Has the Phillips Curve Shifted Outward? Why Labor-Market Cooling May No Longer Be Enough

Has the Phillips Curve Shifted Outward? Why Labor-Market Cooling May No Longer Be Enough

The most consequential feature of the latest inflation data is not simply that core PCE inflation remained above target. It is that inflation rose while a familiar measure of labor-market slack barely changed. Since January 2025, the unemployment-to-vacancy ratio has stayed close to 1.0, yet core PCE inflation reached 3.3% in July 2026. A Phillips curve estimated in 2023 would have associated that degree of slack with inflation nearer 2.5%. The resulting 0.8-percentage-point residual is economically large: it is the difference between an economy approaching price stability and one in which inflation is again moving away from target.

The tempting conclusion is that the Phillips curve has shifted outward. At any given unemployment-to-vacancy ratio, the economy now appears to generate more inflation than it did in the earlier sample. But that statement should be treated as a hypothesis, not a verdict. Eighteen months is a short window in macroeconomic inference, especially when tariffs, commodity costs, immigration policy, productivity, measurement revisions and lagged contract resets may all be moving simultaneously. A curve can appear to shift because its intercept changed, because expectations changed, because the relevant measure of slack became less informative, or because temporary supply shocks entered the residual.

Even with that caution, the policy implication is uncomfortable. If the missing 0.8 percentage point comes mainly from supply constraints and cost pressure, weaker labor demand alone may not return inflation to 2%. Rate cuts intended to protect employment could validate the cost shock through stronger nominal demand, while keeping rates restrictive could impose job losses without quickly producing the needed supply response. The Federal Reserve would then face a less favorable short-run trade-off: more inflation for the same labor-market configuration, and less disinflation from each unit of demand destruction.

Reading the 0.8-Percentage-Point Residual

A reduced-form Phillips curve can be written as pi_t = alpha + beta s_t + gamma E_t(pi) + delta z_t + epsilon_t, where pi is inflation, s is slack, expected inflation is E_t(pi), and z collects supply variables such as import prices, energy, productivity and taxes. If an older specification predicts 2.5% while actual core PCE is 3.3%, the residual is not automatically a structural shift. It is the combined contribution of omitted variables, parameter instability, data noise and genuine structural change. The discipline is to ask which component can plausibly explain the sign, size and persistence of the gap.

The unemployment-to-vacancy ratio is attractive because it combines workers seeking jobs with firms seeking workers. Near 1.0, it suggests a more balanced matching market than the extreme vacancy abundance of the reopening period. Yet it is not a complete statistic. Vacancies can be stale, duplicated or concentrated in sectors where available workers lack the relevant credentials. Unemployment can rise among new entrants while experienced workers remain scarce. Hours, hiring rates, quits, participation and wage growth may therefore describe effective slack differently from the headline ratio.

Composition also matters. A ratio of 1.0 created by weak entry-level hiring and persistent shortages in healthcare, construction or skilled services need not have the same wage and price consequences as a ratio of 1.0 spread evenly across occupations. The aggregate point on the curve can remain fixed while sectoral curves move in opposite directions. That possibility is especially important when cost pressure is transmitted through a few network-critical sectors such as transport, insurance, housing, energy or imported intermediate goods.

Why the Curve Can Move Outward

Tariffs offer the clearest candidate for an outward shift because they raise the domestic price of imported final goods and inputs without requiring a tighter labor market. The first-round effect is a relative-price increase. The macroeconomic danger comes from second-round propagation: firms pass higher input costs into a broad set of prices, workers seek compensation for lost purchasing power, and nominal demand remains strong enough to accommodate both. Monetary policy cannot remove a tariff, but it can influence whether the one-time level effect becomes a continuing inflation process.

Supply capacity can produce the same pattern. Constraints in energy, housing, logistics, semiconductors or skilled labor make aggregate supply steeper, so a given level of demand generates more price pressure. Regulatory expense, insurance repricing and financing costs can reinforce it. Some of these forces are themselves affected by interest rates: expensive credit may restrain demand, but it can also delay housing construction, inventory investment and capacity expansion. The short-run supply response can therefore make the inflation-employment trade-off look worse before it improves.

Expectations are another intercept shifter. If businesses come to expect 3% rather than 2% inflation, annual price resets and wage negotiations can cluster around the higher norm even without exceptional labor tightness. The process need not involve an unanchored public in the dramatic 1970s sense. A modest rise in the perceived normal rate, embedded in contracts and budgeting conventions, is sufficient to create a persistent residual. This is why the frequency and breadth of price increases matter alongside their average magnitude.

A Harder Federal Reserve Reaction Function

Under a stable demand-driven Phillips curve, labor cooling is the mechanism through which restrictive policy reduces inflation. Vacancies fall, wage competition eases, nominal spending slows and firms lose pricing power. An outward shift weakens that mapping. The central bank can still reduce inflation by creating more slack, but the sacrifice ratio—the cumulative output or employment loss required for a given decline in inflation—rises when supply pressure accounts for a larger share of the problem.

The Fed therefore has to separate the impulse from the propagation mechanism. It should not promise to offset every direct tariff or insurance increase immediately, because doing so could require excessive unemployment. It also cannot ignore those increases if they broaden, alter expectations or interact with demand. The relevant question is whether underlying nominal spending and wage-setting behavior are accommodating the shock. Policy patience is justified when forward supply indicators improve and expectations remain anchored; restraint is justified when the shock is becoming generalized.

This framework limits the scope for rapid rate cuts even if payroll growth weakens. If core PCE remains near 3.3% and the residual persists, easing based solely on labor softness risks confirming that the reaction function tolerates a higher inflation intercept. Conversely, a prolonged hold is not automatically optimal. If vacancies and hiring collapse while tariff-sensitive prices explain most of the overshoot, additional tightening may deliver poor inflation benefits at high social cost. Data dependence must mean identifying channels, not merely waiting for another aggregate release.

Conclusion: Cooling Demand Is Necessary, but Supply Relief May Be Decisive

The evidence since January 2025 is consistent with an outward movement in the short-run Phillips curve: an unemployment-to-vacancy ratio near 1.0 has coexisted with core PCE inflation of 3.3%, roughly 0.8 percentage point above the value implied by a 2023 relationship. That gap is too large to dismiss, but eighteen months is too short to declare a permanent regime change. The proper inference is conditional: supply constraints, tariffs and other cost pressures appear to be contributing more to inflation, while the reliability of a single slack measure has declined.

For policy, the consequence is asymmetry. Weakening labor demand may still reduce wage and demand pressure, but it may not be sufficient to restore 2% inflation on its own. Supply relief, productivity growth, narrower price breadth and stable expectations must do part of the work. Until those forces become visible, cuts carry more inflation risk than a conventional Phillips curve would imply; if they do become visible, the Fed can ease without demanding unnecessary labor-market damage.

For investors, this is a regime question rather than a one-release trade. Persistent residual inflation raises front-end rate volatility, limits the reliability of duration as a hedge and favors businesses able to absorb or pass through cost shocks without destroying demand. A fading residual would support the opposite configuration: lower real policy rates, a steeper curve and broader equity participation. The next several quarters should therefore be judged not only by unemployment or inflation separately, but by whether the gap between them closes—and whether it closes through lower inflation rather than higher unemployment.