The great sin, the even greater sin, and the enlightened path
By Scott Sumner
The title is my pathetic attempt to imitate Miles Kimball and Noah Smith, who sometimes post on religion. This won’t be about religion; it’s about monetary policy. Oh wait . . .
In the 1970s, US policymakers knew that inflation was too high, but weren’t willing to bring it under control, as they worried about the effect on unemployment. And in some cases (such as 1972 and 1980) worried about the next election. Britain had the same problem, but far worse than the US. German policymakers did somewhat better (but still rather poorly in absolute terms) because they had more self-control than policymakers in the English-speaking world.
Thus the great sin of monetary policymaking is a lack of self-control. It’s putting the “one marshmallow eaters” in charge of the central bank.
OK, if that’s the great sin, then what is the even greater sin? The even greater sin is fear of a lack of self-control.
In the late 1920s and the early 1930s, policymakers feared what would happen if money was not backed by gold. They had seen the German hyperinflation, and not surprisingly had a rather low opinion of fiat money. The gold standard was seen as disciplining irresponsible policymakers. Something similar occurred in the 1990s in Argentina, when they set up their currency board. They also had memories of a hyperinflation about a decade earlier. And something similar happened when the euro was set up, without enough flexibility to easily prevent deflation, and without allowing the safety valve of devaluation. The architects of the euro recalled the irresponsible monetary policies of some of the Mediterranean countries, in the previous few decades. They wanted to make that sin impossible. And so like the greatest of the Mediterranean heroes, they tied their hands to the mast.
In each of those three cases, policymakers who feared a lack of self-control set up a policy apparatus that was extremely difficult to dismantle, with the euro coming closet to a true “doomsday machine.” These regimes did successfully prevent high inflation, but they all had a fatal flaw. None of them could adapt to a circumstance that was not anticipated by the creators of their regime—a sharp increase in the demand for the medium of account (gold, the US$, the euro, respectively), which led to falling NGDP. And because nominal wages are sticky and debt contracts are in nominal terms, these regimes led to mass unemployment and severe financial crises.
Why is fear of a lack of self-control a worse sin than lack of self-control? After all, in ordinary life it would be a lesser sin. But monetary economics doesn’t follow the rules of everyday morality. The damage of being “too responsible” is far greater than the damage of being “irresponsible.” That’s because wages are much stickier in the downward direction. Hence the German deflation of 1929-33 did more damage than the hyperinflation of 1920-23, even if one ignores the elephant in the room (Hitler.) The Great Inflation of 1965-81 did less damage than the Argentine deflation of 1998-2001 or the recent euro depression. That’s just the way the world works.
Are those our only two choices? No, fortunately there is an enlightened path. The first step is to recognize our ignorance. (I believe NGDP level targeting is best, but might well be wrong.) In that case policy should be conducted from a “timeless perspective” and the policy regime should be occasionally updated to reflect new macroeconomic knowledge. Thus the policymakers of the 1970s should have realized that from a timeless perspective low inflation is better, because there is no long run trade-off between inflation and unemployment. But maybe that (natural rate model) is slightly wrong. Some studies suggest that below a certain level (say 2%) there is a permanent trade-off. So maybe you aim for 2% inflation rather than 0% inflation.
Then new knowledge might show that inflation is not the best target, NGDP is better. Which NGDP target is best? Consider the cost of transition. People made plans based on 2% inflation, at least in the long run. So have the changeover occur at a NGDP target path that is expected to produce 2% inflation, on average. Perhaps 4% NGDP growth. That way debtors and lenders won’t feel robbed. You can write up models where those sorts of sunk costs don’t matter, but those models only make sense if NGDP is the final policy adjustment. But it isn’t. If you mess around with the average inflation rate this time, people won’t believe your promises about NGDP.
Don’t think of these policy changes as constantly changing direction, but rather as iterating in closer and closer to an optimal policy. Of course we will never arrive, it is always about making policy less bad. Money is useful for transactions, but monetary systems are always more or less harmful in a macroeconomic sense—they always move you at least some distance away from a frictionless world with a Walrasian auctioneer.
1. Show self-control; use a timeless perspective.
2. Don’t show too much self-control, shift course when you discover your policy regime is fundamentally flawed.
3. Always recognize that your (and my) favorite policy is wrong, and always be on the lookout for “less wrong” policies. Just as Einstein’s general relativity is less wrong than Newtonian mechanics.
PS. I am indebted to Eliezer Yudkowsky for the phrase “less wrong.”
PPS. You might think this is hopelessly utopian; real world policymakers are not enlightened. I don’t fully agree. We have moved from a bad rule (gold) to a discretionary policy regime (also bad), to a sort of Taylor Rule (better) and don’t see why we can’t then move to a NGDPLT policy (still better), and then perhaps a total nominal labor compensation target (if we were to assume that was still better.) Despite everything, I remain an optimist.