Economists and methodology

Where I argue that mainstream economics should think about the methodology of their subject more, but that to study this methodology it is much better to look at what economists actually do than to look at their (occasional) writing on the subject.

Methodology? Why should I worry about that? It’s what all those heterodox people do - lots of ‘isms’ and ‘ologies’ that are totally incomprehensible! Unlike those guys, I get on with doing real economics. After all, doctors do not spend large amounts of their time worrying about the methodology of medicine. So why should economists?

This is a caricature, but not far off the mark for many economists. (When I refer to just economists/economics from now on, I mean mainstream.) Perhaps more of a concern is that very few economists write much about methodology. This would be understandable if economics was just like some other discipline where methodological discussion was routine. This is not the case. Economics is not like the physical sciences for well known reasons. Yet economics is not like most other social sciences either: it is highly deductive, highly abstractive (in the non-philosophical sense) and rarely holistic. This is all nicely expressed in the title of what I think is one of the best books written on economic methodology: Dan Hausman’s ‘The inexact and separate science of economics’.

This is a long winded way of saying that the methodology used by economics is interesting because it is unusual. Yet, as I say, you will generally not find economists writing about methodology. One reason for this is the one implied by my opening paragraph: a feeling that the methodology being used is unproblematic, and therefore requires little discussion.

I cannot help giving the example of macroeconomics to show that this view is quite wrong. The methodology of macroeconomics in the 1960s was heavily evidence based. Microeconomics was used to suggest aggregate relationships, but not to determine them. Consistency with the data (using some chosen set of econometric criteria) often governed what was or was not allowed in a parameterised (numerical) model, or even a theoretical model. It was a methodology that some interpreted as Popperian. The methodology of macroeconomics now is very different. Consistency with microeconomic theory governs what is in a DSGE model, and evidence plays a much more indirect role. Now I have only a limited knowledge of the philosophy of science, and have only published one paper on methodology, but I know enough to recognise this as an important methodological change. Yet I find many macroeconomists just assume that their methodology is unproblematic, because it is what everyone mainstream currently does.  

This reluctance by economists to investigate their own methodology has a consequence which is the main subject of this post. It occurred to me when I recently re-read a methodology paper entitled “Two Responses to the Failings of Modern Economics: the Instrumentalist and the Realist” by Tony Lawson. The paper, written in 2001, starts on the first page with “There is little doubt that the modern discipline of economics is in a state of some disarray.” This is a strong claim. For example, I have previously written that the influence of economists within the UK government at that time may have been at an all time high, and as this account (pdf) shows, economics remains very influential within the civil service. Where is the evidence for the claim about disarray? The answer in this paper is a selection of quotes from economists writing about aspects of their subject. Now any economist would immediately wonder how representative these quotes were. But more fundamentally, are expressions of concern within a discipline equivalent to it being ‘in disarray’? (For example, see the first quote from a physicist here. Would this be a good basis for a paper that asserts than physics is in disarray?)

Even if we ignore these concerns, given the unfamiliarity of most economists with methodological discussion, it may be unwise to use what economists write about their discipline as evidence about what economists actually do.  The classic example of an economist writing about methodology is Friedman’s Essays in Positive Economics. This puts forward an instrumentalist view: the idea that realism of assumptions do not matter, it is results that count.

Yet does instrumentalism describe Friedman’s major contributions to macroeconomics? Well one of those was the expectations augmented Phillips curve. Before his famous 1968 presidential lecture, the Phillips curve had related wage inflation to unemployment, and if expectations about inflation were included (in some way), the coefficient on this expectations term was often empirically determined (see above) and was often less than one. Friedman argued that the coefficient on expected inflation should be one. His main reason for doing so was not that such an adaptation predicted better, but because it was based on better assumptions about what workers were interested in: real rather nominal wages. In other words, it was based on more realistic assumptions. (For a good discussion of the history of the ‘expectations critique’, see this paper by James Forder.)

Economists do not think enough about their own methodology. This means economists are often not familiar with methodological discussion, which implies that using what they write on the subject as evidence about what they do can be misleading. Yet most methodological discussion of economics is (and should be) about what economists do, rather than what they think they do. That is why I find that the more interesting and accurate methodological writing on economics looks at the models and methods economists actually use, rather than relying on selected quotations.

Afterthought

There is a nice self-conformational element to this post. Someone is bound to tell me that, in my comments on Freidman, I do not really understand what instrumentalism means. And that, of course, just goes to make my point that you should not rely on what economists say about their own methodology!

Paternalism and Irrationality

I have talked before about how most economists have an instinctive dislike of paternalism: a dislike of the idea that someone (usually someone in authority) knows better what is good for people than people themselves. I think this is a very good instinct to have, but sometimes it has to be set to one side. Economists should know this, because they often use economic theory in a paternalistic fashion.

How can I write this last sentence? After all, are economists not very careful to focus on agent’s revealed preferences, rather than any objective model of what is good for people? They prefer what people actually choose as measures of well being (like consumption), rather than some scheme of what is involved in the good life dreamed up by some philosopher.  

However, as Daniel Hausman discusses in a recent book, the idea that a preference based measure of individual welfare can ever be a completely adequate measure of individual well-being is deeply problematic. There are many reasons for this, but the one most economists recognise is where individuals clearly act in ways that are not in their own self interest (or in the interest of others). The example I used before is seat belts.  But behavioural economics and experimental data are giving us many more. For example, an individual’s choice may be influenced by the presence of irrelevant alternatives: the choice between A and B may be influenced by whether C is an option, even though A and B are both preferred over C.

These features of behaviour pose an obvious problem for any preference based measure of well being. What economists typically do, faced with this dilemma, is one of two things. The first may be to deny the premise that agent’s choices in these cases are inconsistent with their own self interest. Failing that, the second response is to try and correct the preference measure of welfare to get round the problem. For example, if someone’s preference for not wearing seat belts is due to ignorance about the statistics, then we might be justified in imagining what their views would be if they did have this information. We can talk about welfare as involving maximising preferences that are not based on false beliefs. The idea is that maximising these purified preferences then maximises individual well-being.

There are a lot of problems with this approach, which Hausman discusses in detail. However, it strikes me that once this attempt is made, economists themselves tend to be paternalistic. Because how do we judge whether preferences need correcting? Often we simply ask whether choices are consistent with rational choice theory i.e. the basic axioms of much of microeconomics. If they are not, then any preferences that violate these axioms need correcting. What we are doing here is elevating rational choice theory, or more generally the micro theory we typically use, to an objective theory of well being.

Let me give one final, and rather more complex, example. There is a lot of evidence (and has been for some time e.g. Ainslie (1992), Picoeconomics, CUP.) that individuals discount over time in a hyperbolic way, rather than in the simple exponential manner which captures impatience as a constant discount factor. Hyperbolic preferences mean that if the choice is between A now and B in x periods time, we may choose A, but if it is between A in y periods time, and B in y+x periods, we choose B. Now hyperbolic preferences cause problems because they are clearly time inconsistent. From a welfare point of view, we have to ask who is the individual whose welfare we are maximising: the individual today with one set of preferences, or the individual tomorrow with different preferences. Faced with this dilemma, economists and governments typically ignore the preferences agents actually have, and carry on using a constant discount rate. Yet if this is not what people actually do, does it make sense to discount at all?    

In this and many other ways, economists appear quite happy to ‘correct’ the preferences people have, and instead make judgements on the basis of the preferences they should have. That may be the sensible thing to do, but it seems pretty paternalistic to me.