Freelance Friday: Is ‘The Contract Year’ a Real Thing?

Freelance Friday 2
Freelance Friday 2 /

Freelance Friday is a regular feature at The Nylon Calculus where we solicit contributions from readers. This edition comes from the team Leonardo Pedicelli and Marley Kraft, based on Leonardo’s senior thesis[1. Which can be found here in its entirety.] concerning the existence or lack thereof of a discernable “contract year” phenomenon in terms of player production. Both Leo and Marley recently graduated from American University, with a double major in economics & statistics, and business marketing respectively. Leo currently works for a small non-profit health care surveying vendor based out of DC where he continues to pursue his interest in research and data. Marley can be found on twitter @MarleyKraftNBA.  Questions, comments or submissions should be directed to @NylonCalculus on twitter or via email to TheNylonCalculus at Gmail dot com.

In economic theory there is a behavior that can be observed within the ‘agent-player’ framework referred to as ‘shirking’. This behavior is characterized by the ‘player’ underperforming by the standards of the ‘agent.’ Shirking has been looked at as an example of when ‘players’ in an economic environment purposefully don’t produce to their maximum ability.

In the context of the NBA, one might expect to see shirking when the right incentive structures create a situation in which a player may feel secure in their position enough to off back from exerting full effort. One could further assume that the time a player may feel the most secure is likely following the signing of a long term contract. In an environment where contracts are 100% guaranteed, players could feel they can afford to slack early and pick up their effort down the road with little to no lose of value to their next contract. At least in theory.

While this shirking behavior is intended to define behavior in which the player underperforms, the flipside of shirking would be an inflated performance in the final year of the contract? Thus, if shirking behavior could be observed within the NBA we should be able to measure the statistical difference between the final year of a contract and the first year of a new contract. However this is not the case. As our study, The Effect of Long-Term Guaranteed Contracts in the NBA: A Case for Moral Hazard, shows we have found no statistical evidence that this behavior occurs.

In this study we tracked the performance of 100 players over an 8-year span for 800 total observations. Using John Hollinger’s PER as a baseline performance indicator we compared the last year of a player’s second contract with the first year of their previous contract while controlling for player-specific, team-specific, and contract-specific variables. The first and second categories address any self-deprecating effects due to a player’s age and experience and minutes played per game, team-winning percentage, and the amount of time a player spends on a team, respectively. Finally, the third category—probably the most important—isolates any variation in the model caused by contract length, number of years remaining in a contract, base salary (in millions), and a dummy variable indicating the first and last year of a player’s contract. Below is a table that illustrates the descriptive statistics of such variables used in the analysis:

VariableObs.MeanStd. Dev.MinimumMaximum

In summary, the average player in the sample had a PER of 16.54, was approximately 27 years old, was the 17th selection in the NBA draft, and experienced 6 full professional years in the league. A benefit of using an 8 year cycle is that each player could be observed throughat least 2 contract cycles.

After the data was collected, six fixed-effect regression models were ran as it address any lurking unobservable adverse effects within each model.  Although we did not prove our shirking behavior theory, we did discover some pretty interesting statistics. For instance, there is a significant positive significant effect of winning percentage on PER and the number of years a player spends on a team. This implies that when an athlete plays for a better team, production increases. And similarly, the longer a player stays with a particular team, the more efficient he becomes—which all makes sense.

Another curious statistic is the overwhelming significance of the interaction term between age and the number of years a player spends on a team, which is negatively associated with PER. Put differently, the adverse effect of the number of years a player spends on a team is reduced for older players.

Not only did we run a regression on the full sample of NBA players, but also on a subsample of players who demonstrated at least two consecutive full contract cycles. This gave us the chance to really investigate the question of shirking behavior, as we compared player efficiency levels from the last year of a player’s previous contract to their efficiency levels in the first year of their new contract. If a player were to shirk, we would expect a player’s PER to be significantly greater in the last year of a previous contract compared to the first year of a new contract. However, our results in fact find no such effect, and even mildly suggest the complete opposite effect of increased production in the first year of a new deal. This analysis is summarized in the table below, more particularly in panel B.


VARIABLES Coefficients Std. Errors  
A. Full Sample (N = 712)
B. Sub Sample with two consecutive contract cycles (N = 305)

There are, obviously, some caveats. The study could have used a larger sample size and while PER is a decent catchall statistic it isn’t the be-all end-all of player performance. A closer look at ‘hustle’ statistics and defensive metrics may provide a different perceptive due to our belief that those stats can be effected by the amount of effort a player puts in to them. Regardless, this study gives a thought-provoking look into how incentives affect player performance in the NBA and whether the ill-reputed ‘contract year’ still exists.