Phillips curve

No single relationship in macro is more famous, more useful, or more abused than the Phillips curve, and the brief demands you present its debate fairly, so learn it as a story of a beautiful idea that was half-right, broke spectacularly, was rebuilt, and remains genuinely unsettled.

What was the original Phillips curve, and why did it break?

The original curve (1958–1960). The New Zealand economist A.W. Phillips, studying nearly a century of British data (1861–1957), found a striking inverse empirical relationship between wage inflation and unemployment: when unemployment was low, wages (and prices) rose fast; when unemployment was high, they rose slowly or fell. Samuelson and Solow (1960) reframed it in terms of price inflation and christened it a policy "menu", a stable tradeoff a government could choose along: accept a bit more inflation to buy lower unemployment, or accept higher unemployment to get lower inflation. Through the 1960s this looked like a law of nature and became the organizing framework of macro policy: pick your point on the curve.

The breakdown (the 1970s). Then the curve vanished. The 1970s delivered high inflation and high unemployment simultaneously (stagflation), with points scattered far off any stable downward-sloping curve. The exploitable menu was gone; a government trying to "buy" low unemployment with a little inflation got the inflation and kept the unemployment. The simple Phillips curve had been empirically refuted, and its failure was one of the great intellectual crises of postwar economics.

What is the expectations-augmented Phillips curve?

The rebuild: the expectations-augmented Phillips curve (Friedman and Phelps). The fix had actually been predicted before the breakdown, by Milton Friedman (his 1968 American Economic Association presidential address) and Edmund Phelps independently. Their insight: the original curve forgot inflation expectations. The true relationship is expectations-augmented:

inflation ≈ expected inflation − β × (unemployment gap) + supply shock

where the unemployment gap is actual unemployment minus its "natural rate" (u − u*). Read what this says. The tradeoff between inflation and unemployment is real but only in the short run and only via expectational error. A government can push unemployment below the natural rate, but only by generating inflation higher than people expected. Once people catch on and raise their expectations, they build the new inflation into wages and prices, the short-run curve shifts up, and unemployment returns to its natural rate, now at higher inflation. Try to hold unemployment below the natural rate permanently, and you must deliver ever-accelerating inflation to keep fooling people (the "accelerationist" hypothesis).

What are the long-run Phillips curve and NAIRU?

Therefore the long-run Phillips curve is vertical at the natural rate of unemployment: there is no permanent tradeoff; you cannot buy lower unemployment with higher inflation in the long run, you can only buy higher inflation. The natural rate got a more precise name, the NAIRU (the Non-Accelerating-Inflation Rate of Unemployment): the unemployment rate below which inflation accelerates and above which it decelerates. Robert Lucas then pushed further (the rational-expectations revolution, and the Lucas critique, 1976): if people form expectations rationally, even the short-run tradeoff can only be exploited by genuine surprises, not by any anticipated policy, since a systematic attempt to inflate for jobs gets anticipated and neutralized. This body of work is why every modern central bank is obsessed with anchoring expectations and why the whole inflation-targeting framework exists: the vertical long-run curve means the only thing monetary policy can durably control is inflation itself, so it had better control it credibly.

Is the Phillips curve dead or just non-linear?

The modern debate: is the curve flat, dead, or just non-linear? (present both sides). Here the honesty the brief demands really matters, because the profession genuinely disagrees:

• The "it's flat / it's dead" case. From the 1990s to the late 2010s the Phillips curve appeared to go nearly flat: inflation barely responded to unemployment. In the 2010s US unemployment fell all the way to ~3.5% (2019) with no meaningful inflation pickup, and the curve seemed to have almost no slope. Proposed reasons: well-anchored expectations (a credible central bank pins expected inflation, so the expectations term dominates and swamps the slack term), globalization (domestic slack matters less when goods and labour are globally supplied), and better-measured, more stable NAIRU. On this view the curve is a poor, near-useless forecasting tool, and some declared it "dead."
• The "it's alive, just non-linear (and shocks matter)" case. Then 2021–22 happened: inflation surged. Revivalists argued the curve was never dead but convex / non-linear: nearly flat when there is slack, but steep when the labour market is very tight (unemployment far below NAIRU, unfilled vacancies everywhere), so it looks flat in normal times and bites suddenly at extremes. Others emphasized that 2021–22 was mostly supply shocks plus expectations, exactly the + supply shock and expected inflation terms in the augmented equation, doing the work the slack term alone never could.

How should an analyst use the Phillips curve?

The fair synthesis, the one an analyst should hold, is this: the mechanism is sound and durable (economic slack does push on inflation; expectations matter enormously; supply shocks matter), but the stable, precisely-calibrated, exploitable tradeoff the 1960s believed in does not exist. The NAIRU is unobservable and drifts, the slope is unstable and probably non-linear, and the curve is a weak forecasting tool that has embarrassed confident forecasters in both directions. Use it as a framework for thinking ("is the labour market tight or slack, are expectations anchored, is there a supply shock?"), never as a formula for predicting. One India-specific caveat to carry: the Phillips-curve logic is even weaker in India, where the labour market is largely informal (making "unemployment" and "slack" hard to measure), where inflation is supply-shock-dominated (food and fuel) rather than demand/slack-driven, and where the RBI's own models lean on the output gap rather than a clean unemployment measure. The tradeoff you were taught in an American textbook does not straightforwardly transplant to Indian soil, which is itself a lesson in reading macro frameworks critically rather than importing them whole.