Long-Term Mortality Effects of an NFL Career (Job Market paper)
Professional football is a high-profile dangerous industry. Most National Football League (NFL) players suffer concussions, ligament damage, or broken bones. However, for many young men, the perceived benefits of an NFL career – fame, high salaries, medical and training services, and utility from playing football – outweigh the perceived injury risk. Using mortality data on all NFL players born between 1937 and 1975 with at least three years’ experience, I attempt to quantify this injury risk in mortality terms. Baron et al. (2012) report that NFL retirees have lower mortality than the age-matched general public, but NFL players are positively selected on many genetic characteristics, smoking rates, and physical fitness history. To reduce positive selection bias, I compare NFL retirees with short careers to retirees with long careers. Cox proportional hazard and OLS models yield no significant linear relationship between career length and mortality. Career-shortening injury histories may increase mortality, however, biasing these estimates downwards. To address this bias, I examine the early retirement margin, before most players can accumulate substantial injury histories. I estimate that mortality for players with careers longer than three years is about 40 percent higher than mortality for players with three year careers, assuming a constant proportional effect across all ages. For positions with a high risk of injury, the mortality hazard for players with more than three years’ experience is more than double the hazard for three-year players.
High draft picks are a coveted commodity in the NBA. Teams are often accused of losing intentionally (“tanking”) in order to increase their odds of receiving a favorable pick in the league's draft lottery. This paper answers two questions related to tanking in the NBA: (1) What is the value of receiving the first draft pick? (2) Do teams lose intentionally to secure higher draft positions? We answer the first question by controlling for the probability of winning the lottery in a linear regression framework. The estimates indicate that winning the draft lottery increases attendance by 5 percentage points during the five year period following the draft. Receiving the first pick is also associated with a small increase in win percentage, though this effect is less precisely estimated. To answer the second question, we analyze games played by non-playoff teams near the end of the season. Using a difference in differences methodology that compares games in which a team can potentially change its lottery odds to games in which these odds are fixed, we find strong evidence of tanking. Playoff-eliminated teams are 14 percentage points more likely to lose a game when doing so increases their chances in the draft lottery. In light of these findings, we examine whether teams tank by benching star players (workers), or whether players themselves put in less effort. We examine the incentives for each party in a principle-agent framework.
Soccer teams regularly compete at altitudes above 2,000 meters (6,562 feet) with World Cup qualification or other honors on the line. Media, fans, and players often question the fairness of playing at high altitudes, and FIFA temporarily banned international matches above 2,500 meters (8,200 feet) in 2007. Researchers agree that traveling to higher or lower altitude can harm athletic performance, but the effects on professional athletes may be too small to influence match outcomes. Additionally, many teams try to limit altitude effects by allowing players extra time to acclimatize before a match. To identify the causal impact of altitude change, we compare South American international match outcomes between the same teams but played at different altitudes within the same country. This approach controls for influences such as differences in travel distance for high and low altitude countries. We find that traveling to lower altitude does not affect performance but traveling to higher altitude has negative effects. In particular, away teams perform poorly in Quito, Ecuador (2,800 meters), and La Paz, Bolivia (3,600 meters). However, away teams do relatively well in Bogotá, Colombia (2,550 meters). We conclude that stadium altitudes should not be restricted under 3,000 meters without further justification.
We evaluate the effects of academic achievement awards for first and second-year college students on a Canadian commuter campus. The award scheme offered linear cash incentives for course grades above 70. Awards were paid every term. Program participants also had access to peer advising by upperclassmen. Program engagement appears to have been high but overall treatment effects were small. The intervention increased the number of courses graded above 70 and points earned above 70 for second-year students, but there was no significant effect on overall GPA. Results are somewhat stronger for a subsample that correctly described the program rules. We argue that these results fit in with an emerging picture of mostly modest effects for cash award programs of this type at the post-secondary level.
Moral Hazard, Peer Monitoring, and Microcredit: Field Experimental Evidence from Paraguay
(with Jeffrey Carpenter)
Federal Reserve Bank of Boston Working Paper 10-6, June 2010
Given the substantial amount of resources currently invested in microcredit programs, it is more important than ever to accurately assess the extent to which peer monitoring by borrowers faced with group liability contracts actually reduces moral hazard. We conduct a field experiment with women about to enter a group loan program in Paraguay and then gather administrative data on the members’ repayment behavior in the six-month period following the experiment. In addition to the experiment which is designed to measure individual propensities to monitor under incentives similar to group liability, we collect a variety of the other potential correlates of borrowing behavior and repayment. Controlling for other factors, we find a very strong causal relationship between the monitoring propensity of one’s loan group and repayment. Our lowest estimate suggests that borrowers in groups with above median monitoring are 36 percent less likely to have a problem repaying their portion of the loan. Besides confirming a number of previous results, we also find some evidence that risk preferences, social preferences, and cognitive skills affect repayment.
The Effects of Expectation on Perception: Experimental Design Issues and Furthur Evidence
Federal Reserve Bank of Boston Working Paper 07-14, September 2007
Numerous studies have found that top-down processes can affect perceptions. This study examines some of the issues involved in designing field experiments aimed at discovering whether top-down mental processes affect perceptions, and, if so, how the influence takes place. Lee, Frederick, and Ariely (2006) (LFA) attempt to go further by testing whether expectations affect perception directly, by altering how sensory receptors and/or the brain’s processing centers interpret a outside stimulus—or indirectly, for example, by changing the amount of attention paid to the outside stimulus. In order to test the robustness of the findings in LFA, this paper reports the results of a field experiment similar to the one analyzed in LFA. The field experiment, designed to address some potential confounding factors in this type of research, confirms that expectations can alter perceptions. However, it also shows that heterogeneity across individuals can play a role in determining the nature of this effect, a finding that complicates the interpretation of results such as those in LFA. To frame the analysis, this paper discusses the difficulties in designing this type of experiment, makes some improvements to existing designs, and suggests some ways of eliminating the confounding influences that remain.
Does Player Race Affect Hiring Through the NFL Draft?
Numerous studies have investigated pay differences by race in professional sports and other industries, with mixed findings. Relatively little research exists, however, concerning racial bias in hiring. The National Football League (NFL) is a particularly interesting industry in which to study this issue, since around 70 percent of the League’s players are black and hiring data are readily available. NFL teams hire top young players in an annual draft. Teams take turns selecting players in the draft, and selection gives the team exclusive hiring rights. Therefore, each player’s selection rank in the draft order proxies for teams’ valuation of his skill. Using the 1990 to 2009 NFL drafts, I control for draft position and look for outstanding differences in NFL performance by race. Black players drafted in the 1990s early in the draft (potential stars) tend to outperform white players drafted at a similar position. Although this difference has eroded over time, since 2000, black players drafted late have been more likely to see some game action compared with similarly drafted white players. I discuss these differences in the context of racial bias post-hire, customer-driven discrimination, and racial stereotypes concerning player type and position.
Steroids, Age, and Contracting in Major League Baseball
Anecdotal evidence and changes in batting ability suggest that many Major League Baseball (MLB) players used performance enhancing drugs in the 1990s and early 2000s. Increased testing and penalties starting around 2005, along with various federal investigations, appear to have decreased steroid use substantially. Offensive production has dropped steadily and only a handful of players test positive for banned substances each year. I examine the effects of this rule change on contracting in MLB. I find that teams signed older star players during the steroid era anticipating strong performance through players’ mid-to-late-30s. As supplement use has decreased, performance by older players in the top 25th percentile by salary has also decreased dramatically. I use young star players as a control group in a difference-in-differences framework to show that this change is largely driven by players above 31 years old. While teams may not have anticipated the end of the steroid era in baseball, I find surprising evidence that teams continue to sign older players to unfavorable long-term contracts. I discuss these findings as they relate to team time preferences and the team-player principle-agent relationship.
Crowdsourcing and NFL Betting Markets
Research suggests that short stock picks submitted in an online community can be used to beat average stock market returns (Avery, Chevalier, and Zeckhauser, 2011). I use game picking data from football "pick 'em" leagues to develop a mildly profitable sports gambling strategy. These crowdsourced pick distributions are most useful when they generate score predictions that differ greatly from the betting line. Success rates also improve later in the season, when fans have more information about team quality.