Monitoring access to food and household food security1
1 This article is adapted from Approaches to monitoring access to food and household food security, FAO Committee on World Food Security, 17th Session, Rome, 23-27 March 1992.
Dr Ramesh P. Sharma is an agricultural economist in the Food Security Service of the FAO Commodities and Trade Division.
The objective of this paper is to identify indicators that could be used for monitoring household-level access to food on a regular and sustained basis and with global coverage. The need for improved reporting on household access to food stems from a widely recognized observation that world food security is becoming less a problem of global supplies, overall stability and global stock levels as such, and more a problem of inadequate access to food supplies for vulnerable groups within a country, caused inter alia by lack of purchasing power (FAO, 1991). While the incidence of food insecurity at the household level is known to be severe in many low-income countries regardless of whether they have food deficits or food surpluses, the status and trends of the problem are not routinely monitored because of lack of suitable indicators. In this context, some feasible approaches to monitoring household access to food are reviewed and indicators appropriate for this purpose are identified.
The objective limits the range of indicators that will be appropriate. If the monitoring framework is to have global coverage, a major consideration in the identification of the indicators is the availability of statistics that are comparable and uniform across countries. Second, in order for the monitoring activity to be sustained over time, these data need to be available on a regular basis, for example annually. Finally, the precision of the indicators must obviously be taken into account as well.
CONCEPT OF HOUSEHOLD FOOD SECURITY
In 1983 FAO defined the goal for world food security; "to ensure that all people at all times have both physical and economic access to the basic food they need" (FAO, 1983). While food security for individuals is the main objective, there are also important household, national and global dimensions of food security. Adequacy of food at the global level and an effective trading system are required for ensuring adequate supplies for food-deficit countries. Whether such countries have access to required amounts of food depends upon their ability to import food commercially. Moreover, national-level adequacy does not ensure that all households are food secure. Household-level food security is determined by both physical access to food and adequate purchasing power. While access to adequate food at the household level is needed to satisfy nutrition levels for all members of a household, nutrition security also depends on non-food factors such as satisfactory health and hygiene conditions and social practices. Thus household food security is one but not the only necessary condition for achieving the overall nutritional well-being of individuals.
The scope of this paper is limited to monitoring and assessment of household food security, which in keeping within the above framework has been defined as follows: a household is food secure when it has both physical and economic access to adequate food for all its members and when it is not at undue risk of losing such access. This definition, while broad enough to embrace the essential components of household food security, namely physical access, economic access and sustainability of access, provides a well-focused scope for discussing monitoring and assessment of household food security. Indicators for each of the three components are discussed.
INDICATORS AND MEASUREMENT ISSUES
Monitoring physical access to food
To ensure food access, an adequate amount of food must be within the physical reach of vulnerable households, whether through their own production or through the market. To meet this condition, a country needs to have an adequate supply of food, which will include imports when domestic production is inadequate. When a country's total food supplies are below some statistical aggregate based on normal food requirements, the food needs of every person in the country cannot be satisfied, even if the food that is available is evenly distributed. In addition, food must be physically accessible to all households. Food should be continuously available in the local markets where households do not produce necessary amounts of food themselves.
Nationally, adequacy of food can be monitored on the basis of total supply relative to total requirements. Ideally, per caput food availability would be a good indicator of adequacy, but statistics on changes in stock levels are often imprecise. In lieu of per caput availability, per caput food production and per caput imports often provide good proxy indicators of adequacy. Although domestic production is the major component of food supply in most countries, the level of imports is a key determinant of adequacy at the aggregate level for many food-deficit countries. The ability of these countries to secure sufficient amounts of food is largely dependent upon their commercial import capacity, i.e. the volume of basic foods that the country can afford to import without undue economic dislocation. Such capacity is assessed by an objective analysis of major economic and financial parameters. The most common determinants of a country's import capacity are total export earnings, foreign exchange reserves, the value of non-food import necessities and debt service obligations.
Apart from requiring adequacy, ensuring access also requires that food be physically accessible to all households. Common obstacles to physical access are war, civil strife, poor infrastructure, inadequate logistics for food distribution and market imperfections. Such problems are more likely to exist in regions characterized by difficult terrain and remoteness.
Monitoring of stock positions over time and across major regions of a country can identify abnormalities in the flow of basic foods, especially to remote areas. Some countries, such as the Philippines, have undertaken more quantitative monitoring of decentralized stocks, involving monthly estimates of food stocks held by households, commercial traders and the public sector. Stock surveys annually cover about 14 000 sample households, 10 300 sample retailers and wholesalers and 3 100 sample commercial warehouses. Since this approach is difficult and expensive (especially monitoring household food stocks), it has not been adopted on a large scale by many countries.
An alternative to direct monitoring of decentralized stock positions and flows of food is an indirect approach focusing on price differences across markets. Normally, differences in price levels reflect real costs of transferring food from one market to another. Under this approach, price information for a selected normal period is used as a benchmark against which to assess changes in price differences in the subsequent periods. An increase in the price differential relative to the base period is indicative of a supply problem. Thus an index based on spatial price differentials, corrected for inflation, may be appropriate. Data are not available to measure this index at the international level. However, disaggregated price data should be obtainable for most developing countries. The selection of specific foods for the index will depend upon their importance in the local diet.
Monitoring economic access to food
Household food security can be ensured only when the capability to acquire food exists. A household acquires access to food through its own production, income-generating activities (i.e. wage employment or trade), ownership of assets and transfers from sources external to it. While the relative importance of these sources can vary markedly, a household's command over the total of these resources must be adequate to acquire enough food on a continual basis. This total command may be called a household entitlement to food (Sen, 1981). A food-insecure household may lose access to regular food and even face famine when one or more of the components of its entitlement breaks down.
The concept of entitlement includes a set of endowments of factors of production as well as the possibilities of exchange. Transfers from the state and gifts are included. Though the concept is intellectually powerful for assessing economic access to food, it needs simplification for practical monitoring work.
It can be said that household food security exists when there is adequate effective demand for food. An operational measure of short-term effective demand which can be used to monitor household economic access to food is real income, defined by the variables nominal income and prices. Public transfers and subsidies also strengthen a household's effective demand.
Nominal income. Four approaches are used to assess household-level incomes: household income surveys, use of data on per caput gross domestic product (GDP) and income distribution, use of wage data and analysis of changes in relevant production sectors. Each has advantages and disadvantages, and the approaches can be considered, singly or as complements, in terms of cost-effectiveness in specific contexts.
Most current knowledge about the magnitude of the food insecurity problem in a country (i.e. the absolute number of food-insecure households) as well as the depth of the problem (i.e, the severity of their insecurity) comes from detailed sample surveys conducted at the household level. Such household surveys typically cover food expenditure (and sometimes consumption) and income. A direct measure of household food insecurity is available in FAO's World food surveys (e.g. FAO, 1985), which estimate the number of malnourished people in a country through the use of income distribution parameters obtained through household income surveys. The World Bank's Living Standards Measurement Study was launched to obtain, inter alia, this type of direct measure of poverty and food insecurity at the household level.
Though household surveys can provide good indicators, this is not a feasible approach for regular global monitoring. Typical household surveys often require resources beyond the capacity of many developing countries. For this reason, detailed surveys are conducted only periodically. Such benchmark surveys need to be combined with other indirect and less costly indicators.
Per caput GDP remains the single most widely used indicator for measuring average income at the national level. This statistic is available for all countries on an annual basis and hence is a good indicator for annual assessments of changes in nominal income. As a measure of purchasing power, per caput GDP in local currency unit is considered a better measure than GDP in dollar terms, though it limits cross-national comparability. A useful compromise is to measure GDP in terms of purchasing power parity.
Yet per caput GDP, however measured, is only a national average and hence fails to indicate income levels of specific income groups where income distribution in the country is uneven. In those economies where income distribution is relatively egalitarian, changes in GDP per caput largely reflect similar changes for low-income households, so the precision of this indicator may be relatively high. The opposite is generally true for those economies where distribution is more skewed. In such economies, a drop in average income often leads to a disproportionate decrease in incomes of the poor. A study of Brazil showed that average per caput GDP fell 14 percent during 1981-83 but real incomes of the poor fell 20 to 30 percent (Cortazár, 1986).
Such quantitative information about relationships between changes in average income and group-specific incomes is not widely available, except through frequent household income surveys or estimates based on detailed general-equilibrium models of the economy, nor is it easily generalizable across countries with diverse economic structures. Yet statistics on income distribution are invaluable for the types of problems addressed in this paper.2 Therefore, efforts need to be made to complement average income data with information on the income distribution pattern in the country.
2 Statistics on income distribution are among the key inputs for the estimation of, for example, the number of malnourished people in the world, such as that routinely made by FAO in its World food surveys (e.g. FAO, 1985).
Wage rates, particularly minimum wage rates, are often considered good proxy indicators of the earnings of low-income households. They have been used in assessing the impact of structural adjustment programmes on the poor (see e.g. Cornia, 1987; FAO, 1989). Wage data are widely available at the international level, notably from the International Labour Office.
In spite of these advantages, the wage rate in itself is not a complete indicator of the real income of a household. First, information is needed on inflation rates, which along with the wage rate determine the real wage. Second, data are needed on the employment situation, particularly on whether employment of low-income households improved or deteriorated during an assessment period. Monitoring of price levels is also relevant for assessing changes in real income. One suggestion is a proxy indicator using wage rates and prices, whereby the cost of the food basket is expressed as a ratio of wage rate. This approach is based on the proposition that the poor are those whose food expenditures absorb 70 percent or more of total expenditure (Lipton, 1983).
Focusing on changes in the production sectors that provide most of the income to food-insecure households is an indirect approach to monitoring fluctuations in income. A benchmark survey (often a household income survey) is required to identify these key production sectors. Once these sectors are identified, annual changes in total output or value added by these sectors are proxy indicators to assess income changes. Disaggregated sectoral output or value-added data at the national level are needed for this approach.
Statistics on sectoral value added, which are combined to make aggregate GDP, are usually available. The level of disaggregation may be adequate to monitor changes in the incomes of some food-insecure groups. For example, in many countries the urban poor depend heavily for their income on the manufacturing and construction sectors. In this case, statistics on value added by these two sectors can be precise enough indicators for assessing changes in incomes of the urban poor. For rural areas, the level of disaggregation, i.e. one agriculture sector, may not be adequate.
The social accounting matrix (SAM) is a widely used framework to organize this type of data in a more disaggregated manner. A typical SAM not only includes a more disaggregated picture of the production sectors (in terms of input-output flows) but also reveals how the income generated through each production sector is distributed across several socio-economic groups. For example, if coffee is an important sector of an economy, a SAM would typically include coffee as a separate production sector, (In contrast, the coffee sector would be a part of the agriculture sector in national income accounts.) Then, if coffee is an important income source for the poor, it becomes possible to monitor changes in the income of the poor by monitoring changes in, among others, this sector. An advantage of this approach is that sectoral production data are usually available in a fairly disaggregated form in most countries.
Targeted income transfers. Many developing countries have public transfer and subsidy programmes which are designed to improve the economic access of the poor to food by increasing household entitlement. The size of the net transfer to a household depends upon the total amount of the subsidy and the efficiency of targeting. Such transfers can be substantial, especially for the poorest group of households. For example, in Sri Lanka the replacement of the blanket food subsidy scheme by a food stamp programme, and the subsequent decision to maintain the nominal value of the subsidy constant in the face of rapid inflation, led to a sharp reduction in real transfers to the poor. This decrease, combined with reductions in real wages following economic adjustment in 1977-78, led to a decline in poor households' ability to meet their food requirements. The calorie intake of the poorest 10 percent of the population dropped by almost 10 percent during the period from 1978-79 to 1981-82. Similar changes have occurred in other countries (see e.g. Pinstrup-Anderson, 1988; FAO, 1989).
Changes in net transfers to low-income households are not easily monitored. Much current knowledge about changes in transfer programmes and their impact on the poor is obtained from detailed household surveys, which are seldom carried out with sufficient frequency. Therefore, an indirect approach is essential. One option is to prepare a benchmark of the beneficiary coverage as well as the level of net transfer and then to apply more easily monitorable indicators in subsequent years, for instance total public expenditure on food subsidies and related income transfers. Many studies use this statistic as a proxy for net public transfers. For cross-national comparison, the budgetary level can also be expressed as a ratio, for example the ratio of income transfer expenditure to total public expenditure or GDP (Pinstrup-Anderson, Jaramillo and Stewart, 1987). An indicator of the efficiency of net transfers, e.g. a ratio of total net transfers received by the target groups to total amount transferred by the government, would be useful to improve the precision of the indicators, but it is difficult to obtain these statistics on a wide scale.
Prices. Average levels of food prices or indices based on the composition of the food basket are widely used in the context of poverty and food insecurity assessments. Prices indicate supply stress, are determinants of a household's real income and directly influence the level of food purchases. Because the poor spend a greater share of income on food than other groups, an increase in food prices usually affects low-income households disproportionately. Market prices are also a good barometer in that they respond fairly quickly to changes in the food demand and supply situation. In addition, data on prices are widely available in most countries at a high level of disaggregation.
In assessing real income, price indices specific to low-income households are more direct. A price index could include all items in the consumption basket of the poor or simply two to three main food items purchased by the poor. These indices must be used with information on changes in nominal incomes. For example, real purchasing power could be expressed as the ratio of the cost of the total food basket to the nominal wage rate.
Monitoring sustainability of access to food
The issue of households' ability to ensure continuous access to food over time is often neglected. Food security concerns of adequacy, stability and access are often viewed from a short-term perspective. However, households and societies may achieve temporary food security at the cost of substantial insecurity in the future. For instance, land, forests and other natural resources may be overexploited to ensure short-term food security. Households also resort to a number of strategies for coping with food insecurity during food crises and famine (for a recent review see Davies, Buchanan-Smith and Lambert, 1991). These include sales of assets, borrowing, migration and overexploitation of common property resources. For some households the consequent reductions in endowments may be temporary. Others may not be able to recoup their lost endowments and as a result become increasingly more vulnerable and insecure. Thus it is important to evaluate current household food security within the framework of long-term sustainability of access to food.
Monitoring of long-term sustainability should reveal whether households are obtaining food at the expense of their future security. Continuous monitoring of household assets and other responses to distress situations should show whether land, livestock and capital are being used at unsustainable rates, in turn causing health, education and culture to be neglected. For many low-income households, indebtedness usually begins with food stress and intensifies over time, leading to lost assets.
Sustainability should receive high priority within a household food security information system. Toward this goal, data such as land ownership pattern, value of other assets, indebtedness and quality of human capital should be collected and reviewed periodically.
TOWARD A HOUSEHOLD FOOD SECURITY MONITORING SYSTEM
Some national-level indicators that can be monitored were identified based on the above review of the determinants of physical and economic access to food at the household level. The availability of comparable statistics at the international or the country level was a major factor considered in identifying the indicators. However, household food insecurity is a problem specific to certain socio-economic groups within a country, and therefore more precise indicators need ultimately to be based on disaggregated information. Several studies have shown how the precision or predictive power of indicators can differ across socio-economic groups and regions even within a country (e.g. Reardon, Matlon and Delgado, 1988; Staatz, D'Agostino and Sundberg, 1990). A good starting point for an effective monitoring framework is then identification of the vulnerable groups and their sources of food insecurity at the country level to provide a basis for identifying relevant indicators of food insecurity. More analytical studies as well as associated statistical work would then be necessary for the development of monitorable indicators. This should be the ideal strategy for approaching the problem of monitoring household access to food. In the meantime, it is feasible to initiate monitoring of household food security based on the indicators already identified by utilizing available or easily obtainable statistics. The proposed approach is outlined below.
Household food security is a multidimensional concept and typically embraces concerns such as adequacy, accessibility, stability and sustainability. Thus, it may not be possible to address all of the main concerns with a single indicator in isolation. Therefore, a composite index made up of selected individual indicators is considered appropriate for this purpose.
Composite indices have been used frequently to measure concepts that are generally complex and difficult to quantify exactly, for example human development, quality of life, basic needs and food security. Some recent examples of composite indices are: the United Nations Development Programme's human development index (UNDP, 1990), a national food security index of the United States Agency for International Development (USAID), (Manarolla, 1989), the International Fund for Agricultural Development's four indices of poverty (i.e. a food security index, an integrated poverty index, a basic needs index and a relative welfare index), (IFAD, 1992) and the augmented physical quality of life index and economic diversification index of the United Nations' Committee for Development Planning (CDP).
A composite Index of household access to food
A composite household food security index appropriate for assessing both physical and economic access to food should embrace important determinants of these two forms of access. Adequacy of food at the national level and its physical availability to all vulnerable households are important. Nominal income, food prices and public transfers are key economic determinants. The composite index of household food security needs to be based on these considerations.
Adequacy of food at the national level can be represented through national average per caput supply, in relation to some normative requirement level or through its individual components, namely per caput food production, per caput stock changes and per caput food imports, for some reference period. If the components approach is selected, variability of production may be a proxy for stock changes in the absence of good stocks data. Per caput export earnings may be used instead of estimated per caput imports to reflect a country's import financing capacity, and hence the degree to which national supply will be adequate. An index of spatial price differentials for a reference period can represent physical access to food locally.
Agricultural production is a source of both food supply and income for low-income rural households. Thus a way to embrace changes in both food supply and earnings in the composite index is to include the deviation of per caput agricultural production from its trend. Since inequity in income distribution is the primary reason that food access differs across population groups, an indicator of income distribution, together with statistics on average income, such as the per caput GDP, also needs to be incorporated into the composite index. Public transfers may offset the effects of income inequality to some extent; therefore an indicator reflecting public expenditure for such transfers is also needed.
Food prices, as one of the two key determinants of real income, have an obvious role in a composite index. Since prices typically affect the poor more than the rich, they are important in assessing economic access to food. To be more representative, the prices to be included in the composite index should reflect the food basket of poor households.
A household food security index can comprise a number of indicators (see box opposite). Data for most indicators are largely available at the international level or are obtainable from individual countries without great cost. Moreover, since most of these statistics are assembled annually, it appears feasible to compile the index on an annual basis.
The composite index defined above, like any index of this type, suffers from some unavoidable conceptual and measurement problems. First, the identification of these indicators is based on the general determinants of physical and economic access to food in a typical low-income food-deficit country. Therefore it suffers from an unknown degree of imprecision in specific situations. Second, since the goal is to monitor household food security at the country level with global coverage, practical considerations of data availability on a wide scale also constrain the selection of the indicators. In view of this, some analytical work is required at the initial stage in constructing an index and validating its precision.
The validity of the index needs to be judged according to statistical qualities, notably comprehensiveness, precision, consistency and replicability. The index needs to be comprehensive enough to embrace the main vulnerable groups and their major sources of food insecurity. The index should be precise in measuring household access to food and should perform consistently over time. Finally, it must be replicable in diverse situations, at least for a subset of countries with similar economies. These statistical qualities provide important criteria for improving the composite index.
Initial assessment of the index can be based on statistics currently available at the international level. Following this first round of improvement, some field testing will have to be done. The main purpose of this field testing would be to identify, for groups of representative countries, more appropriate macro-level, monitorable indicators based on the strength of their linkages with micro-level, direct causal factors of food insecurity specific to vulnerable groups in the country.
While progress in household food security monitoring and assessment has definitely been inadequate for both conceptual and statistical reasons, the prospect looks bright. At the present time an intensification of international efforts and corresponding support for household food security monitoring and assessment is under way. Poverty and food insecurity have received unprecedented emphasis in recent years, particularly reflecting the adverse effects of economic adjustment programmes and recurring famines. Thus, action is being undertaken by international agencies on several fronts including data collection, monitoring and special studies. FAO has expanded coverage of indicators under its Global Information and Early Warning System for Food and Agriculture (GIEWS), extended reporting on financial difficulties faced by low-income food-deficit countries in securing adequate food and undertaken studies to assess the impact of adjustment programmes on the poor. Under the GIEWS it has also initiated a risk-mapping project aimed at identifying vulnerable groups and the causes of their vulnerability. UNICEF has undertaken several important studies to assess the impact of economic adjustments on children, while UNDP has initiated a series of annual reports that emphasize human-centred development as the overall goal of economic development. The World Bank also publishes a range of socio-economic indicators useful for monitoring poverty and food insecurity. Finally, the Administrative Committee on Coordination - Subcommittee on Nutrition (ACC-SCN) of the United Nations collects and disseminates information on nutrition. Household food security and its monitoring will also be major themes at the International Conference on Nutrition sponsored by FAO and WHO in December 1992.
Cornia, G.A. 1987 Economic decline and human welfare in the first half of the 1980s. In G.A, Cornia, R. Jolly & F. Stewart, eds. Adjustment with a human face. Vol. 1, p. 11-47, Oxford, UK, Clarendon Press.
Cortazár, R. 1986. Employment real wages and external constraints: the case of Brazil and Chile. International Employment Policies Working Paper No, 8. Geneva, International Labour Office.
Davies, S., Buchanan-Smith, M. & Lambert, R. 1991. Early warning in the Sahel and Horn of Africa: the state of the art. Vol. 1, A review of literature. Sussex, UK, Institute of Development Studies.
FAO. 1983, Report of the Eighth Session of the Committee on World Food Security, 13-20 April 1983. Rome, FAO.
FAO. 1985. Fifth world food survey. Rome, FAO.
FAO. 1989. Effects of stabilization and structural adjustment programmes on food security, FAO Economic and Social Development Paper No, 89, Rome, FAO.
FAO. 1991. Assessment of the current world food security situation and outlook. FAO Committee on World Food Security, 16th Session, Rome, FAO.
IFAD. 1992, The state of world rural poverty: an enquiry into the causes and consequences. New York, New York University Press. (In press.)
Lipton, M. 1983. Poverty, undernutrition, and hunger. World Bank Staff Working Paper No. 597. Washington, D.C., World Bank.
Manarolla, J.A. 1989. A methodology for ranking countries according to relative food insecurity. Paper presented at the AID Economists Conference, 12-17 November 1989, Washington, D.C.
Pinstrup-Anderson, P., ed. 1988. Consumer-oriented food subsidies: costs, benefits, and policy options for developing countries. Baltimore, MD, USA, Johns Hopkins University Press.
Pinstrup-Anderson, P., Jaramillo, M. & Stewart, F. 1987. The impact on government expenditure. In G.A. Cornia, R. Jolly & F. Stewart, eds. Adjustment with a human face, Vol. 1, p. 73-89, Oxford, UK, Clarendon Press.
Reardon, T., Matlon, P. & Delgado, C. 1988. Coping with household-level food insecurity in drought-affected areas of Burkina Faso, World Dev. 16: 1065-1074.
Sen, A.K. 1981, Poverty and famines: an essay on entitlement and deprivation. Oxford, UK, Oxford University Press.
Staatz, J.H., D'Agostino, V. & Sundberg, S. 1990, Measuring food security in Africa: conceptual, empirical, and policy issues, Am, J. Agric. Econ., 72: 1311-1317.
UNDP. 1990. Human development report. Oxford, UK, Oxford University Press.
[Ukrainian] [English] [Russian]