Sunday, 28 July 2013

Why is prospect theory ignored

I have finally got around to reading Daniel Kahneman's book Thinking Fast and Slow. It is a fantastic read. It brilliantly sets out how Kahneman and Tversky revolutionized the way we think about human judgement and decision making. But, here's the question I was left asking myself - why has the work of Kahneman and Tversky had so little impact in economics? That question might sound bizarre given that Kahneman won the Nobel Prize for Economics (Tversky had died). As far as I can see, however, the insights provided by Kahneman and Tversky have largely been ignored. So, what's gone wrong?
        In his book Kahneman points to prospect theory as one idea economists have endorsed. True enough, their paper on prospect theory is one of the most cited papers in economics. Cites, however, are different to real impact. And very, very little research in economics has properly applied prospect theory. Indeed, given that the original paper is a pretty tough read, I would be surprised if many of the people citing the paper have read it. Kahneman, therefore, seems a bit over-optimistic in saying economists have endorsed prospect theory. More apt is to ask why they have ignored it.
       Prospect theory puts together four basic ideas: (i) People judge outcomes relative to a reference point, so we get gains and losses. (ii) People are risk averse for gains and risk loving for losses. (iii) People overweight small probabilities and underweight large probabilities. (iv) The reference point will depend on context and the way the situation is framed. An example can help put things in context.
  • (Option A) gives £900 for sure and (option B) gives £1000 with probability 0.9 and £0 with probability 0.1. Would you choose option A or B?
Most people go for option A which shows risk aversion for gains. The 0.1 probability of getting £0 is also over-weighted. Consider a further example:
  • I'll give you £1000 to start off. Then (option C) you lose £100 for sure and (option D) you lose nothing with probability 0.9 and lose £1000 with probability 0.1.
Most people go for option D which shows risk loving for losses. Comparison of these two choices also shows the importance of framing. A versus B is exactly the same as C versus D. Yet many choose A and D!
       All four of the basic ideas in prospect theory are crucial to take account of when analyzing economic behavior. For instance, anytime that a person buys or sells something outcomes will be judged relative to a reference point. So, the decision made will depend on framing, mental accounting and how gains and losses are coded. For example, people buy a lot more stuff when they use a credit card than if they use cash. And people are far more reluctant to sell something into which they have put sunk costs. None of this could be predicted by the standard economic model. So why has prospect theory been ignored?
         The main reason is probably the marketing of the theory. Prospect theory was sold as a package for modelling risky choice. This was interpreted as meaning it only needs to be used if all the four basic ideas above appear relevant. Thus, prospect theory is relevant for explaining things like tax evasion, but not much else. This interpretation is wrong. It's enough that one of the basic ideas is relevant for prospect theory to be relevant. Wrong or not, the common interpretation gives economists a convenient excuse to ignore the theory. 
         While convenience is probably the main reason economists have ignored prospect theory, there are better reasons to question its usefulness. Any theory is judged by its ability to make testable predictions. And the problem with prospect theory is idea (iv). We now know that the reference point a person has in mind when making a decision can be just about anything. One person's loss can be another's gain. This is a big problem when it comes to making testable predictions. Recent work tries to tie down the reference point, such as a paper by Koszegi and Rabin, but the psychological heart of prospect theory starts to get ripped out. In its pure form prospect theory starts to become anything is possible.
         We also know that prospect theory is not good at explaining some of the things it was seemingly designed to explain. Consider, for example, the disposition effect bias whereby investors are more likely to sell stocks that have made a gain while holding on to stocks that have made a loss. Loss aversion seems a ready explanation for this - people do not want to cash in on a loss. Barberis and Xiong, however, showed that loss aversion can make the disposition effect even more of a puzzle. Basically, loss averse investors should buy more of the rising stock and sell some of the losing stock to best avoid realized losses.
         The fact that prospect theory is not perfect is not a reason to ditch it. The standard economic model is very far from perfect. Again, however, the foibles of prospect theory provide a convenient excuse to ignore it.
         I write all this as a firm believer that prospect theory should be used a lot more in economics. What it can do far exceeds what it cannot. So will things change? In his book Kahneman says that he can understand why prospect theory is not taught in undergrad econ classes. Well I cannot! I think its high time that prospect theory was brought into the classroom. This seems to me the only way to avoid another generation of economists who ignore the theory. And I think it is key that prospect theory, and behavioural economics more generally, makes its way into the core of principles classes, and not as some add on or option to be chosen. So, if you have not already done so, read Kahneman's book!

Saturday, 20 July 2013

What do NIMBYs, the British Open Golf, the battle of the sexes, and the availability heuristic have in common?

It is the weekend of the British Open Golf Championship. And yesterday there was a fair amount of discussion about slow play. Slow play is annoying in golf because one player taking their time holds up everybody else on the course. The referees have the power to counteract this by putting a player on the clock. Which basically means the player will be timed and penalized for taking too long. Yesterday, the referees put lots of players on the clock and penalized Japanese golfer Hideki Matsuyama. A similar thing happened at this years Masters Championship where 14 year old Chinese Golfer Guan Tianlang was penalized.  
    What interested me about the slow play discussion was the reaction of the players and commentators. All were in agreement that play had been to slow and that 'something needs to be done about it'. But mention a name, such as Matsuyama, and all were also in agreement that 'he was treated very harshly'. That sounds contradictory. There were only 40 or so players on the course at any one time. If there was slow play, they cannot all be innocent!
      This failure to see the aggregate as a sum of its parts is an example of NIMBYism. This acronym was coined to capture the common reaction of residents to a proposed new development. Everyone thinks we need more affordable houses, a road by-pass, a bigger airport, better train links. Everyone also says Not In My Back Yard. But, if it is in no one's back yard then it is clearly not going to happen.
       Why causes NIMBYism? Its partly about fairness and strategy. The NIMBY game is essentially a battle of the sexes game like that in the matrix below. In the standard version of the battle of the sexes husband and wife need to independently decide what to do at the weekend (think of them at their desk at work deciding whether or not to order the ticket). The options are ballet and football. They want to coordinate on the same option but the husband prefers ballet and the wife prefers football (this is not quite the standard version!). So, if they both choose ballet the husband gets payoff 2 and the wife a payoff of 1, etc.

        The battle of the sexes is an asymmetric coordination game with two pure strategy Nash equilibria - they both go to the ballet, or they both go to the football. The difficulty is how to coordinate - the husband would prefer the ballet and the wife would prefer football. Solving this coordination problem is by no means easy. (As an aside, an interesting study by Holm showed that the men are more likely to get what they want.)
           To relate the battle of the sexes to NIMBYism we need to re-frame the game to something like that below. There are two players out on the golf course, Adam and Barry. Play is slow and its only going to get quicker if one of them speeds up. The problem is coordinating on who should speed up. Now you might disagree that this game is a good representation of the slow play game - you might say they should both speed up, or that Adam would be at a bigger disadvantage if he speeds up and Barry does not. I could answer those concerns directly: for instance, Golf is a competition and so there are good reasons why they will not both speed up. The key point, however, is that even in this ideal world, where we only need one player to speed up for everyone to benefit, there are still good reasons why things are not going to work. Both players would rather the other speeds up.

          The battle of the sexes does, therefore, nicely capture the idea of  NIMBYism. Everyone would be better off if people coordinate - play speeds up and the new houses are built. But, no one wants it in their back yard. Adam doesn't want to be the one that speeds up, or has the new houses built next to his home. We can easily end up with zero.
          Another important component to NIMBYism is the availability heuristic. This heuristic says that things seem more important the more easily they are available in our memory. When we think of abstract concepts like a course full of golfers, or a city development, the availability heuristic is not switched on because it is not something we can easily relate to. When we think of a specific concept like a Japanese golfer or a person's back garden the heuristic is switched on. We think how frustrated the Japanese golfer must be to receive a penalty, or how angry the homeowner is to loose his peace and quiet. In the context of the battle of the sexes game this means that when we look at a good outcome we focus on the fact that one player gets less than the other. This seems unfair and we easily relate to it. We overlook the fact that both players are better off than they would have been otherwise. 
          Once we recognize the causes of NIMBYism it becomes much easier to solve. Crucial is the framing of the problem. Adam needs to be convinced that he can gain by speeding up. Sure, Barry gains more, but let's focus on the gain to Adam. We also need to overcome the availability heuristic. One way to do this is to personalize the more abstract concept by, for instance, focusing minds on the annoyance of those golfers held up by slow play. The prescription, therefore, as often in behavioral economics, is to focus minds on the gains realized and the losses avoided.

Friday, 12 July 2013

Reference points and Edgeworth Boxes: A tourist's gains from trade

We have just got back from the PET13 conference in Lisbon. On the day of our arrival in Lisbon we were tired and hungry and desperately seeking food. A stroll in the vicinity of the hotel finally revealed a local corner shop selling fruit and essentials, and we were saved. We were also pleasantly surprised by the price. Our expectation was to pay a lot, because we'd bought a lot and because the shop owner was presumably going to rip off the unknowing tourist. It was pleasant surprise, therefore, when the bill came in well below our expectations. 
      In all likelihood the shop owner did add a bit of 'unknowing tourist profit' to the price. But who cares? We, as customers, were very happy to pay the price we did. And the shop owner was presumably happy to charge the price he did. Everyone is a winner. Indeed, this is a textbook story of exchange - a buyer and seller exchange goods for mutual benefit. From a textbook point of view, the really curious thing about this exchange is why it felt so unusual. Any exchange is supposed to be about mutual benefit, but I rarely feel so happy about the price I have to pay in my local supermarket. So, what is the difference between buying goods as a tourist and buying from the local supermarket? 
       On reflection, the answer seems to be about expectations and reference points. To explain, let's start with a bit of textbook basics. The figure below shows an Edgeworth Box diagram. This is the classic way used to illustrate the gains from exchange. From the bottom left upwards we represent the outcome for myself and from the top right downwards we represent the outcome for the shop owner. The initial point is given by the red endowment point. I have no apples and some cash and have a utility of 1. The shop owner has lots of apples and some cash and also has a utility of 1. 
      Can we gain from trade? Yes. The purple 'Edward U = 1' line plots all the outcomes as good as the endowment point from my perspective. Anything above this is a gain for me. The brown 'Owner U = 1' line plots all the outcomes as good as the endowment point from the owner's perspective. Anything below this is a gain for him. If, therefore, I give up some cash for some apples both myself and the owner can gain. Ultimately we can trade to the point where we both have a utility of 2. This representation of exchange perfectly fits our experience in the shop in Lisbon. We bought some apples, and everyone gained.

       Why are things different when I go to the local supermarket? It's tempting to say that its because the price is different, the supermarket has monopoly power etc. But that cannot be the difference. The exchange that takes place at a supermarket is still supposed to fit the story above. I gain from buying apples, and the supermarket gains from selling the apples. My utility still goes from 1 to 2.
    The difference that I experience in the supermarket reflects my expectations. Given that I go to the supermarket every week I take it as given that my utility will be 2 and not 1. Put another way, my 'psychological endowment point' is to have already bought the apples. So, buying the apples doesn't make me feel all that happy. It's just that not buying the apples would make me feel unhappy. On holiday things are different. It's an unfamiliar city and expectations are much less refined. This means that my 'psychological endowment point' is the same as my actual endowment point. I do not take it for granted that I will be able to buy apples at a reasonable price. So, when we did find apples at a reasonable price it felt like a gain. I felt the gain in utility.
        This way of looking at things highlights the important of reference points and expectations. And, it yields a somewhat depressing conclusion: In our routine everyday life it is very difficult to feel pleasure because our expectations are too refined. Buying stuff from my local supermarket, for instance, does not feel as pleasurable as the textbook says it should because I psychologically take as given that it will be done. I can only get pleasure if something unexpected happens such as everything in the shop being half price. But, that doesn't usually happen! So, we are trapped in a world where pleasure is difficult to come by. This is possibly one big reason why buying experience goods (like going on holiday, going for a walk, or going to watch a football match) is known to give more lasting happiness than buying consumption goods (like a car, or the weekly shop). Experience goods are more likely to be surprising.