What is an economic model?

In recent years, much has been written about behavioral biases in the actions of people, biases that contradict (apparently) the "rational choice" model in economics.

At first sight, such apparent irrationalities appear as a comforting thought ("surely people are not calculating utility at every instant!"). The more I've thought about it though, the less I am convinced that such irrationalities constitute a contradiction - a more appropriate term would be behavioral biases, as I've used in the first sentence.

To understand this, it is useful to begin here. It becomes apparent from reading that piece that one of the main motivations to introduce a utility function (probably first by Pareto then rigorously defined by Samuelson, if I get the paper) was to avoid such considerations of individual behavior.

A supposed irrational deviation from a supposed rational model can be criticized as follows - "the rational model is saying that people do what pleases them. Proving this mathematically requires some slightly strong axioms. So, one can see whether these axioms hold in the real world. Let's say you find they don't hold. Is it then fair to claim that the results do not follow through?"

Let me take an example. Imagine a lab experiment that demonstrates one group of people are unwilling to give up a coffee cup for a chocolate bar when they are "endowed" with a coffee cup, while the exact opposite holds for another group "endowed" with a chocolate bar. This is the endowment effect, and it is claimed to be irrational. One can argue however, that such endowment effects point to the fact that the basic preference ordering of any person is affected by what you give them. From here, lets say you derive a particular utility function that predicts exactly the supposed irrational behavior. Will you then say the behavior is irrational?

The criticism from the other side will then be - "okay, but you, rational choice theorist, claimed that preference orderings are fixed and unchanging, and now you alter that assumption. It is not surprising that you then get a different prediction." My reply would then be - "Precisely!"

This points to a basic fact about what a theory model in economics is, a fact that is becoming clearer to me over time. Theory in economics is never to be interpreted literally, it is highly metaphorical, yet mathematically exact. It is a way of thinking, a way to organize thoughts. Getting too deeply into how people make decisions - whether in the lab or field - is going to throw up so many different interpretations that it is going to be hard to make sense of them. Instead of claiming the rational choice model is wrong, I am quite sure that if people took the trouble of understanding exactly what extra dimension they are bringing in, a dimension that was not part of the traditional analysis, and how that affects the traditional analysis, we will all be better off.

Let me take another example. Think of the classic prisoner's dilemma problem - 2 person game. Now, we have a Nash equilibrium outcome (in pure strategies) which is inefficient in the sense that each is worse off than they could be, and another outcome which is efficient in the sense that each is as well off as possible. If you interpret this model literally, you will say "if I give people such payoffs I should never see anyone cooperating". Then you test this in a laboratory and find that actually people do cooperate. And you publish in the American Economic Review, claiming that "people don't play Nash thus they cannot be rational as defined".

My reply is twofold - (a) the utility function has been misdefined and (b) you're missing the point. Take (b) first. The point is not that people will play Nash or not, the point is much vaguer perhaps but more true, and it is this - "Adam Smith told us (by the way, the rest of this argument does great disservice to Smith who had a much more humane view of the world) the invisible hand (what a great metaphor!) of self interest makes everyone better off. Well this may not be true. Self interest alone may make everyone worse off." This, I don't think, anyone can have a problem with. After all think about any country, company, individual who "did well" - self interest played a big role perhaps in this, but surely it wasn't the only thing; we typically rely on many others often selfless contribution.

Point (a) is somewhat harder. If people are not playing Nash despite the (monetary) payoff indicating - as described by theory - that they should, the conclusion arguably is that there exist non-monetary payoffs that are driving the decisions being made. Thus the mapping from the utility function imagined by the authors of the published article to the monetary payoffs received isn't clear. Now, you may object - "yes, but you're just evading the issue by bringing up another cause that we cannot know is operating". My answer is - "Precisely". This is what we were trying to avoid when we said you get utility from what you consume, whatever that may be.

Of course, if neuroeconomics develops enough, we can actually understand (maybe) the mapping from utility to monetary as well as non-monetary payoffs. That may lay all this debate to rest.

But the broader point will remain - a theory in economics is not something where it is straightforward to take predictions to data. How a theory will operate in any circumstance must be understood not by the blackboard model offered by the theorist, but by understanding the broad meaning behind the mathematics, and interpreting that to analyze any situation. This isn't easy, and I believe requires just as much work as coming with the theory in the first place.

In fact, if you think about it, the idea that "people ought to work together to understand how a new observation can fit received theory, does it really contradict the theory or is the theory not supposed to apply there, etc" is just another way of stating that if we work together, we will be better off. Which is the lesson of the prisoner's dilemma.