The man with the price gun
In the first months of 1994, one of the most conspicuous jobs in a Brazilian supermarket was walking the aisles with a pricing gun. Not stacking shelves — re-labelling them. People who lived through it describe the same scene: a shopper reaching for a tin of oil while, two metres behind them, a staff member marks up the row they've just passed.
Payday was a sprint. Wages went in one door and straight out the other, converted as fast as possible into anything that held its value — rice, cooking oil, batteries, toilet paper — because money left in a wallet overnight was money burnt. Brazilians had lived this way for so long that the banking system had industrialised it: overnight accounts that swept your salary into indexed instruments before dawn so your money could earn its own inflation back while you slept.
By the World Bank's numbers, Brazilian consumer prices rose 1,927% in 1993, and 2,076% in 1994. Four years earlier, in April 1990, the twelve-month rate had touched 6,821%. This was not a war or a bad harvest. It was the climate, and had been for most of two decades — an entire generation had reached adulthood without ever seeing a price stay still.
The government, meanwhile, had already tried nearly everything a government can try.
A graveyard of plans
Between 1986 and 1994 Brazil changed its currency five times. The cruzado arrived on 28 February 1986, the cruzado novo on 15 January 1989, a returning cruzeiro on 16 March 1990, the cruzeiro real on 1 August 1993 — that one simply lopped three zeros off the old notes — and finally the real on 1 July 1994. Five names in eight years: a child born in 1985 had lived through five national currencies before learning to read.
Each arrived wrapped in a stabilisation plan, and each plan is worth knowing for how confidently wrong it was.
The Cruzado Plan of February 1986, under President José Sarney and finance minister Dilson Funaro, froze prices, wages and the exchange rate outright, and then did something extraordinary: it deputised the public. Sarney went on television and asked ordinary Brazilians to police the freeze themselves, reporting any shopkeeper who raised a price. They did, in their hundreds of thousands. They were nicknamed the fiscais do Sarney — Sarney's inspectors. For a few months it was wildly popular. Then the shelves started emptying, beef became difficult to find, and the plan collapsed.
The Bresser Plan of 1987 and the Verão Plan of 1989 were variations on the same freeze, with the same ending.
Then came the plan that still makes people wince. On 16 March 1990 the newly inaugurated Fernando Collor de Mello froze roughly 80% of the country's private financial assets — savings, current accounts, investments — and held them for eighteen months. The money supply fell from about 30% of GDP to 9%. Monthly inflation duly dropped from 81% in March to 9% in June. Then the exemptions and loopholes did their work, industrial output fell off a cliff, and by January 1991 inflation was back above 20% a month. A second Collor plan followed in 1991. Collor himself was impeached in 1992.
Five plans. Five failures with the same shape: inflation falls for a few months, then returns, angrier, having cost the government whatever credibility it had left.
The disease was not what everyone thought
The reason freezes kept failing is the best idea in this story, and it came out of the economics department at PUC-Rio, a Catholic university in Rio de Janeiro.
Ordinary inflation is too much money chasing too few goods: you print, prices rise. Brazil had plenty of that. But on top of it sat something the Brazilian economists called inertial inflation — prices rising today for no better reason than that they rose yesterday.
Brazil had spent decades building this into law. Correção monetária — monetary correction — meant that rents, wages, taxes, savings, contracts and bond yields were all formally indexed to last period's inflation. Every price in the country was, by design, looking over its shoulder at every other price.
Now add the detail that makes it a trap. Prices did not all move on the same day. Each firm, union and landlord raised on its own schedule — monthly, quarterly, whenever the contract said. So the real value of any given price sawtoothed: too high the day after you raised it, sagging steadily until your next adjustment, at which point you jumped again. Nobody could afford to be the one who stopped, because stopping meant your average price sank while everybody else's kept climbing. It was a coordination trap — a bank run in slow motion, conducted on the price level.
Seen that way, the freeze is exactly the wrong instrument. A freeze catches every price at a random point on its own sawtooth. The firm that raised yesterday gets locked in at its peak; the firm due to raise tomorrow gets locked in at its trough. Relative prices are scrambled, quietly and unfairly, and everyone knows it. The moment the freeze lifts, everyone frozen low sprints to catch up, and the whole machine restarts with a vengeance. That is the story of 1986, 1987, 1989, 1990 and 1991 in one sentence.
An idea that sat on a shelf for ten years
In 1984, an economist named André Lara Resende published a paper proposing an indexed currency as a way to eliminate inertial inflation. He and his colleague Pérsio Arida developed it into a joint proposal, presented at an international conference organised by the economist John Williamson. Rudiger Dornbusch, of MIT, gave it the name it still carries: the Larida proposal — LARa resende plus arIDA.
The idea was almost perverse in its simplicity. Don't freeze the old money, and don't confiscate anything. Introduce a second unit alongside the existing currency — one corrected daily, so it never loses purchasing power. Let prices, wages and contracts migrate into it. Once effectively everything in the economy is quoted in the stable unit, the sawtooth is gone by definition: every price in the country is now moving at the same rate, on the same clock, at the same moment. Only then do you declare the stable unit to be the money, and retire the old one.
The elegance is in the ordering. The currency switch happens after prices have been realigned, not before — so there is nothing to catch anyone at a bad point in their cycle, nothing to evade, and nothing to be surprised by.
For ten years nobody tried it. It was too strange, and the politics were never right.
February 1994
After Collor's impeachment, his vice-president Itamar Franco took office and, in May 1993, appointed as finance minister a 61-year-old sociologist with no economics training named Fernando Henrique Cardoso. Cardoso's contribution was not the mechanism. It was assembling the PUC-Rio circle — Arida, Lara Resende, Edmar Bacha, Gustavo Franco, Pedro Malan, Francisco Lopes, Winston Fritsch — and then doing the two things none of his predecessors had done: putting the budget in order first, and explaining the whole plan out loud, in public, months in advance. No midnight decrees. No frozen accounts. Everything announced before it happened. He resigned the ministry on 30 March 1994, a month into the plan, to run for president on it.
On 27 February 1994, Provisional Measure 434 created the Unidade Real de Valor — the Real Value Unit, the URV. (It was converted into Law 8,880 that May.)
What the URV actually was: a number. Published daily by the central bank. It was never printed. There were no notes and no coins, no URV bank accounts, no way to pay anyone in it, no way to counterfeit it. In law it was a unit of account and nothing else — a ruler, not a thing.
Its value was pegged at roughly one US dollar, and every day the central bank announced how many cruzeiros reais one URV was worth. On 1 March 1994 one URV was worth CR$647.50. On 30 June it was worth CR$2,750.00. The cruzeiro real had lost more than three-quarters of its value against the URV in four months, which is simply another way of writing "inflation".
Four months of living in two currencies
Prices were to be quoted in URV. Rent, salaries, milk, bus fares, invoices. But you could not pay in URV, because URV did not exist in any physical form. You paid in cruzeiros reais, converted at that day's published rate.
So the shopkeeper stopped re-pricing. Milk cost 1 URV on 1 March and it cost 1 URV on 30 June. What changed was the number of cruzeiros reais that 1 URV turned into at the till — and that number was not his decision. It was arithmetic, published by someone else, applied identically to his suppliers, his landlord, his staff and his competitors.
Look at what that removes. He no longer has to forecast next month's inflation, or price defensively in case his costs outrun him, or gain by raising early and lose by raising late — because he isn't raising at all. The man with the price gun is out of a job. For the first time in twenty years, every price in Brazil was moving at exactly the same rate at exactly the same moment — which is another way of saying that, measured in URV, inflation was already over.
Wages got unusual care, because wages are where every previous plan had made its enemies. Salaries were not converted at whatever they happened to be worth on the day. They were converted at the arithmetic mean of their URV value over the preceding four months — for public servants, November and December 1993 and January and February 1994. The average across the sawtooth, in other words, rather than a random point on it. It is a small, technical, deeply unglamorous rule, and it is a large part of why the plan did not visibly rob one group to pay another.
The economists also went on television, repeatedly, and explained all of this. That was not decoration. A plan that depends on everyone voluntarily re-quoting their prices in a unit that doesn't exist only works if people understand why they should.
The switch that changed nothing
On 1 July 1994 the government simply declared the URV to be money. One real equalled one URV equalled CR$2,750. Trucks of new banknotes had already been distributed around the country. The cruzeiro real was retired.
And on that day, no price changed. A thing that cost 3 URV on 30 June cost 3 reais on 1 July. There was no re-labelling, no re-pricing, no rush, no shortage, no queue — because the country had spent four months quietly re-labelling itself, one contract at a time, in daylight, with plenty of warning.
Monthly inflation fell from 48% in June to 7.8% in July to 1.9% in August. Brazil has never gone back.
The politics were seismic. In an April 1994 poll, Lula led the presidential race 40% to Cardoso's 12%. On 3 October 1994 — three months after the real launched — Cardoso won outright in the first round with 54.3% of the vote, and went on to serve two terms. Few economic policies have ever been rewarded quite that precisely.
So did the URV actually do it?
This is where honest people still disagree, and the disagreement is the best part.
The sceptical case is that the URV was elegant choreography while something less charming did the real work: a genuine fiscal adjustment before launch, and a hard exchange-rate anchor afterwards, defended with brutal interest rates. Anchor to the dollar and squeeze credit hard enough and inflation comes down whether or not you performed a clever ritual first.
The bill for that anchor arrived on schedule. Between mid-1994 and January 1999 the real appreciated by roughly 25% in real terms, hollowing out exporters. Real interest rates frequently ran above 20% and spiked to around 45% in early 1999. The current account went from near balance in 1994 to a deficit of 4.5% of GDP — about US$33 billion — in 1998. Growth averaged under 2.5% a year from 1995 to 1998. In January 1999 Brazil gave up and let the currency float; it fell hard.
And here is the detail that cuts the other way, hard: inflation did not come back. A devaluation of that size in 1990 would have re-ignited the entire system within weeks, because every contract in the country was wired to pass it straight through. In 1999 it didn't. Whatever else the URV phase did, it seems genuinely to have unwired the indexation — and that unwiring survived the loss of the anchor that was supposedly doing all the work.
A third camp argues the monetary mechanics are downstream of something bigger. John Cochrane reads Brazil through the fiscal theory of the price level: prices settle where the real value of government debt matches what the state can credibly raise later, so disinflation arrives on a believable change of regime — often in anticipation of the policies, before they land. On this reading, what changed in 1994 was that Brazilians finally believed the government meant it.
What nobody disputes is the control group. Five previous plans had good economists, harsher instruments and the full coercive power of the state, and every one failed. This one, which coerced almost nobody, didn't. De-indexing an entire economy voluntarily, in public, over four months, using a currency that never existed, had not been done before and has not really been repeated since.