On umbrellas causing rain

By R.A. | LONDON

NOAH SMITH writes on a bit of macroeconomic rebellion that refuses to go away: the view that low interest rates cause low inflation. That is the precise opposite of the conventional view—that central banks which hold interest rates at a low level risk courting soaring prices. But how could the heterodoxy be right? How could macroeconomists have gotten such a basic relationship exactly wrong?

The basic logic of the argument is as follows. The economy has an equilibrium real, or inflation-adjusted, interest rate. The real interest rate is essentially the nominal interest rate minus the inflation rate. So if the central bank pushes nominal interest rates down to a low level, then over the long run the inflation rate must inevitably move toward a level consistent with the long-run equilibrium real rate. That is, inflation must fall. Otherwise, the economy would be out of equilibrium forever, which is not how economies work (in most economic models, anyway).

The secondary argument is empirical: it used to be the case that interest rates were high and so were inflation rates. In recent years interest rates have been very low, which one might have expected to lead to high inflation. Instead, inflation has stayed very low.

Mr Smith offers a thought experiment to help us wrap our minds around how this might work:

Suppose there was no government debt, and the Fed raised nominal interest rates to 20% and held them there forever. What would happen to real rates? Well, they wouldn't rise to 20% forever, because there's just no way that our society can physically, technologically deliver a 20% riskless rate of return to bondholders. Eventually, one of two things would have to happen: either 1) the Fed's control over the nominal interest rate would break down, or 2) inflation would rise (the Neo-Fisherite result).

Make sense? Try this. Imagine the Fed sets the interest rate it pays on banks' excess reserves to 20%. This would immediately cause the economy to fall into deep depression. Banks would liquidate their loan portfolios and raise deposit rates in order to maximise their excess reserves. Consumers would liquidate their asset holdings and put everything in bank deposits since, as Mr Smith notes, there are very few investments in an economy that can deliver a 20% return. It would be the mother of all credit crunches.

But after that, the Fed would be pumping money toward banks, and on to depositors, at a pretty rapid clip, in keeping with the 20% interest rate. Depositors would be getting richer much faster. Those that wandered into the smoking rubble of the economy would soon find prices rising, as this rapidly growing money supply competed for limited goods (even more limited than normal, since the destruction of the credit system would be a nasty supply shock). The Fed's action would—eventually—deliver rapid inflation.

But rational markets should anticipate this development. And if they expect rapid inflation in the future, inflation in the present should rise. High interest rates should lead to higher inflation rates.

So is the heterodoxy right? Not exactly. In this story inflation rises in the present because expectations of future inflation have risen. High interest rates are incidental to the story: just part of a crazy central bank plan to convince the public that inflation will be higher in future. A promise to keep rates at zero until inflation reached 20% would work just as well, provided it succeeded in convincing the public that future inflation would be higher. So yes, a central bank can raise inflation by raising inflation expectations.

Are higher interest rates the most effective way to do that? Consider an argument from one of the heterodox economists, Stephen Williamson; this is Mr Smith writing:

Williamson has stuck to his guns. Last December, he wrote a blog post suggesting that Paul Volcker whipped inflation in the early 80s not by raising interest rates in the short term, but by lowering them in the long term.

I would argue that Mr Volcker whipped inflation by reducing inflation expectations, which lowered long-term interest rates. And how did he reduce inflation expectations? Well by pushing short-term interest rates to sky-high levels, inducing a deep recession. The heterodox view is that cutting interest rates should have worked just as well.

Why does that not seem reasonable? Monetary policy works through many channels, but we can divide these channels into two broad categories: the mechanical and the psychological. In a mechanical sense and holding other things equal, rising interest rates tighten policy. More people are interested in saving and fewer people are interested in borrowing. Mortgage rates rise, which chills housing purchases and construction. The currency may appreciate, reducing external demand. Other things equal, cutting rates has the opposite mechanical effect.

The central bank might possible be able to overcome this with a sufficiently powerful psychological signal. But it will be difficult for a central bank to argue that it's easing, for instance, when it's actually raising rates. Why? Because markets understand that raising rates has mechanical tightening effects. What's more, markets understand that raising rates has in the past been used to signal a tightening bias. The central bank might be shouting "go, go, go!" but if it's simultaneously waving around a stop sign drivers will hesitate.

How does this all relate to present circumstances? Well, the higher rates=lower future inflation, lower rates=higher future inflation framework takes for granted that interest rate moves are not constrained. But if interest rates fall to near zero, then the lowest achievable interest rate may be too high. You get a long recession, a shallow recovery, and an extended period in which low interest rates correspond with low inflation. The central bank could further reduce real rates by raising inflation expectations, but the means by which it traditionally signalled a desire for higher future inflation is off the table.

Now the leap some economists make here is that if the association between low rates and high inflation is giving way to one between low rates and low inflation, then raising rates might be the surest way to boost inflation expectations and lift the economy off of the zero lower bound. But that strikes me as very wrong. Just because markets do not anticipate that low rates forever means high inflation does not mean they have lost their association between rising rates and falling future inflation. It certainly doesn't mean there are no longer any mechanical connections from rising rates to slower economic conditions.

The neo-Fisherite position here is that central banks should take a leap of faith, and count on the unorthodox action of raising interest rates to shock markets into expecting higher inflation. But if the plan is to try something dramatic in order to raise inflation expectations, there are lots of options available, many of which would appear to be far more likely to work. The central bank could, for instance, announce that it wanted higher inflation and would therefore raise its inflation target. It could promise to buy foreign exchange and depreciate the currency until inflation were higher. It could do lots of things, in other words, that did not, to the first order, represent a mechanical tightening of monetary policy.

Or to turn the empirical side of the heterodox argument on the heterodox economists: when has raising interest rates while at the zero lower bound led to higher inflation?