Examination of National Basketball Association (NBA) team values based on dynamic linear mixed models.

Abstract

In the last decade, NBA has grown into a billion-dollar industry where technology and advanced game plans play an essential role. Investors are interested in research examining the factors that can affect the team value. The aim of this research is to investigate the factors that affect the NBA team values. The value of a team can be influenced not only by performance-based variables, but also by macroeconomic indicators and demographic statistics. Data, analyzed in this study, contains of game statistics, economic variables and demographic statistics of the 30 teams in the NBA for the 2013-2020 seasons. Firstly, Pearson correlation test was implemented in order to identify the related variables. NBA teams' characteristics and similarities were assessed with Machine Learning techniques (K-means and Hierarchical clustering). Secondly, Ordinary linear regression (OLS), fixed effect and random effect models were implemented in the statistical analyses. The models were compared based on Akaike Information Criterion (AIC). Fixed effect model with one lag was found the most effective model and our model produced consistently good results with the R2 statistics of 0.974. In the final model, we found that the significant determinants of team value at the NBA team level are revenue, GDP, championship, population and key player. In contrast, the total number of turnovers has a negative impact on team value. These findings would be beneficial to coaches and managers to improve their strategies to increase their teams' value.