Amartya Sen Watched the Bengal Famine of 1943 as a Boy, Argued the Rice Supply Had Barely Fallen, and Faced Critics Who Dispute the Harvest Figures and a Death Toll Put at 1.5 to 3 Million

When the Royal Swedish Academy of Sciences announced the economics prize of 1998 on 14 October, it named Amartya Sen’s “empirical studies of famine” among the grounds and described one finding in particular: famines have occurred when the supply of food was not significantly lower than in earlier years without famine. Sen, born in Bengal in 1933, had watched the Bengal famine of 1943 from Santiniketan, where he went to Tagore’s school. He wrote in his Nobel autobiography that he had been struck by its “thoroughly class-dependent character”: it did not afflict even the lower middle classes, only people much further down the economic ladder, “landless rural labourers” among them.

That finding rests on three sets of numbers, and each has been challenged: how much rice Bengal had, who could afford it, and how many died. The documents behind the famine itself are in our account of the 1943 papers and the shorter explainer. What follows names who disagrees with whom and works the arithmetic.

What Poverty and Famines claimed

The Academy’s press release says the book challenges “the common view” that a shortage of food is the most important, sometimes the only, explanation of famine. It also notes that a few critics had questioned some of the book’s empirical results, and calls it a key contribution to development economics. Sen found that famines had struck when supply was not much below earlier non-famine years, and that famine-stricken areas had sometimes exported food. What decided who starved was how social and economic factors set the opportunities of different groups.

For Bengal, Senjuti Mallik’s 2022 review in GeoJournal summarises his figures: rice supply about 5 percent below the previous five-year average and 13 percent above 1941, a year without famine. In her account of his argument, fishermen, agricultural labourers and transporters were hit hardest, wartime inflation cut the real wages of landless workers, and big farmers, merchants and rice-mill owners gained. Joseph Lelyveld, reviewing Madhusree Mukerjee’s book on Churchill in 2010, quotes Sen finding “very little overall decline” in “food output or aggregate supply.”

The Commission had already counted three weeks

Sen drew on the Famine Inquiry Commission’s report of 1945, though not on it alone. It measured rice in weeks of the province’s requirements. Current supply, meaning the winter harvest reaped at the end of 1942, the smaller crops of 1943 and net imports, covered about 43 weeks in 1943 and about 39 in 1941. The average for 1938 to 1942 was 49. Of the 1943 shortage in the winter-rice crop the Commission wrote: “Actually, it was less serious than in 1941.” That suggested “at first sight” a breakdown in distribution.

The Commission then added stocks. It put the carry-over into 1943 at about 6 weeks, so total supply came to about 49 weeks and the deficit to about 3. It judged the carry-over in 1941 “considerable” and concluded that the supply position was worse in 1943 than in 1941. Of its stock estimates it said: “We have no information about the actual size of the carry-over in any year.” One member, M. Afzal Husain, filed a minute putting the shortage higher.

Three weeks is about 6 percent of a year. The Commission’s two estimates of 1943 requirements are 8.94 and 10.14 million tons, so three weeks comes to roughly 0.5 to 0.6 million tons. The same tables give Sen’s figures a rough check. Against the 1938 to 1942 average, current supply in 1943 was 6 percent lower in the unadjusted series and 9 percent lower in the adjusted one. Against 1941 it was 13 and 12 percent higher. Mallik reports Sen’s figures as 5 and 13 percent. Sen says he used the adjusted statement together with other sources, and the tables alone do not reproduce his 5 percent, though they show the same pattern.

The Commission also described the victims. About 2 million rural families, some 10 million people, lived mainly on agricultural wages, and artisans such as potters, carpenters and weavers did not grow food. “As prices rose in the early part of 1943,” it wrote, “non-producers were the first to suffer.” Sen and the Commission describe the same people. The argument is over whether a shortfall of this size explains what happened to them.

Tauger and the harvest of late 1942

Mark Tauger, a historian at West Virginia University, challenged the premise in a 2003 article in the Journal of Peasant Studies and restated it in a letter to the New York Review of Books of 24 March 2011. He relies on S. Y. Padmanabhan, director of the Central Rice Research Institute at Cuttack, who reported that the brown spot disease striking the late-1942 rice crop was “exceptionally virulent.” None of the fifteen winter varieties grown at two research stations hundreds of miles apart resisted it, and yields often fell to a tenth of the previous year’s. Tauger adds that Sen’s figures rest mainly on the Commission’s crop forecasts, which he calls projections and not definitive.

Sen answered in the same issue, a month after answering Mukerjee on the same data. The forecasts Tauger quoted belong to the Commission’s Statement I, “Unadjusted Current Supply,” while he used Statement III, “Adjusted Current Supply,” which corrects the yield estimates, and other sources as well. Two research stations in two districts cannot stand for undivided Bengal, which had twenty-seven. And he denied ever claiming that the crop shortage was too small to cause famine. The decline in supply was small, he wrote, but demand rose sharply in the war economy, leaving a substantial shortfall against demand. He drew the line between “a decline in food availability and a shortfall of supply with respect to demand.”

The Commission’s own text shows what Statement III changed. It raised the estimated winter-rice acreage by one fifth, which lifts supply in every year, 1943 included. The Commission had also recorded the disease: after the cyclone of 16 October 1942, it wrote, the Government of Bengal said fungus and root-rot hurt the crop even more than the storm did. What Tauger adds is a measure of how severe the loss was, and that is the part Sen disputes.

Ó Gráda adds a geographic point. The cyclone and the disease were, he writes, mainly confined to western Bengal, where the final forecasts show the steepest crop falls, while Tauger holds that the weather favoured the disease uniformly across the province. Estimates Ó Gráda draws from Maharatna put 65 percent of the excess deaths between mid-1943 and mid-1944 in the eastern divisions, now Bangladesh, where the crop fell less but which was normally short of rice.

Five estimates of the shortage

The Commission’s is about three weeks, roughly 0.5 to 0.6 million tons. Ó Gráda reports an assessment by a leading merchant and Muslim League politician that the 1943 shortage, after allowing for carry-over and the likely summer crop, was one million tons. Mukerjee says in a letter that the notes to her book give several estimates, all substantial, and that Sen quoted a government estimate of the shortfall as 140,000 tons when the document says 1.4 million. Sen called it a typo with no bearing on his assessment. Mallik reports that Paul Greenough, in Prosperity and Misery in Modern Bengal (1982), calculated that 90 percent of the usual rice supply was available, a shortfall of about 10 percent, twice Sen’s.

The fifth comes from the Bengal government. In February 1943 its minister for agriculture, the Nawab Bahadur of Dacca, announced rice output of 6.9 million tons against requirements of 9.3 million. The gap is 2.4 million tons, about a quarter. Ó Gráda writes that the data behind it “were notoriously poor” and that the announcement pushed the price of rice up.

Against requirements of 9.2 to 10.4 million tons, the Commission’s own range for 1941 consumption, a shortage of 1 million tons is 10 to 11 percent and 1.4 million tons is 13 to 15 percent. Ó Gráda’s verdict on the quantity: “There was a food availability problem, though its extent cannot be resolved with any accuracy.” Bengal could have coped with a million tons in peacetime, he argues, but not in 1943, with military demands and disrupted trade.

Where the critics side with Sen

Ó Gráda summarises Sen’s account as one in which speculation, set off by wartime conditions, turned a small shortfall into a collapse in the supply reaching markets. On hoarding he disagrees. The food drive of 1943 found stocks of rather more than 300,000 tons in the districts, which a Government of India memorandum of 7 September 1943 reports the Bengal government calling stock and not surplus. Of 622 prosecutions for hoarding and profiteering in July 1943, only 43 concerned rice.

He still accepts the central point: “the shortfall alone would not have caused such a devastating famine.” His regressions find that the size of the crop decline across Bengal’s divisions did not predict the rise in deaths, while population density, the earlier death rate and youth literacy, a rough measure of living standards, did. His own cause is “the lack of political will to divert foodstuffs from the war effort.” Mukerjee, who left Sen’s sufficiency claim out of her book as seriously challenged, writes that his insight about lost entitlement “remains true and profound.”

Counting the dead

The Commission’s figure was about 1.5 million deaths from the famine and epidemics between January 1943 and June 1944, with no allowance for under-registration of deaths in 1944. Sen’s estimate, as Maharatna lays it out, adds West and East Bengal separately. For West Bengal he took registered deaths from a 1951 census publication, set “normal” deaths at the 1941 and 1942 average or at 1942 alone, and multiplied by 1.51, which assumes about two deaths in three were registered. For East Bengal he used the 1.714 million excess deaths in the 1951 Pakistan census. The sums give 2.6 to 3.05 million, and Sen wrote that “we may be inclined to pick a figure around 3 million as the death toll of the Bengal Famine.” Greenough’s 3.5 to 3.8 million, Maharatna writes, is largely based on Sen’s calculations.

Maharatna found registration data for undivided Bengal that Sen had not used. They show 1,908,622 registered deaths in 1943 against an average of 1,203,507 for 1941 and 1942, an excess of 705,115. Adding 1944 and 1945 gives 1.26 million, and Sen’s own multiplier of 1.51 turns that into 1.9 million. A baseline that allows for the falling death rate of the 1930s gives about 2.1 million, within a range of 1.8 to 2.4 million. The top of the range assumes only 60 percent of deaths were registered, and Maharatna knows of no strong ground for that. He also found that the Pakistan census set East Bengal’s normal deaths at 540,000 a year where the registration data imply about 790,000.

Dyson and Maharatna published the 2.1 million figure in 1991. Ó Gráda writes “over two million,” Sen’s Nobel autobiography “between two and three million.”

What is settled and what is not

No source opened for this piece disputes four things. Rice prices rose several times over in 1942 and 1943. The crop reaped at the end of 1942 was short, from a cyclone and a disease the Commission itself recorded. People who bought food on wages and had no reserves suffered first. Every estimate here published since Sen’s, from Maharatna’s 1.8 million upward, is above the Commission’s 1.5 million.

Four points are open. The size of the 1942 harvest loss is one, with two research stations in two districts set against forecasts for all twenty-seven. The carry-over is another, since the Commission had no information on it. Whether hoarding mattered is a third, on which Ó Gráda and Sen differ. The fourth is whether the toll is nearer 2.1 million or 3 million, which turns on how complete death registration was. On Maharatna’s calculation Bengal’s population in mid-1945 was still about 2 million below its mid-1943 level and did not regain that size until around mid-1947.