Home Contact us Site Map

Cette page existe uniquement en anglais

Home > Activities > Committees >Historical Demography (1997-2001) > Seminar Report

The Demography of Famines: Perspectives from the Past and the Present

Les Treilles, 25-31 May 1999
Organised by the IUSSP Committee on Historical Demography and the 'Fondation Les Treilles'


Most dictionary definitions of ‘famine’ equate it with food scarcity and widespread hunger. They tend to remain silent on the demographic aspects, although the extra mortality caused by famines offers one easy and obvious gauge for ranking famines. By this reckoning, for example, the Great Irish Famine of the 1840s was the greatest in nineteenth-century Europe. By the same token, some of the modern famines highlighted in media accounts are ‘small’ by historical standards. Excess mortality, however, is only one aspect of famine demography. Famines typically reduce births and marriages too, and the migrations that they often give rise to can both increase and reduce the death toll. There are differences between how modern famines kill and how historical famines did so. Modern famines differ too in who they kill; they tend to more class-specific and they are even more likely to target males than females than famines in the past. Moreover, famines often have demographic causes as well as consequences; and the consequences may be long-term as well as short-term.

Les Treilles offered a luxuriously incongruous but highly congenial and productive setting for the conference on The Demography of Famines which took place there in late May 1999. The conference, which was largely funded by the Fondation des Treilles, brought together seventeen specialists in the field. It was very much an interdisciplinary gathering, with demographers, historians, economists and development specialists present. The contributions ranged widely both in chronological and geographical terms, from Japan in the first millennium to Madagascar in the 1980s. This report offers a brief account of all the papers presented: titles and authors’ affiliations are listed below. It is hoped that revised versions of most or all of the fifteen papers presented will be published together in book form in due course.

The conference began with an analysis by Tim Dyson of two famines that hit the central Indian province of Berar in rapid succession in the 1890s. Though Berar was relatively advanced by Indian standards, the underlying causes of famine there were as elsewhere in colonial India. Dyson’s account highlighted how food crises often come in pairs and added an intriguing new twist to the issue of post-famine demographic adjustment. His emphasis was on how ‘bang-bang’ famines and epidemics may well have reinforced each other to restrict population growth in the past. Just because the age structure of famine mortality may sometimes promise hope of some limited degree of demographic and economic recovery in famine’s wake should not obscure the possibility that echoes and chain reactions may sometimes have operated powerfully in the opposite direction as a check on recovery.

Kate Macintyre’s contribution addressed an issue which would recur frequently during our deliberations at Les Treilles, viz. why females seemed to survive famine better than men. Given the lower life expectancy of females during ‘normal’ times in famine-prone environments, this is an interesting and important question. Macintyre set out six possible mechanisms that could explain this mortality advantage, using the case of the Great Irish famine (1846-50) as a case study. The mechanisms are: (1) a spurious reading of poor data, which exacerbate the problem of ‘missing’ women so that their deaths are less likely to be reported than deaths of men; (2) a concurrent decline in fertility which lowers the risk of pregnancy and hence maternal mortality; (3) a biological argument that says women are less likely to die from starvation because of higher body fat levels, lower energy consumption, or possibly higher resistance to disease; (4) differential individual and group survival strategies, in particular in relation to migration and to the acquisition of knowledge of famine foods; (5) the particular political and social structures which may discriminate against women in ‘normal’ times, but act in their favour in times of great stress; and (6) emotional or moral factors associated with differential notions of pride, which may mean that women reach places of assistance (the workhouse or the refugee camp, for instance) in a less worse state than men. Macintyre’s reading of the Irish evidence prompted her to opt tentatively for (6).

Kari J. Pitkänen’s paper on Finland also addressed the issue of sex bias. During the Great Finnish Famine of the 1860s and also during several other periods of excess mortality in Finland in the preceding decades, Pitkänen could find no evidence for consistent sex differentials in mortality, with one exception. Only in 1868, the peak year of famine in a decade of famine or near-famine, did men suffer more than women. A closer examination of the data, however, revealed that even in 1868 sex differentials in mortality increases were limited to certain regions and age categories. There were some small consistent sex differentials in proportional mortality increases, but they reflected sex differentials in baseline (standard) mortality. When absolute increases in famine mortality between the sexes were compared, the differentials largely disappeared. Overall, Pitkänen’s evidence suggested that neither sex had an inherent biological disadvantage in resistance to the particular infectious diseases mainly responsible for the mortality increases during these calamities. Only in 1868 were men in certain age categories and in certain regions more likely to succumb than women. Pitkänen argued that this was mostly likely linked to famine-induced temporary migration, which also tended to be age- and sex-selective.

Two of the contributions dealt with the Great Irish Famine of the 1840s. In their paper, Timothy Guinnane and Cormac ó Gráda offered a preliminary analysis of the role of local institutional or ‘agency’ factors in determining variations in mortality across Irish regions. They highlighted one particular aspect of agency, the management of workhouses under the poor law regime. About one-quarter of the excess mortality of one million occurred in these workhouses. The paper offered examples of both the case study and the comparative approach. Individual case studies yield examples of both good and bad management. Close analysis of workhouse registers and qualitative data offers insight into crucial factors such as the quality of health care in the workhouse and the competence of the guardians. The North Dublin union, the particular focus of part of this study, seems to have been well run during the famine. However, management quality is arguably endogenous, and this suggested the usefulness of a comparative analysis. So the paper also included a statistical analysis of Ireland’s 130 unions in order to determine which unions under- or over-performed, controlling for the economic and locational conditions that faced them.

The paper by Joel Mokyr and Cormac ó Gráda was also about the Irish famine. In the 1840s, there was no civil registration of births, marriages and deaths in Ireland, and the enumeration of emigrants was partial. In these circumstances the well-known tables of death constructed by William Wilde for the 1851 Irish census commissioners are a seductive source. Wilde’s tables are impressive in their detail and in their apparent sophistication; retrospectively recorded deaths are cross-tabulated by age, by sex, by cause, by location and by date. Alas, the data suffer from serious under-reporting. However, the under-reporting was rather systematic, allowing Mokyr and ó Gráda to produce a range of estimates that ‘correct’ for it. The results helped to distinguish between excess mortality due to hunger-induced diseases such as dysentery, and that due to diseases such as typhoid fever, which are less the product of hunger than of the social disruption that usually accompanies famine. They then compared the implications of the outcomes for the role of infectious disease and mortality by age and sex in Ireland and elsewhere.

John Seaman’s paper focused on early warning mechanisms. He described the computer model of rural economy which he had developed in collaboration with FAO/GIEWS, based on quantitative descriptions of rural economy obtained by rapid assessment methods. The model has already been widely applied to prediction and to ‘hindcasts’, and gives good approximations to real events in terms of the use of household entitlements and the severity of food shortage. The system was not designed with the demography of famine in mind, but Seaman felt that the approach, in modified form, might be usefully applied to demographic and other aspects of the study of famine.

Three of the conference contributions dealt with famine in Africa. In a the first of these Michel Garenne and Pierre Cantrelle presented a paper (co-authored with Dominique Waltisperger and Osée Ralijaona) which described a ‘mild’ famine in Antananarivo, the capital city of Madagascar, in the mid-1980s. Vital registration data show that mortality underwent major changes during the 1976-1995 period. Mortality increased slowly and regularly between 1974 and 1984, and then went through a major crisis in 1985 and 1986 and to a lesser extent in 1987-1988. It decreased again and by 1991 was close to the baseline level of 1976-1978. Since 1991 mortality stayed roughly at the same high level. The 1985-1986 crisis seems to be due primarily to a severe food shortage following a big increase in the price of rice, the main staple food. The main causes of death responsible for the increase were malnutrition, diarrhoeal diseases, acute respiratory infections, and in addition for adults, tuberculosis and cardiovascular diseases. Furthermore, deaths due to violence other than motor vehicle accidents also increased during the same period. Garenne and his co-authors estimate the crisis of 1985-6, induced by the removal of a food price subsidy, killed about 16,600 persons, among them 7,600 children, 5,100 adults aged 15-59, and 3,900 older persons aged 60 years or more. Excess mortality was much higher for men (9,400 deaths) than for women (7,200 deaths); the sex ratio of excess mortality was highest for young adults aged 15-59, and lower among children aged 0-14 years, and among older persons aged 60+ years. A malaria crisis peaking in 1987-89 exacerbated the food crisis of 1985-6, and was responsible for a further 1,400 deaths.

Christian Thibon gave a highly informative account, based on a rich combination of oral and written sources, of famine in the central African country of Burundi over the past century or so. The picture is a complex one, in which shifting economic conditions influenced the inter-tribal equilibrium between Hutu and Tutsi. Its broad outlines in demographic terms are cataclysmic famine in the late pre-colonial era, followed by severe local dearths in the 1920s, and the last natural famine, referred to as ‘Manori’, in 1943/44. There followed a politically induced famine in 1972 which resulted in huge mortality (possibly 0.3 million).

Markos Ezra’s paper offered an analysis of the demographic consequence of environmental stress. using primary survey data carried out to look at the demographic consequences of drought and famine in the drought-prone areas of northern Ethiopia, this study argues that growing environmental stress and persisting food insecurity have stimulated changes in the demographic behaviours and attitudes of farming communities in this region. Significant increase in acceptance rates of family planning services; changing attitudes towards early age at marriage and towards having a large number of children; actual reduction in fertility; migration (particularly of the youth) out of the communities; and the tendency by many farmers to be involved in non-farm income generating activities as well as in non-farm employment sectors are among the demographic responses observed in the communities investigated.

Serguei Adamets gave an analysis of the geography of famine in nineteenth-century Russia. until mid-century few Russian regions were immune, but from then on famine became an increasingly regional or localised phenomenon. By century’s end, the regions at risk were limited to the Volga region, the northern Caucasus and the southern ukraine. More effective control of epidemics, commercialisation and more effective PUBLIC action all played a role. All nineteenth-century famines fade into insignificance when compared to those of the early 1920s and early 1930s in the Soviet union.

Stephen Wheatcroft’s contribution was enriched by the increasing availability in Russia of demographic data from the Soviet period. He combined some of his own earlier data on the famine of 1918-22 with new detailed regional data on the 1929-33 period. His focus was on the demographic situation in four major cities and the regions around them in both the 1918-22 famine period and the 1929-33 period. Some detailed information on deaths by date, age, sex and cause are now becoming available for a limited number of areas, and Wheatcroft discussed to what extent these partial patterns would apply to the country as a whole.

The analysis of the demography of famine is often inhibited by poor data. This is certainly the case for China in the 1950s and 1960s. Barbara Sands presented some of her findings on China’s Great Leap Forward famine. The conventional wisdom conveys this as a man-made disaster where misguided economic policies precipitated widespread famine and world record-breaking population losses. Sands reconstructed regional population and grain availability data to find more complex patterns than those suggested by classic famine. While allowing for considerable excess mortality in this period, she suggested that portions of it were due to the influenza pandemic of 1957. This prompted an alternative explanation of the Great Leap Forward famine. Sands concluded that China’s post-1949 economy was less centrally controlled and more vulnerable to natural disaster than previously suggested.

The demography of Japanese famines is also rather poorly documented. Osamu Saito‘s paper relied therefore on non-standard sources for its insights. It provided an analysis, based on chronological source books or annals, of famine frequencies in three different time periods of Japanese history (ancient, early modern and modern). Then, by combining such data and climatological data, Saito offered a fascinating account of the role of climate in causing famine before and after c. 1600. Finally, he re-examined the role of famines as demographic correctives in Japan. Saito noted that the decline in famine incidence in the nineteenth century produced not just a decline in mortality but also a rise in fertility. This pointed to a significant role for famine as a brake on Japanese in previous decades.

Arup Maharatna used data on death registration and the price of rice to analyse four major famines in India between the 1870s and the 1900s. In all cases soaring food prices augured in higher mortality with a lag of a few months. Though the seasonal character of excess mortality was in part determined by climatic and environmental factors, Maharatna insisted that the outbreak of epidemics, in particular malaria epidemics, in the wake of drought and famine, could not but be fundamentally attributed to mass starvation and acute malnutrition.

Finally, Violetta Hionidou described a ‘small’ but very well documented famine on the Aegean island of Syros in 1941-2. The famine resulted from the occupation of Greece during World War 2, and resulted in the deaths of about one-tenth of the island’s population of twenty-five thousand inhabitants. Medical records suggest that these deaths were overwhelmingly due to literal starvation; infectious disease played virtually no role. This outcome highlights a fundamental difference between historical famines and war-related famines in developed or semi-developed economies in the twentieth century. Hionidou’s data also show that infant and child mortality were relatively low, and that men were much more likely to die than women. Bearing in mind the findings of Pitkänen, Macintyre and Garenne et al. regarding female advantage, the greater vulnerability of the men of Syros in a context where infectious diseases were absent seemed to lend support to a physiological interpretation.

CORMAC ó GRáDA (university College, Dublin)

IUSSP, 3-5 rue Nicolas, F-75980 Paris cedex 20, France
Tel  +33 1 56 06 21 73 - Fax +33 1 56 06 22 04 - contact us