Supply and demand-side stagnation
By Scott Sumner
Tyler Cowen often has posts entitled “A very good sentence.” Here Tyler dishes up one of his own:
Yes, it is a big mistake to assume Say’s Law always holds but it is an even bigger mistake to think it never holds.
I’d like to discuss the following sentences:
Supply slowdowns are bad for demand, and they likely are bad for credit creation too, which hurts demand further yet.
There is no contradiction in a model where both aggregate demand and aggregate supply curves shift in unfavorable directions! And in the medium run, each of these shifts pushes the other curve around too.
I think this is right, and in the past I’ve discussed how the AS and AD curves are “entangled” in practice. But it’s also important to understand the mechanism(s), because there is nothing “natural” about this entanglement.
Let’s start with the easiest entanglement to explain, the medium term. Suppose NGDP trends upward at 5%/year, with 3% RGDP growth and 2% inflation. Then we are hit by a supply shock that reduces growth to 0% for one year, assuming we stay at full employment. What happens next? If the Fed is targeting NGDP, then inflation probably rises to 5% and real growth falls to zero percent. Employment is roughly unchanged. No impact on demand. But let’s say they are targeting inflation. Then NGDP growth slows sharply. And because nominal hourly wages are sticky, employment also falls sharply. Now RGDP is falling for two reasons, falling productivity (supply shock) and falling employment (due to a demand shock). RGDP turns negative. This isn’t exactly what happened in 2008-09, but it does pick up some of what occurred.
Now let’s move to the very short run. If the central bank is targeting interest rates in the very short run, a supply shock may lead to lower AD. For instance, a sharp rise in the minimum wage or oil prices might depress business investment. This reduces the demand for credit, depressing the Wicksellian equilibrium interest rate. If the central bank is slow to react (keeps targeting interest rates at the same level) then NGDP starts to fall. Thus AS and AD are entangled in both the short and medium run. But how about the long run?
The natural rate hypothesis says that it wouldn’t matter if AS and AD were entangled in the long run, because AD doesn’t matter in the long run. Thus long run stagnation cannot be a demand-side problem. Is this prediction of the standard model correct? The old Keynesians say no, as do new Keynesians who are increasingly drawn to the old Keynesian model (Summers, Krugman, etc.) It mostly hinges on how you feel about money illusion and wage stickiness near the zero level of wage increases. And at an empirical level, it depends what you make of the Japanese experience over the past two decades.
I’m agnostic on this question. Ironically, Krugman just presented some data that seems to slightly undercut his argument. He has a graph that shows the Japanese have not done as bad as people assume over the past few decades, if you adjust for growth in the working age population. First one small quibble with the graph; I’d like to know how many elderly Japanese are working before I assume the working age population is the right figure to use when adjusting GDP. Japan has a lot of old people, and their number is growing really fast. So Japan may have done worse than Krugman indicates.
But let’s put aside that issue, and assume Krugman’s graph is basically correct. When I look at the graph I see a country falling well behind the US for about a decade, then partly catching up. Younger readers should keep in mind that the relative decline in the 1990s was a big shock to most Westerners. As late as 1991 Japan was viewed as a sort of turbo-charged super-economy, with a more efficient economic model than the West. So I think there is at least a pretty strong prima facie case to be made for the claim that it was the sharp slowdown in NGDP growth that triggered the underperformance. But Krugman’s graph indicates that this underperformance lasted for about one decade, not two. This suggests that in the very long run AD shocks might not matter, or (more likely in my view) matter only a very small amount. It’s hard to tell because the baseline (other developed countries) had their own problems more recently.
Returning to Tyler’s great sentence, despite my constant bashing of the ECB, here’s what I would say about Europe:
1. The sharp rise in European unemployment during 2008-09 and 2011-13 was due to tight money that slowed NGDP growth. Note that BOTH slumps immediately followed ECB tightening, by any measure you choose (NGDP, or target short term interest rates.)
2. The enormous difference in the performance of countries like Greece and Italy relative to Germany and Austria is a supply-side story. They face the same monetary policy. And in the long run that divergence may be the biggest story in the eurozone. The rise in unemployment is really important, but the long run failure of certain economic models is really, really important.
So why do I focus so much on monetary policy? Partly because it’s by far the easiest problem to fix, structural reforms are much harder. Note how Abe quickly changed monetary policy in Japan, ending deflation, but wasn’t able to enact structural reforms. And second, it’s my area of expertise; prior to 2008 I had devoted my life to studying issues like the Great Depression, NGDP targeting, and the Japanese liquidity trap. There are lots of $100 bills lying on the sidewalk waiting for the ECB to pick them up. (OK, 500-euro bills.) I’m most useful to society if I point that out.
PS. NGDP growth (or growth expectations) is the proper way to measure the stance of monetary policy. In other words, M*V. But some commenters demand “concrete steps.” So in response I sometimes point out that a certain slowdown in NGDP growth was triggered by a target rate increase at the central bank. Then commenters complain “but you told us interest rates weren’t the right indicator.” I can’t win.