Why do macroeconomists keep changing their model?

I have a new article in Economic Affairs, which discusses the prospects for monetarism in the 21st century. Unfortunately, the journal was not able to use my graphs, therefore I’d like to discuss one omitted graph that illustrates an interesting trait of macroeconomics—it’s lack of fixed principles.  There is a tendency of macroeconomists to shift with the intellectual fashions of the day.

In the paper, I discussed three general approaches to monetary economics.  

1.  Old Keynesianism:  Money supply data uninformative and monetary policy is often ineffective.

2.  New Keynesian:  Money supply data is uninformative and monetary policy is highly effective.

3.  Monetarism:  Money supply data is informative and monetary policy is highly effective.

Old Keynesianism is popular when inflation is so low that nominal interest rates fall close to zero.  In that environment, one often sees large increases in the monetary base coinciding with very low inflation (left portion of the graph).  This leads many to assume that monetary policy is ineffective at the zero lower bound.  Monetarism is least popular during these periods.  Keynes’s General Theory was actually a special theory for an economy with near zero inflation expectations.

New Keynesianism is most popular when inflation is at moderate levels and fairly stable, say from 1983 to 2007.  During these periods, there is little correlation between money growth and inflation, mostly because inflation is quite stable.  It’s not that money doesn’t matter, rather it’s an example of what Milton Friedman called the thermostat problem.  If you skillfully adjust a thermostat to keep the temperature at a constant 72 degrees, it looks like the thermostat is not influencing the temperature.  New Keynesians do regard monetary policy as still being highly effective, but they focus on interest rates, not the money supply.

Monetarism is most popular during periods of high and unstable inflation, such as the 1970s.  During those periods, there is usually a close correlation between long run money growth rates and inflation (right portion of the graph).  In contrast, interest rates become an unreliable indicator of the stance of monetary policy due to the Fisher effect.

Personably, I view it as a major embarrassment that macroeconomists shift between these models according to the inflation trends of the moment—like teenagers changing their style of dress each fall.    We need a monetary model that fits any macroeconomic environment.  My preference is market monetarism, where NGDP futures are both the indicator and instrument of monetary policy.  Where monetary policy is always effective and fiscal stabilization policy is never needed.

I am currently working on a book that will make the case for a truly “general theory”, a monetary theory that explains monetary policy in both Japan and Zimbabwe.