Expansion Won't Help Run Scoring

MLB run scoring is down. A lot: [iframe src="http://cf.datawrapper.de/1SVSX/4/" width="650" height="450" frameborder="0" allowfullscreen="allowfullscreen"]

There are many causes of decreased run scoring. Jeff Zimmernan at The Hardball Times thinks that one solution is to expand the number of MLB teams:

Why should there be expansion? One reason is the run scoring environment. Run production has dropped from from a steroid-era high of 5.1 runs per game per team in 2000 to 4.1 last season. Pitching is now the dominant force in the game. The last four times the majors expanded, runs increased as the pitching talent was spread thinner. Looking at the two seasons before an expansion of teams and the two seasons after, the average increase in runs scored per game was a third of a run. Without expansion and if things remain static, scoring will likely go even lower as pitching talent becomes more and more concentrated.

This is an interesting, testable hypothesis: when there is a more concentrated talent pool of potential MLB players, run scoring decreases. So, I tested it. Below is a basic OLS regression with MLB run scoring as the dependent variable, and the number of MLB teams divided by the US population as an independent variable, controlling for the designated hitter. Here's what came out:


Screenshot 2015-01-30 23.33.05

Okay, don't get scared off by the scary statistics! The important thing on this table is that little P>t value next to 'teamsper.' That test statistic shows no statistically significant relationship between the ratio of MLB teams to the population and the number of runs scored per game. The designated hitter, on the other hand, is responsible for an additional .34 runs per game, all else being equal.

What really matters? Let's add strikeout and walk rates into our model:

Screenshot 2015-01-30 23.39.29


Now, we've got a model. 51.9% of the variation in run scoring can be explained by strikeout rate, walk rate, and the designated hitter. MLB teams per person in the United States still has no effect.

Here's what it looks like on a scatter plot:


Each dot represents a league-year (So, 2014 is split into AL and NL seasons). That's about as much of a non-relationship as you'll find.

An obvious caveat is that international talent increases the talent pool, so US population per MLB team doesn't do the best job of representing the concentration of talent. I agree that in a perfect world we'd include some measure of foreign talent, but I don't think the results would be different for a few reasons. First, athletes in the United States are playing more and more sports, so fewer are going into baseball over time. This offsets much of the increase from foreign talent. Second, the data overwhelmingly signal no relationship. Even with an imperfect measure, that is a strong null result.

I support expansion for a lot of reasons, but Zimmerman is wrong here. There is no evidence that it will increase run scoring.