How financially deterministic is MLS?

One of the theses of Kuper and Szymanski's book Soccernomics is that team performance in a football league is strongly determined by its payroll; in other words, clubs with highest payrolls finish at the top of the league more often than not.  I've seen publications that state the correlation coefficient between payroll and points in the English Premier League is around 75%, and perhaps there are similar results in other European leagues (definitely Spain, to some extent Germany). 

I decided to look at the payroll figures for the recently completed 2011 MLS regular season to determine if the same principle applies here. 

The numbers that I am using are supplied by the MLS Players Union and reflect base salaries of the players.  Point totals for the teams are easily obtained from the league website, Wikipedia, or other site.  To make the numbers easier to work with (and facilitate comparison with other seasons), I divided team payrolls by a million dollars and team points by number of matches played.  Then I performed a simple linear regression:

Below is a scatter plot of the team payroll and points, overlaid with the regression line:

The dots on the far right side of the plot correspond to LA Galaxy (team payroll $12.8 million) and New York Red Bulls ($12.6 million).  Toronto FC have the third highest payroll in MLS at $5.6 million.  It is quite clear that there's not much of a correlation between salary and points per game in MLS, and for completeness here are the regression coefficients:

a0: 1.190 -- about 38 points for zero payroll (34 game season)
a1: 0.033 -- points per game added per $1m payroll, not statistically significant though

The R² term for this regression is 0.1127, which means that 11.27% of the variance in the points-per-game data is explained by the team payroll.  It's a far cry from results seen in European soccer leagues, or even Major League Baseball for that matter where around 35% of the variability in the standings can be explained by team payroll. 

It's important to remember that MLS has had the Designated Player Rule since the 2007 season, which permitted clubs to make purchases for big-name and high-value players and count a fraction of the salary toward the salary cap.  You see more striations in the payrolls in the league then ever before but most of the team payrolls remain within a narrow band.  There are a few follow-on questions -- first, does the low correlation between pay and league standings hold up from 2007 until now?  And second, if payroll does not adequately explain the differences between teams in the standings, what does?

So how financially deterministic is today's MLS?  Not so much.

MENTIONED: I see that this post was mentioned by Bobby McMahon in one of his recent web roundups.  Thanks for reading, Bobby, and welcome to those readers!  Also got some attention from BigSoccer. Someday I will get an account and start commenting over there.

Comments

[Patchstats.blogspot.com]

I think what you got here is something. However, we need to remove the two outliers LA and NY here, because they are on a class of their own. Also, the regression will truly only work if we get some form of Normal data, now we could transform the data to get it to look normal and then do the regression accordingly. If we leave those two outliers in there, the errors will not be normal which would void one of our assumptions, thus making the regression somewhat pointless anyway. The problem is we need to use non-parametric statistics, because the teams dont have the same distribution, they are not iid. There are large clubs, and there are small clubs, and the way they spend is clearly different, and thus their outcomes are different. We need to find a standard way to calculate them, then do it as such, but treating every club the same will create some bias, and large variance.

~Josh Patchus
patchusj@gmail.com

[Howie_OT]

I would be curious to see things without Los Angeles and New York.

I'd also be curious to see if there is anything with median salary . Close to half of Kansas City's roster makes over $100,000. New England is around seven guys. Using that might help offset the teams who have individual players making as much or more than some teams.