What causes recessions and "empressions"?
By Scott Sumner
Before starting this post, let me highly recommend George Selgin’s recent post on NGDP targeting.
Nick Rowe has a post discussing a scenario where a lack of media of exchange disrupts trade, without affecting employment and output:
I would call that a “recession”, even though (by assumption) output, employment, and (aggregate) consumption are unchanged. People are worse off, because of a reduction in the volume of exchange, due to a reduction in the circular flow of money around the Wicksellian triangle.
Nick’s terminology is unconventional; recessions are usually defined more in terms of output and employment. I’d prefer a third definition—a sharp rise in the unemployment rate, regardless of what happens to output. But I don’t get to choose definitions, so rather than fight a losing battle I’ll invent a new word for the concept I’m interested in. Let’s call an employment recession an “empression”.
To see the difference, imagine a primitive economy where all workers are peasant farmers. There is zero unemployment, as all are self-employed. A spell of bad weather would cause a recession (falling output), but not an empression. However, it just so happens that in the US all recessions seem to also be empressions, and all empressions seem to be recessions. In the following graph, the grey bars reflect recession periods. You can see that sharply rising unemployment is a necessary and sufficient condition for a recession. You can’t say that about inflation, stock prices, yield spreads, steel output, or numerous other variables.
Lots of things might cause a higher unemployment rate. These include:
1. Sharply higher minimum wage rates, or a surge in union organizing.
2. A sharp rise in the share of GDP going to capital, meaning less money to pay wages.
3. A sharp rise in hours worked per week, meaning fewer workers are needed to produce the same nominal output.
However, I don’t believe that any of those factors are important causes of the US business cycle, at least since WWII. Rather the problem is sticky nominal hourly wages and unstable NGDP. Before looking at NGDP, lets examine the growth rate of total labor compensation:
Notice that nominal labor compensation growth slows sharply during recessions and empressions—every single time. So the problem does not seem to be a surge in hourly wages, rather the labor market is being starved of funds to pay workers—my musical chairs “model” of unemployment. (Or perhaps “metaphor”, as respectable economists wouldn’t think it rises to the level of being a model.)
So what causes nominal labor compensation growth to slow at various times? Does the corporate sector suddenly grab a bigger share of national income, leaving less money to pay workers? Or does growth in national income itself slow? Not surprisingly (as labor compensation is a big share of national income) it’s the latter. It turns out that falling NGDP growth (combined with sticky hourly wages) is the proximate cause of both recessions and empressions:
Recall that I said that wage spikes don’t seem to be the cause of recessions and empressions. There is, however, one wage spike that made a recession/empression somewhat worse than one would have otherwise expected. Notice that the slowdown in NGDP growth was pretty modest during the 1973-75 recession. And yet that was one of the more severe postwar recessions/empressions. Why? It turns that that 1974 saw an unusual wage shock, something that generally does not occur during US recessions:
Why did wage growth spike during the 1974 recession? I’d guess it was because Nixon phased out his wage controls during 1974, and workers demanded wage increases to compensate for the high inflation of 1973-74. But while that sort of situation may be common in some unstable developing countries, it’s pretty unusual in the USA. And even during 1974, slowing NGDP growth was still part of the story. The 1974 recession was one part NGDP shock and one part wage shock.
So the cause of post-war US recessions is actually quite simple. NGDP growth slows while nominal hourly wages are sticky, and thus employment falls while unemployment rises.
Why don’t workers offer wage flexibility to prevent high unemployment? For the same reason that Wall Street financiers don’t offer indexed bonds to prevent falling NGDP from creating financial crises—it’s a collective action problem. If any one worker agrees to flexible wages, it doesn’t help him preserve his job. He shows up at the factory gate and finds his workplace is closed down. Only if all workers have flexible wages can we avoid a recession/empression during periods of sharply slowing NGDP.
If all workers bargained collectively that might be possible. But a labor union that covered all 150,000,000 workers would create lots of microeconomic inefficiency, which might be even worse than the business cycle. Better to use monetary policy to keep NGDP growing at a steady 4%/year, or something close to that figure.
To summarize, recessions/empressions are quite simple. A combination of sticky nominal wages and unstable NGDP (i.e. unstable monetary policy) causes recessions. At elite universities they have models that don’t even feature nominal wages and NGDP. Rather they focus on price inflation, interest rates, output and other irrelevant variables. Again, sticky wages and unstable NGDP are pretty close to a necessary and sufficient condition for US recessions/empressions—no need to look for microfoundations. No need to make it complicated.
The policy implications are also simple. Adjust monetary policy to keep market expectations of NGDP growing along a 4% trend line. That will mostly solve the problem of empressions in the US, and any remaining movement in RGDP will be an efficient “real business cycle”, not be worth worrying about.