Payroll and Success in Professional Sports: What the Research Shows

Research finds payroll strongly linked to success in English football, where Kuper and Szymanski report wages explaining about 90 percent of long-run league position, and much weaker links in US leagues, especially the capped NFL. The difference comes from salary caps, restricted player markets, outcome randomness, and reverse causality from winning to revenue.

The short answer from the research

Payroll and winning are linked in every major league, but the strength of the link depends on how freely talent is bought and sold, how much payroll varies between teams, and how much randomness the sport's outcomes contain. English football sits at the strong end, the NFL at the weak end, and baseball and basketball in between, with the baseball relationship moving noticeably over time.

English football: wages track league position

The best-known evidence comes from Stefan Szymanski's work. In Soccernomics (2009), Simon Kuper and Szymanski report that for 40 English clubs between 1978 and 1997, wage spending relative to the league average explained 92 percent of the variation in average league position, while transfer spending explained 16 percent. For 1998 to 2007, they report 89 percent for Premier League and Championship clubs.

Two details matter. First, those figures use positions and wages averaged over many seasons, which strips out year-to-year luck. A single season fits far less tightly, and later replications on Premier League data have reported lower shares. Second, the variable is relative wages, logged, not raw spending.

Earlier, Stephen Hall, Szymanski, and Andrew Zimbalist tested direction of causality in "Testing Causality Between Team Performance and Payroll" (Journal of Sports Economics, 2002). For English soccer, Granger causality from higher payrolls to better performance could not be rejected. The authors attribute this to football's open market for player talent.

MLB: a weaker and shifting relationship

The same 2002 paper examined MLB payrolls from 1980 to 2000. It found no evidence that causality ran from payroll to performance across the full period, though the cross-sectional correlation between payroll and performance rose significantly in the 1990s. The authors point to baseball's restrictions on player movement as the reason the link is murkier than in football.

That 1990s rise was the backdrop to MLB's Blue Ribbon Panel on Baseball Economics, whose July 2000 report, by Richard Levin, George Mitchell, Paul Volcker, and George Will, stated that teams in the top half of payrolls had won all postseason games in the previous five years. It recommended revenue sharing and a luxury tax rather than a salary cap.

The relationship then weakened. Michael Lewis's Moneyball (2003) popularised how Oakland competed on a low payroll by buying undervalued skills. Writing for SABR in 2014, Phil Birnbaum noted that the payroll-to-winning correlation in Berri, Schmidt, and Brook's book was around 0.40, while a chart of 2000 to 2014 showed 2012 to 2014 as the lowest years in the series. Late in the 2014 season, he found payroll correlated with winning percentage at about 0.23, and made the point by noting that ranking teams alphabetically correlated very slightly higher.

NBA and NFL: the Wages of Wins argument

David Berri, Martin Schmidt, and Stacey Brook's The Wages of Wins (2006) argued that payroll explains surprisingly little of the variation in wins across the NBA, MLB, and NFL. The authors reported payroll explaining about 12 percent of win variation in the NBA, about 18 percent in baseball, and under 5 percent in the NFL. Their explanation for the NBA was that player pay tracks scoring more than the full set of statistics that drive wins, so teams systematically misprice talent.

Those figures belong to the era and specification the authors used. They are a reference point, not a constant.

Why the correlation differs by league

Payroll variance. A correlation can only be as strong as the spread in the input. The NFL has had a hard salary cap since 1994 and a salary floor, which compresses payrolls into a narrow band. When every team spends roughly the same, payroll cannot explain much. This is a statistical effect called range restriction, not evidence that money is irrelevant.

Cap design. The NBA has used a soft cap since the 1984-85 season, with exceptions and a luxury tax that let some teams spend well above it. The NHL introduced a hard cap after the 2004-05 lockout. MLB has no cap, using a competitive balance tax on payrolls above a threshold. The Premier League has no cap; from 2026/27 its Squad Cost Ratio limits squad costs, including wages, amortised transfer fees, and agents' fees, to 85 percent of revenue plus net player trading, after clubs voted 14 to 6 on 21 November 2025. A separate anchoring proposal, which would have capped spending outright, was rejected.

Contract structure. In MLB, pre-arbitration players are paid far below market value, so some of the most productive players cost very little. Payroll then measures past performance and veteran contracts more than current talent. In football, players have been free to move at the end of their contracts since the 1995 Bosman ruling, so wages track market value more closely.

Outcome noise. A 17-game NFL season is a small sample, so results contain a lot of luck relative to talent. A single baseball game is close to a coin flip between average teams. Basketball outcomes are less random game to game. Noisier outcomes cap how much any input, payroll included, can explain.

Reverse causality. Winning raises revenue, and revenue funds payroll. Granger tests like Hall, Szymanski, and Zimbalist's exist to separate the two directions, and simple correlations cannot.

Using payroll in an analysis

For anyone building models, payroll is a prior, not a forecast. Practical points:

  • Use payroll relative to the league average, logged, and compare within one league.
  • Average over multiple seasons before judging the relationship; single-season fits mislead.
  • Adjust for contracts that do not reflect current talent, such as rookie deals, dead money, and injured players.
  • Expect payroll's explanatory power to fall in capped leagues for structural reasons.
  • Treat any reported coefficient as specific to its sample period and method.

The useful question is rarely whether money buys wins. It is how efficiently each team converts spending into performance, and that residual is where analysis earns its keep.

Frequently asked questions

Does a higher payroll mean more wins?
On average, yes, but how strongly depends on the league. In English football, relative wages track league position closely over many seasons. In MLB the relationship is weaker and has changed over time, and in the NFL, with a hard cap, payroll explains very little. Correlation also runs both ways, because winning teams earn more and then spend more.
Why does payroll matter less in the NFL?
The NFL's hard cap and salary floor keep team payrolls in a narrow band, so there is little spread for payroll to explain differences in wins. A short 17-game season also adds a lot of randomness to records. Both effects lower the measured correlation even if talent still costs money, which is why the low figure should not be read as money not mattering.
What did Soccernomics find about wages and league position?
Simon Kuper and Stefan Szymanski report that wage spending relative to the league average explained 92 percent of the variation in average league position for 40 English clubs from 1978 to 1997, against 16 percent for transfer spending. For 1998 to 2007, they report 89 percent. Those figures use multi-season averages, so single seasons fit less tightly.
Does Moneyball prove payroll does not matter?
No. Moneyball (2003) showed that a low-payroll team could compete by buying skills the market undervalued. That is a statement about pricing efficiency, not about money being irrelevant. Research from the same period found payroll and performance more correlated in 1990s baseball than before, and later analysis found that correlation weakening again in the 2010s.