The Real Plan: How Brazil Ended Hyperinflation Without a Freeze

On the morning of Friday 1 July 1994, Brazilian bank branches opened with two sets of banknotes behind the counter. Customers handed over cruzeiros reais and received reais at a rate of 2,750 to one. There was no queue of police at the door, no order closing the exchanges, no decree freezing supermarket shelves. Cabinet ministers had spent the previous week telling anyone who asked that nothing would be confiscated and no price would be fixed by law, and for once in Brazilian monetary history that turned out to be true.

Four months earlier, in February 1994, consumer prices had risen about 40 per cent in a single month. Over the whole of 1993 the IPCA, the index the Brazilian statistics office uses to measure consumer inflation, had risen 2,477.15 per cent. A Brazilian who kept a month's salary in a current account lost roughly a third of it by payday. By September 1994 the monthly figure was under 2 per cent, and it stayed there.

Eight Currencies in Fifty Years

Brazil did not arrive at 1994 for want of trying. Between 1986 and 1991 five governments launched five stabilisation programmes, each built on some combination of price freezes, wage rules, and the deletion of zeros from the currency.

Fernando Collor de Mello had promised in March 1990 that he would kill inflation "with a single shot." What his plan actually did was seize the savings of the middle class — cruzado novo balances above a threshold were blocked in the central bank for a year and a half — and buy about six months of low readings before prices resumed their climb. Collor was impeached in 1992 on unrelated corruption charges. His vice-president, Itamar Franco, inherited an economy running at 30 per cent a month and a public that had learned to read every new plan as a mugging.

Each freeze failed for the same structural reason. Brazilian inflation was overwhelmingly inertial: prices rose because contracts, wages, rents and tariffs were indexed to yesterday's prices, and every agent set today's price by adding an expected increment to what the neighbours had just done. Freeze the level and you suppress the symptom while leaving relative prices badly misaligned, so that the day controls lift, everyone raises at once. Pérsio Arida and André Lara Resende had written this argument out in 1984, in a paper circulated in Rio and published the following year in a volume edited by John Williamson, and their proposed remedy was strange enough that nobody in office would touch it for a decade.

Larida

Their idea, which Brazilian economists came to call the Larida plan after its two authors, was to stop trying to kill the old money and instead let it die of exhaustion. A government would introduce a second unit of account alongside the existing currency. Prices could be quoted in the new unit, which would be stable by construction, while payment continued in the old one, which would go on depreciating. As contracts migrated voluntarily to the new unit, the relative-price distortions that wrecked every freeze would unwind in the open rather than being bottled up. When enough of the economy had moved across, the government would declare the unit of account to be legal tender and retire the old notes.

No freeze, no confiscation, no conversion table imposed by decree. Arida and Lara Resende (1985) understood that the scheme required something Brazil had never managed, which was a credible commitment that the new unit would not itself be inflated once it became money.

Itamar Franco went through three finance ministers in eight months before turning, on 19 May 1993, to the senator he had made foreign minister the previous year. Fernando Henrique Cardoso was a sociologist by training, a dependency theorist whose academic work had been sceptical of orthodox stabilisation, and he had no economics degree at all. He accepted on condition that he could pick his own team and that the president would not overrule it in public. Around him assembled Edmar Bacha, Pérsio Arida, André Lara Resende, Gustavo Franco, Winston Fritsch, Francisco Lopes and Pedro Malan, most of them from the economics department of the Catholic University in Rio.

Phase One: The Budget

Cardoso's team refused to launch a monetary reform before the fiscal accounts could support one, on the reasoning that every previous plan had failed because the government kept financing itself by printing. In June 1993 they published the Programa de Ação Imediata, an austerity package aimed at cutting some six billion dollars of federal spending, forcing states to settle arrears with the union, and tightening control of the federal banks.

More important was a constitutional device. Brazil's 1988 constitution earmarked most federal revenue to specific transfers and programmes, leaving the Treasury with almost no discretionary room. In March 1994 Congress approved the Fundo Social de Emergência, a revision amendment that de-earmarked 15 per cent of federal tax revenue for two years and handed it back to the executive. It was a temporary fix to a permanent problem, renewed repeatedly under different names, and it gave the plan a roughly balanced operational budget at the moment it needed one.

Cardoso also did something no Brazilian finance minister had done since 1986: he told the country exactly what he was going to do, months before he did it, and then did it in that order. Announcing the steps in advance removed the incentive to front-run a surprise freeze.

Phase Two: A Currency That Was Not Money

Provisional Measure 434, issued on 27 February 1994, created the Unidade Real de Valor. From 1 March the URV existed as an index, initially set at CR$647.50, restated every day by the central bank so that its value in cruzeiros reais tracked the ongoing inflation — and, in practice, tracked the dollar at close to one to one.

Wages were converted into URV at their average real purchasing power over the four months from November 1993 to February 1994, a rule that spread the arbitrariness of any single conversion date across a period and denied unions and employers alike a specific grievance. Rents, school fees, insurance premiums and public tariffs followed. Shops posted two prices, one in cruzeiros reais that changed daily and one in URV that did not.

For four months Brazilians lived inside a working demonstration. The URV price of a litre of milk barely moved while its cruzeiro price climbed every week, and the lesson taught itself without a single inspector. Cardoso resigned from the ministry on 30 March 1994 to run for president, leaving the plan's final phase to Rubens Ricupero.

Phase Three: The Real

On 30 June the government issued the measure that turned the unit of account into legal tender. The next morning one URV became one real, which meant CR$2,750 to the real and, at the prevailing rate, roughly one dollar.

What held the new currency up was not a freeze but two anchors that cost real money. The central bank announced a ceiling of one real to the dollar and let the rate float freely below it, which in a country running a trade surplus and attracting portfolio inflows meant the real promptly appreciated: by early 1995 it was trading near R$0.83 to the dollar. An overvalued currency and a simultaneous cut in import tariffs exposed Brazilian producers to foreign competition at precisely the moment they might have tested the new money with price increases. The second anchor was interest. Real overnight rates ran above 20 per cent for much of the plan's first two years, and reserve requirements on demand deposits were raised to 100 per cent on the margin.

Gustavo Franco, who became the plan's most articulate defender and later governor of the central bank, treated the exchange-rate anchor and trade opening as a single instrument for forcing productivity onto an economy that had spent three decades hiding behind indexation (Franco, 1995). Edmar Bacha, assessing the first year from inside, argued that the URV's real contribution had been informational — it let relative prices find each other before the reform rather than after (Bacha, 1995).

Ricupero did not survive to see it. On 1 September 1994 a satellite feed carried his off-air conversation with a television anchor, in which he said: "I have no scruples. What is good, we take credit for; what is bad, we hide." He resigned five days later. The plan did not wobble.

The Election

Cardoso had been polling around 20 per cent in June 1994, well behind Luiz Inácio Lula da Silva. Prices stopped rising in July. On 3 October he won the presidency outright in the first round with 54.3 per cent of the vote against Lula's 27 per cent.

The distributional arithmetic explains the swing. Inflation is a tax collected most efficiently from people who cannot escape it, and in Brazil that meant wage earners paid monthly in cash, without the overnight accounts that let the middle class index their balances daily. When monthly inflation fell from 40 per cent to 2 per cent, the real purchasing power of the bottom of the wage distribution rose immediately. Supermarket and consumer-durable sales jumped through the second half of 1994; chicken and milk consumption rose measurably. Household surveys through 1995 recorded the sharpest fall in the poverty headcount Brazil had seen in a generation.

What the Anchor Cost

Stability priced in dollars was expensive to maintain. Brazil's current account moved from near balance in 1994 to deficits approaching four per cent of GDP by 1998, financed by short-term capital that could leave. Each external shock forced the central bank to defend the rate with interest rates that crushed domestic activity: after Mexico devalued in December 1994 — see the peso crisis and the Tequila effect — Brazil abandoned the simple ceiling in March 1995 for a crawling band. After the Asian crisis in late 1997 and Russia's default in August 1998, overnight rates went to nearly 50 per cent, and in November 1998 Brazil arranged a support package with the IMF and the BIS worth $41.5 billion.

The banking system had its own reckoning. Brazilian banks had earned enormous revenue from the inflationary float — the gap between receiving deposits and settling payments, worth something on the order of 4 per cent of GDP in 1993. Within two years that income was nearly gone, and institutions built to harvest it began to fail: Banco Econômico in August 1995, Banco Nacional that November. The government created PROER in November 1995 to finance the absorption of failing banks by healthy ones, and PROES to restructure the state-owned banks that had functioned for decades as the financing arms of governors.

The peg itself did not survive. Gustavo Franco resigned from the central bank on 13 January 1999; the band was widened and then abandoned within 48 hours, and the real fell from R$1.21 to the dollar to above R$2 by March. Armínio Fraga took over the central bank and Brazil adopted inflation targeting in June 1999, trading the exchange-rate anchor for a domestic one. Inflation rose to 8.94 per cent that year and then settled. A devaluation of that size in 1990 would have restarted the indexation machine within weeks. In 1999 it did not, because the indexation machine had been dismantled five years earlier and there was nothing left to restart.

The Thing That Lasted

Brazil's earlier plans had asked citizens to police shopkeepers — in 1986 José Sarney invited the public to act as "fiscais do Sarney," inspectors of his own price freeze. The Real Plan asked nothing of the public except that it read two price tags for four months.

Cardoso served two terms and handed the presidency to the man he had beaten twice. Lula kept Cardoso's inflation target, kept his floating rate and kept his primary surplus, which is the clearest measure of what the plan achieved: it turned a monetary regime into a constitutional fact that changing the government no longer changed. Cardoso, asked during the 1994 campaign about the anti-orthodox economics of his own academic career, replied: "Forget what I wrote." His memoir of the period treats the line as a joke that was also a resignation (Cardoso, 2006). Werner Baer's survey of the Brazilian economy dates the country's modern macroeconomic framework to the URV rather than to any of the reforms that followed (Baer, 2008).

Other countries broke hyperinflations by other means — Germany in 1923 by scrapping the currency outright, as set out in the Weimar hyperinflation, and Argentina in 1991 by chaining the peso to the dollar in law, a commitment that took a decade to detonate. Brazil's method was the one that no government had been willing to attempt, because it required admitting in advance that the existing money was finished. A shopkeeper in São Paulo spent the winter of 1994 writing two numbers on the same jar of coffee and watching one of them stop moving. That was the whole mechanism.