Reading Noah's discussion of the neo-Fisherite rebellion - the idea that low interest rates lead to lower inflation and higher rates to higher inflation - I heard the voice of the greatly-missed Andrew Glyn. He would ask: what's the mechanism? How, exactly, might interest rates have such an apparently perverse effect?

Let's start with the process whereby higher rates are meant to reduce inflation. The idea here is that they depress demand, thus opening up an output gap (or raising unemployment for those who prefer observable entities), which in turn reduces inflation.

However one big fact suggests that, in the UK at least, this mechanism is questionable. It's that the correlation between unemployment and inflation in the subsequent 12 months has been positive. Higher unemployment has led to higher inflation, not lower.

This is consistent with - though far from proof of - the possibility that higher interest rates do not lead to lower inflation. One reason for this might be that higher real interest rates act like an adverse supply shock. They not only reduce demand, but also tend to raise prices. I'm thinking of three possibilities here.

One was discussed by Fitoussi and Phelps (pdf) in their analysis of Europe's stagflation of the 1980s. Higher real rates, they say, encourage firms to increase their mark-ups to raise immediate cash rather than pursue growth. This is consistent with Rotemberg and Saloner's discussion (pdf) of why price wars are more likely in booms than slumps.

A second is that higher interest rates, especially if associated with less bank lending, tend to deter new companies from starting up or from expanding. This reduces the external restructuring that is a major source (pdf) of productivity growth, which would tend to add to costs.

The third is simply that interest rates are a cost of production - simply because production takes time and inventories, and higher rates raise the cost of both. (Let's ignore reswitching!)

Through these mechanisms, it's possible that higher interest rates act like higher oil prices or other costs of production. They both create unemployment and raise inflation. They raise the Nairu. Equally, lower rates - at least if associated with more plentiful credit - reduce both unemployment and inflation.

Now, there's an obvious objection to this. If lower rates lead to lower inflation why did Thatcher and Volcker raise rates in the late 70s and early 80s in order to (successfully) reduce inflation?

Here's a possibility. It depends on whether recessions have a "cleansing" effect. If they kill off lame ducks and allow productive firms to grow, their effect will be to reduce inflation. If, however, there's no such cleansing effect, they won't. And of course, whether there is (pdf) or isn't (pdf) might well vary from one recession to another.

In this sense, the question "do lower interest rates lead to lower inflation?" might have the answer: "sometimes yes, sometimes no." Maybe the reality is messier than models claim.