Contents - Previous - Next

This is the old United Nations University website. Visit the new site at

Kinds of data

Although the sources of entitlement failure are well understood in theory, identifying which households are suffering - or even the types of household or number of households that suffer - from food poverty is more difficult. Calculations of the extent of food poverty rely on estimates of household food availability relative to need and measures of absolute poverty.

Estimated numbers in absolute food poverty

The FAO and the World Bank regularly publish estimates of the numbers of chronically undernourished people. Their findings are based on national food availability and information about its distribution across households. Many issues related to data collection and comparability were described in chapter 2 of this volume. The implications of the slightly different methodologies used by the two institutions are described below in the section on trends. Importantly, both institutions base estimates of food poverty on the distribution of food and income rather than simply per capita supply versus availability or consumption versus expenditure. This attention to household command over food results in a picture of hunger that differs greatly from the regional food-availability estimates presented in chapter 3. Distribution-entitlement failures matter for hunger. Table 4.1 shows the numbers and percentages of people suffering from food poverty.

Table 4.1 Proportions and numbers of chronically underfed by world region

Underfed Year Region
Sub- Saharan Africa Near East and North Africa Middle America South America South Asia East Asia China All LDCsa
Proportion (%) 1970 35 23 24 17 34 35 46 36
1975 37 17 20 15 34 32 40 33
1980 36 10 15 12 30 22 22 26
1990 37 5 14 13 24 17 16 20
Absolute numbers (millions) 1990 1970 94 32 21 32 255 101 406 942
1975 112 26 21 32 289 101 395 976
1980 128 15 18 29 285 78 290 846
1990 175 12 20 38 277 74 189 786

Source: Uvin (1994) as taken from ACC/SCN (1992:105). a LDCs: less-developed countries.

Although income data are notoriously difficult to collect, they are still less complicated than food-intake data; this is one of the reasons that estimates of numbers in absolute poverty are often used to estimate the numbers that are hungry. There is also a strong theoretical basis for this method: those who are among the poorest are also the least likely to be able to feed themselves and their families.

Despite the logical simplicity of this approach, non-market sources of entitlement, coupled with measurement error on incomes, could cause large discrepancies between the numbers in absolute poverty and numbers of households that are food poor. Fortunately, this appears not to be the case generally. FAO estimates of the numbers who are hungry correspond to estimates of numbers in absolute poverty, both in magnitude and in their direction of change over time. Changes in food poverty are estimated fairly accurately from changes in prevalence of absolute poverty. This generalization holds true across all areas of the developing world except East Asia, where the incidence of hunger is much higher than the incidence of poverty and where hunger declined between 1980 and 1990 even though the numbers in poverty did not change (Uvin 1994).

Trends in global food poverty

Worldwide trends in food poverty have been downward, despite rapid population growth in most of the developing world. Such trends are based on the most recent methodologies. Over the period from 1970 to 1990, the FAO adjusted several times its methodology for calculating the number of malnourished people, but the estimates were most commonly based on defining minimal food intake at 1.4 times the basal metabolic rate (BMR), or 40 per cent more calories than basal metabolic functions require. Intake of less than 1.2 BMR is considered inadequate, even for sedentary people, because even the most basic processes such as digestion require additional calories; 1.4 BMR allows for only minimal physical activity (WHO 1985). When the FAO recalculated estimates of hunger from 1970 to 1990 using a standard methodology, it selected a cut-off of 1.54 BMR, which allows for light physical activity.

Using this higher and more realistic cut-off produced hunger estimates for the recent past that were considerably higher than those reported previously. The trend in food poverty that emerged from the data was generally more positive. As Uvin (1994) summarized:

Until these data were published, it was commonly assumed that the proportion of hungry people in the world had declined slowly but consistently over the last decades, but that, as a result of population growth, the absolute number of the food poor had continued to grow. According to the new data, the absolute number of the food poor in the [developing] world has been declining since 1975, from 976 to 786 million persons in 1990. Food poverty has also been declining despite increasing numbers of nations which are not self-sufficient in food and the large number which are food short. Clearly, food shortage has not been an overriding determinant of food poverty.

Geographic overview by region

Nevertheless, trends in food poverty are more discouraging in regions where there is also food shortage. It would be surprising if this were not the case, since, if there is food shortage, at least some households must experience food poverty. Table 4.1 shows the trends in food poverty by region, using the same methodology for each time period.

The only region that has not experienced decline in the proportion of malnourished people - sub-Saharan Africa - is also a region where population growth was extremely high throughout the time period in question. The absolute numbers of malnourished people almost doubled between 1970 and 1990.

This is in contrast to South Asia, the other region for which dietary energy supply (DES) was insufficient in 1990 to provide everyone in the region with basic energy requirements (see table 3.2, p. 58). The absolute numbers in food poverty in South Asia have changed relatively little: 10 per cent fewer people in South Asia were in food poverty in 1990 than had been in 1970. Thus, South Asia has not experienced the declines in the absolute numbers in food poverty that most of the other regions have. This may be partly due to huge surges in basic grain production that have largely eliminated food shortage. But food poverty has not increased, even where shortage persists. This suggests there are many determinants of household access to food, other than its availability within the region.

Similar trends in Latin America support this more complex view: percentages in food poverty in Middle and South America from 1970 to 1990 declined a little, but the absolute numbers increased slightly. The trends occurred although the Latin American region as a whole is not food short. But structural-adjustment programmes implemented throughout the region during this era were one of the factors reducing access to food by poor households. Distribution-entitlement failures are most severe in Latin America and in the Near East and North Africa.

Overall, were China not included in the totals for developing regions, the trend in food poverty would not be positive. There were 156 million fewer chronically undernourished people in the developing world in 1990 than there had been in 1970; there were 217 million fewer in China. Sub-Saharan Africa is too sparsely populated to have the kind of impact on world totals that China or even India have. Although it is still accurate to summarize the trends in proportions of chronically undernourished as positive (sub-Saharan Africa was the only region to experience an increase, and this increase was only 2 per cent), the trends in absolute numbers of chronically undernourished are dominated by China and would not be positive if that nation were excluded.

Comparisons across population subgroups

In theory, the available data on food distribution across households could be used to determine which households are the most likely to suffer from food poverty. In practice, national-level data infrequently contain enough information about food distribution and/or information about the social and economic characteristics of the households to link this information. What we know about the types of households that are likely to suffer from food poverty comes primarily from small-scale surveys and community studies. It is very difficult to generalize from these about which households are hungry on a global or even regional basis, but they still provide important clues about the factors that increase hunger vulnerability.

Rural/urban differences

Important differences exist between rural and urban areas in the incidence of absolute poverty and food poverty, but the distinctions between these areas are not as clear as might be expected. Definitions of what constitutes an urban area are often left to the discretion of statistical offices within developing countries. Even employing fairly standard definitions, such as agglomerations of over 5,000 people, the physical size of the sustaining area is somewhat arbitrary. Some areas officially classified as urban have substantial proportions of their populations engaged in agriculture and may have more in common with rural villages than with central cities; this is particularly common in West Africa. Furthermore, circulation of families and individuals between rural and urban areas creates a group of people who cannot be unambiguously classified according to residence.

Although circular migrants' access to food has not been as intensively studied as other indicators of their well-being, such as infant mortality or adult literacy rates, those with urban exposure tend to be better off than their rural counterparts. This is one piece of the larger generalization that living in urban areas contributes to the health and well-being of populations. Health-care facilities are typically easier to reach for urban residents, who are also much more likely to have access to potable water, flush toilets, and refrigerators. Although access is unequal in urban areas, such amenities are entirely absent in many rural communities. Urban areas also (almost by definition) have better infrastructure for market and transport. The concentration of population also ensures more effective demand for food. All of these factors contribute to a picture of urban advantage for child survival, health, and access to food. For instance, when anthropometric measurements of young children are compared in developing countries, urban children frequently show stronger growth patterns than rural children (e.g. Bryceson 1989; Forster and Handelman 1985; Maher 1981). Additionally, larger proportions of the total (adult and child) population in rural areas have been found to be undernourished in many areas (e.g. Hossain 1987).

But food-consumption and expenditure surveys yield inconsistent results: some show that urbanities command better access to food (e.g. Uyanga 1979), while others document better access in rural areas (Walker and Ryan 1990). Throughout Latin America, more urbanized countries have greater food availability per capita, but data on the adequacy of energy intake between urban and rural dwellers still vary greatly (Uauy et al. 1984). Although food-consumption surveys often omit food consumed away from home - an occurrence that is somewhat likelier in urban areas - the failure of intake studies to show a consistent urban advantage highlights the possibility that conclusions based on growth outcomes may be heavily influenced by non-dietary factors such as sanitation, access to piped drinking water, and health services. Activity levels (and therefore caloric needs) may also be far lower for those engaged in urban occupations. In sum, the evidence of urban advantage in nutrition and its relationship to food poverty is mixed.

Within-rural variation

Not all rural households are equally dependent on agriculture for their livelihoods. Households that are involved in a diverse range of economic activities may be less vulnerable to entitlement failure than those with a single or even a primary - source of entitlement. The following sections consider hunger vulnerability among rural groups with varying food strategies and entitlements.

LAND TENURE. Land ownership improves household food security in a number of ways. Owning any land gives households some measure of control over the means of production; owners can decide whether to consume or sell proceeds. It is especially important where markets are underdeveloped (Tschirley and Weber 1994). Owning land gives a guaranteed source of rural employment, although it does not guarantee that the source will be adequate to meet family food needs. Those who rely on cultivating their own land are less vulnerable to changes in rural wages but are still vulnerable to variations in rainfall, in costs of agricultural inputs, and in other forces affecting their productivity.

One of the most consistent findings from the literature on land tenure and nutritional status is that rural households that own any land are better off than those that do not (Alderman and von Braun 1984; Forster and Handelman 1985; Lipton and Longhurst 1989; Melville 1988; Victora et al. 1986; Walker and Ryan 1990). Land ownership serves as a proxy for household wealth, and therefore its relationship with favourable nutritional status does prove that land enhances food security. However, there is also a considerable body of evidence indicating that landed households are less vulnerable to hunger, in cases of production shortfalls, than are rural labourers (Sen 1981a).

However, absolute size of holding is obviously important. What absolute size is required for food security will depend on local ecological, political, and economic conditions; on household needs; and on other sources of income. In areas where landholding has become fragmented, typical holdings are likely to be inadequate. A study of six poor villages in Bangladesh concluded that at least half an acre was needed to provide basic needs for an average family, but very few families had access to that much land (Hossain 1987). Similarly, Griffin (1976) estimated that 90 per cent of the landholdings in Guatemala were too small for subsistence. This pattern is typical in Latin America, where much land is held as large estates (Sheehan 1987).

A number of studies have shown that households that own more land have lower rates of child malnutrition (DeWalt et al. 1987; Fleuret and Fleuret 1980; Gopalan 1987; Grewal et al. 1973; Victora and Vaughan 1985), yet it is difficult to conclude from this evidence that larger holdings increase household food security because plot size may simply be a proxy for wealth, and the definitions of small and large holdings employed vary widely in these studies. Furthermore, other studies show holding size to be uncorrelated with nutritional status (see reviews in Melville 1988 and Victora et al. 1986), and childhood malnutrition can occur independently of household food poverty. Owning more land is most likely to protect against food poverty where holdings are large enough to accommodate a range of production strategies and therefore decrease risk.

Rural households that do not own enough land to meet their needs may gain access to additional land as hired labour or through tenant or sharecropping arrangements. The type of access to unowned land influences household food security, and sharecroppers and tenants both generally have more control over food than hired labourers, who are less likely to have customary patronage protection to fall back on in case of crop failure (Sen 1981b). Secure tenant farmers in India own their own draught animals (Walker and Ryan 1990) and have access to other resources. Those who are better off to start with are able to benefit more from arrangements that give them access to land that they do not own. However, those engaged in wage labour may suffer fewer income fluctuations because they are also likely to be engaged in non-agricultural employment (Walker and Ryan 1990). In addition, sharecropping arrangements are shrinking as landowners adopt modern agricultural inputs (Alavi 1973).

Overall, land ownership can be expected to matter most where access to land is most heavily governed by ownership. This rather obvious conclusion is sometimes overlooked in the generalizations drawn from studies of landholding in various settings. De Waal (1989) showed that, in the Sudan, pastoralists who had access to a great deal of land were better protected in times of drought than their more wealthy agriculturist neighbours. They were able to change herd compositions and they knew more about wild foods than farmers did. Landlessness, and insecure access to land, threaten food security most in rural communities in which productive assets are privately controlled. Although restricted access to land is still far less of a problem in Africa than in other developing areas, landlessness is growing in Africa as a result of privatization (Jazairy et al. 1992), and other forces are increasing landlessness in the Near East and Latin America.

Additionally, the focus on land tenure should not obscure the importance of livestock in family subsistence strategies. Not all farm families own livestock, and many landless households nevertheless own draught animals, milch animals, or smaller livestock. These can be another store of wealth and also an important source of food. During the 1983 drought in Zimbabwe, households that owned livestock were one-third less likely to require food aid than households of the same income level that did not (Gittinger et al. 1990). Being able to sell animals in times of severe drought can make a critical difference for households whose income would be otherwise inadequate to secure entitlement to food. Although selling livestock has the potential to increase vulnerability to hunger in the future, in some settings buyers protect animals by removing them from regions too dry to support them and then reselling them back into these regions later (Rahmato 1988).

CASH-CROPPING. Cash-cropping has long been a suspected contributor to undernutrition. Cash crops are alleged to have the following effects:

1. They cause greater environmental degradation than food crops (ACC/SCN 1989; Schofield 1975; Van Esterik 1984).

2. They increase the maldistribution of land (George 1977; Melville 1988; Taussig 1978).

3. They take better lands away from crops (Güsten 1968; Wright T. C. 1985).

4. They divert labour from food production (Forster and Handelman 1985; Hamilton et al. 1984; Schofield 1979) and preparation (Schofield 1975).

5. They increase the energy that must be expended to obtain the same nutrients (Fleuret and Fleuret 1980).

6. They render cash-crop producers vulnerable to world market prices (Van Esterik 1984).

7. They increase debt through elevated use of credit (Van Esterik 1984).

8. They divert income from food (Brandtzaeg 1982; Rogers and Youssef 1988; Schofield 1979).

9. They increase use of foods with low nutrient value (Suárez et al. 1987).

10. They lead to irregular intakes because of irregular income (Fleuret and Fleuret 1980; Kaiser and Dewey 1991; Trenchard 1987; von Braun et al. 1994).

11. They decrease dietary diversity (Dewey 1985; Schofield 1979; Taussig 1978).

The first three of these reasons refer more to global than household impacts. The use of better lands for marketed crops reduces global food security only if the marketed crops are products such as sisal and tobacco rather than food. But even factors that primarily affect global production and productive potential influence food distribution across households.

Commercial agriculture has the potential to increase household vulnerability to hunger, because access to food then depends on markets and prices rather than production. But extensive literature reviews have not concluded that cash-cropping is uniformly bad for nutrition (Mebrahtu et al. 1995), and the Latin American evidence points to a mix of subsistence and cash-cropping as the best nutritional strategy (Messer 1984). Fairly compelling reasons why cash-cropping could improve nutritional status include the provision of producers with revenue to improve agricultural productivity and the increase in employment opportunities for the rural landless.

Mixed findings on the effects of cash-cropping on household food security are partly explained by the substantial variation in its effects according to local conditions. Bryceson (1989) argued that cash-cropping tends to be somewhat less risky in Africa because better land availability allows cash-croppers to return to subsistence cultivation when market terms of trade deteriorate. Similarly, cash-cropping is less risky where conditions allow farmers to combine food and cash crops, as where coffee and bananas are intercropped (Bryceson 1989). As in Latin America, cash-cropping families that have the most secure nutritional status are those that have diversified their production, particularly those who also grow food for their own consumption (FAO 1986; Melville 1988).

But crop diversification, particularly in Africa, may be constrained by labour availability. Where male labour is diverted into cash-crop production, women are left with the responsibilities of food production, food preparation, and child care. Nutrition suffers where any of these three responsibilities is neglected or compromised. Household access to food may also decline where women's heavier participation in subsistence agriculture limits the time they have available for remunerated activities. In much of sub-Saharan Africa, women are responsible for providing food for their families and men's earnings are not spent on food (Bryson 1981; Desai 1992; Guyer 1984; Lloyd and Brandon 1991; Seidman 1984). A shift to cash-cropping may mean that increased income comes at the expense of increased food expenditure, because the proceeds from cash crops are controlled by men (Bukh 1979; FAO 1987; Trenchard 1987).

Although these factors explain some of the variable findings on the impact of cash-cropping on nutritional status, some of the ambiguity arises from focusing on the nutritional outcomes for different groups. The effects of cash-cropping on those involved in the process differ greatly from those on others in the community.

Where cash crops affect food supply, their impact depends on how diets resulting from purchasing food with the revenues from cash crops compare with diets based on production. Increased production, even of food crops that are then sold for cash, can threaten food supply and nutrition, as where indigenous cultivators of highland Ecuador trade legumes for flour and lard. In many cases, agricultural commodities produced for the market are not consumed locally (Schofield 1979); local diets have been shown to be increasingly dependent on staple grains and therefore lacking in micronutrients (Abdullah 1988). But, in some areas, commercial production is associated with more diverse local diets (e.g. Glover 1994) that are protective against micronutrient deficiencies. Commercialization that brings with it increased investment in agricultural research and development, extension, and improved marketing increase production incentives and total food supply and potentially reduce food poverty indirectly by reducing food shortage.

Nevertheless, cash-cropping has greater potential to influence food poverty through food prices, labour prices, and land distribution. Landowning households that produce a diverse set of crops for the market are less vulnerable to fluctuating market prices and failures of particular crops. Continuing subsistence production protects cash-cropping households with a source of entitlement to food besides income. Cash-cropping tends to be associated with good nutritional outcomes in these households.

The effects of cash-cropping on non-participating households are more complicated. Rural households with smallholdings may continue to specialize in subsistence production because they cannot afford the risk or the investment necessary to compete with larger-scale commercial farmers. If productivity in the rest of the agricultural sector drives prices down, these households will receive less income from whatever surplus they sell. This increases their costs for food purchases and the amounts they need to sell in order to meet their other cash needs, including taxes. Wealthier households especially those with larger landholdings - are the most likely to benefit from subsidies and credit programmes. Cash-cropping tends to augment the nutritional status of participating households, in part because those who participate are already in the best positions to diversify production and therefore to minimize the inherent risks.

The effects of cash-cropping on landless households are mixed. Cash-cropping often increases the demand for agricultural labour and thereby increases the incomes of rural landless labourers, but the availability and financial returns from expanding wage work depends on the labour supply competing for the new jobs. Where smallholders and outsiders enter in search of employment, cash-cropping schemes can lower wages.

Where cash-cropping drives down food prices, urban dwellers can be expected to enjoy improved food security, if earnings remain stable. Where cash-cropping absorbs labour and does not provide an impetus for rural-to-urban migration, the urban poor can benefit from less competition for jobs. But where, instead, capital-intensive cash-cropping displaces rural wage labour, sending more to the city in search of work, commercialization reduces wages and food security for the urban poor.

Measurements of the impacts of large-scale commercial changes in agriculture on particular population groups tend to be purely academic. Von Braun (1994) has argued that subsistence agriculture is ceasing to be a viable alternative for individual producers and for the global economy. Although analysts tend to assess whether farmers would have been better off remaining in subsistence production, the more important policy question is how to protect particular groups during the transition. Social safety nets are very important in protecting groups such as tenants and smallholders, who lose customary sources of entitlement during large-scale commercial transformations and may not yet have acquired new modes of livelihood.

SEASONALITY. A pattern of fluctuating rural intakes across agricultural seasons has been documented in a wide variety of countries and has been well known for decades (Chambers et al. 1981; Schofield 1974). The basic cause for this pattern is simply lower food availability during "hungry" seasons before harvesting reverses the situation into one of plenty. Even if food intake were constant across seasons, there would still be lower intake relative to need during harvesting because energy expenditure is high during peak agricultural work times.

Although seasonality in food intake is directly related to seasonality in agriculture, there are still important questions with respect to the role of seasonality in influencing food poverty. Are there some rural households that have less-dramatic fluctuations in intake? Are there households that, despite fluctuations, still have adequate intake at every point in the year? The answers to these questions are not known on a global scale, but there is some indication that even wealthy landholders do not have consistent intakes across agricultural seasons (Abdullah 1988; Walker and Ryan 1990). Own production does not safeguard against seasonal hunger as well as it does against more chronic hunger.

Entitlement to food on a year-round basis depends on total production and access to adequate storage facilities, and on effective demand to buy extra food. Markets are a cost-effective solution to seasonal food availability where there is sufficient purchasing power. Money stores better than food, but access to markets in remote rural areas is limited by distance and undeveloped infrastructure; food prices can rise quickly during hungry seasons when demand for purchased food is high. As indicated above, even where markets and commercialized production provide greater quantities of cheaper staple foods, they do not necessarily offer optimal year-round nutrition. Greater consumption of rice throughout the year can be accompanied by greater seasonal vitamin and mineral deprivation; and the very poor may still find they suffer chronic, as opposed to seasonal, hunger.

Within-urban variation

Urban nutritional variation tends to be related to employment, although health and dietary choices are other factors. Jobs in the formal sector tend to pay higher wages than productive activities in the informal economy. Formal-sector jobs also provide more stable income, since wages do not fluctuate in the same way as self-employment earnings. But many who migrate to urban areas fail to find formal-sector employment; they return to the countryside, or continue to try to find a job while being supported by their families, or engage in informal activities, or employ some combination of these strategies.

When the wage and income disparities between rural and urban areas are minimal, the incentives to move and remain underemployed in urban areas diminishes, but urban underemployment and demand for government services remain a persistent problem in most countries in the developing world. Rural development is the most promising method to limit migration and urban blight, since it changes the economic opportunity structure and improves access to social services, rather than trying to constrain rational responses to existing structures. But even countries facing urban crowding have failed to invest the high direct costs of developing rural infrastructure and jobs, in part for lack of funds and in part out of a desire to maintain low urban wages to promote industry. Industrialists provide a far stronger power base for politicians than do the rural poor. Although long-run strategies for promoting integrated national development are desirable for a number of reasons, the poor and growing urban class that is vulnerable to hunger also demands short-run attention. The urban lower class is not limited to those living in squatter settlements or those who have recently migrated to urban areas; however, both because these two groups are easily defined and because there are strong reasons for believing that they might be disadvantaged, their hunger status has been fairly widely studied.

SQUATTER SETTLEMENTS. Consumption surveys have shown, not surprisingly, that urban squatters consume less energy and have a greater prevalence of vitamin deficiencies than any other urban class (Gopalan 1987; Rao 1987). In the Philippines, the average family wage among urban squatters covered only 87 per cent of the cost of a minimum basic diet (Eviota 1986).

Squatters living in large cities sometimes have an advantage, in that more government subsidies are available to them (Alderman and von Braun 1984). However, benefits diminish with distance: those furthest from the urban centre tend to have the highest rates of malnutrition (Wright E. W. 1985). The perpetuation of social inequality that leads to food poverty is difficult to overcome in situations like this. Those living on the urban fringes are usually less economically productive, as they spend more of their time reaching markets and jobs. In addition, squatters may be vulnerable to eviction, or to cutoff of water, electricity, or other services. Generally substandard housing, lack of sanitation, and denial of entitlements to social welfare programmes including free or low-cost health services and access to subsidized food - all these conditions perpetuate social inequality and hunger.

MIGRANTS. Nutritional outcomes may vary greatly between recent and established migrants; they may vary greatly between circular and permanent migrants; and they may vary greatly between visitors and those intending to stay. Migrants (urban dwellers who were not urban born) as a group appear to be better off than those living in squatter settlements, even though some migrants may be urban squatters. Migration is rarely undertaken by individuals in isolation. Many stay and eat with urban contacts while they are looking for employment (Skeldon 1990). These urban relations also help them to find jobs, sometimes locating employment for them even before they leave their rural homes (Fawcett et al. 1984). Money and/or food sent to urban migrants from rural relatives also helps support the migrant's search for a formal-sector job (Fuller et al. 1990). Because migration often takes significant resources to launch, migrants also sometimes start off at an economic advantage relative to the urban poor: they may represent the rural wealthy rather than the rural or urban poor. In sum, as a group they are not at as high a risk for food insecurity as many more-established, but poor, urban dwellers. Definitional problems, coupled with the fact that migrant intentions are not always clear (and not always followed through on when clear), make studying food intake among migrants particularly problematic.

Contents - Previous - Next