Blanchard is Still Wrong About Inflation
Blanchard is Still Wrong About Inflation
Let go of the Phillips curve
The American Economic Association annual shoulder rub concludes today. From the reporting, economists are now overwhelmingly sold on Isabel Schnabel’s thesis that the world economy has transitioned from “the Great Moderation” to “the Great Volatility”. The central problem, of course, is price stability.
Olivier Blanchard ignited quite a kerfuffle on Twitter when he raised the question of distributional conflict as the motor of inflation.
Economists paying attention to class struggle is sure to ignite passions on the left — did he really just say what we think he did? Some were no doubt fooled into thinking that putting politics back into the inflation question was a good thing — isn’t that what we want?
No.
The upshot of inflation being a consequence of distributional conflict is that, in order to tame inflation, wage growth must be tempered — central banks must cool the labor market to achieve price stability. This is a bad idea.
It is hardly surprising to see Blanchard insinuate that tight labor markets are to blame for the present bout of price instability. Recall his remark when Brainard questioned the Phillips curve orthodoxy.
The notion that a tight labor market will lead to inflation is impossible to contradict.
Olivier Blanchard. Rethinking Macroeconomics Conference. October 12, 2017.
This is a guy who thinks that inflation is governed by the tightness of the labor market; empirical evidence be damned. The kerfuffle prompted Tooze to question whether anyone serious was actually worried about a wage-price spiral.
Outside the more conservative corners of European central banking is there anyone, anyone at all, who is seriously worried about wage-price spirals?
Adam Tooze. Jan 4, 2023.
I may be less sanguine than Tooze about policy errors. There is a real risk of intellectual regression. Monetary policymakers are indeed reaching for old ways of thinking about inflation. This is evident from talk of wages as “supercore inflation” and other labor market-centric noises coming from FOMC members. This is a big risk.
We’re shown previously that inflation in advanced open economies is not a function of domestic slack. Instead, it is a function of slack in the global production system servicing the Western consumer as a whole. We revisit the empirical evidence and ask whether there’s any reason to do a Bayesian update of our model of the inflation process.
We obtain data on inflation and unemployment from the IMF. We restrict the sample to twenty advanced economies because EM inflation rates are confounded by unanchored inflation expectations. We isolate the global factor in inflation from our AE panel using latent factor analysis.
As is clear from the graph, the global factor closely tracks median inflation in advanced economies. We can think of it as the rate of inflation in the global tradable sector jointly faced by all advanced economies. The global factor thus contains a very strong signal of slack in the global production system as a whole. We expect inflation in the advanced economies to be more sensitive to the global factor than domestic slack.
We fit panel regressions with CPI as the response, and the global factor and domestic slack as features. We proxy domestic slack by the unemployment rate. We detrend by first-differencing and standardize all variables to have zero mean and unit variance. The slope coefficients therefore represent effect sizes measured in standard deviation units. In all our regressions, we control for the price of crude, dollar strength, country fixed-effects, and lagged CPI. Country fixed-effects and the persistence term are included but not shown.
Our estimates show that AE inflation is a function of global slack, not domestic slack. The elasticity of inflation in an advanced economy against domestic slack is statistically indistinguishable from zero at the 10 percent level (P = 0.254). Meanwhile, the elasticity of inflation against the global factor is very large and statistically significant at the 1 percent level (b = 0.45, P < 0.001). We do note, however, that the signs of the slope coefficients of the covariates are in line with theoretical expectations.
What about recent evidence? Have the elasticities changed over time? In order to test the hypothesis of recent changes in the inflation process, we divide the data into pre-Covid and post-Covid samples, and compare elasticity estimates.
We find that AE inflation has become more sensitive to the global factor since Covid, not less. The elasticity against the global factor has increased from 0.47 to 0.66, although the earlier estimate is within the 95% confidence interval of the more recent estimate (0.45, 0.86).
We also find some evidence that domestic slack has become significant. The elasticity of AE inflation against domestic slack has increased in absolute scale from -0.03 to -0.11, although the earlier estimate is within the 95% confidence interval of the more recent estimate (-0.21, -0.01). The effect is small and could very well be spurious. In any case, our point estimates suggest that the global factor is still six times as important as domestic slack in the determination of AE inflation.
In order to better understand the time-variation, we split the data into five year sample periods (three years for the last period, 2020-2022) and reestimate the elasticities. We find that the global factor has increased in importance over the past three decades. Indeed, inflation in the recent past is more sensitive to the global factor that it has ever been. As for domestic slack, again we find some evidence that it may have begun to matter. But what is clear is that the effect of domestic slack is completely dwarfed by that of the global factor.
Our baseline full-sample estimates suggest that a one standard deviation shock to our proxy for global slack increases US inflation by 1.3%, whereas a one standard deviation shock to the US unemployment rate increases US inflation by just 0.1%. If we restrict the sample to 2020-2022, we find elasticities of 1.9% and 0.3% respectively. Thus, global slack is still an order of magnitude more important in the inflation process than domestic slack.
Blanchard may be “one of Europe’s most prominent economists” but he is simply mistaken about the inflation process. It’s hard to see how evidence against the Phillips curve theory of inflation can ever get him to lose his religion. But one poorly-informed technocrat is not so important. What we should be worried about is broader intellectual regression among economists.
Inflation could come down rapidly in 2023, stay stubbornly high, or even start climbing again. Given the substantial uncertainty around the inflation outlook, it is extremely important for central bankers to refrain from imposing their theoretical priors on the data. Monetary policymakers should not base their expectations of the future path of inflation on the very poor signal in the tightness of domestic labor markets.
Other sources of inflation are the "Beer Game", dead old people, and monopolistic effects. Monopolistic effects are part of the Firm's sector; the monopolizers have admitted this in public. The Beer Game is part of supply shock. 1+ million people dead is part of labor shock. A lot of fast food etc. labor is people on Social Security, or people who relied on Grandpa to take care of the kids but Grandpa died. And then there's the proxy war.
In short, we have a lot of things going on that the PMC can't talk about, and economists are high-status members of the Professional-Managerial Class.
"Monetary policymakers should not base their expectations of the future path of inflation on the very poor signal in the tightness of domestic labor markets."
Depends on what the desired outcome is.