The trade-off between inflation and unemployment — and why it fell apart in the 1970s
In 1958, a New Zealand economist named A.W. Phillips published a paper that seemed to show something beautifully simple: when unemployment is low, wages (and prices) rise faster. When unemployment is high, inflation stays low. There’s a trade-off between the two.
Policymakers were thrilled. It looked like they had a menu: want less unemployment? Accept a bit more inflation. Want less inflation? Accept a bit more unemployment. Choose your preferred combination and dial it in.
Then the 1970s happened. Unemployment shot up AND inflation shot up — at the same time. Stagflation. The menu didn’t work anymore. The curve had collapsed.
What went wrong? That’s the most important lesson of this topic.
Phillips looked at UK data going back to 1861 and found a clear pattern: when unemployment was low (workers scarce, bargaining power high), wages rose faster. When unemployment was high, wage growth was slow. Plotting this gave a downward-sloping curve.
Samuelson and Solow (1960) updated it for the US using price inflation instead of wage growth. The simple version looks like this:
π = a − b × u
π = inflation rate, u = unemployment rate, a and b are positive numbers
So: more unemployment → less inflation. Less unemployment → more inflation. Simple trade-off.
The policy implication: governments could “buy” lower unemployment by accepting higher inflation — like choosing where to sit on the curve.
In 1968 — before the 1970s crisis — Milton Friedman and Edmund Phelps independently warned that the original Phillips curve was missing something crucial: inflation expectations.
Their insight: workers don’t just care about the nominal wage (the number on their payslip). They care about the real wage — what they can actually buy with it. If workers expect 5% inflation, they’ll demand at least 5% more pay just to stand still. This means the short-run Phillips curve shifts upward every time expected inflation rises.
π = πe + α(u* − u) + supply shocks
πe = expected inflation, u* = NAIRU, α captures the trade-off steepness
Say the NAIRU is 5% and the government wants to push unemployment to 3%:
This is the “accelerationist hypothesis” — constantly undershooting the NAIRU causes inflation to accelerate indefinitely. It’s why the NAIRU is called the “Non-Accelerating Inflation Rate of Unemployment.”
In the long run, once expectations fully adjust, there’s no trade-off at all. For every attempt to push unemployment below the NAIRU, the economy eventually drifts back to the NAIRU — but at a higher inflation rate. The long-run relationship between inflation and unemployment is a vertical line at the NAIRU.
Think of it like running on a treadmill. You can run faster in the short term — but the treadmill speeds up to match you, and you end up in the same spot, just more exhausted (more inflation, same unemployment).
In 1973, OPEC cut oil production dramatically. Oil prices quadrupled. This was a supply shock — it raised costs for virtually every business in the world. Firms responded by raising prices. But they also cut production and laid off workers.
Result: prices rose AND unemployment rose — simultaneously. The original Phillips curve had no room for this. But the expectations-augmented version does: a supply shock shifts the entire short-run curve upward — higher inflation at every unemployment rate.
| Period | Unemployment | Inflation | What happened |
|---|---|---|---|
| 1960s US | ~4% | ~2–3% | Moving along a stable SRPC |
| 1973–74 | Rising to 7% | Rising to 12% | OPEC oil shock — SRPC shifted up |
| 1981–83 | Rose to 10% | Fell from 13% to 4% | Volcker’s painful disinflation |
Once inflation is embedded in expectations, reducing it is painful. You have to push unemployment above the NAIRU to get inflation to fall — which means a recession.
Paul Volcker, the Fed chair in 1979–1987, deliberately raised interest rates aggressively to crush inflation. US unemployment hit 10%. But inflation fell from 13% to 3%. Ball (1994) calculated the sacrifice ratio for this disinflation at roughly 3 — meaning every 1 percentage point reduction in inflation cost about 3 percentage-point-years of below-trend output. It worked, but it was brutal.
Sacrifice Ratio = Output/unemployment lost / Inflation reduction achieved
In June 2022, US inflation hit 9.1% — the highest since 1981. Unemployment was around 3.5%. This looked like the “hot” end of the Phillips curve — low unemployment, high inflation. But what actually drove it?
Ben Bernanke and Olivier Blanchard (2023) — two of the world’s most respected macroeconomists — dug into the data. Their finding: most of the 2021–2022 inflation was driven by supply shocks (pandemic-disrupted supply chains, energy price spikes from the Ukraine war), not an overheated labour market. The SRPC shifted upward — same story as 1973.
That’s partly why inflation came back down relatively quickly from 2023 onwards, without needing a massive rise in unemployment — the “immaculate disinflation” that most forecasters had said was impossible. Once supply chains normalised and energy prices fell, the SRPC drifted back down.
However, Ball, Leigh and Mishra (2022) argued demand also played a role — the huge US fiscal stimulus ($1.9 trillion American Rescue Plan) boosted demand while supply was constrained. So the full story involves both supply and demand shifting the curve.
Sources: Bernanke, B. & Blanchard, O. (2023). Brookings Papers on Economic Activity. Ball, L., Leigh, D. & Mishra, P. (2022). Brookings Papers on Economic Activity.
An economy has a NAIRU of 5%. The short-run Phillips curve is: π = πe − 0.8(u − u*). Currently πe = 3% and u = 3%. (a) What is inflation this period? (b) If expectations fully update next period and u stays at 3%, what’s next period’s inflation?
(a) π = 3% − 0.8(3% − 5%) = 3% − 0.8 × (−2%) = 3% + 1.6% = 4.6%
(b) Next period: πe = 4.6%, u = 3%. π = 4.6% − 0.8(3% − 5%) = 4.6% + 1.6% = 6.2%
Inflation is accelerating each period because u < u*. This will keep going until either unemployment rises to the NAIRU or the government stops stimulating — the accelerationist trap in action.
“The long-run Phillips curve is vertical, meaning monetary policy cannot permanently reduce unemployment.” Evaluate this statement.
In favour (supporting the statement): Friedman (1968) showed that any attempt to keep unemployment below the NAIRU leads to accelerating inflation as expectations adjust upward. In the long run, the economy always returns to the NAIRU. The 1970s validated this — attempts to maintain low unemployment through expansionary policy ended in stagflation. The long-run curve is vertical: there’s no lasting trade-off.
Against (challenging the statement): (1) Hysteresis — Blanchard and Summers (1986) showed that deep recessions permanently raise the NAIRU itself. Long-term unemployed workers lose skills and connections, effectively leaving the labour market. If monetary policy causes a recession to fight inflation, the NAIRU itself rises — the “natural” rate is not so natural. (2) The 2010s in the US showed unemployment could fall well below typical NAIRU estimates (to 3.5%) without triggering significant inflation, suggesting either the NAIRU fell or the curve had become very flat. (3) Well-anchored inflation expectations (from credible central banks) mean the curve can shift much less for a given unemployment gap.
Balanced conclusion: The vertical LRPC is a useful long-run approximation, but the NAIRU itself is not fixed — it responds to actual unemployment history (hysteresis) and labour market policies. The statement is broadly correct but overstates the constraint on monetary policy by treating the NAIRU as exogenous and fixed.
Game Theory · Classic ProblemsThe Prisoner's DilemmaWhy two perfectly rational individuals can end up worse off by acting in their own self-interest — the most famous problem in all of game theory.Imagine you and a friend are arrested for a crime. The police separate...
Statistics · Descriptive StatisticsWhat is Variance?The average squared deviation from the mean — why it matters, how to calculate it, and how it relates to standard deviation and risk.Two investment portfolios both return an average of 8% per year. But Portfolio A...
Statistics · Probability TheoryThe Central Limit TheoremWhy sample means always tend toward a normal distribution — regardless of the shape of the original population. The most remarkable result in all of statistics.Suppose you're studying the income of taxi drivers...
Game Theory · Classic ProblemsThe Prisoner's DilemmaWhy two perfectly rational individuals can end up worse off by acting in their own self-interest — the most famous problem in all of game theory.Imagine you and a friend are arrested for a crime. The police separate...
Statistics · Descriptive StatisticsWhat is Variance?The average squared deviation from the mean — why it matters, how to calculate it, and how it relates to standard deviation and risk.Two investment portfolios both return an average of 8% per year. But Portfolio A...
Statistics · Probability TheoryThe Central Limit TheoremWhy sample means always tend toward a normal distribution — regardless of the shape of the original population. The most remarkable result in all of statistics.Suppose you're studying the income of taxi drivers...
