November 11, 2013
The Cost of a Cost-Controlled Win
It’s free agent season, a time traditionally reserved for the baseball commentariat to wonder aloud how we’ve gotten to the point where passably decent outfielders are worth $10 million a year. The standard line is that, in the free agent market, one win above replacement retails for about $5 million. More recent research suggests that GMs actually ended up paying about $7 million per win for free agents during the 2013 season.
With free agents being so expensive, it’s often said that the way to prosperity in baseball, especially for “small market” teams (which seems to be a euphemism for “everything north, south, and west of New York”), is to grow a bunch of cost-controlled rookies in the farm system. After all, they are astoundingly cheap to pay for the first three years. Consider that Mike Trout, Paul Goldschmidt, Matt Carpenter, and Josh Donaldson made just over $2 million between the four of them, and all four will get deserved MVP consideration this week. Even in years four through six, young, cost-controlled players are still something of a bargain even after they (usually threaten to) go through the arbitration process. If it’s #snarkweek on Twitter (and, honestly, when is it not #snarkweek on Twitter?) there’s a ready-made retort to any free agent signing. The team is being lazy or foolish or is overpaying for the false certainty of past performance because they probably could pay $500,000 to some 23-year-old who could give them the same performance. It must be comfortable in that armchair.
But is the cost of that 23-year-old really a cool half million? If we view an organization as only its major league payroll, then the answer is yes. But let’s be a little more comprehensive, shall we? At some point, that 23-year-old got a signing bonus. He also drew a small (but non-zero) salary during his stay in the minors, and he certainly did not emerge as an 18-year-old draftee ready to pitch in the minors. There’s the cost of coaches, trainers, and the hotel rooms in Beloit and Albuquerque that he stayed in when he was in the minors. And oh yeah, there’s the cost of the scouting network that found him and recommended him. None of those things is free.
Then there’s the reality of the situation. A lot of rookies, sophomores, and juniors just aren’t that good at playing baseball. Some are, and that’s wonderful for the teams that are lucky enough to have them, but the guys who have been around a while are there because they were good enough not to be jettisoned. These guys might be cheap, but the ones who are MVP caliber are the exceptions. Most of the time, you get what you pay for. Is that 23-year-old really a value when you add everything up?
Today, we’re going to do some back-of-the-envelope math to answer a question that’s important if we’re going to Mike Carp about a win costing $5 million (or $7 million). Compared to what? If cost-controlled players are such a bargain, how big a bargain are they?
Warning! Gory Mathematical Details Ahead!
He’s sent to Rookie ball, does well there, and improves his game. He works out and adds some muscle. Within a few years, he’s knocking on the door in Double-A, and Jason Parks develops a man-crush on him. He spends five years living and breathing baseball, not getting paid a lot, but certainly living on the team’s dime. And then on that memorable night in June, he gets the call he’s waited his whole life for. He gets to squeal in glee when he finds out that an unfortunate accident has befallen another man who, as a result, will not be able to walk for a few weeks. It’s funny what you think you’ve been waiting your whole life for when you’re 23.
Now, exactly how much has the team invested in our 23-year-old to date? It’s a difficult question to answer because, to take the example of the initial contact by the pro scout, it’s not just a matter of saying “Well, the scout saw X number of players that year and his salary was Y, so our rookie cost the team Y divided by X dollars.” The scout’s job is to watch a lot of really bad baseball, and to act as a talent filter. Sure, much of the scout’s time is spent watching players who will never play for the team that signs his pay checks (and who may never play for any MLB team at all), but that’s just a cost of doing business.
Similarly, all teams in Major League Baseball run minor league affiliates as player development laboratories. Most of the players who play for them will never wear the uniform of, nor benefit the parent club, yet the organization still pays the overhead expenses that come with running a minor league team. It seems silly to spend money on players who will not benefit the big league team, but the organizational (read: filler) players turn out to be necessary. If there were no organizational players, there wouldn’t be enough bodies to fill out rosters for competitive games. The team is paying not only the overhead costs for the 23-year-old rookie, but also for the “glad to be here” guy who sits on the end of the bench in Double-A, because both are necessary for the hotshot prospect to develop. And within the current system, you might as well have that “glad to be here guy,” because sometimes miracles happen.
I’d argue that we need to think about the costs of scouting and development as a prie fixe cost. Everyone pays to have a network of scouts and a minor league development system with coaches, spiffy uniforms, and a strange mascot. For that money, they have the right to extract as much “cost-controlled” talent for the major league team as they can. Like an all-you-can-eat buffet, you might get a lot out of it or you might not.
What does that cost? I tapped a couple of people inside the game on the shoulder and asked about what teams generally spend on scouting and development. Some of the costs were obvious (salaries for the scouts and minor league coaches) and some not-so-obvious (at the end of the season, teams pay for plane tickets to send everyone back home). The numbers I got were around $20 million, usually with a warning that some teams spend more and some a little less.
In addition to the $20 million, there are the big-ticket costs of signing bonuses. Under the current collective bargaining agreement, these are capped: in 2013, the average team had roughly $6.75 million to spend on these bonuses. In addition, teams have an international signing bonus pool, which averaged $2.9 million in the 2013-2014 signing period. (The actual numbers vary based on where you fall in the standings, but for now, let’s just stick to league-average, back-of-the-envelope numbers, shall we?) Add it all together, and you get an average number, league-wide of around $30 million (yes, I rounded), all spent on acquiring and developing talent.
Now, let’s look at what kind of return on investment an average MLB team got in 2013 for its cost-controlled players. First, some details. I used Baseball-Reference’s version of WAR for these analyses, mostly because it was easiest to get WAR values and salary into the same spreadsheet working with their data. I also considered only players who either pitched 20 innings or more during 2013 or had more than 100 PA.
Under the current CBA, players who are in their first three seasons can basically be paid whatever the team wants to pay them, which is usually near the MLB minimum of $500,000. The following summarizes the average production and salary outlay (when salaries were available) for hitters and pitchers who were in their first three years in MLB.
(*Note: for these calculations, I removed several notable “early career” international free agents who were effectively free agent signings from other leagues, including Yoenis Cespedes, Yasiel Puig, Leonys Martin, Adeiny Hechavarria, Norichika Aoki, Yu Darvish, Hisashi Iwakuma, Wei-Yin Chen, Hyun-Jin Ryu, and Aroldis Chapman. For hitters, the average salary if these men are included jumps to $698,902, and the dollars per win figure goes to $624,019. For pitchers, average salary including these men is $726,893 and dollars per win moves to $1,093,072.)
In the aggregate, pre-arbitration players really are a fantastic value. Teams tend to return more on pre-arbitration hitters than pitchers. But, there’s a dark side to the all-rookie approach. Roughly one in three of these pre-arb players functions below replacement level, and dollar for dollar, returns less value than the average free agent. Pre-arb players also tend to be low-yield in terms of the absolute number of wins generated. So it’s not as simple as “let the kids play.” Even for a team just trying to get the best dollar-to-win ratio, there’s a one-in-three chance that the kid isn’t the best idea. Absent any other information (a silly assumption to make, because there will be plenty more information, but work with me), it’s still a good idea to let the kids play. On average, you’re right more often than you are wrong. Still, I can see how a one-in-three chance of disaster would make teams nervous.
Now, let’s look at players who are in their arbitration years (not that anyone goes through arbitration anymore). The standard rule of thumb is that first-year eligible players get about 40 percent of what’s due to them, followed by 60 percent in the second year and 80 percent in the third year.
Arb-eligible players are a little bit more expensive, but they are also less likely to be a disaster on the field. That probably has to do with the fact that the bad ones get weeded out, but even they have some risk associated with them.
Now, just for fun, let’s look at players in their seventh year and beyond.
In 2013, there was a disconnect between the amounts paid to pitchers and hitters in dollars per win among free agent types. Also, take a look at those last two rows. We generally assume that a win costs five (or seven) million dollars, but there’s quite a bit of range around what actually happens on a case-by-case basis. If the chances that a cost-controlled player not hitting a $5 million per win target are around one-in-three, the chances of a free agent doing so are one-in-two. Free agency is a scary place.
But back to the lecture at hand. Let’s total up the costs of an average team’s cost-controlled wins. Before we go further: I am fully aware that no such team actually exists, and your mileage may vary. This is meant to be an abstract exercise in back-of-the-envelope math.
The final average cost, in 2013, for a win by a cost-controlled player is $2,668,482, roughly half of what it costs on the free agent market. It is true that “cost-controlled” really does mean something.
The Hidden Superstar
In fact, scouting and development are so efficient that if they took an additional $20 million dollars in overhead costs and lit it on fire (so that the guys in Double-A could have a nice bonfire team-building activity), they would still be producing wins at a cost of $3.6 million, and would be worth an additional six or seven wins to the team. There are players who are getting serious MVP consideration who put up “only” seven wins last year. Scouting and development is the hidden superstar on your team, and I bet you don’t even know the name of your favorite team’s scouting director and player development coordinator. You should.
Of course, the actual yield from year to year and team to team will vary. Some teams will get next to nothing out of their cost-controlled players. (This has the same effect as taking an inner-circle Hall of Fame player and re-assigning him to the grounds crew for a season.). Some teams will have Mike Trout, who single-handedly justifies the cost of what we might guess the Angels have spent over the past few years on player scouting and development. Some systems produce a bunch of guys who function below replacement level, but as we saw above, that’s not something to which free agent signings are immune.
So—a question. Why don’t we study this stuff more often? Yes, the answer has to do with the fact that a lot of development work goes on where things aren’t recorded, and scouting reports are not generally made public until years after the player has retired. Additionally, development is messy and imprecise. But when has that stopped researchers? We’ve found proxies and workarounds for other “unobservable” variables (e.g., clutch, plate discipline). There are statistical techniques that we can borrow from other disciplines. And this seems a largely unexplored area of study, at least publicly. If we will spend countless hours chasing something that we know adds up to a few tenths of a run, why not work on the big thing that could end up being worth a few extra wins of value?