Mainstream macroeconomics has held three broad consensus positions in the past century. The classical view ruled until the 1930s. The Keynesian view ruled from the 1940s to the 1970s. The New Keynesian synthesis has ruled since the 1990s. Two crises did the damage in between. The Depression and the stagflation of the 1970s each produced facts the reigning consensus said could not happen. This post follows that history.
By mainstream I mean the framework taught in the standard textbooks and used by central banks and treasuries to make policy. The subject is stabilisation: what sets output, jobs and inflation over the business cycle, and what policy should do about it. Growth theory and the rest of the field appear only where that story needs them. This is necessarily a simplified history. Each consensus contained disagreements, exceptions and ideas that did not fit neatly into the prevailing model. I have set those aside to follow the central propositions that became accepted wisdom, the evidence that challenged them, and what replaced them. Economists contest almost everything, so disputes inside the mainstream appear only where they changed the consensus. Schools outside of this evolving mainstream are for the most part left out. They exist, and the field is better for people who explore and test alternatives. But the consensus is what policymakers act on, and its history explains where we are now.
The classical view before 1930
Before the 1930s there was no separate field called macroeconomics. There was economics, and the economy as a whole was thought to behave much like any single market. Prices adjust until supply meets demand. The work of Marshall and Pigou at Cambridge was the high point of the tradition, and four ideas carried it.
The first was Say's Law. Production creates income of equal value. A firm that makes and sells goods pays wages, rent, interest and profit, and those incomes buy other goods. One product can be in glut. A bumper harvest leaves too much barley, and its price falls. But buyers who pay less for barley have more to spend on other things, so demand lost in one market turns up in others. While short-term gluts of this kind can occur, a general shortage of demand across the whole economy cannot last. And if total demand cannot fall short, there is no shortage for a government to fill.
The second was a labour market that clears. The wage is a price. If workers are idle, the wage is too high, and competition for jobs will bid it down until everyone who wants work at the going wage has it. Lasting unemployment therefore had a cause outside the market: unions, minimum wages or workers holding out for more. The cure was lower wages.
The third was the interest rate as the price that balances saving and investment, in what the textbooks call the market for loanable funds. Income not spent on consumption is saved. Savings are lent. If households save more, the interest rate falls and firms borrow more to invest. Thrift therefore never drains demand. It only shifts spending from household consumption to business investment.
The fourth was neutral money, which rested on the quantity theory of money. The theory starts from an identity. The stock of money, multiplied by the number of times each dollar changes hands in a year, equals the value of everything sold in that year. In symbols, MV = PY, where M is the money stock, V is its velocity, P is the price level and Y is output. The classical economists added two assumptions. Velocity is stable, because it depends on slow-moving habits of payment. Output is fixed at full employment by the other three ideas. That leaves the price level as the only thing that can respond to a change in money.
Double the money supply and prices double, while output, employment and relative prices stay where they were. Real things and monetary things lived in separate boxes. This is the classical dichotomy.
The classical economists drew neither a production frontier nor supply and demand for the whole economy. Those charts came later, but both sum up the view well. The economy sits on its production frontier, with every willing worker and every machine in use. The questions for economics were about where on the frontier to be: which goods to make, and how much to consume now against investing for later. A point inside the frontier, with resources lying idle, was a passing disturbance and had no place in the theory. On the later supply and demand chart this is a vertical supply curve at full employment. Demand, set by the quantity of money, slides up and down that line and fixes only the price level.
The policy advice followed from the four ideas. The gold standard tied each currency to gold, which fixed exchange rates and stopped governments printing their way out of trouble. With the price of each currency fixed, flows of gold did the adjusting. A country that lost gold had to deflate until it was competitive again. Budgets were to balance every year. The British Treasury argued in 1929 that public works could not add jobs, because every pound the government borrowed was a pound taken from private investment.
The classical economists knew the trade cycle well. They treated slumps as temporary departures from full employment, caused by a shock or a credit panic and ended by falling wages and prices. The task of policy was to stay out of the way while the adjustment ran.
The Great Depression breaks it
The Depression was too deep and too long to fit that account. United States output fell by roughly a quarter between 1929 and 1933, and industrial production nearly halved. Unemployment reached about 25 per cent. Prices fell by about a quarter. In Australia about one worker in five was out of a job in 1932, and close to three in ten among trade union members, which is the figure most often quoted. A decade after the crash, American unemployment was still far above its 1929 level.
The classical remedy was tried in full. Wages and prices fell a long way. Australia is a clean example. The Arbitration Court cut the basic wage by 10 per cent in 1931, and under the Premiers' Plan of the same year governments agreed to cut spending by a fifth. Unemployment kept rising. Falling wages cut the incomes that workers spent. Falling prices raised the real burden of every debt, so farmers and firms that owed fixed sums went under.
Balanced budgets made matters worse. Tax revenue collapsed with incomes, and governments answered by cutting spending and raising taxes. Each cut removed more demand from an economy that was already short of it.
The financial system failed alongside. About 9,000 American banks closed between 1930 and 1933, and the money supply fell by about a third. The Federal Reserve failed to stop it. Depositors lost their savings, surviving banks stopped lending, and credit dried up for sound borrowers as well as weak ones.
The gold standard spread the slump and locked it in. Countries had to hold interest rates up to defend their gold, even as their economies shrank. Recovery came in the order countries let go. Britain left gold in September 1931 and Australia had already devalued. The United States followed in 1933 and France held on until 1936. The earlier a country left, the earlier it recovered.
A theory that says mass unemployment cannot persist had no answer to a decade of it. Nor could it explain why its own medicine had made the patient worse. The profession needed an account of how an economy can stall well short of full employment and stay there.
Keynes rewrites the short run
Keynes supplied that account in The General Theory of Employment, Interest and Money in 1936. His starting point reversed Say's Law. Total spending determines output and employment. Firms produce what they expect to sell and hire the workers they need to do it. If spending falls, output and jobs fall with it, and no law guarantees that spending will be enough to employ everyone.
Investment is the unstable part of spending. Firms invest on a guess about profits years ahead, and the guess rests on confidence as much as calculation. Keynes called this animal spirits. When confidence breaks, investment collapses, and a lower interest rate may not revive it. No firm builds a factory to make goods it cannot sell.
The interest rate does not reliably balance saving and investment either. Keynes took that job away from loanable funds and gave it to the market for money, where the demand to hold money meets a fixed supply. People hold money for safety when the future looks dangerous, and this liquidity preference can hold the interest rate above the level that full employment needs. In a deep slump extra money is simply hoarded and the rate stops falling. This is the liquidity trap.
Saving therefore can drain demand. If every household saves more at once, spending falls, incomes fall, and total saving may end up no higher. What is prudent for one household is destructive when all households do it. This is the paradox of thrift.
Wage cuts fail for the same reason. A single firm that cuts wages gains an edge on its rivals. When every firm cuts wages, workers' spending falls by as much as firms' costs, and prices follow wages down. The real wage barely moves and debts grow heavier. Wages are also slow to fall in practice, because workers resist cuts in money pay.
Put together, these ideas produce an economy that can come to rest with high unemployment. Firms see no demand, so they do not hire. Households have no jobs, so they do not spend. Nothing inside the system pushes it back to full employment. The economy can sit inside its production frontier for years. This was the central break with the classical view.
On the later supply and demand chart this gives the Keynesian supply curve its shape. Below full employment the curve is horizontal. Idle workers and idle machines can be put to work at the going wages and prices, so extra demand raises output and leaves prices alone. At full employment the curve turns vertical. Nothing is left to bring into use, so extra demand raises prices only, as the classical economists said. Their case became the special case of an economy already at full employment, which is why Keynes called his theory general.
The remedy was for government to fill the gap in spending. Public works or tax cuts add demand directly. The multiplier, first set out by Keynes's colleague Kahn in 1931, spreads the effect: the builders spend their wages, the shopkeepers spend their takings, and total income rises by more than the first outlay. Budgets should balance over the cycle, in deficit in the slump and in surplus in the boom.
The war installs the new view
The Second World War followed the Depression without a break, and it did as much to install the new view as the Depression had done to discredit the old one. Governments spent without limit, and unemployment vanished. The American rate fell from about 15 per cent in 1940 to under 2 per cent in 1944. It was not a clean test, because conscription, rationing and direct control of production did much of the work. But the lesson drawn was simple. The slump that a decade of wage cuts and sound budgets could not end was over within three years of deficit spending.
The war also built the machinery. Keynes applied his framework to war finance in 1940, and the British budget of 1941 was the first drawn up from national income estimates. Economists moved into treasuries and stayed. Governments came out of the war far larger than they went in, with broad income taxes to match. In Australia the Commonwealth took over income tax from the states in 1942. And people who had lived through both crises expected their governments to prevent a return to the 1930s.
The postwar Keynesian consensus
The profession quickly made Keynes tractable. Hicks compressed the book into a single diagram in 1937. The IS curve shows the combinations of income and interest rates at which saving equals investment. The classical theory had the same condition but held income fixed at full employment, so the interest rate did all the adjusting. Keynes let income adjust too, and the classical answer became one point on the curve. The LM curve shows the combinations at which the money market balances. Where the two cross gives output and the interest rate, and nothing guarantees that the crossing sits at full employment. For 40 years students learned macroeconomics by shifting those two curves.
The LM curve kept the quantity theory's money market but changed how money demand worked. It still assumed a stock of money fixed by the central bank, meeting a stable demand for money. The change was that the demand for money depends on the interest rate as well as on income, so velocity moves when the interest rate does. The quantity theory became the special case in which money demand ignores the interest rate and the LM curve stands vertical. Every model built on IS-LM, including the open-economy version that followed, inherited this base: a money stock the central bank controls and a demand for money that stays put.
Samuelson built the neoclassical synthesis on top of it after the war. Keynes governs the short run, when prices and wages are sticky and demand sets output. The classical model governs the long run, when prices have adjusted and supply sets output. The classical model was right about a fully employed economy. The job of policy was to keep the economy there.
Solow supplied the model for the long run in 1956, as did the Australian economist Swan in the same year. In their growth model output depends on capital, labour and technology. Capital runs into diminishing returns: each extra machine adds less than the one before. A country that saves and invests more becomes richer, but the boost to growth fades as the returns shrink. Sustained growth in living standards has to come from better technology. When Solow measured it in 1957, capital explained only about an eighth of the growth in American output per hour worked over the period 1909 to 1949. Technical progress explained the rest.
The business cycle was an old idea, and economists had been charting booms and slumps since the nineteenth century. What the growth model added was a theory of the path the cycle moves around. Supply sets that trend: population, capital and technology fix what the economy can produce, and growth pushes that frontier out. The loanable funds market found its new home here, as the account of saving, investment and the interest rate in the long run. Demand sets the cycle around the trend. Every later regime kept that split. The trend is potential output, and the cycle is the gap between actual output and potential.
The Phillips curve supplied the missing account of inflation. The Keynesian supply curve had a sharp corner: no inflation short of full employment, and nothing but inflation beyond it. The Phillips curve rounded the corner. Phillips, a New Zealander at the London School of Economics, found in 1958 that British wages had risen faster when unemployment was low, across almost a century of data. Samuelson and Solow turned the finding into a policy menu in 1960. A government could choose low unemployment with some inflation, or stable prices with more unemployment.
Measurement and modelling grew up alongside the theory. Kuznets in the United States and Stone and Meade in Britain built the national accounts, which gave governments their first regular measures of income, spending and output. Tinbergen and Klein built econometric models with hundreds of equations, fitted to those accounts. Treasuries used them to forecast the economy and to calculate how much stimulus or restraint to apply.
Governments wrote the framework into their commitments. Britain's 1944 White Paper accepted responsibility for a high and stable level of employment. Australia's 1945 White Paper committed the Commonwealth to full employment. The United States passed its Employment Act in 1946. Managing demand had become a core duty of government.
The international system was rebuilt to make room for it. At Bretton Woods in 1944, with Keynes leading the British delegation, the Allies replaced the gold standard with a looser arrangement. Other currencies were pegged to the US dollar, and the dollar was pegged to gold at 35 dollars an ounce. A country in lasting trouble could change its peg. Controls on capital flows let each government set its own interest rates, and the new International Monetary Fund lent to countries short of reserves. The aim was stable exchange rates without the gold standard's rule that a country in deficit must deflate. Governments were to be free to pursue full employment at home. Mundell and Fleming set out the logic in the early 1960s by adding the balance of payments to IS-LM. A country cannot have a fixed exchange rate, free capital flows and its own monetary policy all at once. Bretton Woods gave up free capital flows.
For 25 years it seemed to work. Growth across the rich world was the fastest on record, and productivity rose year after year. Australian unemployment stayed below 3 per cent for most of the period. Inflation was low apart from the occasional breakout, such as the Korean War wool boom of 1951, when Australian prices rose by more than 20 per cent in a year. Recessions were short and mild. Real wages rose, and home and car ownership rose with them.
The boom entrenched a set of expectations. A generation that had known depression and war came to treat full employment and rising prosperity as the normal state of affairs, and to hold governments responsible for delivering both. Economists shared the confidence. By the mid-1960s they spoke of fine-tuning, and the Kennedy tax cut of 1964 was taken as proof that demand management could hold the economy at full employment. The business cycle looked like a solved problem.
Stagflation breaks it again
The weak point was the Phillips menu. The Keynesian model said inflation rises when demand is strong and unemployment is low. It had no room for inflation and unemployment rising together.
Friedman and Phelps attacked the menu in 1967 and 1968, before the facts turned. Workers bargain for real wages and firms set real prices, they argued. Push unemployment down with stimulus and inflation rises. At first workers are fooled, because their pay buys less than they expected. Then they learn. They expect the inflation and build it into their wage claims, and the employment gain disappears.
The argument put two ideas into the core of the subject. One was the natural rate of unemployment, the rate set by the structure of the labour market and beyond the reach of demand policy. The other was that the Phillips curve shifts with expected inflation. The single curve became a family of short-run curves, one for each level of expected inflation, around a vertical line at the natural rate. A government that holds unemployment below the natural rate gets inflation that keeps accelerating. In the long run there is no trade to make.
The pressure had been building through the 1960s. The United States paid for the Vietnam War and new social programs without raising taxes to match, and the Federal Reserve kept interest rates low. American inflation rose from under 2 per cent in the early 1960s to about 6 per cent by 1970. Fixed exchange rates carried it to the rest of the world.
Bretton Woods then collapsed. American inflation and deficits left more dollars in foreign hands than the United States had gold to redeem. Nixon ended the dollar's link to gold in August 1971, and by 1973 the major currencies were floating. The exchange rate became a market price, moved by trade and by differences in interest rates. The last external check on money creation was gone, and each country now owned its inflation rate.
The oil shocks landed on economies already running hot. The price of oil roughly quadrupled in 1973 and 1974 and more than doubled again in 1979. The Keynesian model dealt in demand, and here was a shock to supply. Dearer oil raised costs and prices and cut output at the same time. Stimulus could fight the lost output or restraint could fight the inflation. Policy could not do both.
The textbooks redrew the supply and demand chart to fit. Aggregate supply became two curves. The short-run curve slopes up and shifts with costs and expected prices, so an oil shock moves it and brings higher prices with lower output. The long-run curve is vertical at potential output, which is the classical curve restored. Demand can move the economy along the short-run curve for a while. It cannot move the long-run curve.
The results broke the menu. United States inflation went above 12 per cent in 1974 and peaked near 15 per cent in 1980, while unemployment hit 9 per cent in 1975. Australian inflation went above 17 per cent in 1975. British inflation reached around 25 per cent. Unemployment rose everywhere, and it stayed high after each burst of stimulus faded.
Governments reached for the remaining Keynesian tool, direct control of wages and prices. Nixon froze both in 1971. Britain tried a run of incomes policies. Each held for a time and then gave way to a burst of catch-up claims. In 1976 the British Labour prime minister, Callaghan, told his own party conference that a country could no longer spend its way out of a recession.
The method fell along with the menu. Lucas argued in 1976 that the big Keynesian models were built on relationships that shift when policy shifts. The equations summarised how people had behaved under past policies. Change the policy and people change their behaviour, so the equations stop holding. The Phillips curve was the plain example: it held until governments tried to exploit it. A model fit to past data could not predict the result of a new policy.
The contest for a replacement
No single replacement was ready. Three candidates held the field for about 15 years, and the eventual consensus took something from each.
Monetarism came first. Friedman and Schwartz had argued in 1963 that the Depression was a monetary failure: the Federal Reserve let the money supply fall by a third and turned a recession into a catastrophe. Friedman held that money drives spending, that inflation is always a monetary phenomenon, and that discretionary policy does more harm than good because it works with long and variable lags. His rule was steady growth in the money supply, year in and year out. On Keynes's money market chart, the central bank was to fix the supply line and let the interest rate go where it would.
Central banks adopted money targets in the 1970s, Australia among them in 1976. The decisive trial came in the United States. Volcker's Federal Reserve tightened hard from 1979 and let interest rates go where they had to. The federal funds rate neared 20 per cent in 1981. Inflation fell from near 15 per cent to under 4 per cent by 1983. Unemployment peaked near 11 per cent in the deepest recession since the 1930s.
The episode settled two things. A central bank can end inflation if it is willing to bear the cost, so inflation is the central bank's responsibility. But money targets do not work. Financial deregulation changed how people held money, and the link between the money aggregates and spending came apart. Targets were missed year after year, and central banks dropped them through the 1980s. Australia dropped its target in 1985.
The new classical school came second. Lucas, Sargent and others rebuilt macroeconomics on two assumptions. Markets clear. And people hold rational expectations: they use all available information and do not make the same forecasting mistake twice. With both in place, only surprises move output. Sargent and Wallace showed in 1975 that any predictable monetary policy would be anticipated and would change prices alone. The result needs both assumptions. The New Keynesians would keep rational expectations and drop market clearing.
Kydland and Prescott added a second result in 1977. A government with discretion will always be tempted to spring a little surprise inflation for a short burst of jobs. The public knows this and expects the inflation in advance. The economy ends up with higher inflation and no extra jobs. The way out is to bind policy to a rule or hand it to someone who will not be tempted. This argument later became the case for independent central banks.
Real business cycle theory took the school to its limit. Kydland and Prescott's 1982 model started from the Solow growth model and added random shocks to technology. Booms and slumps were the efficient response of households and firms to those shocks. Recessions were periods when productivity was poor and people chose to work less. Money played no part and there was nothing for policy to fix.
The school won the argument on method. After Lucas, a serious model had to be built from households and firms that optimise and look forward, with expectations consistent with the model itself. It lost the argument on substance. The Volcker recession showed that monetary policy moves output and jobs for years at a time, and few could believe that 11 per cent unemployment was a voluntary holiday. A model with no room for that could not be the consensus.
The New Keynesians came third. They accepted the new method and used it to rebuild the Keynesian result. Their problem was to explain why prices and wages are sticky when everyone is rational. Without stickiness, money has no real effects and demand does not matter.
They found several answers. Firms with some market power set their own prices, and changing a price has a cost. Mankiw and others showed in 1985 that even a small cost can make it rational to leave prices alone, while the combined effect across the economy is large. Fischer and Taylor showed that wage contracts set at different dates and running for a year or more make the wage level slow to move. Calvo gave the field a simple device in 1983: each period only a fraction of firms get to reset their price. With sticky prices, a fall in demand cuts output, and monetary policy moves output and jobs even when everyone sees it coming. Rational expectations survived. The claim that anticipated policy is powerless did not. Demand management was back, built on new classical foundations.
The New Keynesian synthesis
The schools merged in the 1990s. The method came from the new classicals: dynamic general equilibrium models with optimising, forward-looking agents and rational expectations. The frictions came from the New Keynesians. The long run came from the classics and the monetarists: money is neutral, supply sets output, and unemployment returns to its natural rate. Goodfriend and King named it the New Neoclassical Synthesis in 1997. Clarida, Galí and Gertler in 1999 and Woodford in 2003 gave it its standard form.
The core fits in three equations. The first is a forward-looking IS curve. Households spread their spending over time, so spending today depends on the real interest rate and on expected future income. A higher real rate makes saving more attractive and cuts spending now. What matters is the gap between the real rate and the neutral rate. A real rate above neutral holds demand below potential, and a real rate below neutral pushes demand above it.
The IS channel is hard to detect in the data. Households respond only modestly to interest rates, and a central bank that offsets swings in demand leaves little trace of the relationship behind. But it is one of many channels. Policy also moves demand through the exchange rate, asset prices, and the cost and supply of credit. In a small open economy with a floating currency, such as Australia since 1983, the exchange rate does much of the work. In the model, the IS curve stands in for all of them.
The second is the New Keynesian Phillips curve. Firms that reset prices only occasionally must look ahead, so inflation today depends on expected inflation and on the output gap, the distance between actual output and potential output. It is the Friedman and Phelps curve rebuilt from price-setting firms. When output runs above potential, inflation rises. Expected inflation passes straight into actual inflation. Working versions add a lag of past inflation.
The third is a policy rule. Taylor showed in 1993 that the Federal Reserve's behaviour fitted a simple formula: set the interest rate above neutral when inflation is above target or output is above potential, and below neutral in the opposite case. The rule carries one firm requirement. When inflation rises, the central bank must raise the nominal rate by more than the rise in inflation, so that the real rate goes up. The 1970s central banks had failed that test.
The rule replaced the LM curve. That curve assumed the central bank fixes the quantity of money and lets the interest rate fall out. Central banks do the reverse. They set the rate and supply whatever money is demanded at it, which turns the supply of money, and with it the LM curve, into a flat line at the policy rate. The failure of money targets had shown why. Money demand shifted about after financial deregulation, and Poole had shown in 1970 that when money demand is the unstable part, steering by the interest rate gives a steadier economy than steering by money. Romer showed in 2000 that the textbook model runs on an IS curve and a policy rule alone. Woodford's model has no money in it at all.
Together the three equations redraw the supply and demand chart with inflation on the vertical axis. The IS curve and the policy rule make the demand curve: higher inflation brings a higher real interest rate and lower spending. The Phillips curve is the short-run supply curve. The long-run supply curve stays vertical at potential output.
The model runs on a set of benchmarks that cannot be observed directly. There is the natural rate of unemployment, which policymakers estimate as the NAIRU (u*), the rate at which inflation holds steady. There is potential output (Y*), the level the economy can sustain without pushing inflation up. It sits a little inside the production frontier, which is why output can run above potential for a time. And there is the neutral interest rate (r*), which Wicksell named the natural rate in 1898. It is the modern descendant of the classical interest rate: the rate at which saving and investment balance when output is at potential. The classical economists trusted the loan market to find it. The synthesis gives that job to the central bank, which must estimate the rate and steer towards it. Arguments about why the rate moves, such as Bernanke's global saving glut, are loanable funds arguments on a world scale. Policy is tight or loose only by comparison with these benchmarks, and estimating them is a large part of a central bank's work.
Expectations sit at the centre of all three equations, and that changed what central banks do. If people expect inflation to stay at target, a shock to prices fades quickly and unemployment need not rise much to contain it. If expectations drift, inflation feeds on itself and the cost of stopping it is a recession. Managing expectations became the first task of policy. Credibility, clear targets and open communication became tools in their own right.
The institutions followed. New Zealand adopted an inflation target in 1990, Canada in 1991, Britain in 1992 and Australia in 1993. Governments gave central banks independence over the interest rate so that the promise of low inflation would be believed. The instrument is the short-term interest rate, and the quantity of money is left to adjust. Central banks built their forecasting models on the same core.
Fiscal policy took the smaller part. The automatic stabilisers do their work: tax revenue falls and welfare payments rise in a downturn without any decision. Discretionary stimulus was judged too slow and too political for routine use. The task of budgets was to keep debt sustainable over the medium term and leave the cycle to the central bank.
The record supported the consensus for two decades. From the mid-1980s to 2007 inflation was low and stable across the rich world, and swings in output were about half the size of earlier decades. Economists called it the Great Moderation.
Two tests of the synthesis
The global financial crisis of 2008 tested the synthesis in its turn. The standard model had almost no finance in it, so it did not see the crisis coming. The crisis also ended the belief that low inflation keeps the financial system safe: credit and leverage had boomed while inflation sat on target. That was the lesson Minsky and Keen had pressed from outside the mainstream, that stability breeds risk-taking and private debt drives the cycle. Banks, credit and financial frictions went into the models, and regulators were given separate tools to contain lending.
Interest rates then hit zero and could go no lower. This was Keynes's liquidity trap, back after seven decades. Central banks bought assets and gave guidance about future rates to work on longer-term interest rates. The division of labour that left the cycle to the central bank no longer held, and fiscal stimulus returned as the accepted tool when rates are at the floor. It was a Keynesian remedy applied inside the New Keynesian model, which predicts that fiscal policy is at its most powerful when the central bank cannot respond.
The pandemic brought the second test. Lockdowns and broken supply chains cut what economies could produce, while governments and central banks supported spending on a scale not seen since the war. Inflation followed in 2021 and 2022, reaching about 9 per cent in the United States and nearly 8 per cent in Australia. Central banks first judged it a passing supply shock and waited. Then they raised interest rates at the fastest pace in 40 years, and inflation fell back without the deep recessions of the early 1980s. The episode was argued out inside the framework. The questions were how much of the inflation was supply and how much demand, whether expectations would hold, and whether the Phillips curve steepens when the economy runs into its limits. No rival framework displaced the synthesis.
What each consensus left behind
Both breaks followed the same course. A consensus fell when events arrived that it said could not happen. The classical view said mass unemployment could not last, and it lasted a decade. The Keynesian view said inflation and unemployment could not rise together, and they did. Argument alone was never enough. Keynes and others attacked the classical view through the 1920s. Friedman and Phelps set out the flaw in the Phillips menu in 1967 and 1968. The consensus moved only when the facts did.
The replacement was slow to arrive each time. The Depression began in 1929, the General Theory appeared in 1936, and the Keynesian consensus was not settled until after the war. Stagflation broke the Keynesian consensus in the mid-1970s, and the New Keynesian synthesis was not settled until the late 1990s. In both cases a generation passed between the failure and the new agreement, and rival schools fought over the ground in between.
Each replacement kept what the old consensus could explain. The classical long run survived the Depression: supply sets output over time, money is neutral, and saving and investment set the neutral interest rate. Keynes's short run survived the 1970s, and so did his larger claim that the economy does not reliably correct itself and has to be managed.
What stagflation changed was the manager, the goal and the anchor for prices. Under the gold standard and then Bretton Woods, the fixed exchange rate was the anchor. It tied each country's prices to the rest of the world. Bretton Woods used capital controls to give governments room to manage demand at home, and the treasury took the lead. When the system collapsed, governments gave the exchange rate up to the market, and the anchor went with it. Central banks eventually supplied a new one by setting the overnight interest rate against an inflation target, and the management of demand came with the instrument. The aim narrowed too. Governments had set their own full employment targets and pushed unemployment as low as they judged possible. Now the task was to hold unemployment near its natural rate, which policy can estimate but cannot choose, while keeping inflation stable.
The model central banks use today is built from those layers. The long run is classical. The short run is Keynesian. The natural rate and the central bank that owns inflation come from Friedman. The method and the central place of expectations come from Lucas. The sticky prices that make it all hold together come from the New Keynesians. The third consensus has now lasted about as long as the second did. It was extended after 2008, but the events of 2008 and 2020 did not substantially change the consensus.