The Baseball Economist: The Real Game Exposed (15 page)

Table 19 lists the population and league size by decade over the past eleven decades. It is true that the ratio of major-league players to the population was much higher when the league first started, but this ratio did not last for long. By 1940 the population-to-player ratio was greater than 330,000–to–1, which is higher than the sample average. From 1940 until the present, MLB seems to have done a good job expanding the league to keep this ratio constant through expansion. The current population-to-player ratio is very similar to the 1950 ratio; however, this does not mean that the talent dispersion today is equal to the dispersion of the 1950s. While racial integration was well on its way in 1950, following Jackie Robinson’s entry into the National League in 1947, the number of black players did not reach a plateau until 1970. And though Latin Americans did play baseball, they were nowhere near the presence that they are today. Thus, if the population-to-player ratio has remained stable over the past fifty years, it is still possible that baseball
team talent is drawn from a larger pool of talent that will compact talent dispersion.

However, there are several reasons why the simple population-to-player ratio may contain unreliable information about talent dispersion. First, there exists stronger competition in the labor market for players. Baseball is no longer the dominant professional sport available to athletes, as it was in the first half of the twentieth century. Professional football, basketball, hockey, soccer, and other sports all compete for athletic talent to some degree. Also, the growing U.S. economy provides many more nonathletic opportunities than in the past—if you don’t believe me, consider your job opportunities compared to those of your grandparents—which can attract marginal major-league talent away from the game.

Second, the rising general population numbers do not necessarily mean that the age cohort for baseball players as a percent of the population has remained constant. Much of the nation’s population growth is attributable to increasing life spans, not just increases in immigration and births. This is going to overstate the proportion of individuals available to play baseball over time. Whether or not these factors will have a measurable effect on using the population-to-player ratio as a proxy on talent dispersion is an empirical issue that we can investigate further. My bias is that the ratio is not really very informative with regard to the actual distribution of talent across the league.

The best way to analyze talent dispersion is not to hypothesize about the size of the labor pool, but to use Gould’s theory of quality to examine the similarity of performance across players in different time periods using available statistics. To measure talent dispersion, I have chosen two statistics to measure variability across hitters and pitchers for players. For hitters I use the sabermetrician’s shortcut for measuring the run—generation ability of any hitter, OPS. OPS is the sum of on-base percentage plus slugging percentage, and it is a good measure of individual hitter contributions toward offense. For pitchers, I use the earned run average, or ERA. To measure the dispersion of talent in the league—and this might sound complicated, but it isn’t—I calculate the coefficient of variation of these statistics across players who engaged in one hundred or more batter-pitcher contests in a season. The coefficient of variation is a measure of the standard deviation relative to the mean value.
⁵⁰

The standard deviation is the average difference of the observations from the average value of the sample. Thus, the standard deviation of OPSs and ERAs of players tells us the typical talent spread from very worst to very best. A smaller standard deviation means that players are more uniform in ability, and a larger standard deviation means that players are more various in ability. The coefficient of variation normalizes the standard deviation as a percentage of the mean of the statistic, so that players can be compared no matter what era they came from or whether they are pitchers or hitters.
⁵¹
A greater coefficient of variation means a wider distribution of performance among all players. As the coefficient of variation decreases, the difference between the best and worst players shrinks. A greater talent dispersion on one side of the ball gives good players on the other side of the ball more opportunities to dominate bad players. For example, a wide dispersion among pitching talent gives hitters more plate appearances against exceptionally bad pitchers, and a greater dispersion of batters gives pitchers more chances against weak hitters.

Figure 5 shows the fluctuation of both statistics over time. The variation of both statistics is quite high until the 1920s. The population-to-player ratio is not much different from the previous decades, so it is likely that this is not the cause.

In 1920, something significant did happen in baseball that changed the game forever. Ray Chapman was hit by a pitch and killed. The pitcher that day, Carl Mays, blamed the incident on a slippery ball. Prior to the 1920s, the baseball was not the bright white ball of today, but a greasy, scuffed, tobacco-stained germ magnet. In response, baseball forced pitchers to throw with a clean ball, which eliminated one very special pitch known as the spitball. A few pitchers used only the spitball, which was a method distinct from the hard-throwing pitches still used in the modern game. Spitball pitchers were phased out of the
game over the following decade—many thought it was unfair to deprive pure spitballers of their livelihood. Removing this pitching method likely reduced the variance of hitter and pitcher performance, by making all pitchers more similar. Therefore, it is not surprising that the deviation of pitcher and batter performances began to fluctuate less with the spitball removed from play. Since the 1920s, the present hitting and pitching dispersions have varied over time, but patterns are difficult to identify from the yearly data.

For ease of interpretation, Figure 6 displays the average talent dispersion for pitchers and hitters of each decade to the average dispersion from 1920 until 2005. A higher bar on the graph means a greater dispersion of player talent relative to the historical average, and a lower bar means a greater similarity of players. Compared against baseball history, the current era has a greater dispersion of pitching talent in the entire decade, while batting talent is still quite compact. There is also a distinct trend of dispersion increasing for batters and pitchers after the 1980s. In particular, the variation in the ability of batters in the 1980s was at an all-time low.
⁵²
In any case, though the dispersion of talent began to widen some prior to the 1990s, the two rounds of expansion
in this decade dramatically diluted the talent pool. This is a crucial factor in understanding the game today.

What does this say about the current era of baseball history, especially given the intense public focus on steroids as the cause of the increased number of home runs? Pitching talent is more dispersed than it has ever been, while hitting talent is still quite concentrated. It means there are plenty of batters out there who are able to take advantage of bad pitchers. A more dispersed pitching talent pool gives the best hitters greater opportunities against weaker talent, which ought to lead them to perform extreme feats. Since the last round of expansion in 1998, Roger Maris’s home run record of sixty-one has been surpassed six times by three men: Mark McGwire in 1998 (seventy) and 1999 (sixty-five); Sammy Sosa in 1998 (sixty-six), 1999 (sixty-three), and 2001 (sixty-four); and Barry Bonds in 2001 (seventy-three). While there has been much speculation that this outburst was aided by performance-enhancing drugs—and this does not mean it was not—it is not surprising that these great feats occurred given the increasing pitching talent-dispersion of the league.
⁵³

Lending support to this idea is the corresponding surge in hit batsmen, which began in the 1990s. Figure 7 shows that the number of hit batters has risen along with home runs since the early 1990s. The current era of baseball could just as easily be referred to as the hit batter era rather than the home run era. And most certainly, no one believes steroids are responsible for the rise in hit batters, yet it’s probable that they have a similar cause. While there are certainly many other determinants of hit batters, the increase in relatively less-skilled pitchers taking the mound is likely a contributing factor.

Also, if low-quality pitchers are hitting more batters, shouldn’t better pitchers benefit from facing more bad hitters as hitting talent has become less compressed since the 1980s? In fact, they do, as pitchers increased their strikeout rates over the same span (see Figure 8). Since 1993, a pitcher has struck out three hundred or more batters eleven times. Randy Johnson did it six times, Curt Schilling did it three times,
and Pedro Martinez did it twice. In the thirteen seasons prior to 1993, only two pitchers (Nolan Ryan and Mike Scott) reached the three-hundred-strikeout plateau. As we see with home runs for hitters, there are a few people at the extreme end of the talent spectrum reaching grand achievements.

The fact that talent dilution may be part of the cause for the great performances of players does not mean that steroids have not influenced the game. However, it’s certainly incorrect to say that steroids, or other performance-enhancing drugs, are the only explanation.