Convergence and Clustering in Major League Baseball: The haves and have nots?

When studying the effect of time on competitive balance in baseball two distinct views arise. One view, based mostly on the works of academics claims that competitive balance has never been as equal as is it today. The other view largely provoked by the media and industry stress the opposite opinion that the balance of competition in the MLB is far worse today than it has ever been. However, Berri and Schmidt focus less on the change of competitive balance and more on how and why competitive balance has caused convergence clusters to form. In their study competitive balance is defined as the situation in which all teams have reached such extreme similarity that they all have the same probability of winning. Convergence however focuses more upon variables likeliness to converge on identical values at future times, thus creating identical probabilities of winning. Through analyzing the study one can see that the data collected points out a significant decrease in clusters post 1960. Given these findings, two periods of interest were focused upon 1901-1060 and 1901-2001. Although the collected data has proven that clusters have substantially decreased over the years, other factors are cited that could also represent significant factors. Market size, possibly the most noteworthy factor focuses on the ‘have nots,’ or teams with small market size who are predestined to have below .500 records. When analyzing such a large amount of data, contradictions inherently lie in the outcomes. Although in some years the ‘haves’ have had an advantage in on field performance in other years data paints the opposite picture. Ultimately, Berri and Schmidt conclude that market size is not the primary factor but rather a “shortage of quality managers capable of transforming an underperforming organization into a consistent winner on and off the field” is of primary importance.

Variable / Explanation / Source
t / Time / Major League Baseball
y / yearly winning percentage
d / difference in winning percentage between teams
k / number of clusters / source unclear
G / Estimator of the covariance matrix / Newey-West (1987)

“Schmidt, Martin B. and Berri, David J. Convergence and clustering in Major League Baseball: the haves and have nots? Applied Economics, 2004, 36, 2007-20014.”