The following is an excerpt from the author’s senior thesis at Brown University. He will be presenting his research at the Society for American Baseball Research Analytics Conference on March 15. The full paper will be made available later this spring.
Anyone who’s read Moneyball knows the story. At the end of the 2002 season, Billy Beane and Paul DePodesta arranged to trade Beane to the Boston Red Sox in exchange for two minor-league players, including Kevin Youkilis. To most teams Youkilis was an unexceptional minor leaguer with an unathletic body and no noteworthy skills, and it is probably fair to say that few in the game outside of Boston and Oakland saw any promise in him. Youkilis ended up developing into a very good player and went on to thrive with the Red Sox for years, but it is striking that the cost of a GM with the ability to identify a hidden gem like Youkilis was two minor league players in whom most of the rest of the league saw little value.
Though Beane ultimately decided to stay in Oakland, the episode remains a fascinating illustration of how MLB franchises value the people who put their teams together relative to the players they acquire. Those inside the Athletics’ front office knew better than anyone how important Beane was to the organization, yet DePodesta agreed to receive a pair of unproven minor league players in exchange for the man who had built the entire operation. In other words, one of the game’s then-premier experts on inefficiencies in the market for players had no qualms about trading the goose for a single golden egg. According to Michael Lewis, Beane was fully aware of what was happening in the deal: “He could see only one way to exploit this grotesque market inefficiency: trade himself.”
Nine years later, Theo Epstein found himself in a similar situation. Faced with internal discord after the Red Sox suffered an historic collapse at the end of the 2011 season, Epstein left Boston to accept a job as President of Baseball Operations with the Chicago Cubs—thus raising the question of a top baseball executive’s value again. Some speculated that the Red Sox would demand Trey McNutt, the Cubs’ top prospect, in exchange for Epstein. However, they eventually settled for two less prestigious players, and at the time most analysts seemed to agree that Chicago needed not give up a top prospect in exchange for a non-player. Bradley Woodrum summed up the prevailing sentiment in the title of an analysis he wrote in the midst of the negotiations: “Trey McNutt for Theo Epstein: Eh, Maybe.”
This sentiment is difficult to reconcile with the facts that Epstein was generally regarded as one of the best general managers in the game and the Cubs were about to make him the highest-paid baseball operations executive in the history of the sport. Would not an elite talent evaluator be able to find and acquire several more players of McNutt’s caliber to replace the lost prospect? If the perception that Beane’s value was equal to Youkilis’ raises some questions, the implication that Epstein was worth less than McNutt is downright baffling.
Alternatively, one could express this apparent undervaluation in terms of dollars. I estimate the cost of a win purchased by signing a player through free agency to have been $7,032,099 for the 2013 season. Yet the highest-paid baseball operations executive in the game (Epstein) made just $3.7 million, implying that there is not a single non-uniformed MLB team employee who is worth more than about half a win to his or her team per season—clearly out of line with popular perceptions of how much of a difference a single executive can make. “GMs are just on a different scale,” Woodrum observes. But that such a phenomenon exists in the market does not mean that it represents rational behavior.
This potential market inefficiency could also include lower-ranking team employees. If teams are willing to pay $7 million per additional win, that should hold no matter where those wins come from. Consider an advance scout who sees that an opposing pitcher has a habit of tipping his pitches that no one else had noticed. If the scout’s team uses that knowledge to win a game that it would otherwise have lost, that observation is worth $7 million to his employer because its effect on the team’s win-loss record is equal to the boost it would expect to get from spending $7 million on a player.
However, that does not line up with how well team employees are paid. Entry-level front office positions are extremely competitive—millions of people have dreamed about working for an MLB team, and there are only 30 possible employers—and even qualified, well-educated professionals usually start out as interns making close to the minimum wage with no guarantee of future advancement. Every team’s hiring and salary structures are different, but Tom Tango estimates that junior front office executives generally make as little as half what they could get if they worked in another industry, and I would guess that it’s usually even lower.
Deconstructing the Market
The typical explanation for why wages are so low for front office jobs is that employees agree to take salary discounts because they receive substantial nonmonetary utility from working in baseball—or, in economic terms, that the perks of working in baseball (material or otherwise) shift the supply curve for front office labor to the left. Perhaps more importantly, anecdotally speaking the supply curve seems almost perfectly inelastic around the current equilibrium point; that jobs in baseball pay relatively poorly has not squelched the competitiveness of the application processes nor the zeal with which aspiring employees seek to break into the industry. MLB player agent Joshua Kusnick sums up the current state of the market well: “Teams always have the advantage when hiring, because so many people are willing to work for next to nothing just to get their foot in the door.”
But the rationality of this model is not a given. Rather, it hinges on three key assumptions that may not be true: that there are no meaningful differences in ability and value between prospective hires, that the impact of decreasing marginal returns is large enough to render labor demand generally inelastic, and that a potential hire’s willingness to take a salary cut to work in baseball is unrelated to how qualified he or she is for the job.
The most important assumption for explaining the rationality of the predominant model for the MLB non-player labor market is that of relative homogeneity across the population of prospective hires. As an illustration, Kusnick writes that a common occurrence for baseball operations employees is that, “At some point you price yourself out and end up getting replaced by people who are the same age you were when you started.” From the standpoint of a team this strategy makes sense only if the difference in value between the established employee and his or her replacements is smaller than the difference in their wages. “The supply of qualified candidates is so high that I’m not sure that throwing a lot of money at an established guy is actually going to bring you a significant upgrade,” Dave Cameron writes in support of this viewpoint.
Yet anecdotally speaking the assumption that prospective front office employees are generally interchangeable seems dubious. It would be difficult, for example, to read Moneyball without prior prejudices and not come away with the impression that certain key members of the 2002 Oakland Athletics’ baseball operations department were significantly better at their jobs than not just their counterparts with other teams but even some of their peers within the organization. (Granted, the gaps may have been wider over a decade ago.)
Even if this potential heterogeneity were concentrated in a small proportion of potential baseball operations employees, its presence should affect the way teams approach the labor market. Consider how teams think about players. In any given year there are hundreds of players who are plausible candidates to play outfield for an MLB team that season (ignoring the millions of people who are willing but unqualified). Looking at the vast majority of that population—say, after the best few dozen players—it matters very little whom among them a team promotes from the minors or acquires to fill an open outfield spot because the differences between their projected values are quite small.
But the market for outfielders is defined not by the majority of players whose values are roughly interchangeable but by the minority of exceptional outfielders who stand above the rest: that the New York Yankees could have signed any number of inferior free agents to play center field for close to the league minimum did not stop them from signing Jacoby Ellsbury to a $153 million contract this winter. It is unlikely that the best quantitative analyst or minor-league scout is worth as much to his or her team as an All-Star outfielder, but when teams spend several million dollars to win an extra game, even a very slight variation in skill among possible employees should lead to far greater competition in bidding for at least the best job candidates.
Relatedly, that teams do not take greater advantage of the inelastic supply of aspiring baseball operations employees makes sense only if the effect of diminishing marginal returns for front office personnel is so great that even the best unemployed candidate would be worth less to a team than what a minimum-wage internship would cost. This idea works with the assumption of value homogeneity to keep wages down and discourage competition for job candidates—when labor demand is nearly fixed, noncompetitive, and significantly lower than labor supply, the employers have all the leverage in hiring negotiations.
However, this assumption is questionable as well. Considering again the enormous sums of money that teams are willing to spend to make themselves marginally better, if the next hire provides even a fraction of a win’s worth of value each year while making standard market wages, he or she would provide his or her employer with a phenomenal return on investment. An additional employee would have to be almost literally worthless not to be worth his or her salary. Further, this idea ignores the potential agglomeration effects of bringing multiple insightful baseball minds together, which would mitigate the impact of the decreasing marginal utility of additional hires.
Finally, the prevailing model also assumes that those who are willing to accept significantly lower wages to work in baseball are just as qualified as the would-be applicants for whom substantially lower salaries are dealbreakers (or at least that the difference would not be worth what it would cost to hire applicants with higher income demands). This assumption is more believable than the previous two: presumably both being willing to take a large pay cut to work for an MLB team and being sufficiently knowledgeable about baseball to be a top candidate in a highly competitive job market require a strong passion for the game, so there is probably a correlation between a prospective hire’s qualifications for a baseball operations job and the nonmonetary utility he or she would get from it.
However, if the job in question requires skills that are not specific to baseball operations work, the stronger a candidate is in terms of his or her broadly applicable credentials, the better the job he or she would be able to obtain in another industry and the higher the opportunity cost he or she would face by working in baseball. The truthfulness of this assumption thus likely varies based on how marketable the requisite skills for the front office job in question would be outside of baseball—the difference between an elite hire and an ordinary candidate’s willingness to work in baseball for less money might be relatively small among scouts but would probably be quite large among quantitative analysts.
If instead there is substantial variation in value among baseball operations personnel, our conception of the labor market should be radically different. Each prospective employee should be seen as constituting his or her own market with a perfectly inelastic supply curve kinked from not working in baseball to working in baseball at his or her industry reservation wage. Every organization would be represented by its own discrete demand curve, which would be perfectly inelastic and kinked from hiring to not hiring at a wage equal to how much value the team thinks he or she would add; we would expect to see heterogeneity in the demand curves based on each team’s needs and differing estimates of how qualified the prospective employee is. If at least one organization is willing to pay more than the individual’s industry reservation wage, he or she would take a job with the team that offers the best combination of salary and non-material workplace perks. If not, he or she would take a job in another field.
In our discussions on the subject and his attempts to play devil’s advocate, Matt Swartz has offered three good theories for why the inelastic-supply model could be pervasive throughout the league even if it is irrational, though from a team’s perspective none is a rational explanation for the lack of demand-side competitiveness in the non-player labor market if there exists substantial heterogeneity in value among prospective hires. Two of Swartz’ ideas seem like plausible descriptions of how the people who run MLB teams think: Teams fear internal capacity constraints that could render hiring additional analysts unhelpful beyond basic decreasing marginal returns (i.e., the difficulty of coordinating projects among more employees and the possibility of confusing decision-makers with too many opinions and perspectives), and they lack the means to accurately estimate the value of a prospective hire who is trying to break into the industry or whose previous work for other teams was confidential. However, capacity constraints would not be a problem in the long run if teams were willing to restructure their baseball operations departments to accommodate more employees—consider how teams manage their large staffs of scouts—and in the short term the issues could be minimized through teleworking and establishing clear chains of command for new hires.
The uncertainty argument also fails to hold up under scrutiny. No employer in any industry knows exactly how much a job applicant is worth before he or she is hired; if teams (or any firms) are unable to discern substantial differences between job applicants, how do they ever decide whom to hire? Even within baseball this problem is not unique to the front office. Though their work is admittedly more visible than baseball operations employees’, projecting players’ future performances is far from a perfect science, and the physical nature of a player’s job means he is far more likely to be incapacitated due to an injury or other physical issue than a member of the front office staff. In addition, many aspiring baseball operations employees have active online presences in the baseball blogosphere, so teams can get a decent idea of an applicant’s ability by reading his or her record of published work. And the uncertainty argument would not have made sense for someone like Beane or Epstein, who at the times of their respective self-trade negotiations were widely considered to be among the best general managers in baseball.
Swartz’ third explanation makes more sense: Teams may see the choice of whether or not to bid up prices in the non-player labor market at any given time as part of a series of repeated cooperative games in which MLB teams are all better off not upsetting the status quo. “If say the Astros decide they are going to pay a lot more than what everyone else is paying but everyone notices and starts bidding up analysts, they are worse off than when they started,” Swartz writes. Yet just as Beane’s Athletics famously benefitted from taking advantage of the undervalued market for plate discipline before the rest of the league caught on, if the popular conception of the non-player labor market is incorrect, the first team that begins bidding more than the market price for undervalued front office personnel will end up having gained a relative advantage over its 29 competitors—especially if the organization can negotiate the inclusions of non-compete agreements in its new hires’ contracts.
More importantly, this theory mistakenly assumes the existence of either a leaguewide openness to change or a general awareness of the market inefficiency that seem uncharacteristic of the industry at large. That not every team has seriously integrated sabermetric analysis into its player evaluations and decision-making process more than a decade after the publication of Moneyball speaks to baseball’s slowness to conform to new ideas, and if teams thought they were getting such phenomenal returns on investment from their non-player employees, they would expand their baseball operations departments substantially to take advantage of this vast supply of undervalued labor. Finally, the rationality of any one team’s preference to maintain the status quo is contingent upon the assumption that none of the other 29 organizations will ever challenge it either. If a market correction of this potential inefficiency is not wholly evitable, a team’s best response is to be the first to take advantage of it.
Why This Matters
As an unrealistic but theoretically possible example of this potential market inefficiency, consider a team that is deciding whether to spend $7 million on one or more players in a free agent market similar to the 2013 incarnation or on its front office. The expected returns from spending $7 million on free agent players would be one win. But if those funds were put toward the front office, the team could offer $7 million to any general manager (or any other non-uniformed baseball operations employee) in the game and have it represent at least double his current salary. If there is indeed significant heterogeneity in skill among general managers, one would think that adding one of the brightest baseball minds in the world would lead to more than one win on an annual basis.
Or perhaps the team could hire the 100 best available or aspiring front office employees at the well-above-market salary of $70,000 per year. Would the collective fruits of their observations, research, and manpower be worth more than one win over the course of a 162-game season? If so, that represents an important inefficiency in the baseball labor market: The cost of a win is cheaper when it comes from an executive than when it comes from a player. That would mean that an extra dollar is better spent on the front office than on the field itself.