Nick Rowe writes,

Monetary disequilibrium theorists will disagree with those Keynesians. We add monetary exchange to the mix. We don’t buy and sell output for labour. We don’t buy and sell output for bonds. We buy and sell output for money — the medium of exchange. We buy and sell everything for money — labour, bonds, and output. If there’s an excess demand for money, so the desired stock of money exceeds the actual stock of money, people will be trying to sell more of everything than they buy of everything. An excess of AS over AD means an excess demand for money. It’s the real stock of money M/P that must adjust to equilibrate AD and AS.

If I have this right (and the main reason I am writing this post is to get feedback on whether I have this right), then this is a bit different from the Scott Sumner mechanism. In Rowe’s world, when the money supply contracts, producers are caught temporarily off guard and prices stay too high. This reduces M/P, leading to lower spending and output. In Sumner’s world, when the money supply contracts, price-setters read the situation clearly but nominal wages are sticky. Nominal GDP falls relative to nominal wages, and output contracts.

Going out further on a limb, I would say that in Rowe’s monetary disequilibrium world, we would observe output falling first (due to a drop in the money supply or an increase in money demand), leading to disinflation. In principle, we could see output decline without a drop in inflation. In an old-fashioned aggregate-supply world (like the one I use to characterize AS), we would see disinflation first, leading to lower output. The catch is that for Sumner, output drops when expected inflation drops, so that we could observe the output decline before we see the inflation decline. But overall, we would be surprised if output dropped without disinflation.

This might be a good time to bring up the issue of paradigms vs. theories. A theory is something that stands or falls on a single decisive test. You’re Ben Franklin, you have a theory that lightning is an electrical phenomenon, so you fly your kite with a key on the end of it, and hopefully you live to tell about the confirmation of your theory.

A paradigm is more like a language. There is no decisive test that tells me that English is “true” and French is “false.” Overall, I feel more comfortable blogging in English than in French. A computer programmer these days might feel more comfortable coding in C than in FORTRAN, but that does not refute the latter.

In science, comfort with a paradigm depends in part on empirical issues and in part on the overall “fit” with other theories. The AS-AD paradigm can be made to fit any observed data. If prices and output are moving in the same direction, you call it a shift in AD. If prices and output are moving in the opposite direction, you call it a shift in AS.

One of the reasons that people proclaim the Recalculation Story bogus is that they think of it as an AS shift, and yet we see prices and output moving in the same direction. But that is like running a French sentence through an English translating program, getting a nonsensical result, and pronouncing French a bogus language.

For me, the nice thing about PSST (patterns of sustainable specialization and trade) is that it fits so well with normal economics. Normal economics is about entrepreneurs trying to identify opportunities to exploit specialization of labor and comparative advantage. Normal economics (or normal Austrian economics) does not take equilibrium for granted. Instead, the economy is constantly changing, and individuals and firms are constantly adapting.

One can argue that both AS-AD and PSST are about short-run adjustment in an economy that is far away from a desirable equilibrium. However, the AS-AD approach sees adjustment as being all about getting to the correct aggregate relative prices (the version that I am most used to looks at W/P. but if you prefer M/P that does not ease my concern). The PSST approach sees adjustment as much more complex, requiring a lot of entrepreneurial trial and error to resolve.