125. Document No. STAT/FOOD/11, “The FAO Methodology for Estimating the Prevalence of Food Inadequacy (Undernutrition) and the Required Data Processing and Analysis” was presented by Mr. L. Naiken.
126. The Experts were given the background of the method used by FAO for estimating the prevalence of food inadequacy. The methodological framework based on existing food consumption could help the countries to prepare and periodically update the estimate of undernourished people.
127. The Experts were apprised that the plan of Action adopted at the World Food Summit included a commitment by countries to develop a national Food Insecurity and Vulnerability Information and Mapping System (FIVIMS). This system should include the preparation of estimates of the proportion and number of persons undernourished. In this connection FAO had developed a methodology which involved the specification of two elements: the distribution of household per caput calorie consumption and the minimum per caput calorie requirement of the “average individual” in the population. The cumulative distribution up to the level corresponding to the minimum per caput calorie requirement was taken to reflect the proportion of the population undernourished. Hence the multiplication of the latter proportion by the population gave an estimate of the number of persons undernourished.
128. The Experts were further informed that the methodology described was analogous to the “poverty line” approach in estimating poverty incidence. However, the resulting measure was quite distinct from that of poverty in the sense that it attempted to assess the food inadequacy problem on the basis of data pertaining to the distribution of calorie consumption rather than income or expenditure.
129. It was noted that the minimum per caput energy requirement, which was based on energy (calories) requirement norms, was determined by first calculating calorie requirements by sex-age groups and then aggregating the results using the sex-age composition of the population as weights.
130. It was explained to the Experts that the source of information on the distribution of calorie consumption were surveys that collected data on the quantities of food consumed by individuals in a representative sample of households in the population. In this connection a commonly undertaken type of survey was the Household Income Expenditure Survey (HIES), which included food consumption data as part of their enquiry on household consumption expenditures. Such surveys had been regularly conducted in most countries. However, for the purpose of estimating the distribution of calorie consumption, it was necessary to convert the household level food consumption data into calories and hence derive tabulation relating to the distribution of individuals according to household per caput calorie consumption classes.
131. The Experts were informed that the process of converting the food consumption data from the HIES into calories and tabulating the results in the form of distributions by household per caput calorie consumption was not normally undertaken in most countries. Thus for the purpose of estimating the prevalence of food inadequacy using the FAO methodology it was recommended that the respective national statistical organizations carry out the required data processing and tabulation work on a regular basis.
132. The Experts were informed that in using the generated household level data pertaining to calorie consumption for estimating the distribution it should be noted that in practice the sample of households did not truly represent the population due to a number of reasons. As a consequence irregularities were noted in the tabulated frequency distribution. Furthermore the survey concepts and methodology applied were rarely sufficiently precise to yield accurate estimates of the actual consumption at the household level. As a consequence the household per caput calorie consumption figures was imprecise and in many cases unrealistically high or low. These inevitably affected both the mean and variance and hence the inequality in the distribution. In view of this, the following recommendations were made: (i) consideration be given to fit a theoretical distribution, in particular the lognormal distribution to the tabulated distribution data; and (ii) the parameters of the theoretical distribution be estimated through the use of an approach that dampens the effect of extreme values which is the case of the lognormal distribution, e.g. probit analysis might be considered.
133. It was mentioned to the Experts that estimates of the per caput dietary energy supply (per caput DES), derived through the food balance sheet approach, were regularly prepared and updated by FAO on an annual basis for practically all countries. In view of this FAO had developed a methodology that used these estimates as the mean of the calorie consumption distribution in preparing the global estimates of the prevalence of food inadequacy. In this approach the role of the household survey data was relegated to that of providing an estimate of the inequality in distribution. This option might be considered by countries especially in the case where it was believed that the mean calorie consumption estimate derived from the household survey data was as unreliable as that derived from the food balance sheet approach. In any event this option was appropriate in its own right as it provided a mechanism for assessing the effect of changes in per caput food availability on the prevalence of food inadequacy.
134. In the discussion, the Experts appreciated FAO’s role in providing guidelines for estimating the prevalence of food inadequacy and the required data processing and analysis. The Experts also noted that fitting of lognormal distribution to food consumption might not be necessary for estimating the prevalence of food inadequacy but this would facilitate more detailed analysis.
135. Finally, the Experts recommended that for estimating prevalence of food inadequacy, data on both quantity consumed and monetary expenditure should be collected to facilitate checking the consistency of data.
136. It was pointed out that the estimates of poverty should not be confused with the estimates of undernutrition as prevalence of poverty might not necessarily imply prevalence of undernutrition.
