Caught Looking will take a look at articles from the academic literature relevant to baseball and statistical analysis. Mostly this means recent articles from peer–reviewed academic journals, but suggestions are welcome, especially for interesting dissertation or thesis chapters that aren’t always easy to find. Critique will come if it’s warranted, but the goal is to help share with a broader audience where the academic frontier is and to seek ways to move it forward.

This inaugural review looks at Moneyball After 10 Years: How Have Major League Baseball Salaries Adjusted? by Daniel Brown, Charles Link and Seth Rubin in the October 2015 Journal of Sports Economics.

Michael Lewis’ Moneyball: The Art of Winning in an Unfair Game made a big splash when it came out in 2003 and shook up the baseball and sporting world. It might be a bit much to say that it caused a change in the way that executives, coaches and fans watched and understood baseball, but it certainly coincided with one. In Moneyball After 10 Years: How Have Major League Baseball Salaries Adjusted? Daniel Brown, Charles Link and Seth Rubin set out to try to identify whether insights from the book led to changes in the links between player skills and player salaries.

They focus on the contention from the book that on-base percentage was undervalued in the marketplace, and examine whether this changed after 2003. Economists don’t generally expect to see unexploited profit opportunities, especially in an industry as competitive as major-league baseball, and in fact have studied the links between pay and performance since at least Gerald Scully’s work in 1974. The importance of on-base percentage in the production of runs was documented by Bill James and Pete Palmer among others by the mid-80s. Yet along came Billy Beane, armed with some not particularly new information on the link between skills and wins, and on the link between skills and pay, and put them together to turn the A’s into a winner on a low budget.

Brown and his co-authors look at data from 1996 to 2011, both before and after the release of Moneyball, to see whether owners adjusted their behavior, bidding up the salaries of high OBP players relative to their peers, and indeed they find this to be the case. In yearly regressions, the coefficient estimate for on-base percentage on player salaries is sometimes negative, sometimes positive and not a significant predictor of player salaries between 1996 and 2002, except in 1998 for unexplained reasons. But it begins to appear more regularly as a significant predictor of player salaries for free agents beginning in 2003. OBP is statistically significantly linked to player salaries in 2003, 2004, 2005, 2007, 2010, and 2011 and has the expected positive sign in the other years.

The results presented here build on earlier work by Jahn Hakes and Skip Sauer that tested the same hypothesis using data from 1999-2004. The new paper confirms Hakes and Sauer’s finding of a Moneyball effect, and their use of a longer dataset with greater detail on contract lengths and signing dates allows them to examine how the market progressed toward a new equilibrium. Their evidence suggests that the shift to recognize the value of on-base percentage through player salaries in the labor market came quickly after the release of Lewis’ book. They also show, however, that the shift in player salaries is confined to the market for free agents and find no impact of on-base percentage on the salaries of players restricted by the reserve clause or during their arbitration years.

It is always difficult to disentangle correlation and causation, and this paper is no different. Perhaps something else was causing the shift. Around the same time the book was released, baseball was in the midst of the steroid crisis, and perhaps what we observed was a fear of spending big on sluggers that manifested itself in a shift toward patience. Alternatively, it’s possible that a number of organizations were on the verge of making a similar discovery. Information technology and access to data was exploding in 2003 in many industries, not just baseball, and while Lewis shined a light on one particular organization that was a bit ahead of the curve, it is quite possible the revolution was coming with or without the book. Indeed, Brown and his coauthors show that the change in the labor market came rapidly.

The authors nod to the difficulty of assigning cause and effect, but they don’t dig very deeply to try and sort things out. This is more observation than indictment, as finding a good strategy for isolating and identifying the Moneyball effect is a difficult problem without an obvious solution. One modest suggestion would be to try to identify which organizations had someone in the front office dedicated to analytics prior to the release, and to see how the labor demand responses of these organizations differed after 2003.

One puzzle I see from the current study is why there was no seeming change in the labor market for pre-free agent players, even 10 years after publication of the book. While I agree strongly with the author’s contention that the expected results should appear most strongly and most quickly in the free agent market, it would seem by now that the arbitration process would reflect the changes in the relative valuation of skills.

It would also be fun to take a look at whether some other Moneyball insights translated into practice in major-league organizations. Lewis spent a fair bit of energy describing the A’s draft process, and I’d be interested to see whether and how drafting strategies changed after the release of the book.

These, however, are questions for future research. What Brown, Link and Rubin have done in the meantime is to document an important change in the labor demand function in the post-Moneyball era. It certainly feels like data science is driving more and more of the conversations and decisions in baseball, and this paper provides clear support.

Michael Wenz works in the Department of Economics at Northeastern Illinois University.

Thank you for reading

This is a free article. If you enjoyed it, consider subscribing to Baseball Prospectus. Subscriptions support ongoing public baseball research and analysis in an increasingly proprietary environment.

Subscribe now
You need to be logged in to comment. Login or Subscribe
In regards to pre-arb salaries, what analytics are being argued at hearings and in negotiations? More teams may value OBP but are they downplaying that in negotiations? What about the agents who, presumably, would use advanced metrics to find any advantage for their clients (my assumption is that most agents are not using old school stats if they are arguing for a "Moneyball-type" player). Do arbitration panels still favor the "old school" stats and does this play into whether a team or agent is willing to go to a hearing, knowing that their OBP arguments may not resonate with the panel?

These questions make it unsurprising to me that pre-arb salaries haven't spiked the same way free agent salaries have.

Just some thoughts.
Good questions. Arbs should be based on comps though--players ought to bring them up in arb cases, but even if not, teams who value OBP ahold settle those cases more favorably.
My concern is there is no mention of Batting Average and it's relation to OBP. When analyzing pre and post-Moneyball trends you would need to isolate BA from OBP to tell which one teams were really paying for.
Would be interesting. You could pull ISO out of slugging too.
Looking forward to more in this series.