Market size and the demand for talent in major league baseball

Abstract

In this article, we look at how revenues affect the personnel decisions of Major League Baseball (MLB) teams. It is well established that teams with the strongest demands end up with the top stars and deepest benches, thus the best chance of winning. Since a team's demand for talent is its Marginal Revenue Product, the critical test is whether large-market teams have a greater Marginal Revenue (MR). Controlling for the impact of re-distributional efforts by MLB, we find that the MR of a large-market team is about 50% larger than that of a small-market team. Furthermore, we find that re-distributive efforts have a more severe effect on small-market teams.

“Are the New York Yankees a dynasty because they outsmarted everyone? No, they just outspent everyone.” Sam Smith, Chicago Tribune

Acknowledgements

I would like to thank the many individuals who commented on this article, including the participants at the 2002 Western Economics Association meetings, the Chicago Sports Economists, David Surdam, Larry Hadley and David Berri.

Notes

¹This large-market advantage may or may not be detrimental to competitive balance. As will be seen shortly, we show that teams with greater demands for talent will acquire more talent, hence win more games. If competitive balance is to be interpreted as equal wins by all teams (on average), then indeed competitive balance so defined will be harmed.

²By construction, models which define talent in terms of units increasing winning percentage by one point (discussed earlier) implicitly assume that a player's MP is identical across all teams.

³This depressing effect on rich and poor teams’ MRs of the luxury tax and revenue sharing is discussed in greater detail in Gustafson and Hadley (1996) and Fort (2003).

⁴Since we defined winning percent as the team's output, a more intuitive notion of MR might be the additional revenue of winning one more game. Since one additional win raises WINPCT by slightly more than six points, the MR of one more win is six times the quantity in Equation 4.

⁵If the error term in the WINPCT equation is correlated with the error term in the REV equation, however, then a more efficient estimator is Zellner's Seemingly Unrelated Regression (see Zellner, 1962). We thank Dave Berri for pointing this out to us.

⁶Some small-population teams, such as the Atlanta Braves and Seattle Mariners, have fairly large revenues. As such, one might be tempted to define market size on the basis of team revenues rather than population. For estimation purposes, however, such a bifurcation is not appropriate as this would amount to splitting the sample on the basis of the size of the dependent variable.

⁷An alternative to the BRP data is that provided by Forbes magazine (and Financial World, until its demise in 1996). The problem with the data reported in these periodicals is that they are estimates derived from a variety of sources and methods, rather than recorded directly off the teams’ books. While one does not find large differences between these two sources of revenues (see Table 1), the Forbes data tends to give higher estimates for the large-market, and lower estimates for the small-market, teams.

⁸We also considered a series of dummy variables equal to one if the stadium age is less that 1-year-old, less than 2-years-old, etc. We found that new stadiums significantly increased the revenues for small-market teams, and levelled out after about 5 or 6 years; yet it had no appreciative effect on large-market teams.

⁹While the specification given in (3) ignores other potential determinants of demand (e.g. city income, variables associated the uncertainty of outcome), these variables do not vary sufficiently within the sample. To the extent that these variables grow across time, the trend variable will control for these effects.

¹⁰Porter (1992) includes population in his attendance equation and finds that teams in larger cities sell more tickets than teams in small cities. Our results parallel his in that we find that lagged winning has a much larger effect in large cities, probably due to the greater importance of season ticket sales.

¹¹These conclusions come from noting that the coefficient on NEWSTAD for small-market teams is around $30 million per year; over 5 years, this amounts to a discounted value of about $140 million. Large-market teams, however, do not receive anywhere near as large of a windfall – their 5-year discounted value amounts to only about $35 million. In Hadley and Poitras (2004), the authors find that a new stadium brings in roughly $130 to $140 million in additional discounted revenues over 5 years. Coffin (1996) finds that a new stadium generates about $85 and $175 million in incremental revenues.

¹²Their use of local, rather than total, revenues would be expected to inflate the imputed MRs because it ignores the impact of re-distribution payments. Converting their estimates of contender MRs given in their Table 4 to 1996 dollars, and using our list of large- and small-market teams, their estimates of MRs are fairly close to ours.

¹³Admittedly, this attempt at measuring a player's MP is unique only in that it uses a much more complicated method of isolating the player's contribution to team wins. It still implicitly assumes a counterfactual; in this case, the TPR is computed relative to the average player at his position across all 30 teams.

¹⁴Since Rodriguez signed his new contract in 2000, our estimates of MRPs had to be adjusted up by 10% to reflect 2000 dollars (rather than the 1996 dollars reported in Table 2), giving a MR of one additional win to a large-market teams of $2.086 million. Given Rodriguez's TPR of 7.7, this implies a MRP of (7.7) × ($2.086) = $16.1 million.