TED – Ideas Worth Spreading
Professor Richard Thaler:
So we are going to talk about Behavioral Finance, Behavioral Economics, that is what I do, it’s a mixture of Psychology and Economics. We study how people make decisions, and how make decisions in markets. And today I am going to talk about some research I have done with my former student Cade Massey, who is now a professor at Yale University and we are going to use two concepts from Behavioral Finance that are very important to our field – the first is rational choices, do people make good choices or bad, and second the concept of market efficiency , do markets somehow take a bunch of people that may not be all that smart and integrate them into some very rational whole. And I am going to test these ideas in a kind of unusual place, which is the National Football League Draft.
Now you might wonder why I would do that. Well Cade and I are Football fans, that’s one reason, so it’s amusing, but there is a bigger picture. What if my colleague, the famous Nobel Prize Winning economist Garry Becker, in an article that was written about me in the University of Chicago Magazine, said well it doesn’t really matter if 90% of the people can’t do probabilities, the 10% of people who can will be in the jobs that matter. Which was kind of a polite way of saying that there is no real need to pay attention what Thaler is doing because it is not going to matter where there is big money.
So the National Football League is big money. The teams are worth a billion dollars and up, there is a lot of money on the line, let’s see whether what observes are rational choices and efficient markets. So to get started, let’s give you a quiz and see how you do: very year teams have to choose players and here are two quarterbacks, I will tell you who they are later and suppose your team desperately needs a quarterback , you have narrowed it down to these 2 – Mister A and Mister B. You can see that both have quite good completion percentages, yards per attempt about the same, actually let me add Peyton Manning up here, the greatest quarterback of our generation, you can compare them to Peyton Manning. These guys just played 2 years, Manning played 4 but you can see pretty hard to make a call, Mister B has a 21-1 record, maybe go for that, a little fewer interceptions, otherwise pretty hard to choose. But you think about what you would choose and we come back to them at the end.
Now, this is something that every Football Team has to do, and they do it every year in April, in the Draft which is now such a big thing, televised on ESPN for 3 days. The way the draft works is every team picks, the order of the picking is a function of how they picked the previous year, the worst team picks first, the best team picks last, they go through seven rounds. Importantly for my analysis, the picks can be traded, so you can take the first pick and trade it for 2 other picks, so the first thing for us to do in our analysis was to ask what are these picks worth in the market.
So we went and collected data on 20 years worth of trades and here is what we get: this is the value of picks in the NFL draft. So let me show you the raw data. So if you have ever worked with data, you know that life doesn’t look like this. We seem to have estimated this curve with uncanny precision. And for a while we thought to have discovered Newton’s fifth law or something like that, but it turns out that is not what we discovered. What we had discovered was something that everybody in the league refers to as The Chart. Now the back story on the chart is about 20 years ago somebody at the Dallas Cowboys tried to figure out what picks are worth and wrote down some chart and he assigned the first pick 3000 points and we only have two rounds up here but the last pick is worth one point and this tells teams what picks are worth. So the reason that our curve fits so well is that originally the Cowboys were the only Team that had this but it started drifting around the league and eventually everybody had it and they all to use this to make their trades. Now the thing to note from that chart or from this one is that high picks are very highly valued.
The first pick is worth 3,000 points and you can trade that pick with the 7th and 8th picks or for the last 4 picks in the first round. So it must be that picking first is really important. Now here is the 2nd question to ask, how much do you have to pay these guys, well here is a plot of the initial salary by draft order. This one isn’t quite as regular, but pretty regular, and again quite steep. So you can see early picks are valuable in two ways, you can trade them for lots of other picks, and if you choose to use them, that first player is going to cost you a lot of money. Well what we want to do is ask: can this be rational? Is that pick value a rational price? Now how are we going to do that? Well, notice that curve is very steep, what that tells you is that teams must think they have an uncanny ability to make the kind of decisions that I showed you on the first slide between player A and player B. They must think, I can figure out which one of those 2 is the better one.
So here is a little thought experiment: suppose I take all the players picked at some position, let’s say the quarterback and I rank them in the order in which they are picked, let’s take 2 that are picked consecutively, like the 4th one and the 5th one. What is the chance that the 4th one is better than the 5th one. And we will do that for every position and all players picked. Now if Teams are perfect at this you will get a 100%, they’d be always the 4th one will be better than the 5th one, if they are random, they will be 50%. Where do you think it would be in that range? The answer: 52%. Just slightly better than chance. Now that should worry you, when you know that, that should worry you about the steepness of that curve. But we have done something a little more detailed, what we have done is we have computed a cost/benefit analysis for every player that has been picked going back to 1994. The way we do this is we take a player that has been picked in the first round, let’s say a quarterback, suppose he rides the bench the first year, the second year he becomes a starter, the 3rd year he becomes an all-pro and the 4th year he is back to being an average guy, what we do is we look at how much it will cost you to hire an equivalent player in the free-agent market, the market of veterans that aren’t subject to this draft. So for each player we compute what his performance was worth to the team and we know what they are paid, and we subtract so here is what we get.
This is the chart that illustrates the performance value to the Team as a function of draft order. Now I haven’t put the dots up here because you would see it’s a big cloud, it’s now as regular as the other charts I showed you but what this tells you is that these guys know something. All the way, the curve slopes down all the way, so the best player taken tends to be on average better that the players picked subsequently.
Now let’s put another curve up here, this is how much those players cost, now that also goes down, and if you noticed it goes down a bit more steeply. Now what is a player’s worth to the Team? A player’s worth to the team is the difference, it’s how much he performs minus how much you have to pay him, we call that the surplus, if we subtract we get a curve that looks like this. Now let me show you that curve again, that’s the same curve I just showed you, and I am going to put it on the same graph as the first thing I showed you, this is the chart of draft pick values. Now in a rational efficient market, those curves are the same. You can see they are not quite the same and in fact quite strikingly although the market value picks steeply going down, the actual surplus goes up for a while, the very best pick is somewhere early in the 2nd round, not the best player mind you, remember performance does decline, but the best value, the best performance per dollar, the bargains are in the second round, those are the best picks that you want to get.
Now what should we make of this, if we think about this from an Economists perspective and a Behavioral Economists perspective, I can tell you this is the biggest anomaly I have ever found out in any market. I have spent 25 years doing research in financial markets, you find little edges here and there, this one is huge. Notice what it is saying, that first pick that I can trade for 4 2nd round picks or 5 or 6 2nd round picks, each of those are worth more than the first pick I gave up. Right, that is huge. How can that possibly happen? Remember Professor Becker’s claim that in a rationale world where there is money at stake, the people in charge will get stuff right. So how can this exist? Well the reason is, first of all remember that the teams with those first picks are the bad teams. This year actually teams are competing very hard to win no games because there is player called Andrew Luck, who is thought to be the next Peyton Manning and it’s the teams that manage to loose virtually all their games that will have those early picks. The smart teams, the teams that win all the time like the New England Patriots, they never get one of those early picks. So let’s suppose that we have smart teams and dumb teams, the way an efficient market is supposed to work is that the smart players trade with the dumb players and in so doing make prices rational. In this market that is not possible.
So suppose you are a smart team, you know that these high picks are over-valued, what can you do? Well, you don’t have one. So you know, if you got one you would trade it, but you are not going to lose all your games to get one of those picks. And importantly, unlike in financial markets you can’t sell those picks short, so as a professor I know that those early picks are bad investments but I can’t call up my stock broker and short JaMarkus Russell who was the first pick of the Oakland Raiders a few years ago and was a complete bust. So what could we do? Well what we could do is we could try to round-up all of our rich friends and buy a team that is performing badly and then make them smart. Now notice that is a lot harder than making a few trades and the problem is even that might not work and the reason is suppose that a team comes under market, who is going to be the winning bidder? Is it going to be smartest guy, no, might be the guy who is most willing to overpay to get that Team. So there is no reason to think that the smartest guys will end up owning these teams or that the smart guys can somehow drive the prices to being rational.
Now let me go back to the quiz we started with, here are the two players that you were choosing between that looked very close, one, Aaron Rogers is probably the best young quarterback in the League right now. Looking like he could get up there with Peyton Manning and Tom Brady. Alex Smith has suffered although this year he is playing a bit better, but when the San Francisco 49ers had this choice they took a few months and ended up taking Alex Smith. Aaron Rogers dropped all the way to number 23. Now if you take a couple of months to decide knowing everything that I have said so far, what should you be thinking? You should be thinking that is a tough decision, I should get rid of this pick. If they had just gotten rid of this pick traded it for let’s say the 10th pick and the 20th pick, they could have gotten DeMarkus Ware and Aaron Rogers. They didn’t do it!
So there is a take-away here and the take-away goes well beyond Football. If there is a number that I’d like you to remember from this little presentation, it’s 52%. Remember 52% is the chance that they are right in choosing between these two players. Now here is my claim, they have much better information than you ever have in choosing an employee. They have gotten to watch these players play the same game for 2, 3, 4 years, then they take them to the combine it is called, give them all kinds of physical tests and mental tests, they actually take an IQ Test, and then they get to watch them perform afterwards. This is a market that is much better than any of the ones we are used to. Now that means that 52% may be an aspiration level we can hope to achieve in the choices we make. But if it’s only 52% then can we really justify paying somebody 10 or 20 Million Dollars more than our second choice, if we are choosing a CEO or a Portfolio Manager or even a Professor?