There were two posts yesterday. The first was “Is The Economy Self-Correcting? (Wonkish):”
What I want to focus on in this post is the suggestion by Brad DeLong that I missed a failed implication of Hicksian analysis — that demand shocks should be short-term in their effect. Actually, and very unusually, I think Brad has this wrong. The proposition of a long-run tendency toward full employment isn’t a primitive axiom in IS-LM. It’s derived from the model, under certain assumptions. But there’s good reason to believe that even under “normal” conditions it’s a very weak, slow process. And under liquidity trap conditions it’s not a process we expect to see operate at all.
How is the self-correction of an economy to its long-run equilibrium supposed to work? In textbook analysis, the story is that falling prices raise the real money supply, pushing down interest rates, and hence restoring employment.
So how rapidly would we expect this process to work? Let’s take the most favorable assumption, which is that of a constant velocity of money. Under those conditions, holding the money supply fixed would also hold nominal GDP fixed, so that a one percent fall in the price level would raise real output by one percent. The question then is how responsive prices are to the output gap.
Well, Blanchard, Cerutti and Summers have a new paper that estimates an an “anchored expectations” Phillips curve (aka an old-fashioned, pre-Friedman/Phelps curve), and finds the coefficient on unemployment for the US to be about -.25. That’s for unemployment; on output, given Okun’s Law, the coefficient should be only half that. This implies a half-life for output gaps of around 6 years. The long run is pretty long, in other words; we might not all be dead, but most of us will be hitting mandatory retirement.
And that’s assuming constant velocity. With interest rates dropping, part of the fall in prices should translate into a fall in velocity rather than a rise in real output, so the implied speed of adjustment should be even lower.
But wait, it gets worse: at the zero lower bound the process doesn’t work at all. In a liquidity trap, the proposition of a self-correcting economy falls down — in fact, what more flexible prices would do, arguably, is bring on a debt-deflation spiral.
Yes, a sufficiently large price fall could bring about expectations of future inflation — but that’s not the
droid we’re looking formechanism we’re talking about here.
You might ask, given this logic, why actual slumps usually don’t last all that long. The answer is, first, that the shocks causing slumps are often temporary; but second, in practice central banks don’t sit there passively, holding the money supply constant, but in fact push back against slumps with expansionary policy. The economy isn’t self-correcting, at least on a time scale that matters; it relies on Uncle Alan, or Uncle Ben, or Aunt Janet to get back to full employment.
Which brings us back to the liquidity trap, in which the central bank loses most if not all of its traction. Nothing about basic macro models says that there should be a fast return to long-run equilibrium under those conditions, so the failure to see such a fast return is actually a point in favor of the model, not a failure.
Yesterday’s second post was “Hyperglobalization and Global Inequality:”
I’ve mentioned before that I’m a big fan of work by my CUNY colleague Branko Milanovic showing that if you look at income growth by percentile of the whole world population for the past 25 years, you see “twin peaks”: rapid growth near the middle, representing China’s middle class, and at the top, representing the global elite, with a sag in the region representing the OECD working class. But was it always thus? I asked Branko what a similar picture looked like for the previous generation — and he has obliged.
Here’s a version of the original Milanovic history of the world in one chart:
Branko calls this the elephant picture, although I don’t see it; Janet Gornick says it’s a camel. I say it’s very like a whale …
But here’s the picture for the preceding generation — by “ventile”, that is, 5 percent interval, rather than percentile — and it doesn’t look at all similar:
So pre-1990 or so we had no visible tendency toward either inequality or equality at a world level; the elephant-camel-whale is for the later period. That’s interesting, because post-1990 is also the period of hyperglobalization, of unprecedented growth in trade due to slicing up of the value chain.
How big is the causal link? I don’t know. But there is, I’d argue, an important and interesting stylized fact here for us to work with.