Baseball Forbids Pay for Performance – A Lesson for Healthcare?
Posted May 10 2012 3:21pm
Steven Lewis President, Access Consulting Ltd., Saskatoon SK Canada Adjunct Professor of Health Policy, Simon Fraser University Steven.Lewis@sasktel.net
Star players get paid a lot of money in baseball. Alex Rodriguez, the Yankees’ third baseman, signed a ten-year, $275 million deal in 2008, the year he turned 33. Canadian Joey Votto, the slugging first baseman for the Cincinnati Reds, will get $225 million over the next ten years. Eleven ballplayers’ current annual salaries are over $22 million .
Baseball is all about winning, and winning is all about performance. Unlike health care, baseball has uniformly high quality data on a dizzying array of performance metrics . Thanks to the thriving science of baseball analytics, general managers know pretty precisely how a player’s individual statistical profile contributes to his team’s won-loss record. You can actually quantify the extent to which a stolen base or a triple influences the probability of winning a game. The go-to contemporary stat is “wins above replacement” – the number of wins a player contributes to his team’s annual total compared to a replacement player minimally fit to play big league ball.
Given the utter simplicity and clarity of the goal – winning – and the availability and validity of data (arguably spectacular), you’d think pay for performance (P4P) would be hard-wired into players’ pay structure. Baseball is merit-based capitalism in its purest form: you get paid what you’re worth, and what you’re worth is a function of measurable performance. But the major league collective agreement forbids tying contracts to statistical performance. It has recently gone so far as nixing bonuses for reaching certain career milestones, such as 3000 hits.
Is baseball fouling one off its own foot, or are the stewards and players of the game onto something important?
The Major League Players’ Association is a highly sophisticated and successful union. Its main job is to get good pay, good working conditions, good pensions and other benefits for its members. It naturally prefers that the great bulk of pay should be in the form of base salary rather than performance-based incentives. It’s fine to pay me $15 million a year in the expectation that I will hit 30 home runs, drive in 100 runs, and play third base with fearless aplomb. But it’s not okay to pay me $5 million base and $10 million contingent on reaching these statistical targets.
It’s pretty obvious why players prefer this arrangement. But why don’t winning-obsessed owners, armed with great data, fight it to the death?
It’s because they’re not stupid. To begin with, most pay really is based on statistical performance, only the pay is deferred. The Los Angeles Angels of Anaheim aren’t paying Albert Pujols $20+ million a year based on his potential; they’re paying him for his consistently brilliant performance over his first 11 years. For players under 35, performance over the last few years is a pretty good predictor of performance this year. In other words, the baseball market is highly performance-oriented, but the pay is mainly time-lagged.
More importantly, though, individual statistical targets can get in the way of team performance. Suppose you get a big bonus for hitting 35 home runs. It’s the last week of the season, the game is close, there’s a man on second, none out, tough pitcher on the mound. The team needs you to hit a ground ball to the right side to advance the runner to third. But you’re stuck on 33 home runs, and your personal incentive is to swing for the fences. Or maybe your team has clinched a spot in the playoffs and it would be better to rest you for a few meaningless games at the end of the season. Not so great if you’re a few hits short of an incentive target, and the manager has to choose between giving you a shot at a wad of cash and doing what’s best for the team.
Don’t you love a paradox? Baseball pays huge attention to statistical performance indicators, but shies away from target-based payment. Health care basically ignores tonnes of statistical evidence but many are rushing headlong towards target-based payment. And to heap irony atop the paradox, there are composite performance scores in baseball, like the aforementioned Wins Above Replacement, that could form the basis for payment without corrupting the enterprise. Even so, for pretty sound reasons, baseball won’t go there. In health care, there are no similarly valid omnibus measures of the performance of, say, a family doctor. Yet doctors have negotiated bonuses for performing simple tasks like ordering regular tests for diabetics or screening for certain cancers, both of which can and should be done by a computer.
How did this apparent absurdity become commonplace? It is a perfect illustration of the chasm between theory and practice. P4P sounds like a good idea. But health care and health are way too complex to make it work as well as it might in baseball. The link between what providers do and patient outcomes is not nearly so linear and immediately tangible as the relationship between a team’s on-base-plus-slugging average and the number of runs scored. Moreover, you can’t pay for performance unless the indicators are unambiguous and simple to measure. It’s easy to measure whether you’ve done Pap tests or ordered mammograms. It’s hard to measure whether you’ve helped a frail elderly person with four chronic conditions avoid complications over a ten-year period. That’s why P4P typically pays for the former and has no clue about how to reward the latter.
The result? Health care frequently pays extra for achieving targets that require no special skill or effort, and have little impact on the health of really sick patients. It’s a bit like paying outfielders extra for catching routine pop flies. In health care as in baseball, doing things the right way will get you a better result. But once you define “the right way” as a select list of cherry-picked activities and build the P4P scheme around them, you deflect attention from ends to means, and undervalue less tangible and discrete, but more important elements of success.
Baseball wasn’t always so wise. It used to overvalue some metrics, like batting average, and undervalue others, like how many balls a shortstop fields in a year (the greater the ground covered, the more balls fielded, and the fewer that reach the outfield for a base hit). Teams that developed and embraced better metrics ended up getting better returns on their investment (more wins per dollar) – that’s the essence of Moneyball .
Using data to define and improve performance is great. Tying people's pay to their recent aggregate performance may seem counterintuitive, but it makes pretty good sense. Tying this year's pay to achieving this year's individual statistical targets – particular in team sports like baseball and health – is fraught with moral hazard. Health care needs to understand the difference if it wants providers to focus on what they achieve rather than what they do, and to respond creatively to patients’ needs rather than the target-triggered ring of the cash register.