In the book Soccernomics, the authors make the point that there is a correlation between payroll spending and winning (although the same does not hold true for transfer fee spending). I was curious just how much correlation there really was. Remembering the Guardian published the wage bills for EPL clubs from the 2007-08 season, I thought it would be interesting to see how teams fared compared to what they spent on payroll. Here is the 2007-08 table, with the wage bill for each club. Note that the Guardian couldn't get the numbers for Liverpool, and didn't publish the numbers for the clubs that were relegated: English Premier League - 2007/08 1 Manchester United 87 points -- £121.1m 2 Chelsea 85 -- £149m 3 Arsenal 83 -- £101.3m 4 Liverpool 76 – Not Available 5 Everton 65 -- £44.5m 6 Aston Villa 60 -- £50.4m 7 Blackburn Rovers 58 -- £39.7m 8 Portsmouth 57 -- £54.7m 9 Manchester City 55 -- £54.2m 10 West Ham United 49 -- £44.2m 11 Tottenham Hotspur 46 -- £52.9m 12 Newcastle United 43 -- £74.6m 13 Middlesbrough 42 -- £34.8m 14 Wigan Athletic 40 -- £38.4m 15 Sunderland 39 -- £37.1m 16 Bolton Wanderers 37 -- £39m 17 Fulham 36 -- £39.3m 18 Reading 36 – N/A 19 Birmingham City 35 – N/A 20 Derby County 11 – N/A We don't have Liverpool's numbers, but the other Champions League spots are all occupied by teams that spent over £100 million on player wages. At the other extreme, seven clubs spent less than £40, and six of them are at the bottom of the table, just above the relegation line. The only exception was Blackburn, which finished a very respectable 7th on a limited budget. Six other teams spent between £44.2m and £54.7m, that is, something in between, which put them in the top half of the table (well, Spurs were 11th, but close enough) below the Champions League spots. The lone exception where big spending produced tepid results was Newcastle, which spent £74.6m for a pedestrian 12th place finish. Now, it's only one season, but it does look like a payroll over £100 bought a team capable of winning a Champions League spot, while payrolls under £40m bought clubs destined for the lower half of the table and a relegation battle. There's another way to look at this too, who spent above their means? Various proposals have been floated in Europe suggesting player wages be capped at 65% of turnover to keep teams out of financial trouble. The rule was actually adopted by the League of Ireland, although I'm not aware of any other country that has done so. Even so, I relaxed that standard a bit, figuring teams just over 65% probably could have gotten into compliance if they had to, and looked at teams were wage bills were 70% or more (sometimes a lot more) of gross revenues. Those spending a lot of their revenue on payroll are in italics, including Blackburn (just, to be fair), Portsmouth, West Ham, Middlesbrough, Newcastle, Wigan and Fulham. Knowing what we know about some of these clubs today, it isn't surprising that some were overspending in an attempt to keep up and got themselves into trouble. They were swimming against the tide though. Four of these six clubs were also among the 8 clubs that averaged under 27,000 in attendance that year (together with two other clubs that were relegated). Lower attendance and over spending seem to typically go hand in hand. By contrast, it's hard to argue most of the big clubs, with the notable exception of Newcastle and West Ham, were living beyond their means. We don't have the Liverpool data obviously, but of the top 10 drawing teams that year, only Newcastle spent more than 70% of its gross revenue on payer wages. At the very top of the table, Manchester United may be highly leveraged because of acquisition debt, but it isn't spending itself into trouble because of it's high payroll -- at least it wasn't in 2007-08. So, does the amount spent on payroll matter? Yes, for this season it appears it did. Did some of the smaller clubs overspend trying to keep up? Yes, several did. Link to the Guardian: http://www.guardian.co.uk/football/2009/jun/03/english-premier-league-debt __________ Note: I mentioned in another thread that I've got some new professional responsibilities and, as a result, I'm going to be taking a break from Big Soccer starting next month, so forgive me if my replies in this thread are limited.
I'd be interested if payroll spending makes a difference when it comes to the parity-driven MLS. I know we have a cap - presently in the neighborhood of $2.3 million - but how many teams spend all of that $2.3 million? How many spend less? How many spend more (if you include bonuses, etc.)? If you include DPs, how many spend significantly more? And how does all of that relate (if it does) to a team's position in the regular season standings? and, in the end, in their pursuit of the MLS Cup?
Given that the whole premise of the pyramid systems throughout the world shows the larger, well-supported, and higher spending clubs having a slight tendency to finish ahead of those less able to spend, there's little doubt the premise is true. What's less likely is a correlation when there are small differences. Some clubs are just better at picking up those "rough diamonds" than others. Some just have no idea of wage control. There's the famous tale of Seth Johnson signing for Leeds and expecting around £25000 a week and being offered £40000. He was supposed to have looked rather shocked, and Leeds offered him another £20000 a week, thinking their offer was too low.
Short answer: Not a bit (for the bolded part of your quote) Long answer: There is almost no correlation between salaries and regular season points in MLS. Because of LA's high total salaries, there is actually a slight negative trend, meaning higher total salary leads to fewer points. See the scatterplot here: BTW, you can totally ignore the regression statistics in the graphic. They are totally meaningless. How I Did This: I took the salary data for the 2007-2009 seasons from the MLS Players Union Website and the total points from MLSnet for the same seasons. There are 42 team-seasons (13 teams in 2007, 14 in 2008 and 15 in 2009) with associated salary data. I did not consider pool players (keepers) or players on Seattle's 2008 payroll, as the club did not play in MLS that year.
FYI, here is the raw salary and points data I used. Please correct any errors. Code: 2009 2008 2007 Salary Points Salary Points Salary Points CHI 5134229 45 4808167 46 4716304 40 CHV 2449694 45 2487846 43 1789645 53 CLB 2903550 49 2372680 57 2108226 37 COL 2430119 40 2515949 38 2281303 35 DAL 2695827 39 2553060 36 3072764 44 DC 3278865 40 4881668 37 2276390 55 HOU 2614142 48 2457094 51 2434147 52 KC 2246548 33 2551054 42 2637550 40 LA 9766006 48 9177665 33 9322054 34 NE 2778003 42 2215295 43 1945848 50 NY 3549338 21 3367589 39 4928089 43 RSL 2200041 40 2418376 40 1974249 27 SJ 2116203 30 2503382 33 SEA 3407212 47 TFC 4152167 39 2726478 35 2653987 25
Some more analysis: The salary cap number is an accounting fiction, IMHO. 12 of 15 teams last year spent more than the $2.3 million cap. BTW, one of the teams that didn't, RSL, won the MLS Cup. If you assume the salary cap was $2.2 million in 2008, then every team was above it, although New England barely cleared it. If you assume the salary cap was $2.1 million in 2007, again, only 3 teams out of 13 were below it.
One flaw in the calculation though is that there's no way to account for teams picking up portions of salaries for players that they traded away. In the grand scheme of things its probably a minor effect, but for example last season Colorado paid part of Christian Gomez' salary, rumored to be close to 200K of it.
The bigger issue is that you can't spend money efficiently (to improve competitiveness) in MLS. The extra $5 million that LA spends all goes to one player, which limits it's effect on competitiveness. Of course, that's the point. MLS doesn't want a system where money drives success.
That's an interesting point and one worthy of debate. If you exclude LA from the data set, the remaining 39 show even less of a relationship between payroll and points. Personally, I think it shows that every MLS season is a complete crapshoot for every team. Luck, injuries and the form of key players play a tremendous role in determining the season's outcome. According to Soccernomics, payroll spending accounts for 92% of the variation in points earned during the season in the English leagues. That's just staggering. It means you *can* buy your way to the title. Now that there are 5 teams with the money to compete for the Big 4 slots, I wonder if that number will drop. It would be kind of fun to build an econometric model of what works in MLS. Some variables I can think of are: payroll, # of matches played, age of forwards, age of midfielders, age of defenders, age of keepers, goals scored, goals conceded, keeper save percentage, shots, shots on goal
it's going to be higher in the premiership than the other divisions because the wage gap is so great. You only have to look at pre-season title odds, for example, to see that the 24th team's title odds in the championship are usually on par with the 6th or 7th team in the premiership. Leagues here also don't do player trades, which makes it much easier to improve a team by paying more, as you don't have to give talent away when you sign a player. Clubs can punch above their weight either by bringing through young players or finding cheap players - either young players rejected by other clubs higher up, or by buying players from lower down - because such players are often paid less than the league average. It's just hard to maintain that, as as they become established, those players will feel they deserve to be paid the going rate. Some managers are also very good at organisation and motivation, and will put together teams whose whole is greater than the sum of the parts. Perhaps you could also make a case that spending comparative fortunes on star players (in terms of their salary in relation to their teammates), as tends to be more common in the USA, makes correlation less likely. If you blow a large chunk of the budget on one player, that one player really needs to perform.
It would take a while but I thought of weighting salaries by number of games each player was available. This way, an expensive injured player (Renya) won't count against a team's spending. I've also tried using the standard deviation of salaries on a team as an explanatory variable. I expected the coefficient on it to be negative so that as it increased and teams had few high-payed players and many low-payed players, the team would perform worse. It was insignificantly positive I'd like to try again with the weights as I proposed but to make the data will take time.
Something else I found interesting: The regression equation the data produces is points =-0.3452 (wages in $millions) + 41.988 Oddly, the constant is statisically significant. That means for the basis of this comparison, the expected starting value for any MLS team is 42 points, which is higher than the intutive value of 40 points (10 wins, 10 draws and 10 losses). Indeed, the mean of the data set is 40.8 points. Just something I found interesting. Not sure what it means yet.
I've just about finished Grant Wahl's "The Beckham Experiment" (a must read for any big MLS fan btw, fabulous stuff), and in it about 3 times he refers to MLS as "the ultimate Moneyball league". I both agree and disagree with Wahl on this. On the one hand the Moneyball concept is about maximizing the strengths of your team based on a limited amount of resources (i.e.,, money). In MLS, since everyone is limited by how much they can spend, who can squeeze the most out of the lemons everyone is getting would seem to be at the most advantage. On the other hand, Moneyball is about teams with smaller resources finding resourceful ways to keep up with teams with much higher resources/revenues and in that sense something like the EPL is a better indicator of Moneyball philosophies (and becomes linked with a similar in spirit book, Soccernomics) about maximizing a team's strengths while going up against teams with many more recourses. It's just interesting that in MLS there's almost seems to be NO relation between money spent and where a team ends up in a mythical single table at the end of the season. Somebody must be unlocking some consistent keys to success in MLS somewhere out there, and I'm curious as to what those are. One key in the last few years may be not being in other competitions (SuperLiga or CCL) besides the Open Cup. The small squads size just can't put up with it.
The constant here is interpreted as the estimate of how many points the average team in MLS would earn if they speant no money on player salary. In reality, no team would get 42 points by paying thier players nothing. The assumption in accepting these coefficients is that this equation determines points and nothing in the unmeasured error term is correlated with wages. Since we know that's not even close to true, I wouldn't put any weight on the constant term.
MLS case is exception right now, the systems is kind of f*c*ed up right now. it doesn't follow the common sense.
Yes and no. What's driving the constant term is the fact that wages have almost no effect. Set up the regression equation in terms of deviation from wages: Points = a+b(Wages-AveWages) and you'll find almost the same result. Technically, the constant now means the number of points you would earn if you paid the average level of wages. It's a nicer interpretation but it's still the same result. The bottom line is that the explanatary variable doesn't explain anything - b is close to 0. Because of that, the constant is really just giving you the average number of points. In that light, it had better be significant and non-zero. If you want the constant to be 0, run the regression as (Points-AvePoints) = a+b(Wages-AveWages)
The wages should be adjusted for inflation. Right now they're in different units because $1 in 2007 is not worth $1 in 2008 or 2009. It won't change the ceofficients much but just to be rigorous we should have them all in the same units. Inflation calculated by the CPI was 3.85% in 2008 and -0.34% in 2009. So to put all wages in 2007 terms, multiply 2008 wages by 1.0385 and 2009 wages by (1.0385-1.0034). Getting back to the original post, what is the regression coefficient on the EPL data? *edit* So the regression coefficient for EPL is 0.4241 meaning an extra million pounds will buy almost half a point. When you look at a scatter plot, there seems to be a decreasing returns to money spent so I fit a regression equation with a squared term as well and got: Estimated Points = 18.38+0.7509*money-0.0019*money^2. This indicates the value of money spent is decreasing with total money spent, which makes sense. A first place team has little to gain by spending more.
That non-linearity is interesting. My interpretation is that the top teams get get less for their money (in per dollar terms) since top players are more expensive per "unit of skill". Also, I think you meant divide by 1.0385 and (1.0385)(.9966) when adjusting for inflation.
Where did you get the EPL data? I don't know where I repped you before, but BS tells me I have to spread it around before I rep you again.
Lol! Yeah divide.....oops . I was working on my dissertation simultaneously on a separate computer while doing the above calculations. Bad idea. The non-linearity also should occur naturally from the censored left hand side variable (points). That is, there's a maximum number of points any team can get so the benefit of an extra million must decrease eventually. Sachin: I just copied and pasted it from the first post in this thread.
Two things contribute to 'more points' 1. Good management: Good manager, coach, scouts, stable ownership 2. High wages/transfer fee budget: attract the best players Teams that win the league have BOTH. For MLS, it's all about good management because the wages for each team is identical, give or take. Allocation money, best draft picks are given to teams that do the worst the previous year. So for MLS, it's all about equal opportunity of success for all teams.
Good field/game management is like having good chemistry -- when you win, you have it, when you lose, you don't. Also, it should not be confused or grouped with team/club/front office management.