The Phillips Curve in 2026: Unemployment at 4.1%, Inflation Still at 3.4%

In August 2026, the U.S. unemployment rate stood at 4.1 percent, near some of the tightest readings of the past several decades. Consumer prices were still 3.4 percent higher than a year earlier, well above the Federal Reserve's 2 percent target. If the Phillips curve worked the way it is popularly summarized, an economy this close to full employment should be running hotter on inflation than that, or an economy with inflation this contained should have more slack than 4.1 percent implies. Seven months of 2026 data suggest the real relationship is messier than the textbook slope.

What the Phillips Curve Actually Claims

Economist A.W. Phillips found in 1958 that periods of lower unemployment in the United Kingdom lined up with faster wage growth. Paul Samuelson and Robert Solow adapted the idea for the United States in 1960, framing it as a menu: policymakers could accept a bit more inflation in exchange for a bit less unemployment, or the reverse. The mechanism is straightforward. When unemployment is low, workers have more bargaining leverage, employers compete harder for staff, wages rise, and firms pass some of that cost into prices.

Why the Simple Version Broke in the 1970s

Milton Friedman and Edmund Phelps, working separately in 1968, argued that the trade-off only holds in the short run. Once workers and firms come to expect a given inflation rate, trying to hold unemployment below its "natural rate," later formalized as the non-accelerating inflation rate of unemployment, or NAIRU, does not buy a lasting reduction in joblessness. It just pushes inflation higher. The 1973-1975 period, when an oil embargo sent both unemployment and inflation into double digits at once, is the standard historical case for this: the simple downward-sloping curve cannot explain both rising together. In the long run, the Phillips curve is closer to vertical at the natural rate, not a stable slope a policymaker can ride up and down. Research from the Federal Reserve Bank of Kansas City has since found that once a central bank credibly anchors inflation expectations, as the Fed did by formalizing its 2 percent target in 2012, the observable short-run relationship between unemployment and inflation tends to flatten further, because workers and firms stop adjusting their expectations to every wiggle in the jobless rate.

Seven Months That Don't Trace a Single Curve

The plotted path does not slope cleanly downward the way a textbook short-run Phillips curve would. From January through May, unemployment barely moved, holding between 4.3 and 4.4 percent, while CPI inflation climbed from 2.4 percent to 4.2 percent, the largest over-the-year increase since April 2023. A falling curve would expect inflation to rise only alongside falling unemployment; instead the jobless rate stayed flat while prices accelerated on their own. From June through August, the pattern flipped: unemployment eased to 4.1 percent and inflation cooled to 3.4 percent at the same time, again moving together rather than trading off against each other.

The Federal Reserve's own commentary points to a likely reason. The September 16 FOMC statement described job gains as having "kept pace with the workforce" even as it judged inflation "elevated," and earlier releases tied the spring's price acceleration to energy costs and tariff pass-through rather than labor-market tightness. That fits with what the Federal Reserve Bank of Dallas found in a working paper published the same week: import exposure accounts for roughly 40 percent of the long-run flattening of the U.S. Phillips curve, because price pressure that originates overseas doesn't run through domestic wage bargaining the way the original theory assumed. At the same time, the Federal Reserve Bank of New York reported in February that the curve's slope, flat for years, has partly re-steepened since the pandemic. Read together, 2026's loop-shaped path looks less like a broken relationship and more like a domestic labor-market signal that is being temporarily overwhelmed by a supply-side cost shock.

A Second Measure of Slack Tells a Different Story

The Phillips curve's labor-market side depends on how accurately unemployment is measured, and the headline U-3 rate may be missing meaningful slack this cycle. The Ludwig Institute for Shared Economic Prosperity tracks a broader measure, the True Rate of Unemployment, which counts the jobless alongside involuntarily part-time workers and those earning poverty-level wages. That rate stood at 24.9 percent in July 2026, its fourth straight monthly rise, versus the BLS's 4.1 percent headline figure the same month. A broader accounting of labor-market slack shows how wide that gap has grown. If a large share of the workforce is underemployed rather than counted as jobless, the labor market may have less genuine tightness than the 4.1 percent figure implies, which would weaken the case that current inflation reflects a hot jobs market at all.

Part of that gap may trace to how technology is reshaping entry-level hiring. Fed officials have flagged rising AI-related capital investment as a potential source of inflationary pressure on its own, even as what Stanford's employment data shows about AI's effect on entry-level hiring suggests some of that same investment is displacing junior workers rather than adding them. A slowdown concentrated in early-career roles would not necessarily move the headline unemployment rate much, since it can show up as slower hiring rather than mass layoffs, which is part of why how AI-driven layoffs can compound rather than cushion a slowdown matters for reading the official numbers correctly.

What the September Rate Increase Signals About the Trade-off the Fed Sees

The Fed's own actions this year suggest its policymakers do not see the current combination as a stable trade-off worth tolerating. After holding rates at 3.50 to 3.75 percent for most of 2026, the FOMC voted 12-0 on September 16 to raise the target range to 3.75 to 4.00 percent, its first hike in more than three years, with Chair Kevin Warsh telling reporters inflation had been "too high for too long." The Committee's own September projections put unemployment near 4.1 percent at year-end alongside PCE inflation around 3.7 percent for 2026, not falling to 2.3 percent until 2027. Four policymakers' projections still showed room for two additional hikes this year. Whether the cooler June-through-August readings mark the classic trade-off reasserting itself, or only a pause before tariff and energy costs pass through again, is the open question the Fed's own dot plot suggests it is not yet willing to answer.