Estimating Production Efficiency in Men’s NCAA College Basketball: A Bayesian Approach

Abstract

Using Bayesian analysis with Markov Chain Monte Carlo (MCMC) estimation, we generate estimates of technical efficiency for each game played by an Atlantic 10 Conference men’s basketball team during the 2005-2006 season. The flexibility of MCMC, and its ability to provide an objective measure for assessing model fit, makes it preferable to maximum likelihood (ML) estimation of stochastic production frontiers. Within the context of men’s basketball, this article addresses the question of whether technical efficiency necessarily leads to success relative to one’s competitors. Results indicate that (a) technical efficiency does not vary significantly, either across or within teams, implying that teams in the A-10 play at very close and high levels of efficiency and (b) technical efficiency does not correlate strongly with productivity, suggesting that the fundamental quality of one’s resources are more important than an efficient use of those resources. In addition, parameter estimates suggest that a single turnover or offensive rebound could mean the difference between winning and losing.