According to Prad Krulong, macroeconomists had it and then lost it. They lost it some time after 1970. We had this magic formula, call it M70, and somehow we let it slip from our collective memory.

But what was M70? It was multi-equation computer models that embodied a set of beliefs about macroeconomics. These models said that inflation was determined as follows:

1. Nominal wage growth is an inverse function of the unemployment rate (this is the infamous Phillips Curve).

2. Real wages are determined by the price-markup equation.

Because of (2), the attempt by workers to raise wages when unemployment is low is an exercise in futility. They can raise nominal wages, but never real wages. In a sense, workers trying to get higher wages are causing an externality. They are creating inflation that hurts everyone and helps no one. Given this externality, the logical policy response is to impose wage and price controls.

Oops.

One reason that macroeconomics as of 1970 is not still around today is that the macroeconomics of 1970 was tried in 1971-1973, with adverse consequences.

If we were to go back to thinking in terms of M70, sooner or later we would again come around to the point of view that inflation is a silly result of market failure. We would once again try to use wage and price controls.

I think that the computer models suffer from a k>n problem. That is, the number of plausible right-hand-side variables exceeds the number of observations. Other people have other issues with the computer models. In any event, the models are a vehicle for articulating a set of beliefs, but they do not provide evidence about the validity of those beliefs.

I think we are left with a much more casually empirical approach to evaluating countercyclical policy. How often has it worked? How often has it failed?

What I suggest is that, without any prejudice, look at the instances over the past 50 years in which countries tried countercylical fiscal policy. What were the successes? What were the failures? My impression is that there were many more of the latter than the former.

Look, I take second place to no one in my dislike for representative-agent, rational-expectations macroeconomics. But I think that the failure of M70 was not just that it lost favor with people who were obsessed with microfoundations and fascinated by math. Much of the failure of M70 was and remains empirical.