한국 프로야구의 구단 승률에 대한 분석

초록

We investigate at the team level the effects on the percentage of wins (to the total number of matches) of team's offense power, defense power, style and execution capability of strategy, leadership of the head coach, composition and characteristics of players, size of the average audience, distribution of players' salaries, type of funding for operation and management, hometown, and physical size of the home ballpark. Using the unbalanced panel data covering all 9 professional baseball teams from 2002 to 2012, we construct a fixed-effect panel data regression model and concurrently apply fixed effects vector decomposition to accommodate both time-varying and time-invariant variables in the set of determinants. We arrive at the best model based on the AIC criterion. Long hit and on-base ratios are found to affect team’s winning percentage positively, while WHIP (Walks plus Hits divided by Innings Pitched) affects negatively. Defense ratio measuring fielders’ defense power is found to be statistically insignificant. For offensive power, defensive power, and style and execution capability of strategy, sabermetricians’ assertions seem to prevail while traditional views seem to falter. The relationship between the tenure of the head coach and team’s winning percentage is U-shaped. Players’ average years of experience are found to be insignificant. In the composition of players, only the proportion of pitchers is found to be significant, impacting team’s winning percentage positively. The average number of spectators is found to have positive effect. We measure salary dispersion by HHI index and found that pitchers’ salary dispersion is insignificant, supporting neither tournament theory nor cohesiveness theory. On the other hand, batter’s salary dispersion impacts team’s winning percentage negatively, supporting cohesiveness theory. FA effects are found to be statistically insignificant.

키워드