Joe Gagnon on *25 Years of Excess Unemployment* and the Phillips Curve Debate

Episode Summary

Executive Summary: David Beckworth and Joe Gagnon revisit forecasting mistakes during the post-COVID inflation surge, discuss why bond markets and forecasters missed it, and assess lessons for Fed policy. Gagnon argues inflation is highly nonlinear, with a very flat Phillips curve at slack and near-vertical response at capacity, implying long periods of excess unemployment and the need for more policy humility and possibly a higher inflation/NGDP target.

Main Topics: Forecasting inflation and accountability (Priority: 5/5): Beckworth admits his 2021 inflation forecast error; Gagnon reflects on his own correct and incorrect calls and explains why bond markets, professionals, and households were poor at predicting inflation turning points. Why inflation surged after COVID (Priority: 5/5): The conversation attributes the inflation shock to massive fiscal stimulus, easy monetary policy, and supply disruptions, with Gagnon stressing that standard macro should have predicted more inflation than was anticipated in real time. Lessons from the zero lower bound (Priority: 4/5): Gagnon argues the economy escaped the ZLB because of a large shock, but also notes a long-run decline in the natural real rate and uncertainty about whether today’s fiscal environment will keep rates elevated. Excess unemployment over 25 years (Priority: 5/5): Gagnon’s paper claims advanced economies spent most of the past 25 years below potential, with unemployment higher than necessary because policymakers misread a very flat Phillips curve and overestimated U-star. Nonlinear Phillips curve and wage rigidity (Priority: 5/5): He argues the inflation-unemployment relationship is highly nonlinear: very flat when slack is abundant due to downward wage/price rigidity, but very steep when the economy strains capacity, helping explain prolonged low inflation without deflation spirals. Policy implications for the Fed (Priority: 4/5): Gagnon recommends more humility in estimating slack and natural rates, periodic experimentation with lower unemployment, and possibly a higher inflation target or NGDP level target to reduce the flat-region problem.

Key Arguments: Bond markets, professional forecasters, and households are not reliable predictors of major inflation turning points; they react after inflation changes rather than anticipating them. The 2021–2022 inflation episode was driven by an unusually large fiscal impulse interacting with highly accommodative monetary policy and supply-side shocks. The natural rate of unemployment (U-star) and the equilibrium real interest rate (R-star) have likely fallen over time, partly because of demographics. Advanced economies appear to have operated below potential for much of the last 25 years, not because the economy was perfectly balanced, but because the Phillips curve was so flat that inflation gave little signal. A nonlinear Phillips curve explains why large labor-market slack can persist without deflation and why output gaps are hard to measure in real time. Central banks’ reliance on linear models and stable inflation around target led them to overestimate potential output and the natural rate of unemployment. A higher inflation target could reduce downward wage rigidity’s grip and make it easier to identify and exploit labor-market slack. Nominal GDP level targeting would improve robustness, but it would work better with a somewhat higher growth path than a 2% inflation-equivalent path.

Data Points: Inflation forecast error (2021): Beckworth says his February 2021 op-ed titled “Stop Worrying About Inflation” was “way off.” - Opening accountability discussion about inflation forecasting mistakes. Fiscal package size: $1.9 trillion - American Rescue Plan cited as large enough to be inflationary relative to the estimated output gap. CBO output gap: $400 billion - Beckworth notes early 2021 CBO estimates of slack were far smaller than the fiscal impulse. Nominal GDP gap / level overshoot: $1.3–$1.4 trillion - Beckworth argues the economy’s nominal size was well above its pre-pandemic trend by this amount. Gagnon’s early post-COVID inflation forecast: about 4% - He says he anticipated demand-driven inflation around this level before supply shocks pushed inflation much higher. Actual inflation peak referenced: 8% - Gagnon contrasts his 4% expectation with roughly 8% realized inflation. Countries studied: 11 large advanced economies - Gagnon and coauthor examine excess unemployment across these economies. Time span studied: 25 years - Paper covers the period from the mid-1990s to the post-COVID era. Inflation targeting era: mid-1990s onward - Gagnon says advanced economies broadly adopted 2% or lower inflation targets in this period. Fed framework discussion: 2024–2025 - Beckworth references the next Fed framework review. COVID shock characterization: once-in-a-century - Gagnon says COVID is a uniquely large and unusual macro shock. Nominal GDP target example: 4% or 5% nominal GDP growth path - Beckworth and Gagnon discuss NGDP level targeting paths tied to different inflation goals.

Pivotal Quotes: "the inflation process, in my view now, is about as nonlinear as an economic relationship can be and still be sensible" — Joe Gagnon: Explaining why the Phillips curve can be flat for long periods and then become very steep near capacity. "we just were below potential for most of the time" — Joe Gagnon: Summarizing the paper’s finding that advanced economies had persistent excess unemployment over 25 years. "the Fed can keep for decades the real economy artificially low?" — David Beckworth: Raising the classical-dichotomy objection to long-running excess unemployment estimates and the need for a nonlinear Phillips curve.

Implications: The episode argues for more humility in measuring slack, less faith in linear Phillips-curve models, and stronger consideration of higher inflation or NGDP level targeting. For central banks, it suggests persistent underemployment may have been missed for years.