Microfoundations: a beautiful theory was killed by an ugly fact

Modern economics looks like macroeconomics. Not much is important when lots of people are out of employment apart from getting those people into employment. Getting macroeconomics right is therefore very important and this is why recently the Lucas Critique has been receiving extra attention.

Lucas argued in the 1970s that macroeconomics could offer little useful advice were it built on looking at specific experiences, during specific periods, under specific policy regimes exactly because such models would only be useful in those specific episodes. Macroeconomics in the 1970s, he argued, wasn’t able to teach us anything useful about the future, which is what matters.

Disciples of Lucas had a very big win in the 1970s when they predicted successfully that the long run payoff between inflation and employment would break down. That we were heading for previously unseen combination of high inflation and high unemployment. They won that fight, and I think have kinda coasted since. So, even conceding this success, I would like to put forward three critiques of the Lucas critique.

Firstly, and most obviously, it doesn’t seem very useful, it hasn’t helped anyone in the developed world recover quickly from the little depression. In fact, it seems to have actively caused inaction in some economists who have thrown up their hands when asked for policy advice and said “its all very complicated!” The sort of people who really wound up Keynes in the 1930s.

The most useful policy advice I’ve come across macroeconomically speaking is from Scott Sumner. The advice is very simple and is, essentially, for central banks to keep very sticky prices moving along a predictable trajectory. This means keeping nominal gross domestic product growing at a steady rate along a predictable level. Employment is subdued in countries which have not seen NGDP recoveries like the UK and US and employment is catastrophically low in countries which have seen NGDP plummet like Greece and Spain.

Not a lot of “deep parameters” were necessary to arrive at a useful policy. Imagine, policy changes in a way that might affect the usefulness of Scott’s advice, for example, allowing the widespread proliferation of competing currencies as favoured by the free banking crowd, would this mean Scott’s theory is not “policy invariant” and therefore useless? Possibly, possibly not, it doesn’t matter though because right now what he proposes would be useful. That is how theories should be evaluated.

Subsidiary to this, is my confusion as to what would qualify as policy invariant, “deep parameters.” All aspects of the modern economy are unnatural to an extent; the whole edifice is upheld by the state. I’m not sure what a theory which ignored everything apart from “deep parameters” would like like, but I suspect Hobbes got closest and that wasn’t very close; I remain suspicious of what this approach has to teach us about day to day macroeconomics.

Lastly, Noah and Nick‘s posts defending Lucas’s insistence on microfounding models were very, very fact light. In fact, neither cited anything that suggested the Lucas critique had been empirically verified. This led me to find out whether there was any evidence to support the Lucas critique. It turns out In 1995 “an extensive search of the literature reveals virtually no evidence demonstrating the empirical applicability of the Lucas Critique.” [pdf] A cursory search of more recent literature hasn’t thrown up anything which says “look, proof!” to me. I’d say the last four or five years reveal a similar dearth of useful policy advice.

In summary, the most major problem of microfounding macro models is that it doesn’t seem to have helped us produce anything useful for macroeconomics. Modernmacro has cost me a lot of money in forgone earnings. It has produced some nifty statistical techniques and I’m a big fan of nifty statistical techniques, but they aren’t very useful at putting my friends and I to work.