Thursday, 2 August 1990. Two in the morning, Kuwait time.
On Saddam Hussein’s order, Iraq’s army is crossing into Kuwait. Thousands of Indians live and work in the emirate. One of them, Agnel Rebello, has been here since 1980. Their wages go home as remittances, and remittances are one of the ways India pays for the oil it buys.
Soon after the invasion, India’s foreign minister, I. K. Gujral, flies to Baghdad to ask for Iraq’s cooperation. Indian community groups organise buses, and with Baghdad’s tacit consent the buses carry people out through Basra and Baghdad to the Jordanian border, and on to Amman.
In Amman, Air India’s regional director for the Gulf and the Middle East, M. P. Mascarenhas, is running an evacuation out of a hotel room with very little space. When his team lands, five to seven thousand Indians are already in the city, and the number swells by the day. Some have found hotel rooms. Others are camping at the airport. The team has no mobile phones, and it keeps misjudging how many people will turn up.
The airlift lasted fifty-nine days. Air India flew 488 flights and carried 111,711 people home. Rebello later recalled a man in Kuwait holding a gun to him and demanding his car. It was the Indian state doing one hard thing for a great many people, quickly and well. Hold on to that, because within a year the same state was pawning its gold, and the two facts are connected.
The evacuees were the visible part of a bill. Crude oil rose from about $17 a barrel in July 1990 to about $36 in October. In the financial year 1990–91 India’s oil imports cost $2 billion more than the year before, $5.7 billion in all, partly because its own oilfields were producing less. Remittances from the Gulf fell. Bringing workers home and resettling them cost money. And the war disrupted trade with the Soviet Union, one of India’s main export markets. The economists Valerie Cerra and Sweta Saxena, who later reconstructed the crisis in the IMF’s Staff Papers, list all four shocks.
But a shock topples only a structure that is already leaning. India had spent the second half of the 1980s growing on borrowed money. Its external debt nearly doubled, from about $35 billion in 1984–85 to $69 billion in 1990–91. Commercial loans rose from $3 billion to $13 billion, and the deposits of Indians living abroad from $3 billion to $10.5 billion: money that could leave as fast as it had come. Servicing the debt took about 21 percent of the country’s current earnings abroad in 1990–91 by the government’s count, and close to 30 percent by Cerra and Saxena’s. On the government’s own count the Centre’s fiscal deficit had passed 8 percent of GDP, against 4 percent in the mid-1970s.
Then the politics shook. From August 1990 Moody’s had India on watch for a downgrade. In November V. P. Singh’s coalition fell amid caste and religious conflict, and Chandra Shekhar formed a minority government. India borrowed from the IMF in July 1990 and again in January 1991, and still the reserves drained away. By January, by one count, they stood at about $1.2 billion. On 6 March Chandra Shekhar resigned, forcing an election. The rating agencies moved that month: by one account Standard & Poor’s cut India to the bottom rung of investment grade the very next day, and India’s ratings were cut again before the crisis was over. Foreign commercial banks stopped lending. Creditors declined to roll over loans as they fell due, and Indians abroad began to take their deposits out.
1991Sriperumbudur · A fork the book draws and does not fill.A little after ten on the night of 21 May 1991, at an election meeting at Sriperumbudur, near Madras, a woman who had come forward with a garland bent to touch Rajiv Gandhi’s feet and set off the explosives she was carrying. Seventeen people died, Gandhi among them. The last two phases of the election were put off.
The page draws a fork here and leaves the other branch blank. This book does not guess what Gandhi, the front-runner, would have done with the crisis. It can record who did. P. V. Narasimha Rao had not stood in the election and had very nearly retired from politics. On 29 May he became president of the Congress party, which went on to win 244 seats, short of a majority, and on 21 June, a week before his seventieth birthday, he was sworn in as prime minister of a minority government. His first choice for finance minister, the former central bank governor I. G. Patel, declined. His second was Manmohan Singh.
Before any of that, in the last days of May, with the election still unfinished, the caretaker government had sold twenty tonnes of confiscated gold through the Union Bank of Switzerland, with an option to buy it back. Accounts of the date run from 24 to 30 May, and of the price from about $200 million to $240 million. It had also decided, its successor would tell Parliament, to use part of the Reserve Bank’s own gold to raise money abroad. The new government inherited that decision along with everything else.
as Singh later told it (2018)
Singh said in 2018 that he had planned two doses, a mild one to test the water and a stronger one after it. In the report of his remarks, Rao asked him to call off the second step, and by the time he told C. Rangarajan, then a deputy governor of the Reserve Bank, to stop, it was too late.
as a magazine told it (2022)
A magazine account, reposted online in 2022, says the second step went ahead without Rao's consent. It rests on a single report.
The rupee was lowered twice: 1 July, then 3 July
The Reserve Bank's chronology gives the two dates and one total. It gives no split between the steps, so the two treads are drawn equal and nothing here is to scale. What happened between them depends on who tells it.
Source: Reserve Bank of India, chronology 1991–2000; The Quint, 18 Dec 2018; Outlook, 5 Feb 2022
See the numbers
| Date | Event | Source |
|---|---|---|
| 1 July 1991 | First step: the rupee lowered against the dollar | RBI chronology |
| 3 July 1991 | Second step: lowered again | RBI chronology |
| 1 July to 3 July 1991 | Both steps together: about 18 percent in all | RBI chronology |
Ten days after the oath, it moved. On 1 July the Reserve Bank lowered the rupee against the dollar, and on 3 July it lowered it again. Together the two steps came to about 18 percent.
What happened between the steps depends on who tells it. Singh said in 2018 that he had planned two doses, a mild one to test the water and a stronger one after it; that President R. Venkataraman was not in favour of devaluing at all; and that he still had the letter in which Rao allowed it. The report of his remarks adds that Rao asked him to call off the second step, and that by the time Singh told C. Rangarajan, then a deputy governor of the Reserve Bank, to stop, it was too late. A magazine account, reposted online in 2022, says the second step went ahead without Rao’s consent, and that Rao had been unsettled by an article arguing that the devaluation of 1966 had split the Congress party. That last claim rests on a single report. The book records it and builds nothing on it.
Also in July, the Reserve Bank shipped 46.91 tonnes of gold abroad and pledged it with the Bank of England and the Bank of Japan for a loan of about $405 million. Officials kept the move as quiet as they could, for fear of setting off a panic.
The pawn ticket, 1991
Two weeks of imports, and the nation's gold in pawn
To borrow, India pawned its gold. Weeks later the finance minister ended his budget speech with Victor Hugo.
Source: 1991 Indian economic crisis (Wikipedia); RBI accounts via press
See the numbers
| Item | Figure |
|---|---|
| Foreign-exchange reserves, January 1991 | about $1.2 billion |
| Import cover, summer 1991 | about 2 weeks (of 52 in a year) |
| Confiscated gold sold through UBS with a repurchase option, late May 1991 | 20 t (10 ingots drawn, one per 2 t) |
| Gold flown abroad, pledged with the Bank of England and the Bank of Japan, July 1991 | 46.91 t (23 ingots and a half drawn), about $405 million |
On 9 July Rao spoke to the country. “Desperate maladies call for desperate remedies,” he said. He blamed eighteen months of paralysis and two governments that had put off decisions, and said the country had had no foreign exchange to import even diesel, kerosene, edible oil and fertiliser. The cheaper rupee, he explained, would sell more garments, leather goods, gems and jewellery and farm produce abroad, and create jobs at home. He gave the credit for starting the change to Rajiv Gandhi, who had come to power in 1984, and called his own measures a continuation. He did not pretend, he said, that the repair would be easy, quick or smooth.
On 24 July Singh presented the budget. The deficit, he told the Lok Sabha, was more than 8 percent of GDP; interest alone took almost a fifth of the Centre’s spending; inflation had reached double digits. There could be no adjustment without pain. In the same speech he announced a National Renewal Fund, a safety net meant to retrain and protect the workers on whom the cost of modernising industry would fall. He closed with Victor Hugo: no power on earth can stop an idea whose time has come.
The same day the Ministry of Industry published a new industrial policy, and most of the licence raj for industry came down with it. Licensing was abolished for every industry outside a short list kept for reasons of security, hazardous chemicals, the environment and what the statement called items of elitist consumption. Foreign firms could own up to 51 percent of companies in a list of priority industries with automatic approval. Almost every sentence pointed outward. One pointed the other way:
Industries reserved for the small scale sector will continue to be so reserved.
Garments were on the reserved list. Fifteen days earlier Rao had named garments as the first thing the cheaper rupee would sell.
Chain X191991 → 2015The list survives the crisis19472026The list outlived the crisis by twenty-four years. Garments came off it during the 2000s, and the last twenty items in 2015. The rules that decided whether an Indian factory could ever grow large were not part of the emergency, and so they were not part of the cure.
Door D1 · openedOn the imagined night that frames this book, a little before twenty to ten, the Chief Statistician laid a sheet of incomes on the desk with two rows folded under, 1990 and 1991, and said, “You’ll want that one later.” Here they are.
Tap, or use ← → to read a year
In market dollars, China passes India in 1991
GDP per head in current US dollars. India leads in 1990, $371 to $319; in 1991 China does, $334 to $306.
Source: World Bank, World Development Indicators (NY.GDP.PCAP.CD); PPP (NY.GDP.PCAP.PP.CD); conversion factors (PA.NUS.ATLS), fetched Oct 2026; Maddison Project Database, via Our World in Data
See the numbers
| Year | Market $: China | Market $: India | PPP: China | PPP: India | Maddison: China ahead by |
|---|---|---|---|---|---|
| 1985 | $295.01 | $300.93 | – | – | – |
| 1986 | $282.45 | $314.90 | – | – | – |
| 1987 | $252.26 | $344.94 | – | – | – |
| 1988 | $284.02 | $358.48 | – | – | – |
| 1989 | $311.43 | $349.92 | – | – | – |
| 1990 | $318.50 | $371.09 | $991 | $1,177 | about 43% |
| 1991 | $334.13 | $305.57 | $1,105 | $1,203 | – |
| 1992 | $367.82 | $319.18 | $1,276 | $1,271 | – |
| 1993 | $378.94 | $302.88 | $1,471 | $1,335 | – |
| 1994 | $475.68 | $347.73 | $1,680 | $1,424 | – |
| 1995 | $612.68 | $375.18 | $1,884 | $1,533 | – |
In 1990, converted into dollars, the average Indian produced $371 a year and the average Chinese $319. In 1991 the order flipped: China $334, India $306. China passed India in the year India pawned its gold.
Now turn the Other Clock, because the crossing moves when you do. India counts its years from April to March, so its “1991” is the year of the crisis and the year after it. Over that year, by the World Bank’s conversion factors, a dollar came to cost about 37 percent more rupees, and only about 10 percent more yuan. Measured in current rupees, the average Indian income rose by about an eighth. Measured in dollars, it fell by about 18 percent, and most of that fall was the exchange rate. The real gap did widen that year, though by less: India’s economy grew by about 1 percent, China’s by more than 9. Adjust for prices instead, and the World Bank’s purchasing-power series keeps India ahead in 1991, $1,203 to $1,105, and puts China in front only in 1992, by five dollars. The Maddison Project, the economic historians’ series, has China ahead already in 1990, by about 43 percent. The crossing came in 1991, in 1992, or before 1990, depending on the clock. No clock shows India passing back.
It is worth remembering how it looked from 1990. Asked then which of the two would shape the next century, a sensible person might well have bet on the democracy, with its English-speaking elite, its common-law courts and Asia’s oldest stock exchange. The alternative was a country that within living memory had starved tens of millions of its own people, spent a decade persecuting its intellectuals, and a year earlier had sent the army and its tanks to clear Tiananmen Square.
Why did that democracy keep running out of dollars?
The root runs back to a model. In 1953 the statistician P. C. Mahalanobis worked out a two-sector theory of growth: put more of the nation’s investment into the industries that make machines, and growth would start slowly and then run faster than it otherwise could. The model assumed a closed economy, with no trade and steady prices. In 1954 Nehru and the Planning Commission asked him to start planning studies, and his model became the frame of the Second Five-Year Plan, which from 1956 put the public sector and heavy industry first. In 1955 a panel of economists was appointed to appraise the plan. One member, B. R. Shenoy, submitted a note of dissent, the only one. India, he argued, could not find all the money that rapid industrialisation needed, and the power to award licences would breed corruption.
The plan needed imported machines that India had not earned the foreign exchange to buy. A foreign-exchange crisis in 1956–57 ended a decade of stop-go loosening of import rules, and comprehensive import controls came back and stayed until 1966. In September 1958 an American intelligence estimate found that India’s reserves had already fallen about three times further than planners had expected for the whole Plan and were nearing the danger point; it treated $500 million as the floor below which confidence in the rupee would be at risk. The Plan had been cut back. A meeting of creditors sponsored by the World Bank had just indicated its intention to provide about $350 million in new aid, with $250 million more expected later. India had learned to ration its dollars, and to fill the gap with other people’s.
At 5.26 in the afternoon of 13 July 1965, Lyndon Johnson received India’s ambassador, B. K. Nehru, in Washington. India’s agreement to buy American grain under the PL 480 programme had lapsed on 30 June, and the ambassador carried a letter from Prime Minister Shastri. The note of the meeting, kept by Johnson’s aide Robert Komer, records the President telling him that some members of Congress did not understand why a country like India would go for steel mills instead of fertiliser. A month earlier Johnson had ordered that no new loans to India or Pakistan be decided without his approval, and asked officials how to get more leverage for American money.
Chain I151965 → 2020Planted: the steel-mills line returns in A Highway19472026Then the monsoon failed. India’s net output of foodgrains fell from 78 million tonnes in 1964–65 to 63 million in 1965–66, nearly a fifth, and in March 1966 the two governments’ joint statement spoke of the previous year’s unprecedented drought. On 6 June 1966 Indira Gandhi’s government devalued the rupee by 36.5 percent: a dollar that had cost 4.75 rupees now cost 7.50. Panagariya records that the move was made under pressure from the World Bank, that the loosening of import licensing that came with it lasted about a year, and that by 1968 an intense reaction at home had reversed almost all of it. In 1975 Bhagwati and Srinivasan drew the lesson: outside pressure for reform, applied to a weak government with an election coming, can be fatal.
Put the two roots beside the crisis and a chain appears. A plan assumed dollars the country did not earn. Import controls rationed them. A devaluation pressed on India by its lenders was reversed in anger. A boom in the 1980s was paid for with short-term money, and a war in the Gulf called the money in. It is a tidy story, and the tidiness is the warning.
The gold was freed sooner than the legend suggests. According to the Reserve Bank’s annual report for 1991–92, the loan was repaid between September and November 1991 and the gold redeemed, though for the time being it was kept abroad. On 29 February 1992 Singh told Parliament that the previous government had already decided to use the Reserve Bank’s gold, that his own had not thought it prudent to reverse the decision, and that it had promised to redeem the gold and kept its word. The decision to pawn had been made by one government and carried out by the next, which is how a crisis bargain should be read: as the state’s, not a party’s.
1994The same 1992 speech announced that the National Renewal Fund had been set up, with about Rs 1,000 crore expected from the World Bank’s concessional arm, and promised schemes for unorganised workers too. In February 1994 Singh put its provision for the year at Rs 1,020 crore, and the purposes he named were voluntary retirement in public sector undertakings, training and counselling. Both speeches describe a fund for workers leaving industry. Neither speaks of places: the districts that would feel the tariff cuts are not in it.
2024In 2024 the Reserve Bank moved more than a hundred tonnes of gold from the United Kingdom to India. It is tempting to call this the gold of 1991 coming home. It is not. That gold was redeemed thirty-three years earlier. What came home in 2024 is a rhyme, not a return, and the book keeps it for its last chapter.
The pawn ticket, 1991 · closed
2024: a rhyme, opened in chapter 33
Source: 1991 Indian economic crisis (Wikipedia); RBI accounts via press
See the numbers
| Item | Figure |
|---|---|
| Foreign-exchange reserves, January 1991 | about $1.2 billion |
| Import cover, summer 1991 | about 2 weeks (of 52 in a year) |
| Confiscated gold sold through UBS with a repurchase option, late May 1991 | 20 t (10 ingots drawn, one per 2 t) |
| Gold flown abroad, pledged with the Bank of England and the Bank of Japan, July 1991 | 46.91 t (23 ingots and a half drawn), about $405 million |
So the reform India made under the gun was real, and so was its boundary. The political scientist Ashutosh Varshney later sorted India’s reforms into two tracks. Those that touched mainly elites, such as industrial licences and foreign investment, went through. Those that touched masses of voters directly, such as labour law, the reserved list, farm subsidies and the sale of state firms, stalled. In 1996, on his account of a survey, only 19 percent of Indian voters had heard of the reforms at all.
it happenedit stalled or was reversed
The elite track
Reforms that touched mainly elites
- 1
24 July 1991 Industrial licensing ended
Abolished for every industry except a specified list (paragraph 23).
- 2
24 July 1991 Foreign equity up to 51 percent allowed in priority industries
The statement promised approval with no bottlenecks of any kind (paragraph 25).
The mass track
Reforms that touched masses of voters directly
- 3
2015 The last 20 items freed from the small-scale reserved list
Garments had come off it in the 2000s.
- 4
2015 Land acquisition amendment passed the Lok Sabha, not the Rajya Sabha
Passed on 10 March 2015. The Rajya Sabha did not pass it, and the ordinance behind it lapsed on 31 August 2015.
- 5
2020 to 2021 Farm laws passed in 2020, repealed in 2021
Three laws, passed by Parliament in 2020 and repealed by it on 29 November 2021.
- 6
21 November 2025 Labour codes take effect: the line for prior permission to lay off or close rises from 100 workers to 300
In national law; states issue their own rules under the codes.
Two tracks of reform, 1991 to 2026
Varshney sorts India's reforms by who they touched. A filled marker is a change that happened; a hollow one stalled or was reversed. Each date is from the source named beneath it, and where a source gives only a year, so does the figure.
Source: Varshney, “India’s Democratic Challenge”, Foreign Affairs, 2007 (the two tracks); dates and sources under each event
See the numbers
| Track | When | What | Outcome |
|---|---|---|---|
| The elite track | 24 July 1991 | Industrial licensing ended | Done |
| The elite track | 24 July 1991 | Foreign equity up to 51 percent allowed in priority industries | Done |
| The mass track | 2015 | The last 20 items freed from the small-scale reserved list | Done |
| The mass track | 2015 | Land acquisition amendment passed the Lok Sabha, not the Rajya Sabha | Stalled: Lok Sabha only |
| The mass track | 2020 to 2021 | Farm laws passed in 2020, repealed in 2021 | Stalled: repealed in 2021 |
| The mass track | 21 November 2025 | Labour codes take effect: the line for prior permission to lay off or close rises from 100 workers to 300 | Done |
The elite track ran fast. Licensing ended on 24 July 1991. The mass track waited: the reserved list until 2015; in national law, the 100-worker line until 21 November 2025, when it rose to 300. Rao had called his measures a continuation, and the 1980s had begun them. What the gun added was speed, and a limit. Under duress, a democracy changed what no organised crowd would fight for, and left alone what one would.
China had been reforming for thirteen years by then, one trial point at a time. India’s pawned gold bought it a cheaper rupee, an open door for foreign capital and the end of the licence for most of industry. It did not buy the step that India had sawn off. That would take a different kind of courage, the kind that does not arrive on a deadline.