How Much Does Payroll Matter?
This article was originally published on WahooBlues.com.
Everyone knows that money matters in baseball. I’m a Cleveland fan, so you don’t need to convince me that my small-market Indians are at an unfair disadvantage when competing against teams like the Yankees and the Bank of Steinbrenner (in the immortal words of Ken Tremendous, “It’s like Scrooge McDuck’s gold coin-filled pool”). There’s no question that franchises with the financial flexibility to retain their stars, import new ones, and remain contenders even when their well-paid players bust have a leg-up from the get-go.
But we all know that money isn’t everything. Omar Minaya and the New York Mets gave us a years-long crash course in what happens large payrolls are spent poorly. Meanwhile, plenty of underfunded teams have had success, including last year’s Rangers and the Rays of both 2008 and 2010. Check my facts on this, but I’m pretty sure one or two low-budget Oakland A’s teams have had some mild success in the past too.
Of course, citing these extreme examples isn’t enough for an honest analysis of the impact of financial inequity. There are 30 teams who take the field each year, and to discuss only the extreme, memorable examples is to ignore how the vast majority of other teams operate.
And so, when I found myself wondering how much money matters in the major leagues relative to the recent past, I decided to find an answer more methodically. Using USA Today’s salary database for payroll info, I found the correlations between how much money teams spent and how many games they won for every season since 1988. The results were quite interesting:
Or, for those of you who prefer the aesthetic simplicity of graphs to sifting through large amounts of repetitive numbers: (click to zoom in)
In general, the data point to a clear, if inconsistent upward trend (the correlation of the years and correlations is 0.417). The impact of money was inconsistent before the strike of 1994, skyrockets to its peak in 1999, then plummets immediately after. The added bonus of a dollar has remained relatively stable for most of the last decade at a level higher than 22 years ago but lower than it was 12 years ago.
The biggest surprise to me was the complete irrelevance of money in 1990 and 1992. I had assumed that the correlation would be lower in the first years I tested, but the idea that the relationship between payroll and success could ever be as low as it was in ’92—until I inputted the numbers for the last team, the correlation was actually negative—is mind-boggling.
The flaws with this simplistic analysis are highlighted by the major year-to-year fluctuations—I don’t think the differences in correlations between 1991 and 1992 and 1999 and 2000 necessarily represent the massive shifts in the financial role of the game that they imply. So I smoothed the curve by factoring the two previous and subsequent seasons’ correlations into each year’s score using a 1-2-3-2-1 weighting system. Here’s how it came out with my adjustments: (again, click the image for a better view)
This looks like a much more plausible description of how the role of money in the game has changed over the years. We’ve still got the relative unimportance of money around 20 years ago; the rapid rise in the 90’s; the peaks in 1998-9 and 2004-5; and the slow decline since.
Obviously this analysis is somewhat crude. I make no claim that the payroll-to-wins correlation described here fully captures the impact of cash on teams’ successes, especially considering some of the large yearly fluctuations. My weighting system for the smoothed curve was admittedly fairly arbitrary. And, for simplicity’s sake, I used only Opening Day payrolls in this study, thus discounting midseason signings and trades. Still, I don’t think there’s any reason to doubt the basic pattern we see here.
The initial increase in the mid-90’s can probably be explained simply by the rapid increase in payroll league-wide. In 1992, the Toronto Blue Jays’ $45.7 million Opening Day payroll was more than twice the Los Angeles Dodgers’ league-leading $21.6 million budget from just three years earlier. By 2000, both the Yankees and Dodgers more than doubled what the spendthrift Blue Jays spent eight years earlier, and half of all MLB teams shelled out at least $59 million in payroll.
As for the rebound circa 2004, I would posit that the increased acceptance of sabermetric thought—or at least, the basic principles of properly valuing plate discipline and power—helped teams spend their money more effectively. I’m not sure what would have caused the dips around 1989 and 2000 or the slow but steady decline in the payroll/wins correlation that has persisted for the last six years, though.
With this adjusted model, it is even more apparent that the impact of money in baseball has been significantly diminished in recent years—the weighted correlation for 2010 was just .397, the lowest since 1994. Money still matters much more in today’s game than it did in the pre-strike era, but low-payroll teams can take solace in knowing that the financial advantage enjoyed by richer teams is at least getting smaller.
Finally, as a bonus, here’s what the year-to-year correlations look like if we exclude two teams who had major influences on the curve: one line represents the year-to-year correlations without the World Series-buying Yankees, and the another takes out the A’s. Check out the difference in the early 2000’s:
Lewie Pollis is a freshman at Brown University. For more of his work, go toWahooBlues.com. He can be reached at LewsOnFirst@gmail.com. Follow him on Twitter: @LewsOnFirst or @WahooBlues.
Lewie Pollis is a sophomore at Brown University. For more of his work, go to WahooBlues.com. He can be reached at LewsOnFirst@gmail.com. Follow him on Twitter: @LewsOnFirst or @WahooBlues.
Solid, thorough (yet simple), with nuggets of insight. Thanks!
There definitely is an under lying trend here. It would be interesting to see if this trend levels out (becomes linear) or if it continues downward toward pre-strike years (quadratic). This sounds difficult, but is it possible to give more weight to wins in a year, theoretically what is the difference between 85 and 95 wins. Wins need to be given more power. Keep up the good work love this site
Those early low years came just after the collusion period of the late 80’s, which would have had a huge impact on team spending. Of course, I’m not really sure if that would push those numbers up or down — it probably depends on the details of when and how the money really started coming back once the scandal was settled. But it wouldn’t surprise me if spending patterns were seriously weirded for several years around that time as a result.
A big problem you have here is year to year correlation. Teams that are good tend to be good the next year, and bad team tend to stay bad.
It makes it impossible to even guess at causation. Are team good b/c they pay their players a lot, or do they pay their players a lot b/c they’re good?
I imagine it’s a mix of both.
I’d try to do a regression analysis including current wins and lagged wins (e.g. records from 1,2,and 3 years ago).
Doesn’t the Yankees’s success or failure in a given season have a huge influence on the overall correlation? I think that their payroll compared to other teams have been relatively high, but that didn’t keep them from being terrible in the early 90s. So maybe the new consistency in correlation is that the Yankees are managing their money well.
Conor,
The Yankees’ success or failure counts fairly substantially toward the correlation. Let’s use 2009 figures for comparison.
For a typical non-Yankee team, an extra win would move the correlation coefficient 0.0019.
For the Yankees, an extra win would move the correlation coefficient 0.0093.
No other team has an effect half as big as the Yankees.
Teams that have an effect 40%+ as big as the Yankees:
Marlins (49%)
Padres (43%)
Mets (41%)
Cubs (40%)
Teams with an effect on the fit less than 10% as big as the Yankees (i.e., close to average payrolls):
Brewers (9.9%)
Blue Jays (9.7%)
Indians (9.0%)
Mariners (6.7%)
White Sox (5.7%)
Braves (5.2%)
Giants (2.5%)
Cardinals (1.7%)
The change to the correlation coefficient for an additional win for one team (with a generic team taking the corresponding loss) can be calculated via a partial derivative. The formula works out to: delta_r = Z / (N sigma).
Z is the z-score of the team’s salary (3.32 for the Yankees).
N is the number of teams (30).
sigma is the standard deviation in wins (11.9).
I think this just goes to show how terrible teams were at managing payroll and predicting future performance in the 90s. Weren’t Cecil Fielder and Bobby Bonilla among the highest paid players in the game in ’92 or ’93?
What does the curve look like if we take three-year win-loss records and three-year salaries? Since there’s a lot of variation year-to-year in baseball, I wonder if taking longer stretches will increase the strength of the correlation. Or it might not, since we know that money can be poorly spent.
I realize this may be too simplistic but I’d be interested in seeing a similar chart of $/win.
The “mainstream media” as a whole assumes that winning is directly attained by spending. Thankfully this kind of study shows that it is not completely accurate.
I love the example of the Minaya Mets. From what I read during his time, he was constantly heralded as some brilliant GM (outside of NY anyways). That was a disaster that is still being felt.
Simply spending money doesn’t make you great. Spending money wisely does.
Philosofool:
The three-year salary/wins correlation is a little higher.
Salary in year Y correlates with wins in year Y with R=0.459 (R^2=0.216).
Salary in year Y correlates with wins in year Y+1 with R=0.400.
Salary in year Y correlates with wins in year Y-1 with R=0.530.
Salary in years Y to Y+2 correlates with wins in years Y to Y+2 with R=0.607 (R^2=0.371).
Still, that’s 63% of the variation in wins determined by things other than payrolls.
Hardscrabble:
Doing a plot of dollars per win would be quite deceiving. It would make the Yankees ($2M/win in 2009) look horrible, the Padres ($570K/win in 2009) look amazing, and belittle the achievements of the 2008 Rays ($450K/win).
Instead, consider what a typical league-minimum team would look like ($10M for 67.5 wins). Now look at what the 2009 Yankees ($207M, 103 wins) bought with that extra payroll: $5.6M/extra win. The 2009 Padres ($42.7M, 75 wins) paid a little less: $4.4M/extra win. The 2008 Rays ($43.7M, 97 wins), meanwhile, got fire-sale prices: $1.2M/extra win. (The 2009 Rays were still efficient, but less ridiculously so, at $3.4M/extra win.)
I think part of the problem, though this is a very good article, is that there are only so many more wins you can realistically achieve. Let’s say you have a payroll of 1 trillion dollars. You probably still wouldn’t win every game. It’s just the nature of baseball. A 90 win team only wins about 11 out of 20 games on average.
So, it’s not that money gets you so many more wins, it’s that it gets you more wins consistently. Small market teams scout, draft, develop, and then a guy is ready, he plays for 3-6 years, then they start all over again. The payroll swells and falls through this cycle. Large market teams just keep reloading with Free Agents.
I’d like to see how the teams who are consistently in the top 5 in payroll perform over the same period. Maybe a moving average, or you could group teams together.
Good stuff, Lewie. I wonder if “the slow but steady decline in the payroll/wins correlation that has persisted for the last six years” has anything to do with the increased value placed on young stars with regards to buying out arb and FA years, which would pay them peanuts relative to their worth.
Examples off the top of my head:, Longoria has been “worth” roughly $117 million based on fWAR and has been paid $4 million so far. Ryan Braun’s been worth $110 mil and has made roughly $6.5mil, and Lincecum has made $25 mil and has been worth $122 mil.
I’m sure there’s more, albeit less extreme, examples probably including the likes of Prince, David Wright, certainly Matt Kemp, etc.
I will make it much simpler. I have to ask the Cleveland fan, are current times with the Tribe as interesting as they were in the 90’s? The obvious answer is no. An ownership has a choice. Spend money to put a good team on the field,or not, leaning on the tired excuse that a city is hogtied by a so called ‘”small market”. Cleveland proved in the 90’s that they are not,by selling out every game throughout most of the decade. Gabe Pauls prophesy that Cleveland was a “sleeping giant” was proven true.
Once Charles Dolan bought the team, it was the beginning of the end, as he went about dismantling the team. . Instead of the thrilling baseball landscape around the Jake,during this period, you have now slipped back to the bad old days. A louzy team with a small payroll, with no chance of competing.
The moral of the story? If you want to be a Big League owner, you have a civic responsibility to the community to present a contending team, every year.. It must be frustrating for long time fans to have tasted succsess, only to have it slip away because of cheapskate carpetbagger named Charles Dolan
Could the graph be altered due to aging players with large contracts? They get signed and do good at first, then decline as they age and so produce less wins, but they still might have a large contract: Perhaps, for example, Beckett this year? Could this cause the graph to have a slow decline that corrolates with the age of contracted players or would the number be statistically insignificant?