January 14, 2003
The 1987 Free Agent Market
In an article that appeared on Baseball Prospectus recently, I concluded that, in spite of an across-the-board decrease in player salaries, the winter's market has done a very efficient job of equating free agent salaries with performance. Players are being paid less, but more so than in the recent past, they're being paid in proportion to what they're worth. I went on to suggest that this constitutes compelling evidence that ownership is not colluding to restrict the market:
The one hypothesis that probably can be ruled out on the basis of this evidence is that of an explicit attempt at collusion. A collusive market, although favorable from the perspective of ownership, does not operate efficiently. Because collusion imposes an artificial constraint on the number of bidders for any given player, the market is not as well equipped to estimate this player's value accurately. The fact that this year's market appears to be doing a better job of aligning salary with performance suggests that collusion is unlikely.
Inspired, perhaps, by the news that a group of player agents are considering filing a collusion grievance against the owners, a number of readers wrote to ask that I verify my claim by analyzing the collusion year markets of 1986-1988. Thanks to a variety of electronic newspaper archives, and a fantastic new transactions database put together by the good folks at Retrosheet, I was able to do just that.
I restricted my analysis to the 1986-87 off-season, which featured the strongest free agent class among the collusion years. Tim Raines, perhaps the best player in baseball at the time, led the way, along with a variety of near-great performers like Jack Morris, Willie Randolph, Lance Parrish, and Andre Dawson.
Blissfully ignoring a variety of intelligent feedback I received on the discount rate and other matters, I decided to apply the same methodology as in the previous study. Salary was calculated based on a net present value approach, discounting future seasons at a rate of 5% and averaging income across all contract years; value was approximated based on a weighted, three-year average of Clay Davenport's WARP statistic.
Working with the 1987 data presented a number of challenges that hadn't surfaced in the original analysis. Under the fog of collusion, contemporaneous data on contract terms and dollar figures were difficult to come by, especially in the case of marginal players. In order to remain consistent with the 2001 and 2002 studies, which considered only guaranteed income, I did my best to track down and exclude any income in the form of incentive payments. Even where I could find such information, however, the result was problematic: The 1986-87 market, if nothing else, was tremendously risk-averse, and incentive clauses constituted a larger share of income than in today's market. Players received bonuses for things like not suffering a chronic back injury (Parrish), or making a pedestrian number of starts (Doyle Alexander), which they were very likely to achieve. Nevertheless, I considered only base salary in the analysis.
I was able to find reliable information for 69 free agents. Here's how everything shakes out:
Keep in mind my hypothesis: In a collusive market, because there is no mechanism to ensure that a player is paid what he's worth, salary should not align very closely with performance. The best way to test that is with the R2 values of the two regression lines that appear in the chart, and on that account, the evidence is mixed.
Something seems fishy, certainly. Beyond a reasonable doubt? On the basis of this evidence, probably not.
Indeed, after further consideration, I'm not so certain that my proposition (competitive labor market = efficient labor market) is correct, even in theory. Teams differ in the way that they assess a player's value. For that reason, it is certainly true that a market with a greater number of bidders does a better job, on the average, of performing that assessment accurately.
The problem is that a player's salary is not determined in accordance with the average assessment of his talent. It is determined, rather, in accordance with the most favorable assessment of his talent. (This phenomenon has a name, the winner's curse, that is apropos to baseball). A player need only accept one free agent bid, and more often than not, he's going to accept the one that pays him the most. The really absurd contracts in recent memory - Darren Dreifort, Mike Hampton, Jay Bell - all came in highly competitive markets in which a single owner had an unreasonably favorable view of the player in question.
Does that mean that there's no conclusive way to distinguish 1987 from 2003?
Flash back to January 1987. Walk Like an Egyptian is at the top of the pop charts. The Dow Jones Industrial Average has coasted past 2,000. John Elway has broken Cleveland's heart for the very first time. And in baseball, the free agents are getting utterly and completely shafted.
What's remarkable, pouring through the newspaper stories circulating at that time, is how obvious all of it was. Far from being a well-kept secret to be revealed only years later, collusion was pervasive, profound, and explicit. Collusion and drug suspensions were the dominant stories of the off-season. The owners, under Peter Ueberroth's direction, were crying poor. Don Fehr was bitching up a storm. Arbitration on a grievance over collusion in the 1986 market was going on at the very moment that the 1987 free agents were soliciting offers.
It is worth recounting the experiences of some of the prominent free agents from that year's class:
The only free agents who had anything resembling a normal winter were those like Brian Downing, Bob Welch, and Willie Randolph, who acceded to their former team's initial offer with little resistance. To a man, players who tested the market were punished - forced to take a pay cut, or sit out until May, or both. No contemporary player has had an experience anything like what the free agents of 1987 went through as a matter of course.
The evidence is just as clear if we forget the war stories, and stick to the data. Taking the top 30 free agents for 1987 and 2003 as measured by WARP', I've run a series of comparisons that appear below:
Table 1: Placement of Top 30 free agents Re-signed Signed Unsigned/ w/old club w/new club Japan 1987 22 7 1 2003 11 16 3
The vast majority of 1987 free agents (73%) resigned with their original club, a clear indication of the lack of competitive bidding. The figure was even higher in 1986, when only five of 62 free agents changed teams. In contrast, roughly 60% of the free agents who have signed thus far this winter have new addresses.
Table 2: Timing of Top 30 free agent signings Nov Dec Jan Feb Mar Apr May Unsigned 1987 3 8 8 2 2 1 5 1 2003 3 24 0 N/A N/A N/A N/A 3
The disparity between the two markets is even more apparent if we evaluate the timing of free agent signings. The conspiratorial lack of interest in the 1987 free agent class is reflected in additional time those players spent waiting by the phone. Only about a third of the free agents had signed contracts by the new year; it's fortunate that we weren't publishing Baseball Prospectus back then, because Chris Kahrl would have had a cerebral hemorrhage. This year, thankfully, only three top free agents were without a contract as of the first of the year (Ivan Rodriguez, Jose Hernandez, and Chuck Finley), and Chris is doing pretty well.
Table 3: Contract length for Top 30 free agents 1 yr 2 yr 3 yr 4+ yr Unsigned Average 1987 18 10 1 0 1 1.41 2003 10 6 5 5 4* 2.27
* Includes Greg Maddux, who has agreed to arbitration with the Braves but not yet settled.
Following the 1985 season, an ownership group known as the Player Relations Committee (note the conspicuous acronym PRC) circulated an analysis detailing substantial drop-offs in performance among players who had been signed to long-term contracts. Without a competitive market structure in place, the result was a de facto moratorium on contracts of more than three years in length. In the 1987 market only one player (Tim Raines) achieved even that threshold; everyone else was signed to two guaranteed years or less. In spite of the concern over insurance premiums, contract length was significantly higher this off-season than what was observed in the collusive 1987 market.
While the players would eventually recover nearly $280 million in damages as a result of the owners' collusive behavior, it is hard to overstate the profound long-term psychological impact of past shafting. The market is down this winter, there are millions on the line, and it's easy to understand if the players want to conduct their due diligence on collusion.
But collusion isn't happening, not this time around, and to suggest otherwise is irresponsible. The structure of the 2003 free agent market, in terms of timing, contract lengths, team changes, and the premium paid to elite players, has far more in common with the competitive markets of the past several years than with 1987.