LJANUARY 2000 THE CHINESE FAMINE OF 1959-61 137 crop failures initiate the chain of effects, the proponents of this approach argue that the best way to understand famines is to look at what happened food availabil Although Sen has amply demonstrated that famine could occur without reduction in per capita food supply, his measures of availability are defined at either national or highly aggregated regional levels, which may not directl refute the propositions of FAD proponents who emphasise local conditions. A formal assessment of the entitlement approach is further obstructed by other data limitations, such as the lack of records on personal property and detailed market commodity prices. More importantly, previous research has not meas ured separately the contributions of food supplies and other entitlement arrangements to a famine. As a result, we still do not know the relative importance of the famine causes o In this paper, we analyse the Chinese famine of 1959-61, where the drop in od availability and different arrangements of rights to food were both important factors. Under the centrally planned regime, China had an effective, urban-biased ration system in which city residents were given legally protecte rights to acquire a certain amount of food. In contrast, compulsory grain procurement quotas were imposed on the farmers. As a result, farmers were entitled only to the residual grain. In years of poor harvest, there was barely enough grain left in the village for the farmers after they fulfilled the quotas During the Great Leap Forward in 1959-61, Chinese agricultural production collapsed because of a sudden institutional change, natural calamities and a series of policy mistakes. The grain output dropped by 15%o in 1959 and reached only about 70% of the 1958 level in 1960 and 1961. Careful studies of the newly released data reveal that this crisis resulted in widespread famines and caused about 28-30 million excess deaths (Peng, 1987 and Ashton et aL. 1984). To analyse this catastrophe, we apply Sens entitlement approach to the centrally planned system. We formulate a framework that is amenable to empirical testing and that simultaneously considers per capita food supply and the right to food as determinants of famine. A panel data set for 28 Chinese provinces for the period 1954-66 is used for the empirical analysis. We use the percentage of rural population and pe capita grain output in a province as proxies for the degree of urban bias and 2 Sen's entitlement approach de food availability decline as a cause famine. What the entitlement approach opposes is to take food availability as the See Osmani(1995) for further discussions on the differences between the FAD approach and the entitlement approach. We owe this clarification referee 3 Due to data limitations, Sen(1981a)used national per capita grain output as the indicator for food ailabil of Ethiopian and Bangladeshi famines. Presumably, per capita food supply region is a more relevant indicator. For the great Bengali famine, Sen used food availability for the district, a more local indicator, as per capita availability measures. However, Bowbric questioned the reliability of the production figure Our focus is on the causation of famine. See Coate and Ravallion (1993) for discussions on insurance arrangements and Dreze and Sen (1989, 1990)for the role of government prevention and relief. Economic Society 2000crop failures initiate the chain of effects, the proponents of this approach argue that the best way to understand famines is to look at what happened to food availability.2 Despite the clarity of views at the theoretical level, the usefulness of the two approaches has not yet been rigorously examined in empirical analysis. Although Sen has amply demonstrated that famine could occur without a reduction in per capita food supply, his measures of availability are de®ned at either national or highly aggregated regional levels, which may not directly refute the propositions of FAD proponents who emphasise local conditions.3 A formal assessment of the entitlement approach is further obstructed by other data limitations, such as the lack of records on personal property and detailed market commodity prices. More importantly, previous research has not meas￾ured separately the contributions of food supplies and other entitlement arrangements to a famine. As a result, we still do not know the relative importance of the famine causes. In this paper, we analyse the Chinese famine of 1959±61, where the drop in food availability and different arrangements of rights to food were both important factors. Under the centrally planned regime, China had an effective, urban-biased ration system in which city residents were given legally protected rights to acquire a certain amount of food. In contrast, compulsory grain procurement quotas were imposed on the farmers. As a result, farmers were entitled only to the residual grain. In years of poor harvest, there was barely enough grain left in the village for the farmers after they ful®lled the quotas. During the Great Leap Forward in 1959±61, Chinese agricultural production collapsed because of a sudden institutional change, natural calamities and a series of policy mistakes. The grain output dropped by 15% in 1959 and reached only about 70% of the 1958 level in 1960 and 1961. Careful studies of the newly released data reveal that this crisis resulted in widespread famines and caused about 23±30 million excess deaths (Peng, 1987 and Ashton et al. 1984). To analyse this catastrophe, we apply Sen's entitlement approach to the centrally planned system. We formulate a framework that is amenable to empirical testing and that simultaneously considers per capita food supply and the right to food as determinants of famine.4 A panel data set for 28 Chinese provinces for the period 1954±66 is used for the empirical analysis. We use the percentage of rural population and per capita grain output in a province as proxies for the degree of urban bias and 2 Sen's entitlement approach does not oppose food availability decline as a cause (hypothesis) for famine. What the entitlement approach opposes is to take food availability as the only cause for famine. See Osmani (1995) for further discussions on the differences between the FAD approach and the entitlement approach. We owe this clari®cation to an anonymous referee. 3 Due to data limitations, Sen (1981a) used national per capita grain output as the indicator for food availability in the study of Ethiopian and Bangladeshi famines. Presumably, per capita food supply in a famine region is a more relevant indicator. For the great Bengali famine, Sen used food availability ®gures for the district, a more local indicator, as per capita availability measures. However, Bowbrick (1986) questioned the reliability of the production ®gures. 4 Our focus is on the causation of famine. See Coate and Ravallion (1993) for discussions on insurance arrangements and Dreze and Sen (1989, 1990) for the role of government in famine prevention and relief. # Royal Economic Society 2000 [ JANUARY 2000] 137 THE CHINESE FAMINE OF 1959±61