J.C. Bradbury emails me on the allocation of talent

I hope you are doing well. I have a Micro III question that I thought might interest you. I often have such Tyler questions, but keep them to myself, yet this morning I decided to share with you.

What does Jeremy Lin tell us about talent evaluation mechanisms? This article ( http://www.wired.com/wiredscience/2012/02/what-jeremy-lin-teaches-us-about-talent/?utm_source=dlvr.it&utm_medium=twitter ) argues that the standard benchmarks for evaluating basketball and football players at the draft level are flawed. The argument is that Jeremy Lin couldn’t get the opportunity to succeed because his skill wasn’t being picked up by the standard sorting procedure. This got me thinking.  Baseball sorts players in a different way than basketball.  In professional basketball (and football), college sports serve as minor leagues, where teams face a high variance in competition (the difference between the best and worst teams in a top conference is normally quite large), with very little room for promotion. There is some transferring as players succeed and fail at lower and higher levels, but for the most part you sink or swim at your initial college. This is compounded by the fact that the initial allocation of players to college teams is governed by a non-pecuniary rewards structure with a stringent wage ceiling, which likely hinders the allocation of talent.  At the end of your college career, NBA teams make virtually all-or-nothing calls on a few players to fill vacancies at the major-league level.  In baseball it’s different. Players play their way up the ladder, and even players who are undrafted can play their way onto teams at low levels of the minor league. At such low levels, the high variance in talent is high like it is in college sports; however, promotions from short-season leagues through Triple-A, allow incremental testing of talent along the way without much risk.  I have looked at metrics for predicting major-league success from minor-league performance and found that it is not until you reach the High-A level (that is three steps below the majors) can performance tell you anything.  Players in High-A who are on-track for the majors are about-the age of college seniors.  Performance statistics from Low-A and below have no predictive power. Baseball is also much less of a team game than basketball, so this should make evaluation easier in baseball but it is still quite difficult by the time most players would be finishing college careers.  Also, a baseball scout acquaintance, who is very well versed in statistics, tells me that standard baseball performance metrics in college games are virtually useless predictors of performance (this is contrary to an argument made in Moneyball).  Even successful college baseball players almost always have to play their way onto the team.

Back to Lin. He played in the Ivy League and his stats weren’t all that bad or impressive in an environment that is far below the NBA. If Lin is a legitimate NBA player, he didn’t have many opportunities to play his way up like a baseball player does.  In the NBA, he experienced drastic team switches, and even when making a team he received limited opportunities to play. MLB teams often keep superior talent in the minors so that they can get practice and be evaluated through in-game competition.  An important sorting mechanism for labor market sorting is real-time work.  Regardless of your school pedigree, most prestige professions (lawyers, financial managers, professors, etc.) have up-or-out rules after a period of probationary employment where skill is evaluated in real world action. Yes, there is a D-League and European basketball, but the D-league is not as developed as baseball’s minor-league system, and European basketball has high entry cost and may suffer from the same evaluation problems faced by the NBA.  Thus, I wonder if the de facto college minor-league systems of basketball and football hinder the sorting of talent so that the Jeremy Lins and Kurt Warners of the world often don’t survive.  Thus, another downside of these college sports monopsonies is an inferior allocation of talent at the next level.

J.C.’s points of course apply (with modifications) to economics, to economies, and to our understanding of meritocracy, not to mention to how books, movies, and music fare in the marketplace.  Overall I would prefer to see economics devote much more attention to the topic of the allocation of talent.

Here is J.C. on Twitter, here are his books.


I was just pondering this a few minutes ago. But of course, I always ponder this.

I think the Jeremy Lin question is not a question regarding allocation of talent, i.e. is Jeremy Lin a quality NBA player. I think the question is/was "is there a system and team that showcases his talents, hides his flaws, and allows him to grow?" To me, it's clear that the D'Antoni Knicks have that system and team.

Steve Kerr thinks that about 60% of NBA players have the success of their careers determined by teammates and coaches. Bruce Bowen, Steve Kerr, and Steve Nash are great examples of this. Ricky Rubio might win rookie of the year, and he couldn't even get playing time on his European team.

Well, I think the question is really about how players fit certain teams. I agree with Kerr and I would go even further: no matter how talented you are, you always depend on teammates. It is probably impossible to put a % on this but I think this is pretty obvious. There is also the point about fluctuations in a career. Maybe Rubio was not playing because he needed that extra experience that he got just now by playing on the NBA.

By the way, I think this kind of problem applies to most technical professions. I've worked with software developers who a few years ago were just ok and now for whatever reason are stellar.

If you look ar All-Star teams, you will see they are highly skewed towards first round picks. Therefore, the predictive value of scouting (versus just scouting statistics) is not negligible.

A basketball and football minor league system would clean up (anc probably ruin) big money college sports.

But All-Star teams are skewed towards the players with the best individual statistics. Shane Battier will never play on an All-Star team, but depending on the other players surrounding him, I might prefer Battier to say.... Carmelo Anthony.

I was more thinking about football and baseball since basketball has only two rounds of drafts.

Football has the same problem to an even larger degree. What round was Tom Brady picked in? Was Terrell Davis predicted to be an All-Pro?
Remember when Denver used to turn 4th round draft picks into thousand yard rushers under Shanahan?

I guess what I'm getting at is that marginal product in team sports is really difficult to tease out. Tom Brady might have been a ZMPer if he had played for a team that couldn't keep him upright long enough to throw.

Look at what round the players in the pro bowl were picked and see if it looks random to you.

"A basketball and football minor league system would clean up (anc probably ruin) big money college sports."

I believe this is how hockey works in Canada. Most skilled hockey players are identified by their early teens and they go straight into junior hockey at 16 or 17 with the hope of bring drafted by the NHL. If they aren't drafted by age 20, they can give up and go to College (or University as a Canadian would say) or try and make out a career as a pro hockey player in the minor leagues with the hope of someday being called up to the NHL as an undrafted player (probably to replace an injured player).

Tyler, the analysis of experience goods addresses the problems posed by the discovery of talent. You may start with the idea that "people are experience goods" as discussed in this paper


Alternatively you may read Hayek to understand the market as a mechanism to know people. Hayek liked to use the word discovery rather than allocation because he emphasized knowledge (and decisions presume some knowledge). You may say that for Hayek all goods were experience goods.

This is something I find lacking in my perception of Keynesianism. We just learned that people shouldn't overextend for houses, that's what the economy has the capacity to produce, and the Keynesians note that the economy isn't producing to capacity. Well, yeah.

Two things to keep in mind: a professional baseball career (counting minor league time) is much longer than a pro basketball career. The amount of time available to evaluate someone will naturally affect how they are available.

Secondly, there are costs and benefits to evaluating workers (or athletes). Decisions about personnel will always be made under imperfect information (even more imperfect than might be feasible) because of the costs of collecting information. Failing to identify some talent does not always mean the evaluation process is suboptimal.

Yes, people don't understand that we don't have infinite time, resources, and knowledge with which to judge people. Obviously, even less will be devoted to those who are initially projected late rond or undrafted since most of those people won't even make it out of training camp.

It is easier to evaluate individual talent in baseball than in football or basketball.

I wonder if having your draft televised on ESPN hurts player selection. In football at least it leads to an obsession with QBs.

NFL teams are obsessed with QBs because QB play is really important in the NFL.

It is, but not so much that a QB needs to be your first pick.

Why should we believe Jeremy Lin represents a systematic failure of talent evaluation between college and the NBA? One reason he's so popular is that such "out of nowhere" cases are so rare. You'd have to believe that there's a large pool of talent that's simply never had the opportunity in the NBA and then just exits the basketball labor market entirely. I'm skeptical.

The case is much stronger that Jeremy Lin suffered from Asian and Harvard stereotypes, rather than suffering from a systematic failure in talent allocation caused by a wage ceiling. Other players would've overcome his failures to succeed in practice evaluations, but his success in actual college games was discounted. J.C. dismisses Jeremy's college performance from the Ivy League, but when Jeremy had his opportunities against top 25 opponents like BC and UConn, he shined. These success gained plenty of media recognition, including a lengthy profile in Sports Illustrated. Jeremy showcased his skills, but for an Asian guy from Harvard, these performances were dismissed by the establishment as flukes.

There are obvious reasons why talent evaluation from college basketball is superior to college football. In college basketball teams play 30+ games rather than 12-14 per season. There are 300+ D-I basketball teams but only about 100 teams in the top tier of college football. There are more nonconference games in college basketball than college football, so there is plenty more opportunity for college basketball players to play up. There's a tournament in college basketball, not top tier college football, which gives mid-major players a larger opportunity to showcase themselves. There is plenty of opportunity for college basketball players to rise to the top. Jeremy Lin seized his opportunity in college, and it was ignored.

Lin is a one handed b-ball player who has trouble going to his left. When forced to go left he often turns the ball over at an unacceptable rate. He is a below average defender.

He has a better then average shot, is a good playmaker.

Perhaps if he had gone to a better college b-ball program he might have learned how to play with his left hand. But many scouts would looked at Lin and seen a turnover prone one handed b-ball player - they don't survive in the NBA for long. Nor is the NBA the best place to learn these basic skills.

So the general talent evaluation of Lin was correct. Lin is more an example of how small the marginal differences are between players you make the NBA Vs those that fail. In some respects Lin has NBA level talent. But as mentioned before he lacks the complete set.

Not every player in the NBA is a star. Many are just a bit better at some skill that a team needs to complement their star players. Management skill in matching those player skills into a complete team is why some organizations are always near the top. i.e. a marginal player may have more impact on one team and not another.

These comparisons overlook the fact that the maturation proces is different in different sports. The average rookie in the MLB is over 24 years old. It takes a baseball player a long time to learn the game and become physically mature enough to play. By contrast, basketball is an easier game to learn and requires less physical maturity, so rookies tend to be much younger. Football is somewhere in between. It requires a high degree of physical maturity, but, with the exception of a handful of skilled positions, it does not take a long time to learn.

In selecting players, this must be kept in mind. The longer the maturation process is, the more time you need to evaluate players. A common mistake in player evaluation is confusing maturity for one's age and talent.

Not generally true. The difference is instead the contract structure. In baseball the teams have players under contract and can pay them peanuts for years while burying them in the minors. Then they are promoted when they are near their prime.

The same used to be true in soccer, but after implementing a somewhat free market, it was no longer possible to bury talents too long. Thus, rookies in soccer are younger now than 30 years ago. Does this indicate that the games is easier now? Hardly.

David Pinto at Baseball Musings also comments on this (http://baseballmusings.com/?p=79106). He makes the argument that the ratio of available talent to available positions is much greater for the NBA and NFL, so they can afford to have suboptimal talent evaluation and talent development systems.

Basketball is a relatively easy game to predict success in because sheer size matters so much. Many Hall of Fame talents, such as LeBron James, Kareem Abdul-Jabbar, and Wilt Chamberlain were nationally famous by age 16.

As for Jeremy Lin, he did best in college in non-conference games against major programs where he was the only player on Harvard who could compete physically against big time talent. But, against his Ivy League regular rivals, where his opponents knew his game and his teammates expected him to fit in, he was merely all conference, but never Ivy League Player of the Year.

Because he's Chinese, a lot of people are assuming that he's an unselfish team player who knows his role, etc etc. Yet, so far, the evidence is that he's most effective in a Give Me the Damn Ball and Get Out of My Way playground mode.

He was ineffective in the NBA until he wound up in a situation where all the stars were out so he could take over the game and make like Allen Iverson.

"Because he’s Chinese, a lot of people are assuming that he’s an unselfish team player who knows his role"

Uh, what? I think you are letting your bias get in the way here. He has a good number of assists, but all I have heard is about how he needs to dominate a game to be effective, etc. Certainly no mentions of role-playing.

Right. Lin needs the ball to do his thing. He wasn't succeeding as a role player, but when the Knicks were in complete disarray they let Lin take over the entire offense for part of one game and suddenly discovered a Star.

Looking backward, his Harvard coach should probably have just oriented the entire offense around him and let him do his thing. His Harvard statistics are reminiscent of Michael Jordan's at North Carolina. Who was the only man who could hold Jordan to no more than 20 points per game? Dean Smith, his college coach.

So, Lin is indeed kind of like Tim Tebow in the sense that he's not very useful unless you turn your team over to him. Tebow is a great guy and all that, but, for various reasons, you can't use him on special teams or as a tight end or as a safety. You have to use him as quarterback or nothing, and then you have to change your offense around to suit Tebow's talents. A lot of NFL teams said "No thanks" to that proposition, and it's by no means clear yet that they were wrong. Is it all that surprising that NBA teams, who only get 2 draft choices, didn't want to make a Tebow-like commitment to Lin? Heck, his Harvard coach didn't make as big a commitment to Lin as the Knicks are doing.

Lin is nothing like Tebow. Tebow is not useful under any circumstances. He is a good running QB, but he had horrific stats and the bottom line is his team got lucky to get as far as it did.

Lin is a legitimately good NBA player with the stats to back it up, and was predicted to be a solid NBA-er pre-draft (by some).

But Lin never showed anybody in the NBA that he could be a useful subordinate role player, somebody in the rotation. He has only flourished as The Man.

Tebow would be like Lin if he threw for 300 yards and 3 touchdowns every game (and won). Lin would be like Tebow if he was shooting 25 percent from the field while his teammates' defense was pulling out spectacular last-minute victories against mediocre teams.

Also, I suspect Ivy League basketball has slowly gotten a lot better over the years, due to the Ivy League's role as the gatekeeper to Wall Street jobs, which have gotten so much more remunerative than any other career. In turn, that drives up Ivy League endowments, which in turn improves financial aid.

If you are a Division I talent and have got a good head for numbers, why not go Ivy League now that Ivy League financial aid is so lavish? Say you've got full ride offers from Iowa State, UTEP, and Valparaiso, or you could go to Yale, which doesn't have athletic scholarships but does let in athletes with sub-Yale test scores, for $5,000 per year. And the Yale coach introduces you to some ex-Yale players who now work for Goldman Sachs?

So, Ivy stats shouldn't be discounted as much as in times past.

With Lin, he tended to post big numbers against big name out of conference opponents, then recede somewhat in league play. With the best players in the Ivy League sticking around for four years to get their valuable diplomas, in contrast to big time college ball where one and done is the norm for top talent, the quality of Ivy League regular season play might now be a lot higher than is assumed.

The WSJ had an article about all the guys working on the Street who suddenly are famous to their co-workers for having shut down Jeremy Lin in league play.

This is demonstrably not true, and most Division I talent does not have a good head for numbers and could not get a Wall Street job regardless of whether they went to Harvard or not.

Obviously, most don't, but a few schools have done well for themselves focusing upon above-average intelligence athletes, like Duke in basketball or Stanford recently in football (Luck, Gerhart, Marecic). The Ivy League has upped its game in basketball recently because it's financial aid is so generous that it's the equivalent of a full ride for many prospects.

I'm familiar with a fine high school quarterback with SATs around 1850 out of 2400 who was heavily recruited by Harvard. He chose USC, but could never beat out all the NFL quality QBs on the depth charts. Rather than transfer, he uncomplainingly soldiered for four years as the opposition QB in practice, pretending to be Terrelle Owens or whomever USC was playing on Saturday. He picked up a master's in real estate development and I suspect some USC booster in the rather closed world of big time SoCal real estate development has not forgotten his sacrifices for the program. But real estate development isn't the really big money compared to the Street.

not to mention to how books, movies, and music fare in the marketplace

I wrote a post called Amazon.com, Daniel Kahneman, and the future of publishing on how books fare given the scouting / talent development situation described above; with the rise of ebooks, it looks like many, many more people will ultimately face the market, instead of the kinds of up-or-out situations that are analogous to basketball (like: finding an agent, finding a publisher, ensuring that publisher can market your book / get it in stores, and so on).

I think it's a bit strange to have this discussion without focusing on the fact that NBA talent evaluators for the most part ignore the single best statistics system for evaluating players: Wins Produced. David Berri's Wins Produced system correlates with actual wins with a superior degree of accuracy, and demonstrates that many players capable of producing wins are not given opportunities to play because coaches and general managers (and owners) have a cognitive bias favoring players that score a great deal, with insufficient regard to how inefficient the scoring is and the lack of contributions in other key areas (such as rebounds).

Under the Wins Produced metric, Lin outperformed the average NBA player's performance in college, which would seemingly make him at least a good prospect, and during his rookie year performance he also greatly outperformed not only the average NBA rookie but was well above the performance of the average NBA player. During a greater sample size of minutes in the D-League, Lin achieved an elite Wins Produced score.

As such, it's hard not to think of the NBA as operating in a pre-Moneyball state using poor statistical measures. For more information on Wins Produced, Berri's detailed analysis and numbers breakdown on Lin appears here: http://wagesofwins.com/2012/02/12/who-could-have-known-about-lin/

Re: Tylers request for more work on how to allocate (and select talent). I agree, I have hired many people over the years for professional positions, most of my picks have worked out well, but I always have the uneasy feeling that, once I have weeded out the obvious fakes and crazies, and set minimum qualifications, I could have done just as well picking at random.


I guess this is the most cited paper more attention to the topic of the allocation of talent, from the abstract it claims that they have empirical evidence of the success of their system. My guess is that, like most psychology, the evidence is made up, or uses very small data samples, so isn't valid, but I would be interested in others views.

Lin is the exception that proves that the current system actually works very well. His rise from bench warmer to near all-star is unprecedented--that's why it's "Linsanity" or whatever--and this means that the NBA's talent allocation mechanisms are actually fairly impressive.

I have long thought that our schools in both sports and academics can miss late maturers. I guess on the academics side, the up side is that there are some really great auto mechanics out there.

It also misses, shall we call them, savants, that is those who are very good at one thing and not so good at school. It works both ways a person can for example be be very good at school and yet be a very bad doctor/engineer/computer programmer etc.

Comments for this post are closed