MLS: The parity league, even in the Designated Player era

Continuing my look at the finances and on-field results of Major League Soccer, I expand my earlier financial determinism study to the period between 2007 and 2011.  This period covers the entire Designated Player era.

As in the previous study, I use base salary information provided by the MLS Players Union, but converted club payroll figures to 2007 US dollars.  See this post for further details.  Incidentally, it's been brought to my attention that the salary figures from the Players' Union may not necessarily cover the true payroll costs of MLS clubs.  Is that accurate?  You're welcome to illuminate me if you're in a position to know — correspondence kept in strict confidence, of course.

Below is the scatter plot of payroll (in 2007 USD) and points per game for all teams that competed in MLS regular seasons from 2007 to 2011, a total of 75 observations.  Overlaid on the scatter plot is the regression line.


The intercept is 1.281, which corresponds to 44 points in a 34 match season, and the slope is 0.028, which means that for every $1 million in payroll (real 2007 dollars), the increase in points-per-game is 0.028.  That figure is not statistically significant, however.  Or greater interest is the R2 term, which indicates that just 2.7% of the variation in PPG over the past five seasons can be explained by team payroll.

In my previous post a number of commenters suggested that I remove the outliers from the regression and observe the changes in the terms.  Well I did do that and didn't see much of a difference, but for the sake of the argument let's remove the team payrolls above $5 million (2007 dollars) and recalculate:


With the reduced data set, the R2 term drops to 0.000.  That is, if you only look at teams below $5 million in real dollars, payroll has no relationship with on-field results.

Now the next question is whether there exists a payroll-performance relationship below $3 million.  Below is the plot for that data subset:


There is very little correlation between the payroll of this subset of teams and their on-field performance.  In fact, there is a very slight negative correlation — R2 is about 0.011.

For comparison with other American sports leagues, here is the payroll vs performance scatter plot for the 2010 NFL season (data from the Pat Hruby post on ESPN):

NFL_2010The R2 of the 2010 NFL regular season is .105, not too different from the .113 of the 2011 MLS season.  It must be stated that in 2010 the NFL was operating without a salary cap while a new CBA was being negotiated, so there is a much wider range between largest and smallest payrolls than in a typical season. (There is a salary floor of $108 million and a cap of $120 million in the 2011 season.)  So it's possible that in a given season the correlation between payroll and performance in the NFL is even smaller.

One could do similar analyses for the NBA and NHL, but I'll leave that to analysts who focus on those sports.

So whatever way you slice the data, Major League Soccer remains very much a parity league in the Designated Player era.  One suggestion has been to remove the Designated Players from the payroll analysis, but then you would have to remove their impact on league results, which is not at all a simple task.  Nevertheless, the data go a long way toward demonstrating that success in MLS depends on many factors, very few of which have to do with payments to players.