The findings point to the importance of expectations, relative performance, and "regret" in affecting health, which are not highlighted by standard models of health capital. I also discuss potential implications for employment contracts in terms of a trade-off between ex post health and ex ante incentives for productivity.