Skip to main content

What is the Phillips curve?

The Phillips curve is the empirical relationship between unemployment and wage inflation (or price inflation): lower unemployment is associated with higher wage growth, and vice versa. Named after economist A.W. Phillips, who observed this pattern in 1958 British wage data, the Phillips curve became one of the central relationships in macroeconomics, shaping monetary policy for decades.

The intuition is straightforward. When unemployment is very low, the labor market is tight. Workers are scarce, so firms must offer higher wages to attract and retain them. Competition for scarce workers bids wages up. Higher wages are passed through to prices, causing inflation to rise. When unemployment is high, labor is abundant, workers have few alternatives, and firms can hold wages flat or cut them. Low wage growth translates to low price inflation.

The Phillips curve inverts the trade-off: policymakers must choose between low unemployment (and high inflation) or high unemployment (and low inflation). This trade-off shaped monetary policy in the 1960s and early 1970s, until the emergence of stagflation in the 1970s shattered the simple relationship and forced economists to revise their understanding of how expectations affect inflation.

Quick definition: The Phillips curve describes the inverse relationship between unemployment and inflation: lower unemployment is associated with higher wage and price inflation.

Key takeaways​

  • The Phillips curve is an empirical relationship between unemployment and wage inflation, observed in data across countries and decades.
  • In the original 1960s version, the relationship was steep and stable; a 1% drop in unemployment was associated with 1–2% higher annual wage inflation.
  • The 1970s stagflation (high inflation and high unemployment simultaneously) seemed to break the Phillips curve, until economists realized inflation expectations needed to be included.
  • The expectations-augmented Phillips curve (1960s) shows that the trade-off between unemployment and inflation holds only for unexpected inflation; expected inflation shifts the curve.
  • When inflation expectations rise, the entire Phillips curve shifts outward; when expectations fall, it shifts inward.
  • The Phillips curve has flattened significantly since the 1990s, meaning inflation responds less to unemployment changes than it once did.
  • The flattened modern Phillips curve has made monetary policy more complicated; the unemployment rate is a less reliable guide to inflation pressure.

Phillips's original discovery: 1958 U.K. wage data​

A.W. Phillips was a New Zealand economist studying wage behavior in the United Kingdom. He gathered historical data on wage inflation and unemployment from 1861 to 1957 and plotted the relationship. The pattern was striking: years with low unemployment had high wage inflation; years with high unemployment had low wage inflation. The relationship held across nearly a century of data. (See Federal Reserve economic research on the Phillips curve.)

Phillips estimated the relationship mathematically. He found that wage inflation could be approximated as a function of unemployment, with the form:

Wage inflation = 2% + f(unemployment)

Where f is a decreasing function—as unemployment falls, wage inflation rises.

More specifically, when unemployment was 5%, wage inflation averaged about 2% annually. When unemployment was 3%, wage inflation was about 4% annually. When unemployment was 7%, wage inflation was near zero or negative.

The relationship was not random scatter but a clear, reproducible pattern. This surprised many economists. For decades, labor economists had assumed wage inflation was determined by union power, labor supply growth, and institutional factors—not by something as simple as the unemployment rate. Phillips showed that unemployment itself was a powerful determinant.

The Phillips curve in 1960s macroeconomics​

Phillips's finding arrived at an opportune moment. In the late 1950s and early 1960s, economists were searching for the key trade-off facing policymakers. Phillips offered the answer: unemployment and inflation were inversely related. You could have low unemployment and high inflation, or high unemployment and low inflation, but not both.

This trade-off immediately captured policymakers' attention. President Kennedy's economic advisors—including the economist Paul Samuelson—used Phillips curve estimates to argue for expansionary policy in the early 1960s. They estimated that the U.S. Phillips curve allowed unemployment to be pushed down from 6.5% (in 1961) toward 4%, with only modest inflation rises (from 1% to 2% or 3%). The trade-off seemed favorable: accept a 1% inflation rise to get a 2.5 percentage-point unemployment drop.

This reasoning drove policy. The Federal Reserve kept rates low, Congress passed tax cuts, and the economy expanded. By 1965, unemployment had fallen to 3.8%, and inflation had risen to 3%. The Phillips curve trade-off appeared to work.

Economists formalized the relationship. The "menu of choice" concept emerged: policymakers could select any point on the Phillips curve, trading unemployment for inflation in whatever proportion they preferred. The society could choose 4% unemployment and 4% inflation, or 5% unemployment and 2% inflation, or 3% unemployment and 6% inflation. The Phillips curve was the constraint.

Stagflation and the breakdown of the simple Phillips curve​

This comfortable world shattered in the 1970s. In 1973, the OPEC oil embargo caused oil prices to quadruple. Simultaneously, inflation had been rising since the late 1960s (partly due to the Vietnam War spending) and inflation expectations had risen. Unemployment and inflation both rose at the same time—a phenomenon called stagflation that should not exist according to the simple Phillips curve.

In 1974, U.S. inflation reached 12% while unemployment was 7%. In 1975, inflation was 9% and unemployment was 8.5%. The simple Phillips curve predicted that high unemployment should accompany low inflation, but instead both were high. The curve seemed to have shifted outward, offering a worse menu: every point on the new curve involved more inflation for a given unemployment rate.

Economists and policymakers were shocked. Some questioned whether the Phillips curve existed at all. Others realized the missing element: inflation expectations.

The expectations-augmented Phillips curve​

Economists Milton Friedman and Edmund Phelps independently argued in the 1960s that the Phillips curve relationship should include inflation expectations. They distinguished between the short-run Phillips curve (the trade-off policymakers observe in real time) and the long-run Phillips curve (which is vertical at the natural rate of unemployment).

Here is the key insight: workers and firms negotiate wages based partly on what they expect inflation to be. If workers expect 3% inflation, they demand wage increases of at least 3% just to maintain real (inflation-adjusted) purchasing power. If they expect 6% inflation, they demand 6% wage increases.

The expectations-augmented Phillips curve incorporates this:

Inflation = Expected inflation + f(unemployment gap)

Where the unemployment gap is actual unemployment minus the natural rate.

This revised relationship explains the 1970s puzzle. In the 1960s, inflation expectations were anchored near 2% because inflation had been low. So the observed Phillips curve looked like 2% + f(unemployment). When unemployment fell to 3%, inflation rose to 3 or 4%.

But as inflation climbed in the late 1960s and early 1970s, inflation expectations rose. Workers and firms began expecting 5%, then 7%, then 10% inflation. The Phillips curve shifted outward. Now when unemployment fell to 3%, inflation rose to 8 or 9%, because firms were raising prices by the expected 7% plus extra for tight labor markets.

In the 1975 example: expected inflation was about 8%, unemployment was 8.5% (above the natural rate of roughly 4.5%), so the unemployment gap was about -4%. The expectations-augmented Phillips curve predicts inflation near 8% + f(-4%) = 8% - (some reduction) = roughly 9%. This matches the data.

The expectations-augmented Phillips curve also explains why the Volcker Fed's assault on inflation in the early 1980s required such high unemployment. To bring inflation from 14% down to 2% required crashing inflation expectations. That meant sustained, high unemployment—above the natural rate—for years. Unemployment rose above 9% in 1983. Only once inflation expectations had fallen back to 2% could unemployment be allowed to fall.

Shifting the Phillips curve through expectations​

The expectations-augmented Phillips curve reveals that the entire trade-off curve shifts when inflation expectations change. Here is how:

Step 1: Low inflation environment. Inflation is 2%, and expected inflation is 2%. The Phillips curve is based on 2% + f(unemployment gap). When unemployment is at the natural rate (no gap), inflation is 2%. When unemployment is 1 percentage point below natural, inflation is 3%.

Step 2: Surprise inflation shock. OPEC cuts oil supply, driving oil prices up. Firms raise prices. Inflation spikes to 5%. But inflation expectations have not yet adjusted; firms and workers still expect 2%.

Step 3: Expectations begin adjusting. After a few months of 5% inflation, workers and firms realize inflation is persistently high. They revise expectations upward to 4%. The Phillips curve now shifts: when unemployment is at the natural rate, inflation is 4%. When unemployment is 1 percentage point below natural, inflation is 5%.

Step 4: Further adjustment. If inflation stays at 5% for years, expectations continue rising toward 5%. The Phillips curve shifts again, now based on 5% expected inflation. When unemployment is at natural, inflation is 5%.

This shifting is the key to understanding why the 1970s stagflation occurred and why bringing it down required years of high unemployment.

The modern Phillips curve: flattening and anchored expectations​

Since the 1990s, the Phillips curve has flattened. The same unemployment drop that would have produced 1–2 percentage points of inflation in the 1970s now produces 0.1–0.3 percentage points. (See the Bureau of Labor Statistics analysis of inflation trends.)

Why has the curve flattened?

Inflation expectations anchoring. The Federal Reserve's credibility and successful inflation control in the 1980s and 1990s have led firms and workers to expect inflation to remain near 2% indefinitely. This is an anchor. Even when unemployment falls and labor markets tighten, the expected inflation component does not move much. Without rising inflation expectations, the observed Phillips curve relationship weakens.

Globalization and import competition. Firms competing with cheap imported goods cannot raise prices even when domestic labor markets tighten. Wage growth is suppressed by the threat of offshoring. This weakens the Phillips curve relationship.

Labor market changes. Declining union membership, rise of gig work, and weaker worker bargaining power mean wage growth is lower than tight labor markets would otherwise suggest. Workers cannot demand higher wages despite scarce labor.

Measurement issues. Some economists argue the Phillips curve has not actually flattened; we just have measurement problems or the right specification of the Phillips curve is different from what we are using.

The practical result: by the 2010s, unemployment fell from 10% (in 2009) to 3.5% (in 2019) without significant inflation. The simple Phillips curve suggested this should have triggered 3% to 4% inflation. Inflation remained near 2%.

This flat Phillips curve created a monetary policy puzzle. If the Phillips curve is flat, then unemployment is a poor guide to inflation pressure. The Fed could not rely on the unemployment rate to signal when to tighten policy. Other inflation signals (expectations surveys, wage growth indices, price growth) became more important.

The Phillips curve in the 2020s​

The 2020s has tested and clarified the modern Phillips curve. After the pandemic shutdown, demand surged while supply was constrained. Inflation spiked to 9% by mid-2022. Unemployment fell to 3.5%.

The key question: was the Phillips curve reasserting itself, or was inflation being driven by supply constraints rather than demand-driven unemployment?

Evidence suggests both were at play. Unemployment falling below 4% did contribute to wage pressures—wage growth accelerated to 5-6% by 2023. But the primary driver of inflation seemed to be supply constraints (semiconductors, energy, shipping) rather than demand. As supply recovered, inflation fell from 9% to 4% in 2023-2024 despite unemployment remaining around 4%.

This experience suggests the modern Phillips curve is weak but not absent. Tight labor markets do contribute to inflation, but the relationship is so attenuated by anchored expectations, globalization, and other factors that supply shocks dominate over demand pressures.

Real-world examples of the Phillips curve​

1960s United States. The Phillips curve was steep. Unemployment fell from 6.5% (1961) to 3.8% (1965), and inflation rose from 1% to 3%. The trade-off was favorable to policymakers, and policy was expansionary. By 1966, inflation had reached 4%, and the Fed began tightening.

1970s United Kingdom. The Oil Crisis and high inflation expectations led to stagflation. In 1975, unemployment was 4% and inflation was 24%—far worse than the 1960s Phillips curve suggested. The curve had shifted catastrophically outward.

1980s Volcker disinflation. Paul Volcker, Fed chair, pushed unemployment to 9.7% (1975) to break inflation expectations that had reached 13% (1980). Inflation fell to 2% by 1983. The transition required sustained, high unemployment as inflation expectations adjusted downward.

1990s-2000s. The Phillips curve flattened as inflation expectations became anchored near 2%. Unemployment fell to 4%, then 3.8%, without much inflation rise. By 2000, inflation was 3% despite 3.8% unemployment. The trade-off had become much less favorable to policymakers—moving down the Phillips curve no longer traded unemployment for inflation in the old proportion.

2015-2019 expansion. Unemployment fell from 5.3% to 3.5%, but inflation remained near 2%. The flat Phillips curve prediction held. Wage growth was only 3%, not the 5% inflation expectations might suggest.

2021-2023. Unemployment fell from 5% to 3.5%, and wage growth accelerated to 5-6%. But inflation was driven more by supply constraints than the Phillips curve relationship. As supply recovered, inflation fell despite tight labor markets, suggesting the curve remained flat.

Common mistakes​

Mistake 1: Treating the Phillips curve as a stable, exploitable trade-off. The 1960s policymakers thought they had found a permanent menu of choices. They were wrong. The Phillips curve shifts when inflation expectations shift, and the trade-off disappears once expectations adjust. There is no permanent trade-off between unemployment and inflation.

Mistake 2: Assuming a flat modern Phillips curve means inflation is no longer related to unemployment. The modern Phillips curve is flatter than in the 1960s-1970s, but the relationship is not zero. Tight labor markets do contribute to inflation, just more slowly and less directly than they once did. The relationship is attenuated but not absent.

Mistake 3: Ignoring that the Phillips curve can shift suddenly. Inflation expectations can shift rapidly if credibility is lost or a major shock occurs. A central bank with low inflation credibility can see the Phillips curve shift outward quickly if inflation spikes. Central bank credibility is crucial.

Mistake 4: Using 1970s Phillips curve estimates in the 2010s. The shape of the Phillips curve changes across decades. Estimates from the 1970s do not apply to the 2010s. Policymakers must regularly update Phillips curve estimates as institutions and inflation dynamics change.

Mistake 5: Forgetting that supply shocks affect the Phillips curve. A supply shock (oil price spike, pandemic supply chain disruption) can raise inflation without tight labor markets. When supply shocks dominate, the Phillips curve relationship is obscured. Policymakers must distinguish between demand-driven inflation (related to the Phillips curve) and supply-driven inflation (not directly related to unemployment).

FAQ​

Why is the Phillips curve important for central banks?​

The Phillips curve tells central banks how much slack exists in the labor market and how close inflation is to accelerating. A central bank can use Phillips curve estimates to gauge whether it should tighten or loosen policy. If unemployment is below the natural rate and the Phillips curve suggests inflation is rising, the central bank should raise rates.

How does the Phillips curve differ from Okun's law?​

Okun's law links unemployment to GDP growth (output gap to unemployment change). The Phillips curve links unemployment to inflation. Okun's law answers "how much unemployment changes when growth changes." The Phillips curve answers "how much inflation changes when unemployment changes." They are related but distinct relationships.

Can the Phillips curve ever be upward-sloping?​

Rarely, but theoretically yes. In extreme cases of supply shocks combined with policy errors, unemployment and inflation could rise together. This is stagflation, observed in the 1970s. But it is not a stable upward-sloping Phillips curve; it is a temporary shift in the relationship due to expectations or supply shocks.

Why did inflation expectations matter in the 1970s?​

Because workers and firms expected inflation to remain high, they incorporated high inflation into their wage and pricing decisions. The Phillips curve relationship held, but it shifted outward to reflect high expected inflation. Once expectations were high, only sustained high unemployment could bring inflation down.

How does the Fed use the Phillips curve today?​

The modern Fed uses the Phillips curve as one input among many. It monitors inflation surprises, inflation expectations surveys, wage growth, and labor market tightness. It does not rely solely on the Phillips curve to signal inflation pressure, because the curve is flat and inflation has remained anchored despite tight labor markets.

Is the Phillips curve still used by economists?​

Yes, but with caveats. Economists still estimate and use Phillips curves, but they are more skeptical about its reliability and place less weight on it than in the 1960s-1970s. Most central banks have diversified their inflation forecasting tools rather than relying primarily on the Phillips curve.

Summary​

The Phillips curve is the empirical relationship between unemployment and inflation, with lower unemployment associated with higher wage and price inflation. Discovered by A.W. Phillips in 1958, the relationship became central to 1960s monetary policy, suggesting policymakers could trade unemployment for inflation in a stable, exploitable way. The 1970s stagflation proved this wrong, revealing that inflation expectations shift the entire Phillips curve—workers and firms incorporate inflation expectations into wages and prices, shifting the trade-off outward when expectations rise. The modern Phillips curve is flattened by anchored inflation expectations, globalization, and labor market changes, meaning inflation responds less sharply to unemployment changes than it once did. The relationship remains relevant but is now one of many tools central banks use to gauge inflation pressure rather than the primary guide to monetary policy.

Next​

→ Modern Phillips curve developments