Keynesians vs. monetarists: whose framework did the 1970s vindicate?
Two frameworks ran the postwar economy. One said you could trade a little inflation for lower unemployment forever. The other said that trade was an illusion that would self-destruct the moment you tried to use it. The 1970s settled the bet — but not the way either side’s partisans remember.
See the lineage as a graphThe consensus that ran the 1960s
“In order to achieve the nonperfectionist’s goal of high enough output to give us no more than 3% unemployment, the price index might have to rise by as much as 4 to 5% per year. That much price rise would seem to be the necessary cost of high employment and production.”
— Paul Samuelson & Robert Solow, “Analytical Aspects of Anti-Inflation Policy,” American Economic Review, 1960
Read it again: a cost the economy could simply pay. By 1960 the leading economists of their generation believed inflation and unemployment sat on a stable menu, and a government could pick the point it liked. This was not a fringe view or a crude one. It was the most sophisticated macroeconomics on earth.
To see why the menu looked solid, you need the framework underneath it: the postwar neoclassical synthesis. Keynes had shown in 1936 that an economy could get stuck below full employment, with idle workers and idle factories, because total spending fell short. John Hicks turned the argument into the IS-LM apparatus, and a generation of economists fused it with conventional microeconomics into a single working machine. The state had a lever — aggregate demand — and a target — full employment — and the math said the lever reached the target.
The policy program followed directly. If output was too low, raise demand: cut taxes, raise spending, ease money. The fiscal multiplier meant a dollar of government spending raised output by more than a dollar as it circulated. Fine-tuning was the watchword — the economy was a machine with dials, and trained hands could hold it near full employment year after year.
Then came the piece that made it a menu. In 1958 A.W. Phillips plotted nearly a century of British data and found a stable inverse relationship between wage inflation and unemployment: when unemployment was low, wages rose fast; when it was high, they rose slowly or fell. Samuelson and Solow imported the curve to the United States and read a policy choice into it. You could have lower unemployment if you accepted higher inflation, or lower inflation if you accepted higher unemployment. Pick your point.
And the 1960s data cooperated. Through most of the decade, US inflation and unemployment did trace out something like the curve. The Kennedy-Johnson tax cut of 1964 was the program in action — deliberate demand stimulus, followed by falling unemployment and the long boom. To a policymaker in 1968, the framework was not a theory awaiting a test. It was a working tool with a track record.
The formal home of this apparatus — the Keynesian-cross multiplier, the IS-LM machine, the AD-AS model that the Phillips curve sits inside — is the intermediate-macro core. The intellectual lineage, from Keynes through the Hicks transformation to the Samuelson-Solow Phillips menu, is its own chapter.
The framework at its strongest
“The Phillips curve... seems to have provided a remarkably stable description of the relation between wage changes and unemployment over a long period.”
— The empirical consensus of the early 1960s, as the synthesis read it
This is the Keynesian framework arguing its own case. The lineage runs through Samuelson, Solow, and James Tobin — the architects of the synthesis that turned Keynes into an operating manual. Their claim was modest and powerful at once: we have a stable empirical relationship, a coherent theory of why output can sit below potential, and a lever that demonstrably moves it. The state can stabilize the business cycle and hold the economy near full employment. Two decades of postwar prosperity were the evidence. To call this naive in advance is hindsight; in 1968 it was the frontier.
“All of our discussion has been phrased in short-run terms... it would be wrong to think that [the menu] is so reliable that it will hold... over the long pull.”
— Samuelson & Solow’s own caveat, American Economic Review, 1960
The sharpest critic of the menu was the menu’s own authors. They knew the curve was a short-run relationship and said so. What they did not have was a theory of why it would break — a mechanism that explained how trying to exploit the trade-off would dissolve it. That mechanism was about to arrive from Chicago. The fiscal-multiplier piece of this framework, argued in its own dedicated walkthrough, lives at Does government spending help the economy?
Where this leaves us
The Keynesian framework was the best macroeconomics of its era, and its 1960s record was real. It had a working theory, a working lever, and twenty years of prosperity behind it. But its load-bearing assumption — that the inflation-unemployment trade-off was stable, invariant to the very policy that tried to exploit it — was exactly the assumption a decade was about to test. The menu was real until someone tried to live on it.
In December 1967, a Chicago economist stood before a roomful of his peers as the incoming president of their professional association and told them the menu was a mirage — that the moment policy tried to hold unemployment below its natural level, inflation would not stay put but accelerate without limit. Almost no one in the room believed him. His name was Milton Friedman, and the decade was about to take his side.
Friedman’s challenge
“There is always a temporary trade-off between inflation and unemployment; there is no permanent trade-off. The temporary trade-off comes not from inflation per se, but from unanticipated inflation, which generally means, from a rising rate of inflation.”
— Milton Friedman, “The Role of Monetary Policy,” AEA Presidential Address, American Economic Review, 1968
This is a prediction, made in advance, in front of the whole profession. Before the decade broke, Friedman told the room exactly how the menu would fail: not a stable trade you could hold, but a temporary one that demanded ever-faster inflation to maintain — and collapsed the instant policy stopped accelerating it.
Friedman’s argument has one moving part: expectations. The Phillips menu works only if workers and firms are fooled. When the central bank stimulates demand, prices rise, but nominal wages are set in advance, so real wages fall — firms hire more, unemployment drops. That is the trade-off. But it depends on workers not noticing that their real wages have fallen.
They notice. Once people expect inflation, they demand wages that account for it. The bargain that bought lower unemployment evaporates: real wages return to where they were, hiring returns to normal, and you are left with the higher inflation and nothing to show for it. To keep unemployment below its natural rate — the rate consistent with the economy’s real structure — the central bank has to keep surprising people, which means inflation must keep rising. The menu is not a place you can stand. It is a treadmill that has to speed up forever.
Formally, the expectations-augmented Phillips curve replaces the naive trade-off with one that includes expected inflation $\pi_t^e$:
$$\pi_t = \pi_t^e - \beta(u_t - u^*) + \varepsilon_t$$where $u^*$ is the natural rate of unemployment and $\varepsilon_t$ is a supply shock. In the long run, expectations catch up to actual inflation, so $\pi_t^e = \pi_t$. Set them equal and the inflation terms cancel: $0 = -\beta(u_t - u^*)$, which forces $u_t = u^*$. The long-run Phillips curve is vertical at the natural rate. To hold $u_t < u^*$, you need $\pi_t > \pi_t^e$ every period — inflation must accelerate without bound. There is no stable point below $u^*$ to choose.
You can fool the workers once. You raise prices, their real wages secretly fall, firms hire, unemployment drops. But you can only do it again if you fool them more — a bigger surprise than last time. Keep it up and the surprises have to grow forever, which means inflation has to grow forever. The instant you stop accelerating, the trick stops working and unemployment snaps back to where it started. The lower unemployment was never bought; it was borrowed against an inflation that has to keep rising to pay the interest.
Behind the natural-rate move sits the older monetarist core: Friedman’s claim that “inflation is always and everywhere a monetary phenomenon” — that the price level is governed, over any horizon that matters, by the growth rate of the money supply, not by fiscal fine-tuning or union militancy or oil. The quantity theory, restated for the modern economy, is the engine room of the framework. The natural-rate hypothesis is what it implies for the Phillips menu specifically.
“There is always a temporary trade-off between inflation and unemployment; there is no permanent trade-off.”
— Milton Friedman, AEA Presidential Address, 1968
Was the trade-off real, or always an illusion?
Friedman and Phelps argued the Phillips menu was not a stable relationship at all but a temporary artifact of fooling people — one that would self-destruct the moment policy tried to exploit it. And they said so before the decade that tested the claim.
A prediction made in advance
“To state [the] conclusion differently, there is always a temporary trade-off between inflation and unemployment; there is no permanent trade-off.”
— Milton Friedman, AEA Presidential Address, 1968
This is the monetarist framework at full strength. Friedman is not offering a competing curve to fit alongside the old one — he is denying that the long-run curve has a slope at all. The lineage runs through Friedman and Phelps, and the canonical diagram — a short-run Phillips curve that ratchets upward as expectations adjust, sliding along a vertical long-run curve — is the framework’s signature. What makes the claim formidable is its timing. It was a forecast, not a postmortem. The framework put its prediction on the record three years before the decade that would judge it.
“The hypothesis [is] that there is a ‘natural rate of unemployment’... consistent with equilibrium in the structure of real wage rates.”
— Friedman’s own framing of the load-bearing premise, 1968
The vulnerable point is the natural rate itself. The whole argument hinges on an unobservable — a “natural” unemployment rate set by real factors, around which actual unemployment oscillates. Critics asked the obvious question: how do you know what it is? If you can only infer the natural rate after the fact, the hypothesis risks becoming unfalsifiable — whatever happens, you can say unemployment was above or below natural. The defense is that the accelerationist prediction is testable independent of pinning down $u^*$ precisely: if sustained stimulus produces accelerating inflation rather than a stable lower-unemployment point, the menu is broken regardless of where exactly the natural rate sits. The political backstory of how this Chicago argument won its audience — the broader rise of free-market economics in the Cold War climate — is its own question.
Where this leaves us
The monetarist framework made a sharp, falsifiable, in-advance prediction the Keynesian framework did not: sustained demand stimulus below the natural rate produces accelerating inflation, not a stable higher-inflation point. The two frameworks now disagree on the record about what a decade of trying to hold down unemployment would do. Whether the prediction was right is not something either side could settle by argument. It needed data. The data was about to arrive — and it arrived carrying a complication that neither framework had placed at the center of its story.
The test came faster than anyone expected. By 1974, inflation and unemployment across the rich world were rising together — the one thing the stable menu said could not happen. But the decade did not test the two frameworks cleanly. Twice, in 1973 and 1979, the price of oil exploded, shoving the whole economy in a direction neither the Keynesian nor the monetarist apparatus had built its case around. Which framework called the decade — and how much credit does each get when oil did some of the work?
The decade as the test
“The menu had one rule: you couldn’t have both. The 1970s had both.”
— The single fact the decade put on the table
The Phillips menu forbade exactly one combination: high inflation and high unemployment at the same time. Across the G7 in the 1970s, the two rose together — the misery index, inflation plus unemployment, climbing year after year. The forbidden box was where the whole decade lived. For the lived chronology of how it got there — the gas lines, the embargo, the Nixon shock — see the case-up walkthrough on the decade itself; this stage uses the data as the verdict, not the story.
Open the panel below to drive the actual G7 misery path against the expectations-augmented Phillips engine — the exhibit the debate turned on.
Put the two frameworks on the same diagram and watch them diverge. The Keynesian reading expects the economy to slide along a stable downward-sloping Phillips curve: more demand, less unemployment, a bit more inflation, and you stop where you like. The monetarist reading expects the short-run curve to shift upward year after year as expected inflation climbs — the economy tracing an outward spiral, not a stable point. The 1970s data traced the spiral. Each year the short-run trade-off looked worse than the last; the same unemployment rate came paired with ever-higher inflation. That is the accelerationist path, drawn in real time.
On the dynamic AD-AS diagram, the short-run aggregate supply curve is the expectations-augmented Phillips relation rearranged. As actual inflation runs ahead of expectations, expectations ratchet up: $\pi_{t+1}^e = \pi_t$ under adaptive expectations. Each period the short-run curve shifts so that the same output gap now sits at higher inflation:
$$\pi_t = \pi_{t-1} - \beta(u_t - u^*) + \varepsilon_t$$With expectations equal to last period’s inflation, holding $u_t < u^*$ makes $\pi_t > \pi_{t-1}$ every period — inflation accelerates. And the supply-shock term $\varepsilon_t$ is not idle: the 1973 and 1979 oil shocks were large positive $\varepsilon_t$, pushing inflation and unemployment up together independent of the expectations channel. The decade is therefore not a clean monetary experiment — but the direction it moved, both variables rising, is what the monetarist framework predicted and the stable-menu Keynesian framework forbade.
Imagine the Phillips menu as a price list that quietly reprints itself every year. You buy a little less unemployment with a little more inflation — and next year the same purchase costs more inflation, because everyone now expects last year’s rate. Keep buying and the list reprints faster and faster. Then add oil: two enormous price shocks that raised inflation and killed output at the same stroke, with nothing to do with money or expectations. The honest reading is that both engines ran at once — the expectations treadmill the monetarists described, and the supply shocks neither framework had centered. But only one framework had ruled out the box the decade lived in.
“The menu had one rule: you couldn’t have both. The 1970s had both.”
— The G7 misery index, 1971–1985
Did the 1970s settle the natural-rate argument?
The retrospective reading is clean: the decade was the decisive experiment, inflation and unemployment rose together, the stable menu broke, Friedman was right. But “won the prediction” and “won the argument” are not the same claim — and the gap between them is the rest of this walkthrough.
Both frameworks read the same decade
“The predictions [of Keynesian models] were wildly incorrect, and the doctrine on which they were based is fundamentally flawed... the task now facing contemporary students of the business cycle is to sort through the wreckage.”
— Robert Lucas & Thomas Sargent, “After Keynesian Macroeconomics,” 1979
This is the counter-revolution claiming the decade. Lucas and Sargent hardened Friedman’s expectations move into full rational expectations: agents do not just adapt to past inflation, they anticipate policy itself, so any trade-off a policymaker tries to exploit dissolves the instant it is announced. On this reading the 1970s were not a surprise but a confirmation — the predictable result of running a model whose coefficients were never stable to begin with. The framework called it, and the wreckage was the proof.
“The ‘accelerationist’ or Friedman-Phelps hypothesis... is by now accepted by the majority of economists. I, for one, see no point in disputing it... [but it] does not imply that monetary and fiscal policies are powerless.”
— Franco Modigliani, AEA Presidential Address, 1977
This is the load-bearing voice of the stage — a leading Keynesian conceding the data. Modigliani, in his own AEA Presidential Address, did not deny the natural-rate point or cling to the stable menu. He accepted it outright: the accelerationist hypothesis was correct, the long-run curve was vertical, the trade-off the synthesis had banked on was gone. The best Keynesians lost the menu and admitted it. But Modigliani then drew the line that survives into the synthesis: conceding the long-run vertical curve does not concede that stabilization policy is useless, because nominal rigidities make the short run real and a deep recession is not self-correcting fast enough to leave alone. He gave up the menu and kept the foundation — which is exactly the move Stage 4 turns into the resolution.
See the two lineages set against each other in the intellectual-genealogy graph: the Keynesian-revolution cluster (Keynes, Samuelson, Solow, Tobin) and the counter-revolution cluster (Friedman, Phelps, Lucas, Sargent) — the two frameworks this walkthrough holds against each other, drawn as lineages.
Where this leaves us — the first layer
Monetarists won the prediction. The decade confirmed the accelerationist natural-rate framework and falsified the stable-trade-off Keynesian framework — at the level of the empirical claim the two actually disagreed on. The stable Phillips menu is gone from serious macroeconomics. Mark this as one layer of the verdict, not the whole of it: winning the prediction is not yet winning the argument. The monetarist policy program still had its own test ahead, and that test ran the other way. As cross-country confirmation that the break was not a US peculiarity — that the misery rose across the rich world at once — the per-capita-GDP trajectories of the G7 over 1971 onward sit on the GDP map.
Winning the prediction is not the same as winning the policy. Monetarism did not just diagnose the disease; it prescribed a cure — stop trying to fine-tune, fix the growth rate of the money supply, and let the economy find its own natural rate. The decade had tested the diagnosis and confirmed it. Now it was about to test the prescription. That part did not go the way the diagnosis did.
The resolution: what each framework got wrong
“By emphasizing the supply of bank reserves and the growth of the money supply, the new procedures... placed greater reliance on fluctuations in interest rates to bring the demand for money into line with a predetermined target for monetary growth.”
— The Federal Reserve’s October 6, 1979 statement announcing the shift to a money-quantity operating target (the “Saturday Night Special”)
On a Saturday in October 1979, the monetarist program stopped being a diagnosis and became a policy regime. Paul Volcker’s Fed announced it would target the quantity of bank reserves and let interest rates go wherever that required. This was the k-percent rule meeting the real economy — the moment monetarism got to prove that controlling money was the same as controlling inflation.
The monetarist policy claim was distinct from the diagnosis. The diagnosis said the price level is governed by money growth in the long run. The policy said: therefore the central bank should abandon discretionary fine-tuning, announce a fixed, low rate of money-supply growth — the k-percent rule — and stick to it. Remove the discretion, anchor expectations, and inflation has nowhere to hide. On paper it is elegant: a transparent rule that takes the wheel out of fallible human hands.
It ran into a wall the diagnosis had not anticipated: money-demand instability. The k-percent rule assumes a stable relationship between the money supply and nominal spending — a predictable velocity. But just as the experiment began, financial innovation and deregulation tore that relationship apart. Money-market funds, interest-bearing checking accounts, and new payment technologies meant that “the money supply” was a moving target whose link to inflation kept slipping. The Fed would hit its reserve target and watch inflation and output do something other than what the target predicted. The dial monetarism wanted to turn was no longer connected to the engine.
So Volcker broke inflation — through the brutal 1981–82 recession the high interest rates produced — but he did it less by hitting money-growth numbers than by accepting whatever interest rate was required to crush demand. By 1982 the Fed had quietly abandoned strict reserve targeting and returned to managing the funds rate directly. Through the rest of the decade, central banks across the rich world downgraded the monetary aggregates from policy targets to background indicators. The diagnosis traveled into every textbook. The policy rule did not survive contact with the data.
What came next kept the diagnosis and discarded the rule. The New Keynesian synthesis absorbed the natural-rate hypothesis and rational expectations — the long-run Phillips curve is vertical, expectations are forward-looking — while keeping the Keynesian insight the monetarists had downplayed: nominal rigidities are real. Prices and wages do not clear instantly (Calvo and menu-cost pricing model exactly this), so there is a genuine short-run output-inflation trade-off and a real job for stabilization policy. The synthesis is named here, not re-derived; its internal machinery is its own chapter.
“Monetarism died sometime in the early 1980s, when central banks discovered that the money supply was no longer a reliable guide to anything.”
— The verdict the operating record forced on the policy rule
Did monetarism win the 1970s?
Yes and no, and the “no” matters as much as the “yes.” The monetarist diagnosis became textbook orthodoxy. The monetarist policy rule was abandoned within three years of getting its real-world test. Conflating the two is the most common over-claim about the whole episode.
The policy rule at its strongest — and the foundation that survived
“Inflation is always and everywhere a monetary phenomenon... and can be produced only by a more rapid increase in the quantity of money than in output.”
— Milton Friedman, the monetarist policy claim at its sharpest
Take the policy at full strength before judging it. If inflation is fundamentally monetary, the cleanest cure is a rule: fix money growth, stop improvising, let the economy settle at its natural rate. It removes the discretion that lets politics inflate, and it gives the public a number to anchor on. The Volcker disinflation is the case that seems to vindicate it — a Fed that adopted a quantity target, accepted the deepest postwar recession, and broke a decade of inflation. As a demonstration that a credible central bank committed to price stability can win, it is the canonical success. Drive the disinflation engine yourself in the live exhibit below.
“[The accelerationist hypothesis] does not imply that monetary and fiscal policies... are powerless to stabilize the economy.”
— Franco Modigliani, defending the surviving Keynesian foundation, 1977
This is the Keynesian insight that the decade did not kill, argued at its strongest. Prices and wages do not clear instantly. That is not an ideological preference — it is why Volcker’s disinflation cost millions of jobs instead of happening overnight. If wages had adjusted freely, breaking inflation would have been painless; it was agonizing precisely because nominal rigidity is real. And nominal rigidity is exactly the foundation the New Keynesian synthesis kept: a short-run trade-off that gives stabilization policy a job, sitting underneath a vertical long-run curve that grants the monetarist diagnosis. Keynesianism did not lose this decade. It lost a naive curve and kept its load-bearing wall.
The internal contest the synthesis then had with itself — New Classical versus New Keynesian over how much stabilization policy can actually do — is the next round of this argument, and its own walkthrough.
Where this leaves us — the full verdict
The 1970s adjudicated the dispute, and the verdict has three layers, each reported with reasons. The monetarist diagnosis won: the Friedman-Phelps natural-rate hypothesis is now textbook — no exploitable long-run trade-off, the long-run curve is vertical, sustained stimulus below the natural rate accelerates inflation. Friedman predicted it in 1968, in advance. The monetarist policy lost: stable money-growth targeting proved unworkable because money-demand instability broke the velocity link; the Volcker experiment won via the funds rate and the recession, not the money numbers, and the aggregates were abandoned by 1982. And Keynesianism was not simply the defeated side: its nominal-rigidity foundation survived and was absorbed into the New Keynesian synthesis. The image to keep is the one every modern central bank’s model carries: an expectations-augmented Phillips curve that is monetarist in its long-run shape and Keynesian in its short-run slope. This is a consensus with components, not a draw and not a wholesale monetarist win. Neither framework won outright; the synthesis kept the best of both.
For the live modern argument over how much a central bank can really control — the Volcker credibility regime and what came after — see Can central banks control the economy?
The verdict, in three layers
We started with two frameworks and a bet. The Keynesian synthesis said the inflation-unemployment trade-off was a stable menu a government could pick a point on; the monetarist challenge said the menu was an illusion that would self-destruct the moment policy tried to exploit it. The 1970s settled the bet, but not into a clean win for either side. The honest verdict has three layers, and naming all three is the whole point of holding the two frameworks side by side:
- The monetarist diagnosis won. The natural-rate hypothesis is now consensus: there is no exploitable long-run trade-off, the long-run Phillips curve is vertical, and sustained demand-management below the natural rate accelerates inflation. Friedman called it in 1968, before the decade that proved it.
- The monetarist policy lost. The k-percent money-growth rule could not survive money-demand instability; Volcker broke inflation with the interest rate, not the money numbers, and the monetary aggregates were quietly retired across central banks within a few years.
- Keynesianism kept its foundation. Nominal rigidities are real, the short-run trade-off is real, and stabilization policy has a job — the insight the New Keynesian synthesis built on, underneath the monetarist long-run curve.
So neither framework won wholesale. The synthesis that runs every modern central bank took the monetarist diagnosis and the Keynesian foundation and fused them into a single curve — monetarist in its long-run shape, Keynesian in its short-run slope. The most common mistakes are the two over-claims this comparison exists to refuse: “monetarism won” (it lost the policy) and “it was a draw” (the diagnosis decisively won the frame-level argument). The truth is more interesting than either, and you can only see it by holding the two frameworks against each other, because the win and the loss both belong to the same framework’s different parts.
This is the middle round of a longer argument. The great macroeconomic framework duel runs in three acts: Hayek versus Keynes on the 1930s (the Austrian liquidationist case against demand-management), Keynes-the-synthesis versus the monetarists on the 1970s (this walkthrough), and New Classical versus New Keynesian after 1980 (the synthesis arguing with itself over how much stabilization policy can do). Each round inherits the unfinished business of the last. The 1970s did not end the argument. They moved it to the next ground.