CONVEX
Glossary/Macroeconomics/Phillips Curve
Macroeconomics
8 min readUpdated Apr 12, 2026

Phillips Curve

ByConvex Research Desk·Edited byBen Bleier·
inflation-unemployment tradeoffNAIRUwage-price tradeoff

The historical inverse relationship between unemployment and inflation, when unemployment is low, inflation tends to rise, and vice versa, a core framework underpinning central bank policy decisions.

Continue reading on Convex
Current Macro RegimeSTAGFLATIONDEEPENING

The macro regime is unambiguously STAGFLATION DEEPENING. The hot CPI print (pending event, 24h ago) is not a surprise — it is a CONFIRMATION of the pipeline signals that have been building for weeks: PPI accelerating faster than CPI, Cleveland nowcast at 5.28%, breakevens rising +10bp 1M across the …

Analysis from May 14, 2026

What Is the Phillips Curve?

The Phillips Curve is the most influential, and most controversial, relationship in macroeconomics: the empirical observation that unemployment and inflation tend to move in opposite directions. When the labour market is tight (low unemployment), inflation rises. When the labour market is slack (high unemployment), inflation falls.

Named after New Zealand economist A.W. Phillips, who documented the relationship between UK unemployment and wage growth in a 1958 paper, the Phillips Curve became the intellectual foundation of modern central banking. Every time the Fed decides whether to raise or cut interest rates, it is implicitly making a judgment about the Phillips Curve, whether the current level of unemployment is generating too much, too little, or just the right amount of inflation pressure.

For traders, the Phillips Curve matters because it determines how the Fed interprets labour market data. A strong NFP report or a declining unemployment rate isn't just economic news, it's an inflation signal that feeds directly into the Fed's reaction function and therefore into every interest rate, equity, and currency trade.

The Original Phillips Curve

Phillips's 1958 Finding

Phillips analysed 97 years of British data (1861-1957) and found a stable, inverse relationship between unemployment and wage inflation. The curve was non-linear:

  • At very low unemployment (<3%), wage growth accelerated sharply, a steep portion of the curve
  • At moderate unemployment (4-6%), wage growth was modest, a flat portion
  • At high unemployment (>7%), wages stagnated or fell

Samuelson and Solow's Extension (1960)

American economists Paul Samuelson and Robert Solow adapted Phillips's wage curve into a price inflation curve, arguing that wage increases translate into price increases as firms pass through higher labour costs. They presented policymakers with a seemingly straightforward menu:

Unemployment Rate Expected Inflation Rate Policy Choice
3% 5-6% Accept more inflation for less unemployment
4% 3-4% Moderate trade-off
5-6% 1-2% Low inflation at the cost of higher unemployment
7%+ 0% or deflation Price stability at the cost of mass joblessness

This "menu" implied that policymakers could choose their preferred combination, and for a decade, this framework guided economic policy.

The Evolution: From Simple to Complex

Stage 1: The Original Trade-Off (1960s)

The 1960s seemed to validate the Phillips Curve. Kennedy and Johnson administration economists used it to justify expansionary policies: push unemployment down, accept a bit more inflation, and enjoy the prosperity. Unemployment fell from 7% (1961) to 3.4% (1969). Inflation rose from 1% to 6%. The curve worked, or seemed to.

Stage 2: Stagflation Destroys the Simple Curve (1970s)

The 1970s shattered the original Phillips Curve. Both unemployment AND inflation rose simultaneously, a combination the curve said was impossible:

Year Unemployment CPI Inflation Phillips Curve Prediction
1973 4.9% 8.7% Should be low inflation at this unemployment
1975 8.5% 6.9% Should be very low inflation at this unemployment
1980 7.1% 14.8% Completely off the chart

What went wrong: The original Phillips Curve ignored inflation expectations. When workers and businesses expect higher inflation, they demand higher wages and raise prices pre-emptively, shifting the entire curve upward. The 1970s oil shocks combined with un-anchored inflation expectations produced a new, higher Phillips Curve at every unemployment level.

Stage 3: The Expectations-Augmented Phillips Curve (1970s-1990s)

Milton Friedman and Edmund Phelps independently predicted the 1970s breakdown. Their key insight: the Phillips Curve trade-off is only temporary. In the long run, unemployment returns to its "natural rate" (NAIRU) regardless of inflation. Attempts to permanently lower unemployment below NAIRU simply produce ever-accelerating inflation.

The expectations-augmented Phillips Curve:

Inflation = Expected Inflation − β(Unemployment − NAIRU) + Supply Shock

This framework says:

  • Inflation equals expected inflation when unemployment is at NAIRU (no trade-off in the long run)
  • Inflation rises above expectations when unemployment is below NAIRU
  • Inflation falls below expectations when unemployment is above NAIRU
  • Supply shocks (oil prices, supply chains) can shift inflation independent of the labour market

This became the dominant framework at central banks and remains the intellectual basis of the Fed's approach today.

Stage 4: The Flat Phillips Curve (2000-2019)

From the mid-1990s through 2019, the Phillips Curve appeared to flatten to near zero, unemployment fell from 6% to 3.5% over two decades, yet core inflation barely budged from 1.5-2.0%. The "β" coefficient (sensitivity of inflation to unemployment) shrank toward zero in every empirical estimate.

Why it flattened:

Factor Mechanism Period
Globalisation Cheap imports from China suppressed goods prices regardless of domestic labour markets 1995-2020
Anchored expectations Post-Volcker credibility kept inflation expectations at 2% 1995-present
Technology/automation E-commerce + automation reduced pricing power; Amazon effect 2000-present
Gig economy Reduced worker bargaining power; suppressed wage growth 2010-present
Measurement error Headline unemployment understated true slack (U-6, participation) 2010-2019

The policy consequence: The flat Phillips Curve gave the Fed license to keep rates near zero for years. If low unemployment doesn't cause inflation, there's no reason to raise rates pre-emptively. This contributed to the extended zero-rate period of 2008-2015 and the cautious hiking cycle of 2015-2018.

Stage 5: The Non-Linear Revival (2020-Present)

COVID revealed that the Phillips Curve is not flat, it's non-linear:

  • Flat at moderate unemployment (3.5-6%): The curve is nearly horizontal. Inflation is insensitive to unemployment changes in this range. This is what prevailed in 2010-2019.
  • Steep at extreme tightness (below ~3.5%): When unemployment drops well below NAIRU and the labour market is extremely tight, inflation surges. This is what happened in 2021-2022 when unemployment fell to 3.4%.
  • Steep at extreme weakness (above ~8%): During severe recessions, deflation risks emerge as mass unemployment destroys demand.

The non-linear Phillips Curve reconciles the 2010s (flat, because unemployment was in the moderate range) with 2021-2022 (steep, because unemployment was at historic lows).

NAIRU: The Equilibrium Unemployment Rate

What Is NAIRU?

NAIRU (Non-Accelerating Inflation Rate of Unemployment) is the unemployment rate at which inflation is stable. Below NAIRU, inflation accelerates. Above NAIRU, inflation decelerates. At NAIRU, inflation stays constant (at whatever level expectations are anchored).

NAIRU is the Phillips Curve's equilibrium point, and like the neutral interest rate (r*), it is unobservable, time-varying, and surrounded by enormous uncertainty.

NAIRU Estimates Over Time

Period Estimated NAIRU Key Influence
1960s ~4.0% Strong labour bargaining power
1970s-80s ~6.0-6.5% Structural unemployment from oil shocks + regulation
1990s ~5.0-5.5% Welfare reform; globalisation reducing wage pressure
2000s ~5.0% Continued globalisation; tech productivity
2010s ~4.0-4.5% Declining unionisation; gig economy
2020s ~4.0-4.4% (Fed estimate) Debate: has it risen post-COVID?

The NAIRU Uncertainty Problem

The Fed's NAIRU estimate has a confidence interval of approximately ±1.5 percentage points. If NAIRU is 4.0%, unemployment of 3.5% is only slightly inflationary. If NAIRU is 5.5%, unemployment of 3.5% is a dangerously overheated labour market. The difference between these two scenarios is the difference between holding rates steady and hiking aggressively.

This uncertainty is why the Fed has moved toward "data-dependent" policy rather than targeting a specific unemployment rate. They watch inflation itself, not just the labour market, because NAIRU is too imprecise to rely upon exclusively.

Trading the Phillips Curve

The Labour Market → Inflation → Rates Chain

Every labour market data release triggers a Phillips Curve calculation in the market's mind:

Labour Market Signal Phillips Curve Implication Rate Implication Asset Impact
NFP strong + wages accelerating Unemployment well below NAIRU; inflation pressure Hawkish, fewer cuts, potential hike Bonds ↓, USD ↑, Stocks mixed
NFP strong + wages moderating Near NAIRU; growth without inflation Goldilocks, gradual cuts Bonds flat, USD flat, Stocks ↑
NFP weak + wages decelerating Moving above NAIRU; inflation receding Dovish, more/faster cuts Bonds ↑, USD ↓, Stocks regime-dependent
NFP weak + wages still strong Supply-side inflation (not demand-driven) Stagflation risk, Fed paralysed Bonds mixed, Gold ↑, Stocks ↓

The Average Hourly Earnings Trade

The most direct Phillips Curve trade focuses on the Average Hourly Earnings (AHE) component of the NFP report:

  • AHE MoM > 0.4%: Wage inflation accelerating → Phillips Curve pressure → hawkish → sell 2Y Treasuries, buy USD
  • AHE MoM 0.2-0.3%: Consistent with target → no Phillips Curve alarm → neutral
  • AHE MoM < 0.2%: Wage growth decelerating → Phillips Curve easing → dovish → buy 2Y Treasuries, sell USD

When the Phillips Curve Breaks Down (Trading the Surprise)

The highest-value trades occur when the Phillips Curve doesn't work as expected:

  1. Low unemployment + falling inflation (2023-2024): Supply-side improvement → Phillips Curve shifts left → the Fed can be more dovish than expected → buy risk assets aggressively
  2. Rising unemployment + persistent inflation (stagflation): Phillips Curve fails entirely → the Fed has no good option → buy gold, sell both stocks and bonds
  3. Low unemployment + stable low inflation (2017-2019): Phillips Curve is flat → the Fed can stay easy → buy growth stocks and duration

What to Watch

  1. NFP Average Hourly Earnings: The most direct Phillips Curve variable. Track MoM and YoY wage growth relative to expectations.
  2. JOLTS job openings-to-unemployed ratio: A broader measure of labour market tightness than unemployment alone. A ratio above 1.5 signals extreme tightness.
  3. Atlanta Fed Wage Growth Tracker: A more stable measure of individual wage growth (tracks the same individuals over time), reducing composition effects.
  4. FOMC "longer-run unemployment rate": The Fed's NAIRU estimate, published in the SEP. Shifts signal changing Phillips Curve assumptions.
  5. Unit labour costs: Wages adjusted for productivity. Rising unit labour costs = inflation pressure even if productivity is growing. This is the "true" Phillips Curve variable that central banks care about.

Frequently Asked Questions

What is NAIRU and what is the current estimate?
NAIRU (Non-Accelerating Inflation Rate of Unemployment) is the unemployment rate below which inflation begins to accelerate and above which it decelerates. It is the Phillips Curve's "neutral zone." The Fed's current NAIRU estimate is approximately 4.0-4.4% (published in the Summary of Economic Projections as the "longer-run unemployment rate"). However, NAIRU is unobservable and highly uncertain — the confidence interval spans roughly 3.5% to 5.5%. The estimate has shifted over time: it was ~6% in the 1980s, ~5% in the 1990s-2000s, and drifted down to ~4.0-4.4% in the 2010s-2020s as structural changes (demographics, globalisation, technology) reduced inflationary pressure at lower unemployment levels. The post-COVID debate centres on whether NAIRU has risen (due to reduced labour supply and deglobalisation) or remains near 4%.
Why did the Phillips Curve seem to disappear from 2000-2020?
From the mid-1990s through 2019, the Phillips Curve appeared flat to nonexistent: unemployment fell from 6% to 3.5% yet inflation barely budged from 1.5-2.0%. Several explanations: (1) Globalisation — cheap imports from China and other emerging markets acted as a deflationary force that offset domestic wage pressures. (2) Anchored expectations — after Volcker's 1980s disinflation, long-run inflation expectations stabilised near 2%, preventing wage-price spirals from developing. (3) Gig economy/automation — reduced worker bargaining power, suppressing wage growth even at low unemployment. (4) Measurement issues — headline unemployment understated true labour market slack (U-6 and prime-age participation told a different story). (5) Amazon effect — e-commerce increased price competition, keeping consumer goods deflation running alongside services inflation. The flat Phillips Curve gave the Fed license to keep rates near zero for years, under the logic that low unemployment was "not inflationary." COVID proved this thinking wrong — or at least incomplete.
Did COVID prove the Phillips Curve is real after all?
Yes and no. The post-COVID experience showed that the Phillips Curve is non-linear — flat at moderate unemployment levels but steep when the labour market is extremely tight. When unemployment plummeted from 14.7% (April 2020) to 3.4% (January 2023) — driven by both demand recovery and supply constraints (early retirements, reduced immigration, long COVID) — wages surged 5-6% and services inflation reached multi-decade highs. This was classic Phillips Curve logic: excess labour demand driving up wages and prices. However, the subsequent disinflation from 2023-2024 occurred without a significant rise in unemployment (from 3.4% to only 4.2%), which shouldn't happen if the Phillips Curve is steep. The resolution: supply-side healing (supply chains normalising, labour force expanding from immigration) allowed disinflation without demand destruction — a supply-side Phillips Curve shift rather than a movement along the curve.
How do Fed officials use the Phillips Curve in practice?
Fed officials use the Phillips Curve as one input into their inflation forecasting framework, though they are careful not to rely on it mechanically. In practice, FOMC members watch the "labour market tightness" indicators — unemployment rate vs. NAIRU, job openings-to-unemployed ratio (from JOLTS), quits rate, and wage growth — as signals of whether the labour market is generating inflationary pressure. When these indicators suggest the labour market is "too tight" (unemployment well below NAIRU, quits rate elevated, wages accelerating), the Phillips Curve logic pushes toward tighter policy. When they suggest slack (unemployment above NAIRU, job openings falling, wages decelerating), it pushes toward easier policy. Powell has explicitly cited the Phillips Curve framework in multiple press conferences, noting that "the labour market needs to come into better balance" for inflation to sustainably return to 2%. However, the Fed's models have been repeatedly wrong about the Phillips Curve slope, contributing to both the "too low for too long" error of 2021 and the overshoot fears of 2023.
How should traders use the Phillips Curve?
The Phillips Curve provides a framework for interpreting labour market data releases in terms of their inflation and rate implications. When NFP is strong and unemployment is below NAIRU: the Phillips Curve logic says inflation pressure is building → hawkish for rates → bearish for bonds, growth stocks. When unemployment rises above NAIRU: inflation pressure is easing → dovish for rates → bullish for bonds, growth stocks. The most tradable Phillips Curve signal occurs when the relationship breaks down: if unemployment falls sharply but inflation doesn't respond (as in 2017-2019), it suggests the curve is flat and the Fed can be patient — bullish for equities. If unemployment rises but inflation stays sticky (as some feared in late 2024), it suggests supply-side inflation that the labour market can't fix — a stagflation signal that is bearish for both stocks and bonds. Watch the wage-growth component (Average Hourly Earnings in NFP) as the most direct Phillips Curve variable — it links the labour market to inflation through the cost channel.

Phillips Curve is one of the signals monitored daily in the AI-driven macro analysis on Convex Trading. The platform synthesises data across monetary policy, credit, sentiment, and on-chain metrics to generate actionable trade recommendations. Create a free account to build your own signal layer and see how Phillips Curve is influencing current positions.

ShareXRedditLinkedInHN

Macro briefings in your inbox

Daily analysis that explains which glossary signals are firing and why.