Battlepanda: We need a hippocratic oath for economists


Always trying to figure things out with the minimum of bullshit and the maximum of belligerence.

Wednesday, June 07, 2006

We need a hippocratic oath for economists

The unpronouncable Knzn defends the concept of the Non-Accelerating Inflation Rate of Unemployment reluctantly. The idea is, if unemployment is too low, then wages will rise across the board, which in turn leads inflation. Thus when the unemployment rate fall below a certain threshhold (the NAIRU or "natural rate"), the fed should kick into action by raising the interest rate to forestall inflation. The problem is, how does one figure out this natural rate? From what I can see, the economists guess. If their guess is proven wrong by subsequent events, then they go back to fudge their initial assumptions before guessing some more.

Here's Knzn's lukewarm endorsement of the NAIRU:

The theory hangs on today mostly because nobody has a better alternative (kind of like what Winston Churchill said about democracy). It’s very hard to defend the NAIRU theory, except when somebody comes up with another theory that’s supposed to replace it. Then it usually becomes clear that the other theory is even worse.

Personally I am quite loyal to the Keynesian paradigm in which the NAIRU theory is embedded, but I’d prefer to do Keynesian economics without the NAIRU. There are two issues I often concern myself with in this context: (1) the pre-NAIRU orthodoxy may not have been entirely wrong; and (2) the U in NAIRU may be the biggest problem. That is, there may be a “non-accelerating inflation rate of labor market slack,” but unemployment (or any single statistic) may be the wrong measure. Maybe I’ll do a post about this at some point. [snip]

I’m not sure it’s possible to have “no theory.” If you want to draw policy conclusions from empirical results, you need to have some theory – expressed or implied – about how those results relate to policy. It is helpful to be able to describe the theory rigorously. In a sense, all theories are wrong (just approximations of reality), but some are wronger than others. (I suppose you could have a completely intuitive approach to policy – Greenspan taken to the logical extreme – but then you’re outside the domain of economics.) The NAIRU theory, even though it’s hard to defend, seems to have been a reasonably successful guide to policy[.] [snip]

I feel a bit strange being considered a NAIRU person. In my real life personality, I’m on record having sided with the “far lefties” (yes, those communists like James Tobin) on this issue. Lately I find myself in the position of defending the NAIRU because I’m afraid of throwing out Keynesian economics with the bathwater.

At this point I have to ask Knzn -- Is economics a science? Because if it is its theories have got to be capable of being proven wrong when it is contradicted by real-world results to the contrary. When an hypothesis is proven wrong in chemistry or biology, the discredited theory is not kept hanging around in the wings until a better one comes along to take its place. Knzn argues that we need a theory whether it does any good or not because we need a guide to shape policy. My take is the exact opposite -- if you are that uncertain about a theory, how can you possibly apply it to policy? Imagine a doctor saying "Well, I know all that theory about humors is bunk, but we don't have a better theory yet, so leeches it is!"

From what I can see, all the evidence supporting the NAIRU is inductive rather than deductive, meaning that starting from first principles, an economist can sit down with a pencil and a scratch pad and demonstrate how there simply must be a NAIRU. It is only at this point when he reaches for the real-world data so that he can approximate a figure for the NAIRU. If the unemployment rate falls beneath the rate he calculated and inflation does not kick in, the economist will ball up his approximation and start on another, then another. But there is pretty much nothing the real world can throw at him that will cause him to go back to that first sheet on the scratch pad -- real world results can't disprove what was inducted without real world imputs.