In late 2007, the Shanghai stock market was going through a boom. A Professor of Economics was visiting China and there he learnt that everyone, everyone was expecting the stock market to crash after the Olympics in September 2008. In other words, they assumed the market to keep rising like a Diwali rocket till then, and then fall dramatically.
The professor was amused that no one had reasoned backwards from there. Everyone expected the market to fall in September. So, what would happen in August? Wouldn’t they think, “Hey, I don’t want to get too greedy. I know that the index will rise another month, but what if it suddenly starts falling and I am caught short? I had better sell right now and book my profits.”
Now, if many people think so, prices will fall, not in September, but in August. But if everyone knows that everyone else is thinking that way, why would they wait till August? Won’t they start selling in July? And so on, by induction, if people expect the market to tank in September, prices should fall right now.
This is what traditional economics suggests. Behavioural economics argues that the professor’s analysis is incomplete. Obviously, not everyone will decide to sell in August, but some people will. What will happen in the market is the classic battle between greed and fear. Some people, overcome by greed, will hold on in the hope of making even more gains, while some people, overcome with fear, will sell. At some point, fear will overcome greed and the market will crash.
So, behavioural economics is useful. It helps us understand why booms and busts occur. But is it useful in the sense that we can do something about it? One way of doing something about it is to profit from our knowledge of behavioural economics. If I can use this new science to predict exactly (or even approximately) when the boom will turn into a bust, I can keep buying just before that point and then sell. If I can do that, then I can make consistent supranormal profits, and the Efficient Markets Hypothesis would have broken down. Of course, I will also need to keep my knowledge and actions secret from others, because if everyone starts doing what I did, then my knowledge will be useless.
So, when Ritwik says that behavioural economics will falsify the EMH, this is what he means (or should mean). But is it anywhere close to doing so? While it is good at explaining, in terms of actually predicting, it is no better than a trader’s gut feel. This is not surprising. Even after the behavioural scientists have a working theoretical model of a human society (which I doubt they have yet) they will need to come up with the data that they can feed into their models. This data is inevitably voluminous and difficult to obtain.
So, instead of asking me whether scientific models have to be perfect to be used, Ritwik needs to explain what stage of usefulness his preferred models are at. Right now, we have some theories that have been validated by surveys of actual human behaviour, and through experiments performed in controlled laboratory settings. Occasionally, we have seen them validated in pilot studies, but even here, conditions are fairly controlled. What we need to see is a working model of human and social behaviour that puts together most of the theories that we have come up with.
In other words, what we have right now is equivalent to a situation where Newtonian laws of Physics are being developed, and it is being claimed that developing complex machines is a trivial step that can be left as an exercise to the reader. Yeah right. Someone just has to put together Newtonian mechanics, corrections to them that need to be performed when you apply coefficients of friction and air resistance, the laws of electromagnetism that will come into play when electricity is involved, and then predict how the machine will behave in real life. None of these will require Einstein-like genius, but it does require a lot of work. Engineers these days can put together a machine entirely on computer and be able to predict very well how it will behave, but it has taken a couple of centuries of trial and error to get to that stage. It takes astonishing hubris to claim that we have reached that stage with a science that has developed in the past 3 decades and which deals with systems as complex as the human mind and human society.