The Phillips curve is a compact way of talking about a stubborn policy trade-off: when labour markets are tight, inflation often runs hotter; when unemployment is high, price pressure often eases. It is not a law of nature, and it has disappointed anyone who treated a single historical scatter plot as a permanent menu of choices. Used carefully, though, it remains one of the best teaching tools for connecting unemployment, inflation, and expectations.
Where the idea came from
In 1958, A. W. Phillips documented an inverse relationship between wage inflation and unemployment in the United Kingdom over a long historical sample. Later work reframed the idea in terms of price inflation and unemployment for many countries. The early message was seductive: perhaps policymakers could “buy” lower unemployment with a little more inflation, or cool prices by accepting higher joblessness.
That reading collided with the 1970s, when many advanced economies experienced high inflation and high unemployment together — stagflation. The missing ingredient was expectations. If workers and firms come to expect rising prices, they build those expectations into wage bargains and price-setting. Then the short-run trade-off can shift upward: the same unemployment rate is associated with higher inflation than before.
Short run versus long run
Modern textbooks usually draw a short-run Phillips curve that slopes down: lower unemployment, higher inflation, other things equal. In the long run, many models assume a vertical line at the natural rate of unemployment (sometimes called the non-accelerating inflation rate of unemployment, or NAIRU). Along that vertical line, inflation can be high or low depending on expectations and monetary policy, but you cannot permanently hold unemployment below its natural rate just by accepting higher inflation.
Think of the short-run curve as temporary bargaining room, and the long-run line as a reminder that expectations catch up.
When demand unexpectedly strengthens, unemployment can fall and inflation can rise for a while. As people revise their inflation expectations upward, the short-run curve shifts up. Returning unemployment to its earlier level then requires a period of tighter policy — the disinflation pain of the early 1980s in the US and UK is the classic case study.
What “tight” labour markets mean in practice
Unemployment is only one labour-market gauge. Vacancies, quit rates, wage growth, and participation also matter. A low headline unemployment rate with weak wage growth may signal less inflation pressure than the same rate with rapid pay rises in services. UK readers often watch ONS average weekly earnings alongside the Labour Force Survey unemployment rate; US readers pair the BLS unemployment rate with employment cost indexes and payrolls.
The Phillips curve is therefore better treated as a family of relationships between labour-market slack and price pressure than as one fixed equation. Different inflation measures (CPI, CPIH, PCE, core services excluding housing) can tell slightly different stories in the same month.
Why the curve seems to flatten or shift
Several forces can change how strongly unemployment maps into inflation:
- Anchored expectations. Credible inflation targets can keep expectations near 2% even when unemployment moves, flattening the short-run response for a time.
- Global supply. Cheaper imported goods can mute domestic wage pressure in consumer prices.
- Labour-market institutions. Bargaining coverage, minimum wages, and migration alter how quickly pay responds to vacancies.
- Measurement. Changes in how unemployment or inflation is measured can look like “curve shifts” when they are really data changes.
After the global financial crisis, many economies saw unemployment fall without a sharp inflation pickup for years — the so-called flat Phillips curve debate. The post-2021 inflation surge reminded everyone that supply shocks and rapid demand recovery can still move prices quickly, even if the wage channel is only part of the story.
How central banks use the idea
Neither the Federal Reserve nor the Bank of England steers policy from a single Phillips-curve chart. They do use the underlying logic: estimate slack, watch wage and price dynamics, and judge whether inflation expectations remain anchored. Forecasts embed some version of a Phillips-type relationship, then stress-test it against alternative scenarios.
Suppose unemployment sits near an estimated natural rate and core inflation is close to target. A demand boom that pushes unemployment well below that estimate may raise the odds of above-target inflation with a lag. The numbers and lags are empirical — check FOMC or Monetary Policy Report materials rather than treating any classroom diagram as a forecast.
UK and US framing
US discussions often emphasise the Fed’s dual mandate and the role of PCE inflation. UK discussions emphasise the BoE’s 2% CPI target and the interaction of pay growth with services inflation. Both countries learned in the 1970s–80s that expectations matter, and again after 2021 that energy and supply shocks can dominate the first stage of an inflation episode.
Related reading: inflation, unemployment, real interest rates, and glossary entries for natural rate and inflation.
Common misreadings to avoid
First, the Phillips curve is not a promise that higher inflation permanently buys lower unemployment. Expectations and credibility sit in the middle of any serious version of the story. Second, correlation in a scatter plot is not a structural policy menu — identification problems abound when supply shocks move inflation and unemployment together. Third, “the” curve is really a shifting cloud of relationships across sectors and measures; services inflation may track wage pressure more closely than goods inflation in some episodes.
If you take only one habit from this explainer, let it be this: when someone waves a Phillips-curve chart, ask what inflation measure, what slack measure, what sample period, and whether expectations are assumed anchored. Those four questions separate classroom cartoons from policy-relevant discussion.