We analyze the effect of player salary, a sunk cost, on player utilization in the National Basketball Association (NBA). According to economic theory, rational agents make decisions based on marginal expected benefits and costs, and non-recoverable costs should not influence decision-making. Therefore, NBA teams should be playing their most productive players, regardless of salary. Whether decision-makers in the real world uphold this normative theory and ignore sunk costs has been the topic of much empirical work. Previous similar studies have looked at whether NBA teams irrationally escalate commitment to their highest drafted players by giving them more playing time than their performance warranted, coming to mixed conclusions. We build upon these studies by using salary to measure the impact of financial commitment on playing time, by using a fixed-effect panel data model to control for unobserved individual heterogeneity which may have been biasing previous results, and by using a spatial econometric model for a robust check of playing time dependence among players within each team. Our results indicate that a small but significant sunk-cost effect is found.