In 1936, John Maynard Keynes came up with a macro model that was a product of its time. That’s the wrong way to do macro. Models should be based on the empirical facts of all countries and all time periods.

If your macro model cannot explain why the US experienced a major deflation (with NGDP falling nearly 30%) from mid-1920 to mid-1921, the model is useless. If it cannot explain why industrial production suddenly fell by 30% after mid-1920, the model is useless. If your model cannot explain why the 1921 depression was followed by a very quick recovery, whereas the 1929-33 depression was followed by a long 8-year recovery, the model is useless. You need a truly “General Theory”, applicable to all times and all places.

Unfortunately, most macro models cannot explain these important facts, and many others.

My previous post on the US and Hong Kong Phillips Curves got some positive feedback:

Brian Donohue
Jun 21 2019 at 4:45pm
Very clever.

Jun 22 2019 at 3:49am
Agreed. With empirical data like this, who can deny the monetary framework is the right way to do macro-economics. The theory is sound but that doesn’t always transfer into empirics.

When I wrote the post I overlooked this point.  It seems obvious to me that macro models must be monetary models.  But lots of people disagree. So Chris is right, it does constitute a powerful argument for the monetary approach to macro, which many people still do not accept.  What other explanation is there for the difference between the US and Hong Kong Phillips curves?  Surely there is no alternative explanation as simple and elegant.

Think of the following two-part model:

NGDP = M*V(i, bank failures), where velocity is positively related to nominal interest rates and negatively related to risk of bank failures.  (Since 2008, you need to add IOR.)

And unexpected NGDP shocks affect real output:

This is a simple but incredibly powerful framework for analyzing a wide range of macro issues.  Of course it’s just a framework; we still need to model all of the factors that impact aggregate supply.  But this framework can easily explain the stylized facts that I discussed above.  Thus the big decline in NGDP during 1920-21 was mostly caused by a huge decline in the monetary base.  Note that the monetary base does not even appear in most New Keynesian models.  What is their explanation for 1920-21?  The quick recovery was due to the fact that nominal hourly wages were more flexible in those days.  The contraction of 1929-33 was caused by a fall in the monetary base during 1929-30, and a fall in velocity during 1930-33, caused by lower interest rates and bank failures.  The slow recovery from the Great Depression was caused by the NIRA, the Wagner Act, and minimum wage laws, which sharply contracted the supply of labor, and also the contractionary monetary policy of 1937-38.

The framework also applies to modern times.  The deep recessions of 1982 and 2009 were caused by NGDP growth falling far below expectations, due to tight money policies.