In his Bloomberg collum “How Were So Many Economists So Wrong About the Recession?” Tyler Cowen piles in on a popular question. His answer is that is what their models told them. And Cowen is right that in ordinary New Keynesian models, “needing” a recession to bring inflation back down from over-target to target levels is a standard result. The conventional wisdom of a year+ ago predicting a rescission was indeed just that, conventional. He does not, however get at why experience this time was different. If it was the models, why were they wrong?
The models used to predict unemployment/recession have an implicit Philips Curve [see The Rise and Fall(?) of the Phillips Curve (substack.com) ] built into them, as is appropriate for analyzing macroeconomic shocks that are uniform enough across sectors to enable treating the economy as if it produces one good with one labor input. In that kind of model there is only one relative price that can get away from its income-maximizing ratio, the wage/good price ratio. If the Fed is engineering/accommodating inflation in this scenario and decides to change course, it has but one point of entry, to reduce demand for and the (rate of increase in the) price of goods, relative to labor. If the relative price of labor cannot fall (fast enough) it will rise above a market clearing level and unemployment will result: the typical Phillips Curve result.
What was (somewhat) different about this time was that the shock that the
Fed was accommodating with temporarily over-target inflation in 2021 was a (series of) sectoral shocks: the shift from demand for services to demand for goods because of COVID, the volatility of grain and petroleum prices (and natural gas prices in Europe) after the Russian invasion of Ukraine, and the effects of these on transportation logistics. It was relative prices among goods that became askew, not between labor and the one “good.” This meant that reducing aggregated demand with monetary policy instruments – the higher EFFR – did not necessarily lead to raising the relative piece of labor relative to the decelerating price level. Demand for labor could continue to expand and unemployment to fall even as demand for some kinds of good contracted and others increased. The Fed policy response that in a one-good economy would produce recession did not in the multi-good economy.
Cowen scarcely acknowledges, however, that there WERE recession doubters. He only reluctantly credits Krugman for NOT predicting recession even though he gestures at Krugman’s reason for being correct, “unkinking” of supply chains (readjustment of relative prices among goods). Cowen fails even to mention other members of “team transitory, like Brad Delong. [https://braddelong.substack.com/p/the-feds-remarkable-feat]
Amusingly, Cowen “credits” Larry Summers for consistency in continuing to be a recession pessimist. 😊
[Standard bleg: Although my style is know-it-all-ism, I do sometime entertain the thought that, here and there, I might be mistaken on some minor detail. I would welcome what the diplomats call “a frank exchange of views.”😊]
There is an atavic attachement to the Phillips Curve by prominent economists. Even after they die, their "legacy" will live on...
https://marcusnunes.substack.com/p/the-mirage-of-spiking-unemployment
Nice post. It's interesting that, for all his fulminations against the representative agent, the "representative good" and "representative wage earner" problem you point to goes right back to Robert Solow's aggregate production function. He clearly did not like its being tricked out via the Cass-Koopmans-Ramsay model; he liked even less the production function's use as a hapless pillar of RBC and New Keynesian models. If he could read your post, he would have said, maybe, "As this shows, toy models should stay home."