All over the world, governments at every level are debating the parameters of enormous spending plans to support the post-Covid recovery.
In the midst of all of the talk about economic stimulus, a familiar debate has re-entered the headlines: will too much stimulus overheat the economy and create “out-of-control” inflation? These fears reverberate across the political spectrum, from members of the Democratic establishment to those on the right. New research from Jonathon Hazell, a postdoctoral researcher at the Julis-Rabinowitz Center with a joint appointment in the Department of Economics' Griswold Center, sheds light on a critical aspect of the economic underpinnings of this topic, adding a new layer to an old debate with far-reaching implications for policy.
In their October 2020 working paper, Hazell and co-authors, Herreño, Nakamura, and Steinsson, focus on the Phillips curve, which they describe as “a formal statement of the common intuition that, if demand is high in a booming economy, this will provoke workers to seek higher wages, and firms to raise prices.” Rooted in the empirical observation that unemployment and inflation have had an inverse relationship over certain periods, the workhorse model of the Phillips curve has evolved over the years as a tool for thinking about the relationship between unemployment (or the output gap), inflationary expectations, and price-level movements.
Notably, during the Volcker disinflation of the early 1980s, when the Federal Reserve, with Chair Paul Volcker at the helm, “sharply tightened monetary policy,” the economy witnessed a steep decrease in inflation accompanied by rising unemployment, confirming the idea of a policy trade-off and supporting arguments for a “relatively steep” Phillips curve. However, in recent decades, economists have observed “missing disinflation” both during and after the 2008 economic crisis as well as “missing reinflation” in the past few years, with very low unemployment levels before the 2020 economic crisis. What might account for these phenomena?
A Novel Approach
Hazell et. al address the question by using a novel dataset to examine regional changes in the slope of the Phillips curve from the early 1980s to 2018 and tease out more explicitly the role of long-term inflation expectations versus unemployment in affecting actual inflation rates across the U.S. This regional approach is key. Given the large variation in unemployment rates across states, all existing under the same national monetary policy regime, a national estimate of the slope of the Phillips curve misses the significant discrepancies between states, as evidenced by unemployment from California, Pennsylvania, and Texas.
The researchers’ approach also overcomes the constraints of previous empirical studies in two critical ways. First, they analyze changes in the prices of non-tradable goods in order to protect against regional Phillip curve slopes of zero, as tradable goods are more likely to be set at the national level. Second, the authors’ use state-level consumer price indexes for the U.S. dating back to the 1970s and a new tradeable demand spillovers instrument, based on the idea that “supply shocks in tradable sectors will differentially affect demand in non-tradable sectors in regions that are differentially exposed to the shocked tradable sectors: e.g., an oil boom will increase demand for restaurant meals in Texas.”
Expectations Matter Most
By leveraging these data and methods, Hazell et. al find no “missing disinflation” or “missing reinflation,” but rather that the Phillips curve is “very flat, and was very flat even during the Volcker disinflation of the early 1980s.” Comparing the Core Consumer Price Index (CPI) inflation and the long-term expectations of CPI inflation from the Survey of Professional Forecasters and Blue Chip, the paper estimates that of the large 6% drop in inflation during the Volcker years, the rise in unemployment accounted for only slightly more than a third of the drop, with the other two-thirds attributable to the 4% decline in long-run inflationary expectations from 1981 to 1987.
The results show that long-term inflationary expectations have historically been much more important in explaining inflation than changes in unemployment. This empirical evidence makes it harder to ignore the elephant in the room of inflationary expectations as a key driver of actual inflation. “The stability of inflation since 1990 is due to long-run inflationary expectations becoming more firmly anchored,” the paper concludes.
This work, already cited in The Economist and Bloomberg View, has far-reaching implications for policy. Inflation today should remain relatively stable, as long as expectations about future inflation remain anchored at their level since 1990. At the heart of inflationary expectations–which the paper confirms are critical in influencing actual inflation rates–is monetary policy and the political process behind it. The independence from political pressure maintained by Volcker is no longer the norm, and former President Donald Trump’s rocky relationship with the Federal Reserve and its Chairs is well-documented. With monetary policy playing a large role in governments’ efforts to recover from the Covid-19 crisis, these findings should help inform central banks’ policy frameworks.
As Paul Krugman, Nobel laureate and Professor Emeritus of Economics and International Affairs at the Princeton School of Public and International Affairs wrote in a recent blog post, this research shows that “low unemployment does lead to higher inflation, but even a very hot economy only leads to modest inflationary overheating.” To Krugman, these findings imply that not only should economists resist criticizing the ambition of the Biden administration’s spending plans due to inflationary fears, but also that the research calls into question central banks’ understanding of the relationship between inflation and unemployment.
This research is a key addition to the data and empirical evidence informing policy discussions and an exciting and valuable addition to a longstanding debate. The paper is available from the National Bureau for Economic Research and Hazell’s website, along with a link to a brief Twitter explanation from co-author Jon Steinsson.