Is winner-take-all bad or good for the middle class? Evidence from baseball

A “winner-take-all” market is one in which the top stars get paid much more than anyone else. It’s an apt description of the American economy in recent decades. Top financiers, CEOs, entertainers, and athletes now routinely earn more than ten million dollars a year, and the share of all income (after taxes) going to the top 1% of households jumped from 8% in 1979 to 17% in 2007.

What impact does the rise in the share taken by those at the top have on the incomes of those in the middle? On one view it’s bad: if the additional millions going to the “winners” had instead been spread among those in the middle, the latter would have been better off. Others suggest the impact is good: winner-take-all markets help make the pie bigger than it otherwise would have been, and a larger pie means a larger slice for the middle class in absolute terms, even if that slice has shrunk relative to the slice of those at the top.*

Pay in major league baseball is a good test case. Since the 1970s professional baseball has had the two defining characteristics of a winner-take-all market: owners’ and/or consumers’ judgment that top stars are much more valuable than the next best, and stars’ ability to exit if offered better pay elsewhere. Salaries for baseball’s top players have skyrocketed. Also helpful: unlike in pro football and basketball, baseball teams’ total pay is not limited by a salary cap.

Here are the two contending hypotheses:

1. Winner-take-all is bad for middle-pay players. Stars’ big paychecks come largely at the expense of their teams’ mid-level players.

2. Winner-take-all is good for middle-pay players. Teams that pay big money for top stars enjoy greater revenue growth via higher game attendance, richer TV deals, better jersey and hat sales, and so on. The stars collect a growing share of these teams’ total payroll, but this is more than offset by the degree to which they help boost the payroll. As a result, salaries for the middle players on these teams increase more than on other teams. has data on the salaries of all major league players since the mid-1980s. I’ll examine change from 1989 to 2007, as both are business-cycle-peak years. (I exclude the four teams created after 1989. The Cincinnati Reds also are left out, due to missing 1989 salary data.)

Does paying big money for top stars enlarge the pie? On the horizontal axis of the following chart is change in the share of each team’s total pay that goes to its top three players. Consistent with what we would expect in a winner-take-all market, for most teams that share rose. For example, in 1989 the best-paid trio of players on the San Francisco Giants got 22% of the team’s total pay. In 2007 the Giants’ top three got 40% of the total pay, an increase of 18 percentage points. On the chart’s vertical axis is 1989-to-2007 change in each team’s total pay, in millions of inflation-adjusted dollars. The hypothesized positive association isn’t there. Teams that increased the portion of their pie going to their top three players haven’t gotten a faster-growing pie in return.

That points us toward hypothesis 1, which says a rising share of a team’s pay going to its top stars is bad for those in the middle. As the next chart shows, that’s indeed how things have played out. The chart plots the change in pay for each team’s middle five players from 1989 to 2007 by the change in the top three players’ share of the team’s total pay. Middle-player salaries have tended to grow less rapidly on teams in which the top three’s share has risen more.

The following set of charts elaborates a bit. It shows changes in top players’ pay and changes in middle players’ pay for four teams. The first two teams, the San Francisco Giants and Toronto Blue Jays, are on the right side of the second chart above. Pay for their top three players exploded. It rose for their middle players too, but much more modestly. The next two teams, the Baltimore Orioles and Milwaukee Brewers, are on the left side of the second chart above. Pay for their top three players rose sharply, but less than for their counterparts on the Giants and Blue Jays. Their middle players, by contrast, did better.

But that’s not the full story. To the two hypotheses listed above we should add a third:

3. Winner-take-all is bad for middle-pay players, but its harm is outweighed by other developments.

The “other” development that has mattered most is the growth in team payrolls. Total pay for the median team soared from $23 million in 1989 to $89 million in 2007. This has been the key determinant of salary growth for middle-pay major league players. On average, the pay of the middle five players rose by $300,000 less on a team with a ten-percentage-point increase in the top three players’ pay share than on a comparable team with no change in the top three’s share.** But salaries for the middle five players nevertheless increased on almost all teams, in many instances handsomely so. Across all teams, the average increase for the middle five between 1989 and 2007 was $1 million, nearly a 200% rise. Even among the six teams on which the top three players’ share of pay rose the most — those to the right in the second chart: Houston, Pittsburgh, San Francisco, Toronto, Oakland, and the L.A. Angels — the average increase for the middle five players was $540,000.

What accounts for the sharp jump in team payrolls? One element is enhanced revenues due to expanded demand for tickets, TV rights, and team paraphernalia. Another is a shift in the balance of power away from owners in favor of players. These developments have enabled some teams — the New York Yankees are the paramount example, as you can see in the second chart above — to concentrate a growing share of pay on their top three ballplayers and simultaneously provide a large rise in pay for their “middle-class” players.

Implications for the broader economy probably are limited. One, though, is that even if winner-take-all hurts middle-class incomes, if we had very rapid economic growth it might not matter much. Alas, figuring out how to get that isn’t so easy. A good substitute might be moderately strong growth coupled with strong unions (as in the 1950s and 1960s) or low unemployment (as in the late 1990s). But I’m not too optimistic about that either.


* Some recent analysis and commentary: Andrews-Jencks-Leigh, Cowen, Drum, Kenworthy, Klein, Thoma, Yglesias.

** This is based on a regression of change in middle-five players’ pay on change in top-three players’ share of team pay, change in total team pay, and 1989 level of middle-five players’ pay.

10 thoughts on “Is winner-take-all bad or good for the middle class? Evidence from baseball

  1. Should the minor leaguers be included in this analysis? It would not be easy, but these are professional baseball players too.

  2. Always nice to read about economics and baseball in the same article! I think the key is the shift in power towards players– and also the “multiplier effect” of increasing revenues through cable, paraphernalia, etc.

  3. Interesting. But no inferences of general applicability should be drawn. Salaries in baseball are highly constrained by rules established in their CBA. Many salaries are multi-year, especially at the top. Younger players have little bargaining power. It takes a couple years to reach arbitration and several more to reach free agency. Also those rules have varied over time so there may not be an apples to apples comparison throughout your time frame. For all I know, changes in those rules could be a significant contributing factor to the changes you report. There is no way of knowing what would have happened if every player throughout this period had been a free agent at the start of each season, a la the more typical worker in the US economy.

  4. Among the many constraints in MLB salary, baseball established minimum salaries for all players based on tenure and the initial minimum more or less tripled in the years of your study. You might find the post “The Economic History of Major League Baseball” by Michael Haupert interesting at

  5. You might be interested in a paper that asks the (sort of) opposite question: Helmut Dietl, Tobias Duschl and Markus Lang (October 2010) “Executive Pay Regulation: What Regulators, Shareholders, and Managers Can Learn From Major Sports Leagues” (it’s available on SSRN).

  6. “A good substitute might be moderately strong growth coupled with strong unions (as in the 1950s and 1960s) or low unemployment (as in the late 1990s).”

    Lane, I’m surprised you don’t mention expanding the EITC, which you make such a good case for in the last chapter of Egalitarian Capitalism.

    Your suggestion of “moderate growth” seems to accept implicitly that strong unions would preclude fast growth (?). That may well be true.

    But I know of no good evidence (only rather hoary theory. i.e. Okun) suggesting that redistribution per se is antithetical to growth — that equity and economic efficiency collide in an inevitable zero-sum tradeoff.

    In fact, quite the contrary:

    1. In your paragraphs on “Left government” in the Oxford Handbook of Comparative Institutional Analysis you cite multiple studies showing that left governments correlate with faster economic growth. You cite no studies to the contrary.

    2. A. The econometric literature (including yours) states pretty resoundingly that in prosperous countries, equality and long-term growth either correlate positively, or have no correlation. My rather amateurish analysis of the Luxembourg Wealth data suggests a quite strong correlation between wealth equality and long-term growth. B. You’ve shown that that equality results from redistribution, not market forces.

    3. There is not a single prosperous, modern country that does not engage in massive quantities of redistribution. If that redistribution was a constraint to growth, should we not expect to have seen more “efficient” countries emerge, and surge ahead of the rest? Hasn’t happened.

    Now it’s possible that prosperity breeds bleeding hearts, and that explains the ubiquitous redistribution — that if humans were more steely-eyed, we’d all be better of. But the absence of even a single exemplar makes that surmise seem . . . libertopian. (And we all know how planned utopias have turned out over the centuries.)

    Here’s the possibility I’d like to suggest, that I have yet to see even considered in the professional economics literature: that a certain level of redistribution is in fact necessary for a modern, technological, high-productivity economy to thrive. The theory would probably have much to do with:

    1. Maintaining and increasing aggregate demand — keeping the log rolling, and growing.

    2. Increasing money velocity in the real economy.

    3. Allocation of financial capital into productive pursuits, especially fixed capital. (IOW, money flows being directed to producers by consumers based on their wants and needs, rather than by the omniscience of a supplier elite. Wisdom of the crowds.)

    4. The increasing inability in advanced economies of less cognitively blessed individuals to find work which gives them an economic claim on a decent share of our country’s spiraling prosperity, hence an income to contribute to aggregate demand and money velocity.

    5. Wider distribution of the opportunity that economic security delivers — providing a safe and solid springboard and platform for tens of millions to move into more productive pursuits.

    If this notion holds any water, as you’ve pointed out, the EITC could be the best vehicle for that redistribution. Friedman’s negative income tax with a big improvement: incentive to work. (You’ve shown in your work that more employment is key to more prosperity.)

    (Should also do a campaign teaching EITC recipients that they can receive their payments on their regular paychecks. Only about 1% do. This would increase “salience,” increasing the incentive effect.)

    If the EITC was sufficient to provide a guaranteed income for all who work (and especially if all had guaranteed health care [and to some extent education]), we would have much more freedom to implement more economically efficient labor and trade policies that would otherwise be draconian — at least locally (temporally and geographically).

    Your work does much to put in question some key truisms of neoclassical macroeconomics — truisms that may not be truths, but that I nevertheless find lurking in, haunting, and in my opinion sometimes misdirecting your work.

  7. Steve: In the current (Jan-Feb 2011) issue of Boston Review there’s a forum on full employment. In my comment I suggest sharply expanding the EITC as a way of ensuring income growth for middle-class households. I’ll link to it when it comes online.

  8. Thanks, Lane, look forward to seeing it.

    I’d just add: the EITC doesn’t just serve to ensure income growth for middle class households – improving equity. It also encourages employment — improving economic efficiency.

    The incidence is dual: workers have more incentive to work because they’re paid more, and businesses have more incentive to hire because they can get a higher-quality worker without paying the full ticket.

    If my notions above have any merit, it makes for an equilibrium that is both higher and more equitable. Viz Bartels:

    This recent set of exchanges, on a logical hole in Adam Smith’s vision of ever-increasing prosperity and wages, might give further weight to my notions:

    Especially the very cogently argued opening discussion by Weintraub:

Leave a Reply