Salary…. Again

In a NY Times article on the lockout, an economist is quoted as saying “The statistical correlation between payroll and win percentage is practically nonexistent”.  This was picked up on by the APBR board as well as the RealGM forum (which I have to say I don’t typically read).  I honestly, 100% don’t understand how this issue keeps coming up.  Maybe one more walk-through will help.

Is there a connection between payroll and winning? Yes.  Across seasons there seems to be a correlation of at least .2 between team spending and wins, depending on if you use relative salary or not (to account for differences in the salary cap across years) and some other details; the correlation seems to be higher in some years than others but nothing consistent.  So, taken literally, what Zimbalist said was wrong.

Does spending more make you win more?  This is really what people want to know.  Teams like the Mavericks and Lakers run over the cap seemingly every year and this presumably gives them an unfair advantage compared to teams that can’t or won’t spend as much.  Taken at face value, the correlation just mentioned might lead you to believe that spending more does mean you win more.  However, as I discussed in detail before, this is misleading.  Teams win because they have better players than other teams.  In theory, teams would also thus pay better players more since there is more demand for their services and more profit from their output.  But the size of the correlation (i.e. never larger than .5 and change) tells us that salary doesn’t explain a lot about why teams win; three quarters or more of the differences across teams is due to factors beyond salary.  This is true both using some statistical knowledge (salary is an imperfect stand-in for team quality, and so it has some correlation but wouldn’t if quality were accounted for) and using a simple thought experiment.  If the Pistons gave each player on their team a raise, would they win more games next year?  The answer is obviously “no”; the Pistons would still have crummy players.  So spending money by itself obviously cannot make teams win games.

What did Zimbalist really mean?  I don’t want to put words in his mouth or give him credit if he isn’t due, but I assume he meant something like what I said above: if you account for the players on a team, there is no connection between salary and wins.  Additionally, this does not seem to be an NBA-specific problem.  As the Wages of Wins points out, the correlation between salary and winning is low in the NFL, MLB, and (I think) the NHL as well.  The issue is that teams apparently aren’t especially good at evaluating talent (they can to some degree, but not well; otherwise you would expect a higher correlation than what we see), or to be more generous they may pay for things beyond talent (perhaps they pay for players perceived as superstars to draw season ticket buyers even if they know the player isn’t superstar quality, or they pay players who have been on the team a long time out of loyalty even though their production might drop off).

Is this the Knicks’ fault?  The Knicks have, at least in recent years, been notorious for spending a lot without winning a lot (for example).  If we ignored them, maybe salary explains wins then?  I ran the numbers across 10 seasons and the correlation does increase, but only from .256 to .377.  That seems like a sizable jump, but it still means that salary can only explain 14% of the differences across every team that isn’t the Knicks.  This isn’t an outlier issue.

Is this Oklahoma’s fault?  One of the complaints I saw a few times on the RealGM forum is that teams like the Thunder, who have good players but low salary because those players happen to be rookies, skew the results.  This is silly; they are part of the results.  Part of the reason that salary has a low connection to wins is a poor eye for talent, but part of it is also the artificial boundaries placed on player contracts.  If a team picks an immediately productive player in the draft, they don’t have to pay him market value.  So the Hornets can benefit from Chris Paul, the Spurs from Tim Duncan, the Cavs from LeBron James, and so on, for a few years before they have to pony up more money.  If you draft one of those guys, you don’t even have to get many other productive players to win a fair number of games; you can be competitive with a low team salary.  So yes you could win a lot of games by identifying good players and paying them a lot, but you can also win a lot of games by drafting good players and paying them like crap.

I’ll note that there were a couple of useful posts in the RealGM forum.  Someone looked at changes in spending versus changes in winning from year to year.  You would expect that if teams could identify good players and then get them, teams that raise their salary should win more.  But the correlation was only about .1.  Again, pretty tiny.  Presumably this is again because teams have trouble identifying good players, and also probably because players’ salary increases as they’ve been in the league but they get worse with age.  This obviously doesn’t apply to rookies, since young players will earn a bit more and get better in general, but there are only so many young players to counteract the decline of older players, and I would guess (although I don’t know for sure) that year-to-year jumps in salary are larger for players in second and third as opposed to rookie contracts.

So if the owners as a group wanted to increase parity for some reason (even though it would probably not affect revenue and may be impossible), maybe they could focus on better methods for evaluating free agents and college players.  Then they could enact a crazy draft like the one I described in a previous post and try to give each team a better chance of winning.  Or, the teams that don’t want to spend as much could try to exploit knowledge the other teams don’t have, like the Moneyball A’s.  If you can evaluate talent better than the other guys, you don’t have to pay as much to get it.

So let’s recap, so hopefully the key points stick.  Salary is correlated with winning, sure.  But it isn’t large, because salary is really a poor proxy for talent.  Teams that pay a lot will tend to win more games because broadly speaking they’ll have paid for a good player or three, but it isn’t a guarantee.  Reducing how much teams can spend on players is not going to even the playing field, it’s just going to reduce how much bad players get overpaid while they still take up unreasonable percentages of a team’s salary cap.  And that’s that.


This entry was posted in Uncategorized and tagged . Bookmark the permalink.

One Response to Salary…. Again

  1. Pingback: What Can the NHL Tell Us About the Salary Cap? | Sport Skeptic

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s