JC Bradbury – of Sabernomics fame – has offered two comments this week related to items posted in this forum. The first is on the discussion of payroll and wins in baseball. The second comment is on the latest from Michael Lewis – Blind Side.
Bradbury on Payroll and Wins
Yesterday Bradbury chimed in on the debate – and it is hard to believe this is a debate – about the link between payroll and wins in Major League Baseball.
Just to review – as Stacey reported earlier in the week – we find in The Wages of Wins that payroll explains about 18% of the variation in wins. On the flip side, this means that 82% of the variation in wins is not associated with payroll. JC offers a bit more sophisticated analysis than our simple regression of winning percentage on relative payroll, but his conclusion is the same. JC finds that payroll does not explain much of the variation in wins and furthermore, the predictive power of payroll is quite low.
As JC notes, this does not mean that payroll and wins are not statistically related. It simply means that there is much more to wins than how much a team spends on its players.
In The Wages of Wins we offer an explanation for the weak link between pay and wins. Our explanation centers on predictability of future performance in baseball.
Now this is a different point from the Moneyball observation. The Moneyball story centers on the ability of decision-makers to understand productivity. We do not touch upon that issue with respect to baseball. We do comment on this issue with respect to the NBA, and for those interested in an academic treatment of this subject in baseball, please examine the work of Jahn Hakes and Skip Sauer (recently noted at The Sports Economist).
What we do observe is that even if you evaluate productivity correctly in baseball, predicting future performance is still difficult. Baseball players are not perfectly consistent across time. Although we could discuss the regressions showing this point, just think about the Detroit Tigers this year. Did anyone – including the Tigers – think that team was going to come close to 95 wins?
Basically, we argue that if you have trouble forecasting then payroll – which is based upon expectations of the future – will not be strongly linked to wins.
Bradbury on Blind Side
JC also commented this week was on the latest from Michael Lewis – Blind Side. A few weeks ago I offered a comment on this work. Once again, though, Bradbury’s analysis is just a bit better. In addition to offering a better summary of the work, Bradbury gets at the larger issue. Specifically, the NCAA reacted to the tragic story of Michael Oher and his rescue by the Tuohy family by launching an investigation. Given the life story of Oher – which is brilliantly recounted by Lewis – this appears to be a heartless reaction by the NCAA.
As Bradbury notes, the NCAA appears to be primarily about the exploitation of labor. Malcolm Galdwell and Jim Peach noted a few weeks ago that the NCAA does not promote competitive balance in sports. It does impose policies that underpay its top athletes. This underpayment results in a transfer of revenues to the coaches and schools that employ the so-called student athletes.
Bill Cherry has a website called College Sports by Charlie (he notes the name has no meaning so don’t ask who Charlie might be) that reports the salaries paid to the top college coaches. As Cherry reports, more than a dozen coaches are paid $2 million or more and Charlie Weis and Pete Carroll make more than $3 million. To put this in perspective, the Chancellor of the 23 California State University system – which I believe is the largest state system of higher education in the country – makes less than $400,000. Yet there are football coaches who make five times this amount. Clearly football is generating some revenue and currently a significant share of this money is going to the coaches.
Of course people don’t tune in on Saturday afternoons to watch the coaches. People are watching the players. But the players do not get much of the money their labor generates. By definition, this means these players are exploited. Given the disadvantaged background of many of these athletes, the NCAA is basically conducting a reverse Robin Hood – taking money from the poor (the athletes) and giving it to the rich (coaches and schools).
Bradbury on Being Civil
JC makes one last comment in concluding his discussion of payroll and wins.
I’d also like to ask for politeness among those arguing. The tone of the response has been too harsh to be productive. If we are all after the truth, it’s best to keep things civil.
In my experience, academics tend to respect each other when engaged in debate. This behavior is probably motivated by self interest. If you are not nice to other people – who may very well be much smarter than you – you might feel very stupid when they decide not to be nice to you.
I will note that when comments are not particularly civil, we are not inclined to respond. Debates are exchanges of information, not contests. Try and stick to your point and to your evidence. And remember, this is just a discussion of sports. As we note in the first paragraph of The Wages of Wins: Sports are entertainment. Sports do not often change our world, but serve as a distraction from our world.
In other words, although this research is interesting and does shed light on how people make decisions, sports by themselves are not actually “important.” So if you are getting upset by what people say about sports, well, maybe you just need to expand your horizons.
– DJ
Wages of Wins Journal 
Guy
November 10, 2006
“it is hard to believe this is a debate”
I think the authors find it hard to believe it’s a debate because they and those of us who have made criticisms of this analysis are talking past each other. No one is challenging the accuracy of the .18 R2, nor the resulting conclusion that payroll “explains” 18% of the variation in wins. So when you, or JCB, restate those conclusions — no matter how plainly — it is not (to us) responsive to the points being made.
The question being raised is whether those results support the broader conclusion that “you can’t buy wins” in baseball, or your contention that Costas is wrong when he says that money is a primary determinant of “success” in baseball. We could just say these are subjective judgments, and agree to disagree. But there are real-world implications that logically follow from your analysis, and I don’t think they’re correct.
One is that payroll disparities among franchises shouldn’t worry us, since they have little impact on team success — a point you’ve made explicitly. Let’s look at that. If we rank the 30 teams by avg. payroll over the years 1992-2005 (borrowed from Tom Tango at The Book website), and divide them into quintiles, we get these W%;
Top .538
2nd .510
3rd .504
4th .481
Bottom .459
The top 6 spenders are outspending the bottom 6 by over 2 to 1 (138% vs. 64% of average payroll) to win about 13 more games. To me that looks like a substantial payroll impact, but one could certainly argue that’s not very many wins for all that money. But now lets look at the average number of division championships (and WCs) teams won over 13 seasons by those same payroll quintiles:
Top 6.2
2nd 3.7
3rd 2.7
4th 2.5
Bottom 1.0
Now the story looks different: spending twice as much money increases my chances of winning my division six-fold. Spending 38% more than average doubles my chances. That sounds like a pretty good investment. And I think 9 in 10 baseball fans will tell you that it is winning championships — divisions, pennants, and the WS — that defines “success” in baseball.
These results are why, if you told the fans of the six franchises at the bottom that “you can’t buy love in baseball”, you’d probably find it a hard sale. Or if you imposed a 50% payroll tax on the teams at the top, should they just laugh it off and feel confident it won’t reallyhave much affect on their performance? I don’t think so.
Obviously, there’s a lot of luck in baseball. A team will often finish +/-6 wins compared to its true talent just by chance. What that means is a team that wants to regularly make the playoffs needs to build a substantial true talent advantage over other teams. A huge payroll makes that far easier to do (though not guaranteed, of course). A very small payroll makes it almost impossible. In the middle, a combination of luck and skill can definitely yield success (though not as often as for the top payroll tier). Still, we’re not just talking about Tampa and Pittsburgh here: only one team with a payroll less than 94% of league average has a record over .500 in this time (Oakland).
Another implication of your thesis, I would think, is that when teams want to improve, there are things other than raising payroll that are more likely to help them succeed. One of those is “get lucky”, but I think we can all agree that isn’t helpful advice. (And if that is the only other major determinant of success, then teams should indeed start spending — what else could they do?) But if there are factors other than luck, that would be very interesting and important to know. But as best I can tell, your analysis doesn’t shed any light on this. And an R2 of .18 doesn’t necessarily prove there even ARE any such factors that are under the control of baseball teams. For example, a huge explanatory factor is of course the performance of non-free agent players, whose compensation is not proportional to their talent. But if finding and developing such players is itself a crapshoot, then expanding payroll is STILL the best way to improve your team. So unless/until you can identify other factor(s) that are as or more important than payroll, I don’t think you’ve made your case.
Guy
November 10, 2006
P.S. The r for payroll and championships in these years was .68 .