In this post, we summarize and discuss three new papers (“Oil Prices, Monetary Policy and Inflation Surges” by Gagliardone and Gertler, “It’s Baaack: The Surge in Inflation in the 2020s and the Return of the Non-Linear Phillips Curve” by Benigno and Eggertsson, and “What Caused the U.S. Pandemic-Era Inflation?” by Bernanke and Blanchard). These papers have drawn a lot of attention in recent weeks. The papers explore theoretically and empirically the questions “why did inflation get so high?”. The answer of the research question is “because of a perfect storm of shocks, together with rigidities in the economy and some bad policy responses”. From an empirical perspective, these papers are a confirmation of our in-house research and forecast (i.e. here and here). In the US, going back to target requires monetary policy to remain restrictive at least for another few/several months.
(We are glad to see that markets are not pricing as many cuts in 2023 as they did only a few weeks ago. Indeed, this is what our research has been suggesting..).
What the papers do
Paper #1: Gagliardone-Gertler (2023) develop a NK-DSGE model that features complementarities (in the use of oil for both firms and households) and real wage rigidities which introduce a short-run inflation/unemployment trade-off. In the event of an oil price shock, complementarities reduce substitution from oil to labor, enhancing the decline in labor demand. Similarly, for households complementarities reduce spending on other goods in the wake of the oil shock. Consequently, the commodity shock has a larger effect on nominal prices (it generates a higher inflation) than in the case without complementarities. The intuition is pretty simple: the increase in marginal cost is much larger with complementarities, owing to the larger reduction in the marginal product of labor implying greater inflationary pressures. The main result of Gagliardone-Gertler (2023) is shown in Figure 1. The figure reports core and headline PCE inflation (the solid black line), the model fit (the black dashed line) and the model contributions. According to the authors, the decompositions resolve a puzzle as to why inflation was low during the period 2014 to 2019 despite low unemployment, while high in recent years despite the same low unemployment level. In the former period, shocks that reduced oil prices in conjunction with tight money shocks helped keep inflation low. In the current period, just the opposite has happened: positive oil shocks in conjunction while easy money shocks (the purple area) have placed upward pressure on inflation. Put it simply: we got a perfect storm of shocks amplified endogenously by some rigidities and not offset appropriate monetary/fiscal policy.
Figure 1. Main result of Gagliardone-Gertler (2023).
Paper #2: Benigno and Eggertsson (2023) propose a non-linear New Keynesian Phillips curve to explain the surge of inflation in the 2020s. The key ingredient (on top of the usual price rigidities) is that economic slack is measured as firms’ job vacancies over the number of unemployed workers (V/U). The contribution is to derive the so called “Inv-LNK Phillips curve” (an-inverse L-shape AS curve) when the measure of economic slack is given by V/U, which also engenders non linearities when there is labor shortage. Benigno and Eggertsson (2023) also provide empirical evidence of a non-linear relation between V/U and consumers’ prices. In particular, they show that when V/U goes above 1 (that is when the number of vacancies is greater than the number of unemployed), core inflation accelerates as much as 4 percentage points above the target (per each pp of V/U). This relation is presented in figure 2, which shows V/U (the blue line “labor market tightness”) and the US CPI inflation rate (the red line). Finally, the paper has a few policy implications, the main one of which is that appropriate monetary policy can bring inflation down without a significant recession and that the recent inflationary surge was mostly generated by a labor shortage — i.e. an exceptionally tight labor market.
Figure 2. Labor market tightness (V/U) and US CPI inflation rate.
Paper #3: Bernanke and Blanchard (2023) estimate a simple dynamic model of prices, wages, and short-run and long-run inflation expectations that approximates a structural vector autoregression (SVAR). With the use of a principal component analysis, Bernanke and Blanchard (2023) show that at least until the invasion of Ukraine, US inflation came primarily from too strong aggregate demand, largely due to the large fiscal packages, reinforced by a relative demand shift from services to goods, and by shortages in a number of markets. Against the interpretation circulated on social media, the authors provide evidence that aggregate demand played a central role in explaining price shocks by focusing on the common component of movements in commodity prices. As for headline inflation, Bernanke and Blanchard (2023) main result is presented in Figure 3, which shows the estimated contributions of the model to headline inflation. The reader should notice that net of food and energy prices (that is in core space), the contribution of V/U is roughly as much as the one of “shortages” (a measures of negative supply shocks) in recent quarters. Because the transmission from V/U to core inflation is typically slow, the authors argue that core inflation might remain above the Fed target, unless V/U falls significantly.
Figure 3. Bernanke and Blanchard (2023) main result.
Our comments/critiques of the papers
In this section we point out some of the weaknesses of the papers.
- The main weakness of Gagliardone-Gertler (2023) is that part of their results are “obvious” because they are implicit in the assumptions of the model. Indeed, as we pointed out above, the main mechanism in the model is the complementarity between oil and other inputs. Because firms cannot switch away so easily from oil, a commodity shock results in a larger shock to the marginal cost and, ultimately, in a higher inflation than otherwise (isn’t this obvious/forced by the assumptions?). Also, what the authors call “good fit” (that is the difference between the solid and the dashed line in Figure 1) is pretty arguable, as the model predicted deflation in core PCE in 2015-2019 as opposed to low and stable inflation. Unfortunately, these issues are not discussed in the paper.
- The main weakness of Benigno and Eggertsson (2023) is the pre-Covid period. The main claim of the authors is that the non-linearity in V/U kicks-in above 1 (a situation they label as “labor shortage”). As mentioned above, in this circumstance, inflation is expected to accelerate sharply. Having said so, the authors do not really discuss why in 2018-2019 inflation remained stable despite V/U being well above their threshold (see Figure 2). Not only, but the empirical results are based on 57 observations (see Table 1 of the paper), on a specific form of non-linearity, and a specific form of Phillips curve (i.e. the authors do not really impose the canonical “anchored” restrictions on the coefficients). In our research (not shown for brevity) some of the authors results do not survive to robustness checks, as they are too sensitive either to the threshold chosen or the functional form. Nevertheless, in all regressions, V/U remains a good measure of “slack” as pointed out in the literature and fits pretty well the data (including using non-linear versions).
- The main weakness of Bernanke and Blanchard (2023) is that the identification strategy demand/supply is pretty weak. The authors claim that the first component of food and energy prices captures the demand side but in reality, there are good reasons to think it can also capture the supply side, unless other restrictions are imposed (the same thing applies to V/U, which can increase for both demand or supply factors). It would have been easier to repeat the interesting analysis on core inflation only. The impression is that the results in Bernanke and Blanchard (2023) for core inflation would have confirmed previous papers which have argued that about half of the runup in core is due to demand-side and half to supply-side factors.
Ok but where do we go from here?
The three papers confirm our previous findings. The three papers have something in common: they all argue that going back to 2% (in core space) requires a significant burden. For instance, the model of Bernanke and Blanchard (2023) expects inflation to remain well above 2% in 2027 if V/U does not fall below 1.2 (see Figure 14 of Bernanke and Blanchard (2023)). From our point of view, we have reached the same conclusion about 1 year ago when we commented on Figura and Waller (2022) – see our note here. Indeed, in mid-2022 our models delivered a 3.6% core PCE price inflation forecast for 2023 (Q4/Q4) under the assumption that V/U would fall to 1.5 (see Table 1 of our note). One year later, today, the forecast has changed only marginally: monetary policy cannot turn dovish any time soon if the labor market remains so tight.
(The reader can see the latest run of our “main” model using V/U as a measure of slack, assuming that V/U falls to 1 in 2024 and remains at that level in 2025 here).
