The inflation–unemployment relationship resembles a policy menu only when expectations and economic structure are held constant. The history of the Phillips curve shows why that assumption matters: a short-run correlation can shift when expectations, supply shocks or the monetary regime change.
A curve conditional on expectations
This experiment uses a simplified expectations-augmented relationship:
denotes inflation, expected inflation, unemployment and a natural-rate reference. Holding expectations constant, greater unemployment lowers predicted inflation. Increasing shifts the whole curve upwards. Increasing moves the point where to the right.
The simplified model uses as a NAIRU reference. In empirical work, the natural rate and the non-accelerating-inflation rate of unemployment need not be identical, and neither is directly observed.
- Illustrative slope β
- —
- At u = 5%: inflation (%)
- —
- Annual observations
- —
Dots are observed data; the curve, expectations, β and uₙ are illustrative scenarios, not estimates. Changing the period resets the sliders. The dashed line marks uₙ.
Historical data and sources
Unemployment: annual mean of the 12 seasonally adjusted monthly observations (BLS). Inflation: annual consumer-price inflation (World Bank). Snapshot: October 31, 2026; values rounded to 3 decimals. FRED · UNRATE · FRED · FPCPITOTLZGUSA
| Year | Unemployment (%) | Inflation (%) |
|---|
Three periods, no immutable law
Dots are actual annual US observations from the 1960s, the 1970s–80s and 2000–2020. Inflation is the World Bank annual consumer-price inflation indicator. Unemployment is the annual mean of twelve monthly, seasonally adjusted BLS observations. Both series are available through FRED and bundled locally, requiring no runtime data requests.
The 1990s are left out on purpose. They form the core of the so-called Great Moderation: low, stable inflation alongside falling unemployment. As a transition between the 1980s regime and the flatter recent curve, including them would blur the contrast between the three scenarios.
The slope in each scenario is illustrative, not estimated from the displayed observations. A smaller slope represents flattening in the recent scenario. Neither sliders nor historical presets reconstruct actual expectations; they help distinguish movement along a curve from a shift of the curve.
Empirical warning. A stable curve does not imply a permanent policy trade-off. Supply shocks can increase inflation and unemployment together; the equation omits them to isolate expectations. The displayed curve is neither a forecast nor a NAIRU estimate.
Sources: BLS monthly unemployment via FRED and World Bank annual inflation via FRED.