Payroll and Wins

Dave Berri, coauthor of The Wages of Wins, is today’s DH at The Baseball Analysts. Berri, an economist at Cal. St.-Bakersfield, answers the question: Can Money Buy Love in Baseball?

Contrary to popular perception, payroll in professional sports is not strongly linked to wins. A $100 million team does not win twice as many games as a $50 million team – not even close. Our own work has shown that only about 18% of a team’s regular season wins can be attributed to its payroll. In other words, more than 80% of a team’s regular season record cannot be tied to team spending. We would add that this is what we see when we look at teams in Major League Baseball from 1988 to 2005. In other words, the lack of a link between spending and wins is not a recent phenomenon. Across time more spending is not an elixir that leads automatically to success on the field. As the saying goes, games are not won on paper. Moreover, they are not won just because you spent a pile of paper.

Also, check out the site for the book. I received a copy of the book recently, and I have been loving it. Berri, Schmidt, and Brook do some of the most interesting work in the field of sports economics. I think the book is going to be a big hit, so pick up a copy of the book when it comes out in May. And look out for the authors’ blog, which is just getting started.

4 Responses “Payroll and Wins”

  1. Cyril Morong says:

    Where do they get the 18%? Is that an r-squared from a regression? What went into the regression?

  2. Marc says:

    But wouldn’t 18% be a significant amount in baseball, where there tends to be regression toward the mean (to use my limited statistical vocabulary)? That would be the difference between 100 and 82 wins. That seems pretty significant to me.

  3. David Gassko says:

    Okay, the 18% isn’t explaining “wins” but rather variance in wins. In 2005 (I don’t know if this is representative of other seasons), the standard deviation in wins was 10.834, meaning the variance was 117.38 (any differences due to rounding in this post). Multiplying by .18 gives us a figure of 21 wins, which is extremely significant. If we convert that back into standard deviation terms, that’s 4.6, meaning that payroll explains less variance than random luck, which happens to explain 34% of variance (1 SD = 6.35 wins). I haven’t actually loooked at the study, of course, so I don’t know how “right” the authors are in their estimate.

    Here’s what I will say: A decent general manager should be able to get an extra win on the free agent market for $3 million (even less if it’s a bad team). One SD in payroll last year was $34.234 million. If we change the Yankee payroll to a more reasonable $130 million, it goes to $25.416 million, which is the figure I’ll use. In that case, one SD = 25.416/3 = 8.5. Whereas the authors find that an increase of 1 SD in payroll is equal to 4.6 wins, in a “smart” market, it would be worth more like 8.5 wins. Bad GMs/the Yankees, are the reason the r^2 is .18 and not .62 or so.

    (Note: I’m not a mathematician so my math might be off in some parts here. Hopefully, JC can correct me if it is.)

  4. Vince says:

    Our own work has shown that only about 18% of a team’s regular season wins can be attributed to its payroll.

    I realize that to get this down to 0% is probably impossible, but that should still be the goal. I interpreted this as evidence that high-payroll teams have a decided advantage and baseball’s economics are still screwed up.