Peter Bowbrick      0777 274 6759

Does it matter?

The main economic traditions have argued that most famines were caused by sudden declines in the amount of food available per capita. Recently, the entitlement approach has become popular, arguing that most famines occur when there is no such decline. It argues that famines may occur when one group of people suddenly eats a greater proportion of what is available than previously, leaving less available for other groups, which starve. I have shown elsewhere that this is theoretically untenable. Alternatively, it argues, famines may occur because speculators remove grain from the market, so there is not enough food to go around and people starve. If this is the case, then a famine can be stopped instantly by Government's seizing speculative stocks and distributing them, possibly introducing a rationing system and relief schemes. Government may import small quantities and dump them on the market to frighten speculators into releasing their stocks. The difference in diagnosis is of the greatest practical importance: if a famine is in fact caused by a decline in food availability per capita, and the Government responds by seizing and distributing speculative stocks, the death rate will be higher than if they had done nothing at all. For example, in the Bengal famine of 1943, Government was obsessed with speculation and the need for rationing and chose to believe that there was no real shortage, with the result that two million people died. (Famine Inquiry Commission 1945a,b).

This paper asks two questions. First, can speculation actually cause a famine? Second, if a famine is caused by a sudden reduction in availability per head, does speculation increase mortality?

The English Classical Economists

Contrary to the statements of commentators like Devereux (1993) and Ravallion (1997) the English classical economists did not take a laissez-faire approach to famine. Adam Smith set out the view of the English classical economists famines were nearly always caused by a decline in food availability per head, arising from a fall in supply or an increase in the number of mouths to feed. Government should deal with this by increasing effective demand, so that the free market would import the necessary food. He felt that the speculator was a necessary and desirable part of this process

This view was put into practice during the Irish famine of 1845-1849, when potato blight reduced the amount of food available per head (Woodham-Smith 1962, 1991). The British Government introduced a work programme to give people without incomes the cash to buy food. However, it took several months for a price signal to result in imports from America, and in the meantime, the only effect was higher prices, and higher profits for speculators. People who had enough money to buy food at normal prices sold their assets to survive, impoverishing themselves to the extent that many died when the potato crop failed a second or third time. The Government gave money for free food and soup kitchens, but high prices meant that most of this money went to speculators. Government decreed that the Irish landlords should pay massive poor law rates to finance additional relief, but the money obtained was a drop in the ocean. The economists then concluded that since all the demand side measures had failed, nothing could be done. The only way to prevent future famines was to reduce demand, by letting nature take its course. "'I have always felt a certain horror of political economists,' said Benjamin Jowett, the celebrated Master of Balliol, 'since I heard one of them say that he feared the famine of 1848 in Ireland would not kill more than a million people, and that would scarcely be enough to do much good.' The political economist in question was Nassau Senior, one of the Government's advisers on economic affairs." (Woodham-Smith 1991 pp 375-6). He need not have worried: 2.5 million people died out of a population of 8 million.

The famine could have been stopped without spending any more than was spent over the period. Instead of spending nearly all the money on relief and public works, Government could have banned grain exports and imported large quantities - in fact it imported less than 1 kg per head over the whole period and it actively encouraged exports. The amount available would then have been enough to keep the whole population alive. Prices would have been within the reach of those people not directly affected by the crop failure, and grain could have been supplied cheaply for charity - there were enormous charitable donations from Britain and the USA as well as Poor Law relief. Also important is the fact that people could have worked on their farms producing food, instead of being forced to work on relief schemes or to live in the workhouse.


The Entitlement Approach

A variant of the English classical approach is the entitlement approach, popularized by Sen (1976, 1977, 1980, 1980b, 1981, 1981 b, 1984) and seen to be the dominant and correct approach by commentators like Devereux (1993) and Ravallion (1997). Like the classical economists they concentrate almost entirely on the demand side, though they do not present it in any greater breadth or depth than for example Hunter (1873), Frere (1874) and Mahalanobis, Mukkerjee, and Ghosh (1946), nor do the prescriptions go much further than the Bengal Famine Code (1897).

The main novelty of this approach is that it asserts that most famines occur when there is no sudden decline in total food availability. For example, a boom or inflation may mean that some people eat a lot more food than before, so there is not enough to go round for the others and some die. In all famines some people continue to eat well, and the starving may be excused for thinking that they were starving because the rich were eating more than usual. The theory appeared sporadically in textbooks, but was introduced to the wider economic literature by Sen. Similarly, the possibility that speculation often caused famines was always popular, though dismissed by Adam Smith, but was introduced to the wider economic literature by Sen.

The entitlement school asserts its difference by saying, first, that its predecessors adopted the Food Availability Decline (FAD) approach, which states that all famines are caused by declines in absolute availability (not in availability per head) and which ignores the demand side, and, second, by attacking the FAD approach violently (e.g. Sen, Devereux). This approach does not appear in the literature, and is wholly mythical. The entitlement theorists are so eager to distance themselves from the FAD approach that they tend to avoid looking at availability at all - indeed Sen is highly critical of the Bengal Government because they wanted to determine the degree of shortage while a famine was raging. One result of this is that in their discussions there is a confusion between a famine caused by a redistribution of available food between groups, and the uneven impact of shortages on different groups, which is seen in any famine, however caused. The fact that some occupation groups starve while others hardly suffer is taken as evidence that the famine is caused by redistribution (e.g. a boom famine) rather than by a reduction in supply, which is a non sequitur. Similarly, there is a confusion between a famine being caused by speculators removing food from the market, and a speculators making a profit in a famine caused by a fall in supply (Adam Smith's case). It is argued that because speculators are seen to make a profit, the famine must have been caused by speculation, which is a non sequitur.

Because of this confusion, I consider separately whether famines can cause speculation, and what the effect of speculation is when a famine is caused by a fall in supply.

Ravallion (1986), working in this tradition, argues at book length that speculation was a cause of the Bangladesh famine. Unfortunately the book contains little economic analysis of the market and no statistics whatsoever. It is an object lesson on what happens if one embarks on econometrics without getting the economics right. He argues that speculators acting rationally, but acting on inaccurate crop information, stored too much and caused starvation where it was not necessary. This, if true, is trivial: if they had not acted on the information available, and it had proved correct, the death toll would have been far higher. However, it will be shown below that the statistics which do not appear in his book refute his argument.

The entitlement approach is based very largely on Sen's and others analysis of five famines, in which they assert that speculation and redistribution (e.g. boom famine) were the causes. The factual base of these analyses has been widely challenged. Sen's data has been criticized in some detail by Goswami (1990), Dyson and Maharatna (1991), Dyson (1991, 1996), Basu (1984, 1986), Kumar (1990). Dyson concludes ". . while they are far from being complete explanations, FADS were probably involved in all five of Sen's famines." Dyson, T. (1996). Bowbrick (1986, 1998), examines Sen's paradigm case, the Bengal Famine of 1943, and argues that his facts are in conflict with those in his sources. Seaman and Holt (1980) while not mentioning Sen, explode one of his cases: "In the Ethiopian case dealt with in this paper, the inevitability of a massive food shortage in the Wollo Province was known to the Imperial Ethiopian Government from a detailed qualitative crop survey completed in November 1972. The government was reluctant to declare an emergency until March 1973. But when international relief was called forward in April, it took until November for bulk food shipments to begin to arrive."

Since the theory lacks a factual basis, and its theoretical approach to boom and distributive famines has been shown to be wanting, this paper examines its other theoretical novelty, that a famine can be caused by speculation, rather than a fall in supply. It asks whether it is possible, even in theory, for a major famine, to be caused in this way in the countries for which it has been postulated as a cause.


Can Speculation cause famine?

Initially a model is presented here of a food economy similar to those where it has been seriously argued that speculation caused a famine, notably to the situation in Bengal in 1943. When one looks at these conditions, it is clear that the speculation hypothesis is absurd. Accordingly, I introduce a string of increasingly unrealistic assumptions, to create a model of a situation where there is, at first sight, some possibility of a famine being caused by speculation. There are extremely restrictive assumptions and stated preconditions on availability of storage, finance and transport, and the non-availability of Government stocks, for instance. These are put in because a famine cannot be created without them, however irrational the speculators. The fact that they are totally unrealistic in most economies means that it is quite impossible for speculators, however irrational, to create famines in these economies.

Degrees of shortage

People die in famines because they do not eat enough food, so the primary economic analysis must be in terms of food. Prices and money are of interest in so far as they affect the quantity of food available, and the ability of individuals to get food. (cf. Ravallion (1986 p18) Ahigh and unstable food prices were major contributing factors to excess mortality. Accordingly the degree of shortage is of primary importance. The following degrees may be distinguished.

These definitions are in terms of the nutrition of the population, not in terms of the market situation.

The Model

Let us take the example of a least-developed economy consuming 100x tonnes of rice a year in a "minimum no shortage" situation, when everybody has enough food for long-term survival, and the poorest people have only just enough. Rice is the only food crop. 90% of the population is in the rural sector. Rice is produced by owner-occupiers and tenants. Most of these do not end up with enough for subsistence and have to do paid work to survive. Wages are mainly paid out of the rice accumulated by landlords. In a "minimum no shortage" year the agricultural sector produces a surplus, 10% of the total, which goes to feed the urban sector. By assumption, if the total supply rises by 10% everybody increases consumption slightly, and there is still 10% of the crop going to the urban sector.

Most of the rice going to the urban sector, 60%, is handled by commercial grain traders in a normal year when the production is 110x to 120x tonnes. The remainder may be marketed by farmers, landlords, lorry drivers acting as occasional traders, or by firms like textile manufacturers or jute traders whose management decides to make a small speculation from time to time. a considerable amount is supplied direct from the rural sector to relations in the cities.

It is assumed that the leading commercial rice traders form a tight cartel and operate together for the good of the cartel. It is not possible to get farmers, landlords, consumers or non-grain businesses to join this cartel: they would each act in their own best interest, buying or selling whenever they like. It is in each individual's interest to act as a "free rider" letting the cartel bear the expense of withdrawing grain from the market, while they sell everything they can at the highest price possible. Those of the commercial traders who form the cartel normally handle 6% of the total. For simplicity it is assumed initially that all commercial grain traders join the cartel.

The question to be asked is whether speculation can cause a famine in a year where supply is 110x tonnes, 10% above the "minimum no shortage" supply. For a famine to occur, there must be a fall in the quantity of rice consumed to the famine level, with the remaining rice being removed by the speculators. It is assumed that the famine level is 80% of minimum normal supply, or 80x tonnes. At this level there is a first degree shortage, with sufficient rice to provide a barely adequate diet for everyone, provided there is rationing, or else there is a second degree shortage where there is insufficient food for long term survival even with rationing. Since this country does not have a large, incorruptible, bureaucracy, there is no rationing. As a result the poor are particularly badly hit, and many die of starvation, including many farmers.

In order to create the famine, the cartel must withdraw 30x tonnes from the market, to bring the total consumed down to 80x tonnes. They must also buy enough of what is left to have some chance of making a profit from selling it, say another 10x tonnes, making their total purchases 40x tonnes. With the figures assumed so far, this means that they buy virtually all that leaves the farm. This includes grain paid in rent to landowners, which would normally be used to pay for the labour of landless labourers and deficit producers. Indeed, it may require that some farmers sell part of their minimum subsistence stocks.

Preconditions for the famine

It is not possible for speculators to create a famine unless certain conditions apply.

I assume that rice is the only food crop. If we make the more normal assumption that rice accounts for 40% to 60% of the calories consumed in a predominantly rice economy, speculation obviously cannot cause a famine. People get calories from other starches, oil, vegetables etc. Even if speculation reduces the total calories by 30% it will not reduce the number of calories to the extent that there is general hunger and starvation among the rural and urban population (though some groups of people may suffer badly). It is difficult enough to procure this quantity, of grain and remove it from the market, but speculators must also buy a further quantity for resale. This implies that they buy all the rice grown.

It is also assumed that none of the crop goes for animal feed, manufacturing, brewing etc. These normally provide some buffer against famine. The quantity varies from country to country, but even in a poor country, could easily provide an extra 10% food.

It is assumed that there is no carry over of food from the previous year. Normally, perhaps two or three months' supply of grain is carried over. In an emergency this can be eaten, as well as the current crop, and the country goes into the next year with little or no carryover. If this is the situation, then speculation will not work. (This does explain why a famine often occurs when two of the last three years have had poor crops.)

It is assumed that there is no cassava crop. Normally this is eaten when the tubers are mature, at three or four years old. In an emergency, the one-year old, two-year old, and three-year old tubers can be harvested, doubling supply in one year, but leaving no emergency stocks to be carried through into future years. Again, this explains why a famine often occurs when two of the last three years have had poor crops, and there is no carry over of emergency supplies.

The famine can occur if and only if there is a closed economy, otherwise traders from neighbouring countries would soon fill the gap and make all the profits arising from speculators withdrawing rice from the market. The closed economy could arise from transport problems, currency restrictions or physical restraint by neighbouring governments. If there is an open economy, the presumption is that the famine is being caused by something other than speculation. If there is an open economy, but there is a three months' interval between ordering rice from abroad and its arrival, it may appear that there is some opportunity for speculators to create a famine for this short period. However, as will be shown below, the cost of creating a three month famine is as much as the cost of creating a nine month famine, and the returns are much lower, so it is not profitable.

The precondition of a closed economy implies that nobody is willing to give food aid, or that the Government is not willing to accept it.

The famine can occur if and only if there is some price at which the cartel can buy all this grain. It is necessary to assume that there is no "copycat" speculation, where people observe the grain traders buying and decide to speculate themselves. Commercial traders outside the cartel do not try to speculate, but rather sell to cartel members. There are no non-rice traders buying to secure supplies for their staff or customers. Consumers do not buy for their own security. Landlords are willing to sell all their stocks, knowing that it reduces the food supply in the neighbourhood. Farmers are willing to sell something out of their subsistence stocks. This does suggest that traders are able to persuade sellers that it is a short-term shortage, and that prices will fall later.

For the famine to occur, the cartel must be able to raise enough money. Buying this amount of grain can only be done by paying a very high price, probably more than 50% over the normal price. The cartel must buy 40x tonnes instead of its usual 6x tonnes and pay 50% extra. This means 10 times the normal outlay. In a country with a rudimentary banking system, where the grain trade is a significant part of total trade, this will cause problems. Commercial banks are not likely to finance such a speculative, politically dangerous and unethical venture. These problems are particularly serious in a country in which there has been a sudden boom in the manufacturing industry for some other reason.

For the famine to occur, traders must be able to store, export or destroy the surplus. Traders who normally need storage for a maximum of 6x tonnes (less if they take delivery over a season) now have to store 40x tonnes. The grain must be stored well away from landlords and farmers who may have second thoughts about their sale. There is no surplus of storage space in the rural areas of such economies. Similarly, in least developed countries there is not much warehouse space in urban areas - especially if there is a manufacturing boom. Outdoor storage would leave the grain in the sight of the starving, and would probably lead to looting or confiscation. Keeping it in the back rooms of a lot of small businesses scattered round the city, in lots of 50 bags or so may be feasible, but would be expensive and risky. What is required is a large warehouse, outside the residential areas (in case of riot), and easy to protect.

Another precondition is that transport from the rural areas is not a constraint. 40x tonnes must be moved in a month or two, instead of 10x tonnes over the whole of a season, as in normal years. Few least developed countries have this infrastructure. If the speculators are to export the grain they are removing from the country, international transport must be feasible. This precondition is likely to conflict with the precondition that the country is a closed economy and that no food aid will arrive.

There are no state food security stocks. If these exist, the state can break the market by releasing them.

At the beginning of the year there are no private or commercial stocks. If, as is usually the case, there is a carry over of one or two months' stocks, the amount that has to be bought by speculators is perhaps 25% higher.

The Rational Speculator

Before deciding that a famine was caused by speculation, government should be sure that these preconditions apply. Next, government should decide whether the rational speculator would in fact store this quantity of grain in the expectation of making a profit. It should be asked whether the speculators could rationally accept that the preconditions applied, whether the risks to the speculators would be acceptable and whether the enterprise would be profitable.


For the famine to occur, it is necessary that the speculators believe that there is a closed economy and that it will stay closed throughout the famine period. They will not risk their money if they think that there will be a sudden influx of grain. Even if the imports only reduce the shortage, the fall in price may make the speculation unprofitable. The fact that there is in fact a closed economy does not mean that rational speculators buying after the harvest believed that it would be closed later in the season. If rational speculators would not have believed that the economy was closed government should not act on the presumption that they did.

For a famine to occur, it is not only necessary that speculators can buy all this grain. It is necessary that speculators believe that they can, both when they launch their campaign, and as the campaign continues. In addition, they must believe that they can buy the rice at an average price which is not enormously above the normal price, say 50% over the normal price. If the price paid is a scarcity price, there will not be sufficient margin between buying and selling to justify the speculation. Once the cartel believes this, it is likely to stop purchasing, knowing that its members have already lost a great deal of money.

Again, for the famine to occur, the cartel must believe that it will get the finance, and that the banks will continue to support them until they can start paying back. If the bankers get cold feet, because of the political repercussions, say, the enterprise fails.

The cartel must believe that the necessary storage will be made available. It will be secure and affordable.

The cartel must believe that it will be able to obtain the transport to move the rice from the production area and, if necessary, export it, all before the prices rise.


If the cartel was to make a profit from the speculation, it would have to get a total revenue from the disposal of the 40x tonnes greater than the cost of purchasing it, even though only 10x tonnes are sold at famine prices. The cost of storage, bribes, transport from the producing areas and, most of all, risk would have to be taken into account.

The option of destroying the 30x tonnes surplus, or letting it rot, is not attractive. If they are to break even the net selling price (ignoring marketing costs) must be four times the purchase cost. This ignores the bribes, transport and risk. The purchase price, itself, was, perhaps, 50% higher than the normal, so consumers are being asked to pay at least six times the normal price on average throughout the famine. It is questionable whether an average level this high could be assured from this degree of shortage, with this level of purchasing power, though some sales could well be at much higher prices. This calculation ignores the bribes, transport and risk, which could bring the break-even price as high as ten times the normal price.

If the starving people realized that grain was being destroyed one could expect riots and looting followed by Government confiscation of stocks.

The 30x tonnes may be exported instead. This may be done soon after harvest, but not after the famine begins to bite. Transport from the villages and shipping within the time limit are a constraint, especially in a country that does not usually export grain. Transport is also expensive. In many countries an overvalued exchange rate makes export unattractive. It would be most surprising if the exporters recovered as much as half the inflated price they paid. If they did, they would have to charge a price 2.4 times the purchase cost and 3.7 times the normal purchase price for the remaining 10x tonnes to break even. Again, transport, risk, bribes, etc have to be taken into account, so a rational speculator would not act unless there was a considerably higher expected price.

The assumption that export is easy may imply a conflict with the assumption that it is a closed economy.

The option of the cartel's making its money at the beginning of the season, and then releasing the remaining stocks at the end of the season is even less attractive. If there was, say, a 6-month period of high prices instead of 9 months, prices would have to be raised accordingly to make the same revenue. The 30x tonnes are three months normal supply for the whole country, which, taken in addition to the remaining subsistence stocks, of perhaps 21x tonnes, would mean 70% more rice per month than in normal years. It is perhaps 18 months' supply for the non-farm population. The population would have very little money available to buy it with and prices would fall to glut prices. This makes the breakeven price for the 10x tons sold at famine prices absurdly high.

The remaining option is to carry over the surplus 30x tonnes to the next year. If the next crop was again normal at 110x to 120x tonnes, then the total supply would be 140x to 150x tonnes and this alone would suggest a depressed price. However, the farmers and landlords would have plenty for their own use, and would only buy from out of the sector if it was very cheap. The cartel would be sitting on 30x stocks when the city only wanted 10x. If the cartel tried to dispose of the stocks, the market would collapse completely.

In practice, this option would be even less attractive. Storage of grain in non-purpose-built buildings means damage from rats, insect infestation, damp, contamination from whatever was stored there before, as well as theft, so losses of 30% could be incurred. The cartel would still be in a difficult financial situation, as they may not break even on the 10x tonnes sold at famine prices, and might expect to make their profit on the remaining 30x tonnes. This requires a lot of expensive credit. There is the risk that aid agencies will send the famine relief supplies too late to help the starving, but in time to oversupply the market in the following season.

None of these options offer the cartel a certain way of getting a price that would cover their costs. The risk to them is enormous. It would be irrational for them to attempt to create a famine by speculation.



The biggest costs are risks. The speculators are operating in an environment where there is little hard knowledge, and where there are major political and economic risks.

The first risks are on buying and selling prices. For a famine to occur, it is necessary that the cartel are firmly convinced that the prices will rise to levels high enough to cover their cost and give them a very substantial profit. In fact, traders have no solid grounds to believe that prices will rise to the levels hoped. They do not have any experience of past famines. There are many reasons why they will be sceptical of reports or rumours of price levels in past famines or elsewhere. They are well aware that in times of scarcity and high prices their weighted average return over the season is well below the peak price. They know that the press and rumour report the occasional extremely high black market price rather than the going price. Indeed, the traders themselves may have spread the rumours, in order to push up the price of their stocks. Traders talking to each other boast of their occasional good deals, rather than mentioning the average price. They know that retailers, officials and police take a proportion of scarcity rents as bribes. They know that they will be pressed to provide the army, the police, politicians and officials, and even hospitals and schools, at low prices. Finally, they know better than anybody that when there is even a normal scarcity, many of the population spend everything they have to buy food: they cannot pay more just because there is a full-blown famine. In famine situations the poor may use their doors, their corrugated iron roofs, their draught animals and even their children, to buy rice. This is not of much value to the cartel, who need money in the short term. It may mean that a considerable part of the profit goes to rural landlords who have money, and who will recover it by renting out these goods in future years. It is necessary both that the speculators believe that they will make prices high enough for profitability, and that they are confident that they will.

Similarly, they must have a belief that they can buy the grain at an average price which will leave them a substantial margin. The prices they pay at the margin, when they are trying to mop up all grain stored by others, will almost certainly be above the price that they can make a profit at, and may be very near the famine price. If it proves too expensive, some traders may drop out of the cartel, stopping the enterprise.

Throughout the enterprise there is a risk that the cartel will collapse. a group of traders who previously were in competition must be brought together and kept working together under tight control. They may be of different religious, racial or national groups. One very satisfactory long-term result for each individual member of the cartel would be to bankrupt some or all other members of the cartel.

During the purchasing period each must spend ten times the normal expenditure on purchasing. It would pay each member to buy less and let other cartel members buy more. Throughout the selling period, every cartel member would stand to make large profits if they could sell off most or all of their stocks while other members withheld their stocks from the market. The moment it is seen that one cartel member is breaking, there is little speculative profit to be made, and all the others may scramble to unload their own stocks. If there are four equal members and one unloaded the full 10x onto the market, the shortage would be considerably less severe (90% of minimum normal supplies) and the market price would fall from extreme famine levels - unless the other members increased their stocks, making it much less profitable for themselves..

Similarly, throughout the selling period each cartel member would be tempted to sell rather more than the agreed amount, to increase supplies.

There is a threat that the stocks will be looted by rioters, or stolen by the local mafia.

Politicians will certainly blame the shortage on speculators: as they do whatever the cause of the famine. There will be at least lip service paid to the possibility of seizing and distributing speculative stocks. The 40x tonnes cannot be hidden, and the neighbours and the police will know where they are. At the least the cartel will have to pay big bribes. This may not be enough: if the famine threatens the regime, government may act, however much money powerful individuals may make in the short run. Confiscation of stocks and nationalization are real threats.

In most countries the cartel would have to permit supplies to the army and perhaps civil servants at a reduced price. There might also be price control, which would at least restrict the proportion of the supplies that could be sold at black market prices by the traders - even if their customers then resold at black market prices.

In many countries there is a restriction on the movement of grain from one district to another, with the result that traders have to bribe the police at the district boundary. This may be nominally part of normal food policy, or part of a security check, or openly part of an extortion racket by the police. The effect of this, or an octroi, is that there is no way of keeping the movement of grain secret. One risk is that the level of bribes rises. Another is that movement controls will be imposed in earnest. It is a normal and legitimate government policy to restrict inter-district transfer of grain in a shortage situation, so that it can procure cheaply in surplus areas. If government notices the high prices and grain movements early enough, it can block movements of grain. This would cut either the rural or urban market from its grain supplies and prevent export. It would also leave large stocks where they were difficult to store and very difficult to protect.

Imports by government, aid shipments, aid money and commercial shipments could also break the market. In each of these cases powerful politicians and officials would stand to make a lot of money.

There is a possibility that speculators, their families and their communities will be attacked and possibly murdered, especially if they belong to a religious or racial minority.

Hiding the evidence

Any such speculation would produce evidence which could scarcely be missed by government and the local population at the time, some of which would be equally obvious to the economist looking at the historical record. The evidence includes: evidence of crop size etc.; stock levels several times as large as in normal years; grain movements four times as high as usual immediately after harvest, exports one third of total production from a country that does not usually export, large borrowings by grain traders, a collapse of grain prices immediately before the next crop, a collapse in the urban market price in the following year.

In addition there would be strong reason to believe that there was a closed economy, that there was no "hoarding" or informal speculation, that there was no animal feed, brewing or manufacturing use for grain, and that there was in fact a cartel.

The evidence in the literature

Generally there is a lot of evidence to show that there is speculation in a given famine, and that some people cannot afford to pay the going market price and starve. This is to be expected in any famine, however caused, and gives us no reason to believe that the famine was caused by speculation, or that speculation increased the death rate.

The suggestion that the Bengal famine of 1943 was caused by speculation is contradicted by the evidence. There is no suggestion in Sen's sources (e.g. the Famine Inquiry Commission 1945) that speculators removed grain from the country by export, destruction or storing into the following year. On the contrary, there were substantial imports, repeated house to house searches failed to find any substantial stocks, and there was no glut on the market at any time in the two years following the famine. While private individuals were believed to have increased their stocks at the beginning of the war, when Japan entered the war and again when Burma was occupied, there was little scope for further hoarding in the months leading up to the famine. These private stocks would have mitigated the effect of the crop failure.

Ravallion (1986) argues that speculators in Bangladesh stored more than was necessary, putting up prices, and that this was a cause of the famine. As evidence of this, he states that there was a drop in prices at the end of the season, as speculators disposed of surplus stocks, but, as he does not give any statistics, one is left with the impression that he means the major drop in prices discussed here. If there was this major withdrawal of supplies from the market, there would have been a glut when it was released, and prices would have fallen below the prices for normal years. However, Seaman and Holt (1980) give the statistics, and these show that in some parts of Bangladesh there was no drop at all, and in others the drop was small, in spite of the fact that demand is extremely inelastic at these price levels. The implication is that there was slightly too much stored in some areas, but the predictions were more accurate than they are in most years. This begs the question of what is too much. Clearly speculators were correct to store as much as they did if their information suggested that it was needed: to do otherwise would not just lose them money, it would put the population at risk of starvation.

Relaxing the assumptions

Can speculation be used to create a famine for a shorter period? At first sight it might seem that the cartel need not buy so much if they are only trying to create a famine situation for a few months, so the outlay and risks are smaller This is not so. If they bought much less, and they tried to create a famine situation for two or three months, everybody else who had grain stocks would sell them. They would obtain all the profit, leaving none to the traders who kept their grain off the market. The alternative would be to buy the same as in the main example and have a shortage for a shorter period, selling perhaps 3x tonnes at famine prices. This would have all the costs and risks of the main example, with a much smaller payoff.

Can speculation be used to create a less serious shortage, say 90% of "minimum no shortage supply"? The speculators would take 20x tonnes off the market, instead of 30x tonnes. One option would be to buy the same as in the main example, 40x tonnes, and sell off 20 tonnes at scarcity prices. The other would be to buy 30x tonnes and sell off 10x tonnes at scarcity prices. The first option is more attractive in that the traders sell twice as much at scarcity prices, and so get a higher total revenue, but the purchase costs are significantly higher - certainly some of the grain will be bought at a price higher than the selling price. One can imagine supply and demand curves at which these could be profitable options, but it would be difficult to show that they are realistic.

If animal feed, brewing and manufacturing took as much as 20% of the total, speculation would become very risky indeed. In a political or humanitarian emergency this could be diverted to human consumption, and, indeed, some of it would be diverted merely because the price was higher in the other market. If all of it was diverted, prices would fall to the minimum normal supply level. If only some of it was diverted, prices could be expected to fall below the breakeven level. a rational cartel would not attempt to create a famine unless they had good reason to believe that this would not happen.

It is possible for consumers to "borrow" from the supplies of future years. Cassava is a buffer against shortage in many countries. It is not usually harvested for three or four years, when yields are at a maximum. In times of shortage it is possible to harvest one or two year old crops and increase food supplies. Of course this does reduce supply in future years. Similarly it is possible to eat green maize rather than waiting for it to ripen - sometimes a third of the crop is consumed in this way. Consumers can eat rice as soon as it is harvested, rather than waiting a couple of months until its flavour has developed. A rational cartel would take this into account. After a shortage year, this buffer would disappear.

In a multi-crop economy, the price effect of reducing the supplies of a single crop is limited. In a closed least developed economy prices of substitutes would also rise. If rice provided only half the calories, a 20% reduction in rice supplies would result in only a 10% reduction in calories, and it might be expected that there would be a smaller rise in the price of rice than in economies where rice provided 80% of the calories

In an economy where farmers are all self-sufficient or producing a surplus, and where they are owner occupiers or tribal tenants (in parts of Africa for instance), it is not possible to create a famine that hits farmers in the way described. Unless the farmers can be persuaded to sell their minimum security stock or can be forced to give it up the cartel will not be able to secure the stocks, and the farmers will survive. In these economies starvation among farmers is a sure sign that the shortage was not caused by this type of speculation. The cartel could create a famine among the rest of the population by buying up everything over and above these minimum survival stocks, and restricting supply to the non-urban population. Rather less need be bought than in the main example, but the cartel will be selling a much smaller amount, to the urban population only. For example the cartel might buy 20x tonnes, leaving the farmers with minimum normal supplies, and sell 8 tonnes. This relies crucially on the farmers' security stocks being an absolutely fixed amount, with the farmers not reducing the amount under any circumstances, even when they can get famine prices by selling some of it on the market. If they were slightly flexible, willing to sell an extra 5%, or 4.5x tonnes to get a high price, the speculation would also be unprofitable. The cartel could buy up this flexibility, but only by paying such a high price that the speculation becomes less profitable.

The assumption was made that all commercial grain traders joined the cartel. If they did not, the proportion of the total marketed in normal times by the cartel would be smaller, and the task of creating a famine would be larger.


Does speculation cause famine?

Can speculation cause famine when there is plenty of grain available? It has been shown that it is only possible under some conditions. Even when it is possible, it is probably irrational for speculators to try and create a famine situation. Certainly it would be very difficult to hide the evidence.

It would certainly be a great deal easier and more profitable to create a famine when there is already a shortage. The cartel would not have to buy nearly as much, and would not have the problem of disposing of the surplus. Other speculators would not act as free riders to the same extent. It would be irrational for speculators to create a famine in a situation where there was a surplus, when they could do it more easily and more profitably by doing it in a shortage year.

The conclusions derive from the specific assumptions of the model, and it would be wise to substitute assumptions from the economy in question before drawing conclusions rather than to


generalize. The factors taken into account do, however, apply in some measure to most economies threatened by famine.

It should always be remembered that whatever the cause of famine, the remedy of importing large quantities of food and distributing it free or on public works schemes will be effective. It is dangerous to rely on other remedies and other diagnoses without the hardest of evidence.


Speculative prices as a result of famine

What are the effects of the high speculative prices and profits that occur when a famine is caused by a shortage? What effect do they have, if any, on the death rate? There is no shortage of evidence that prices are high during a famine, and that some people die because they cannot buy food, but it is not self evident that more people die because there is speculation. Indeed Adam Smith argues that fewer people die, because the high prices forces individuals to ration their consumption over the famine period.

In normal years with no shortage, speculators buy grain with the intention of selling it later in the year at a price that covers storage costs and gives them a profit. This ensures that supplies are available throughout the year. Their estimates of total supply and the optimum amount to release each month are seldom correct, so prices are usually significantly higher or lower than the average in the month or two before the harvest. This speculation is a necessary part of the marketing system. a similar function is carried out by state marketing boards.

If there is a shortage, speculators carry out the same within-season speculation. If they did not, then supplies would be consumed early in the year, leaving nothing for the months before harvest. Typically, speculators are better informed about national crop prospects than farmers, and are in a better bargaining position, so their purchase prices are high, but not as high as the prospective profit would justify.

If there is a first or second degree shortage, then, by definition, there would be no deaths if available supplies were perfectly rationed. The function of rationing is to reduce consumption by some groups, so there is enough available for the poor. Perfect rationing would reduce prices to normal levels. It would also be necessary to ensure that food was given to those unable to buy these prices. Historically, though, even the best schemes have not been able to stop the rich from buying food on the black market, or buying foods that were not covered by the rationing, even when the population was totally committed to rationing because of wartime patriotism and when there was an excellent administrative system. It is questionable whether the standard of rationing system which can be established at short notice is capable of preventing deaths if there is a first or second degree shortage.

a half-perfect rationing scheme can sharply increase the number of deaths. This can happen if it does not reduce consumption by the largest consumers, but rations what is left perfectly. For example, if half the population eats 60% of the food available, and the remaining 40% is split evenly among the rest of the population, they will get 80% of the minimum survival ration, and all will die.

Speculators aiming to maximize profit will sell at a price which will enable a large section of the population to eat as much as usual, with another large proportion eating less than usual, but enough to survive until the next harvest. The remainder will die. There will be more deaths than with perfect rationing, but not necessarily more than with imperfect rationing. The high prices have the function of determining that it is the poor who die rather than the rich. This does not imply that fewer people would die if it could be determined somehow that the rich died instead.

a state marketing board which sold the reduced supply at normal prices would not be likely to perform better than speculators. It is likely that those who buy from the marketing board and sell to the public would take the speculative profit, openly or on the black market. If, however, there was effective price control at normal prices, then rationing by queuing and rationing by influence would determine who got the grain and who died. This might mean that a different set of people died to that when there was speculation, but there is no reason to believe that fewer people would die.

It is possible, of course, to determine a timetable for releasing grain, which, with a discriminating monopoly pricing system, would produce fewer deaths than the profit maximizing by speculators. This is pointless unless a realistic marketing system in which it would work can be envisaged. It is more to the point to import the necessary food to restore the status quo.

In a Third Degree Shortage, there is not enough food for the population even with rationing. Perfect rationing would result in the death of the entire population. Governments tend to underestimate the degree of shortage, so there is a grave danger that they will apply the rationing remedy to a Third Degree Shortage, in the belief that it is only a First or Second Degree Shortage, with terrible consequences. This danger is all the more acute because there is very little difference between the No Shortage situation and a Third Degree Shortage when there is a relatively uniform distribution of consumption, when, for example most of the population are barely above the subsistence level.

Formally, the way to minimize famine deaths is the horrifying expedient of killing off the surplus mouths at the beginning of the famine, and have perfect rationing for the rest. This is not too dissimilar to the classical economists' response to the Irish famine, and it is similar to an army's decision to kill its wounded before embarking on a difficult and dangerous retreat. The decision may be to abandon isolated areas and concentrate on easily accessible, and politically important areas. It is a drastic alterative to massive imports.

Given that this is politically unacceptable, speculation is a politically acceptable alternative. It determines that some people will die, the rich rather than the poor. The state does not have to take the blame for this decision, and it can blame speculators for a famine which may have been caused by Government incompetence. It can soften the political effect by giving small amounts of food to the starving, and giving this the maximum publicity.

a state marketing board could, in principle, charge the same speculative price, but this is unlikely to be politically acceptable. Alternatively, the people who buy from the marketing board could be allowed to make speculative profits. In both cases the price determines who is to starve. Effective price control at normal prices would be disastrous: many or most of the population would not be able to secure enough to survive. a state marketing board might disburse stocks at a different rate from speculators, but the rate will be determined by political pressures, rather than what minimizes the death rates.

The long-term effect of the famine may be worse than the short-term. Many people sell their tools, their roofs, their land or their children to survive and borrow whatever they can. The high mortality of such a destitute population in normal times can accumulate to reach the death toll of a famine in a few years. In principle rationing can reduce this effect, for First Degree and Second Degree Shortages only. a state marketing board could take full speculative profits and use them for relief after the famine, but this is unlikely to be politically acceptable. It is conceivable that price control would have an impact if prices could be held at a level just above the one where the "necessary" number of people die: the highly inelastic demand means that the market price may be well above this.

Do high prices cause deaths?

It is naive to complain about high prices and speculation during a famine unless there are viable alternatives. For First and Second Degree Shortages there is a theoretical alternative, rationing. In practice it may not be possible to introduce a sufficiently good system and there is a danger that a worse situation may be created. There is also a danger that rationing may be introduced when there is in fact a Third Degree Shortage, with disastrous consequences.

It is naive to model alternative ways of distributing grain, alternative timetables for disbursing it and alternative pricing systems, unless there is some way of implementing these systems in a real world marketing system.

The work of the entitlement school repeats at great length their observation that speculative prices are high in a famine and some people die because they cannot pay high prices (e.g. Sen, Devereux, Ravallion) but there is no analysis of the alternatives and no policy implications are drawn. No attempt is made to show that more people die than would die if prices were lower and the same amount of food was available. Shooting the speculators might be satisfying, but it will not solve the problem.

These complaints draw attention from the true conclusion: the only solution is imports.


It has been shown that it is only technically possible for speculators to cause a famine under extremely restrictive conditions which seldom apply. Even when it is technically for them to cause a famine, it will nearly always be irrational for them to do so. While it is possible to create science fiction scenarios where it is both possible and rational, it is certainly impossible in any of the 25 countries I have worked in, or any country I have read about. Unless it can be shown that for a particular economy at a particular time it is both technically possible and rational for speculators to create a famine it should be assumed that the famine is caused by a shortage and that large imports are required. Activities aimed at making speculators disgorge surplus stocks are counter productive.

This means that one of the two claims to novelty of the entitlement approach is incorrect in theory, as well as being contradicted by the facts in the key examples quoted.

The claim of entitlement economists that high prices are a major contribution to excess mortality is also incorrect. High prices select who is to die, the poor rather than the rich. It is the quantity of food available per head that determines how many people will die.


Basu, Dipak "Food Policy and the analysis of famine" Indian Journal of Economics 64 254: 289-301, 1984.

Basu, Dipak "Sen's analysis of famine: a critique" The Journal of Development Studies 22:3 April 1986.

Bengal Administration, Bengal Famine Code, (Revised edition of December 1895) Calcutta. 1897.

Bowbrick, P., "A refutation of Sen's theory of famine", Food Policy. 11(2) 105-124. 1986.

Bowbrick, P., How Sen's Theory Can Cause Famines, Quality Economics, Nottingham This web site October 1998.

Devereux, S. (1993), Theories of Famine, Harvester Wheatsheaf, New York, London.

Dyson, T. and A. Maharatna, "Excess mortality during the Great Bengal Famine: a Re-evaluation" in The Indian Economic and Social History Review, Vol 28, No. 3, 1991.

Dyson, T. (1996) Population and food: global trends and future prospects, Routledge, London and New York.

Dyson, T., "On the Demography of South Asian Famines, Part II" in Population Studies, Vol 45, No. 2, July 1991.

Famine Inquiry Commission Final Report, Madras, Government of India. (1945).

Famine Inquiry Commission, Report on Bengal, New Delhi, Government of India, (1945),

Frere, Sir Bartle On the Impending Bengal Famine: How it will be met and how to prevent future famines in India, London, John Murray, 1874.

Goswami, O., "The Bengal Famine of 1943: Re-examining the Data" in, The Indian Economic and Social History Review, Vol 27, No. 4, 1990.

Hunter, W.W. Famine Aspects of Bengal, Simla 1873

Kumar, B.G. (1990) "Ethiopian Famines 1973-1985; a case study" in J Dreze and A. Sen (eds) The Political Economy of Hunger 2 Oxford, Clarendon Press.

Mahalanobis, P.C., Mukkerjee, R.K., and Ghosh, A. "A sample survey of after effects of Bengal famine of 1943." Sankhya 7(4),337-400. 1946.

Ravallion, M. (1987), Markets and Famines, Clarendon Press, Oxford.

Ravallion, M. "Famines and Economics" Journal of Economic Literature 35 (3) 1205-1242 Sept 1997.

Seaman, J. and J.Holt "Markets and famines in the Third World", Disasters, 4 (3) 283-297, 1980.

Sen, A., Resources, values and development, Blackwell, Oxford, UK, 1984.

Sen, A., "Famines as failures of exchange entitlements", Economic and Political Weekly, Special No, August 1976.

Sen, A., "Starvation and exchange entitlements: a general approach and its application to the Great Bengal Famine", Cambridge Journal of Economics, No 1, pp 33-59, 1977.

Sen, A., "Famine Mortality: a study of the Bengal Famine of 1943", in Hobsbawm et al, Peasants in History: Essays in Memory of Daniel Thorner, Calcutta, Oxford University Press, 1980.

Sen, A., "Famines", World Development, Vol 8, No 9, pp 613-21, September 1980b.

Sen, A., Poverty and Famines, Clarendon Press, Oxford, UK, 1981.

Sen, A., "Ingredients of famine analysis: availability and entitlements", Quarterly Journal of Economics, August 1981 b, pp 433-464.

Smith, Adam, The Wealth of Nations, Everyman, 1910, 1977.

Woodham-Smith, C.V., The Great Hunger: Ireland 1845-1849, Penguin 1991 (Hamish Hamilton 1962)



To famine page

Home page