With the Holiday Season upon us it is time to start thinking about what gifts to give and what you hope Santa brings to you. The answer to both questions might by Hot Stove Economics.
J.C. Bradbury’s latest combines both economics and the study of baseball statistics in an exploration of the decisions baseball teams make in the off-season.
Of course, most readers in this forum are basketball fans (I think). So why should you read a book about baseball? One issue I have emphasized previously is that we can increase our understanding of what is happening in basketball by looking at other sports.
This point is illustrated below. Recently I sent along a few questions to J.C. about his book. If you are a basketball fan, read these answers and see if you can see the differences and similarities between the two sports. And then, go buy this book (or put it on your list for Santa)!!
1. Let’s start with the obvious question. What is your book essentially about?
I’m trying to estimate what baseball players are worth, or to use Branch Rickey’s description, “put a dollar mark on the muscle.” Building off the work of many other economists who have attempted to value baseball players, most notably the late Gerald Scully, I use player performance and revenue estimates to determine what wins are worth to teams and what players contribute to wins. Once you know those two things, you can impute what individual players are worth to teams. In the book, I fully explain and justify my methodology so that interested readers can observe and understand what I have done.
2. There are many sabermetric books, or books looking at statistics in baseball. How is your book different from the other books in this area?
Well, I use some sabermetrics in my book for valuing players. Bill James, John Thorn and Pete Palmer, and Voros McCracken all made important contributions to helping us understand what things players do to help teams win. I don’t dwell on sabermetric questions. My main goal is to understand the business relationship between play on the field and financial success. Sabermetricians have used some financial models to connect player performance and worth, but these simple approaches are too limited to proxy the impact of performance on revenue. What’s missing from sabermetric value assessments is economics. I approach the problem using common tools of labor economics, which has been missing.
3. When we think of baseball and economics we often think of Moneyball. That book by Michael Lewis argued that baseball’s labor market is inefficient. Do you think that was historically true? Do you think that is true today?
Well, it depends on what you mean by inefficient. I think the baseball market is always trending toward efficiency, but the market process of allocating resources is always going to have bumps. Michael Lewis identified some inefficiencies that existed when he wrote the book. Later, economists Jahn Hakes and Skip Sauer found that while the undervaluation of on-base-percentage did exist at the time, that inefficiency had almost completely disappeared by the time the book came out. There is no honey hole of undervalued talent that teams can return to time and again and expect consistent success. Competition among teams is fierce, which is why teams are always on the lookout for new inefficiencies that they can exploit.
4. Related to the previous question, do most teams in baseball employ “advanced statistics” in making decisions? Do you know which team do (or do not)?
Every team in baseball employs some sort of statistical analysis, and most have for many years. Just the other day I read a note about sabermetrician Craig Wright working for the Atlanta Braves in the mid-1990s. Many people view the Braves as an anti-stats team; yet, in the midst of their heyday, they had a leading sabermetrician working for them. You can’t succeed in building a winning team in baseball without good scouting. Stats can be used to improve scouting and find hidden gems that are not immediately obvious. Teams understand this and have ratcheted up their scouting departments to employ stats-based analysis. This is a trend that predates Moneyball.
5. Your book reports work you have done on the aging of baseball players. How is your study different from the studies offered in the Sabermetric community (in terms of methods and results)?
Within the sabermetric community, 27 has been considered the peak age of players. Some studies of players even found younger peak ages. But, the studies that underpin these estimates are biased, including many players who only got to play multiple seasons because they had lucky initial seasons. When you don’t properly account for this bias, you’re going to have many players in the sample who decline by chance rather than aging; therefore, it looks like players age more quickly than they actually age. When I look at a large historical sample of players over their careers and estimate an aging function, I find the expected peak age for players is 29-30. In the big picture, when players peak exactly isn’t all that important. From their mid-20s to mid-30s players tend to play their best baseball. And while all players improve and decline, good players tend to stay good and bad players tend to stay bad. Talent is more important than aging.
6. In your book you not only report measures of a player’s on-field production but also the value of each player in terms of revenue. What are the steps you follow in these calculations? How correlated are these estimates with player salaries?
First I estimate how performance translates into winning for teams. Much of this work was aided by work of sabermetricians. For pitchers, I estimate how many runs pitchers prevent based on Voros McCracken’s notion that pitchers have little control over hits on balls hit into play. For position players, I use John Thorn and Pete Palmers linear weights measure of offensive performance and John Dewan’s Plus/Minus measure to measure defense. All performances are denominated in runs.
I then use Forbes’ Business of Baseball revenue estimates to measure how much added or prevented runs improve team revenue. This generates a non-linear runs-to-dollars conversion that generates player values.
The estimates explain about 27 percent of the differences in salaries across free-agent pitcher and about 33 percent of the differences in salaries across free-agent hitters. This may seem small, but given the year to year performance fluctuations for players (30 percent for pitchers and 40 percent for hitters), the estimates predict well.
I want to thank J.C. for taking the time to answer these questions. And once again, I encourage everyone to go take a look at Hot Stove Economics.