How Broad-Based is Inflation?

Why a monetary response is absolutely warranted

Adam Tooze is ‘not convinced that the current round of price increases should really be thought of as a general inflation at all.’ This seems to be a common perception in certain corners of the cognoscenti. The idea is that the current bout of inflation is driven by idiosyncratic factors related to the pandemic (as Matt Klein has shown) and is therefore transitory. The upshot of this argument is that the Fed should be patient. A more vigorous line of attack was launched recently by Isabella Weber, who suggested that price gouging by oligopolists may be responsible for the current bout of inflation, and that policymakers should consider post-World War II-style price controls to defeat the enemy in detail. Todd Tucker made a similar argument at the Roosevelt Institute, where Josh Mason and Lauren Melodia have also argued for targeted interventions instead of monetary tightening.

We argue that these analysts are mistaken. We show that the current bout of inflation is extremely broad-based. We argue that what really matters, and has always mattered, is the common component of inflation. We marshal very suggestive evidence that the common component of inflation contains a strong signal of inflation expectations. Furthermore, we argue that inflation expectations are the forcing variable for monetary policy. We therefore conclude that the Fed should and will depart on a hiking cycle in early 2022, as we predicted and as the FOMC has since signaled.

Part of current confusion about inflation stems from a viral paper by Rudd, who argued that economists’ belief that ‘households’ and firms’ expectations of future inflation are a key determinant of actual inflation’ is based on ‘shaky foundations’. What Rudd gets right is the bankruptcy of the Phillips curve model of the inflation process. (The intellectual rigidity was best captured by Olivier Blanchard’s response to Brainard’s blasphemy: “The notion that a tight labor market will lead to inflation is impossible to contradict.”) But although Rudd repeatedly promises to interrogate the empirical evidence, the paper simply fails to present compelling empirical evidence against the hypothesis that expectations are crucial to the inflation process. To be fair, since hard currency-issuing central banks succeeded in anchoring inflation expectations on target (a process that was largely complete by the mid-1990s in the advanced economies), expectations have been somewhat ignorable, meaning that price setters could roughly assume that long run inflation was pinned down at the target level by the response function of the central banks.

Actually, this picture is not exactly right. Inflation expectations did repeatedly threaten to get de-anchored — on the downside. Because policymakers were working with the wrong model of the inflation process, monetary policy introduced a systematic deflationary bias, with the result that they missed their target on the low side in 80 percent of all quarters between 1995-2019. But that was before the regime change in 2020-2021.

---

Enough talk. Let’s get on with the quantitative analysis. We have argued previously that the lowest volatility quintile of the CPI price series contains a strong signal of inflation expectations. The reason is that these are the slowest-moving price setters in the economy who must pay very careful attention to where inflation will be before setting their own prices. Another way to get at inflation expectations is to consider the latent factor of the price series. This is a purely statistical model in which individual inflation series are modelled as linear combinations of the latent factor plus an idiosyncratic error term. If expectations are central to the inflation process, then this latent factor should be highly correlated with the lowest volatility quintile of the price series. We show that, contra Rudd, this is indeed the case.

We obtain all 179 individual price series from the BEA, compute the lowest volatility quintile and the latent factor using sklearn. The following graph shows their variation since 1999. We find a correlation of 0.94 between the two series, suggesting that they contain exactly the same signal. The obvious candidate for the common signal is expected inflation, as economists rightly believe, Rudd’s polemic notwithstanding. At any rate, the comovement of the single factor and the lowest volatility quintile suggests general inflation, whatever the interpretation of the common factor. We can see that this common signal of general inflation is now higher than at any time since at least 1999.

That’s the money shot. But further observations are in order. First, note that the mean correlation of the individual price series has jumped over the past year. The last time we saw such comovement was during the crises years of 2009 and 2012. Recall from the previous graph that, in 2009, expectations threatened to get de-anchored on the downside; while in 2011-2012, as now, they threatened to get de-anchored on the upside.

Is this covariation driven by expectations? We now test this hypothesis explicitly. In order to do so, we compute rolling 24-month regressions with the individual inflation series modeled as linear functions of our common factor. We then compute the mean r-squared across series. The next figure shows the mean percent of variation in the inflation series explained by the common factor. We can see that both our proxies of expected inflation explain an unprecedented percentage of variation since the pandemic.

To sum up, we’ve seen that the current bout of inflation is very broad-based. It is driven by a single underlying factor. That is, it is very much ‘a general inflation’. Moreover, this single factor is tightly coupled to our preferred proxy of inflation expectations. Our interpretation is thus that inflation expectations are getting de-anchored on the upside. The Fed, of course, cannot let that happen. Even if you don’t buy that the common component is expected inflation, there can hardly be any doubt that the current bout of inflation is broad-based and threatening to get endemic.

Monetary tightening, blunt as it is, is meant precisely for this state of affairs.

The Fed knows exactly what to do. It should proceed to do it. And people need to stop panicking about the hiking cycle. It’s a good thing that we’re on the cusp of leaving the ‘New Normal’ of low rates behind. The revival of the monetary cycle should help temper the buildup of financial imbalances that have been growing especially fast since the March panic, as evident in stratospheric asset valuations.

Meanwhile, politicians should stay out of the price stability game. It wasn’t simply ideology that led the advanced economies to abandon price controls. After all, the one certainty associated with price controls is a healthy black market. Instead of trying to do the Fed’s job, politicians should concentrate on public investment and the articulation of institutions that can serve as the eyes and ears of the state—a job that no one else can do and one that is absolutely critical in light of the planetary impasse and the deteriorating world situation.

---

Postscript. Big kerfuffle on Twitter.

---

Post-postscript. Interesting pushback from Steve Houf on twitter. He buys that inflation is broad-based but remains convinced that it may be transitory. The exchange motivated me to look at the trend component (using HP filter) of our broad-based measures.

The key graph here is the trend component of latent inflation. It is higher than it has been since 1999.