Previous Page Table of Contents Next Page

4. Food security measurement: discrepancies and definitions

Food energy defi ciency, child malnutrition and low body mass index (BMI) in adults: differences and anomalies between Africa and Asia

Maarten Nubé
Centre for World Food Studies (SOW-VU), Vrije Universiteit
Amsterdam, The Netherlands

There are two main approaches for the assessment of undernutrition prevalence in a country or region. In the first method, utilizing information on food availability and distribution, an assessment is made of the prevalence of food energy deficiency by estimating the numbers of people whose dietary energy supply is likely to fall below a certain physiologically determined threshold. The other main method for the assessment of undernutrition prevalence is based on direct information on the nutritional status of individuals, mainly in the form of anthropometric data on height and weight, generally in combination with information on age in the case of children.

When the assessment is made on the basis of food energy deficiency, Sub-Saharan Africa is the region in the world with the highest levels of undernutrition. However, when made on the basis of anthropometry, it is South and Southeast Asia where the levels of undernutrition are highest (Svedberg, 1999). Also at the country level, there are large discrepancies between the two types of undernutrition estimates (Nubé, 2001). Analysis of data on 39 countries reveals very low correlation coefficients between country levels of food energy deficiency as reported by FAO, and levels of undernutrition in adults (BMI <18.5) and children (weight-for-age < median - 2 SD) as reported in household surveys. Also within geographically confined groups of countries (South/Southeast Asia, Sub-Saharan Africa), relationships between prevalence of food energy deficiency and underweight in children or adults are weak. Figure 1 shows graphically the relationship between food energy deficiency and low BMI in adult women for the 39 countries. It may be noted that the relationship between food energy deficiency and prevalence of children with a low weight-for-age yields a similar pattern.

Figure 1 also reveals large differences in the relationships between food energy deficiency and anthropometry between the three main developing regions. Most strikingly, for several countries in South and Southeast Asia, the prevalence rates of low BMI in adults are much higher than in Sub-Saharan Africa and Latin America, despite the fact that levels of food energy deficiency are in the same and largely


Source: FAO and Demographic and Health Survey (DHS)

Results call for efforts to identify the main causes of the observed discrepancies. Among factors that may intervene in the relationships between nutritional input (food availability) and nutritional outcome (anthropometry) are the occurrence of diseases, patterns of physical activity (workloads) and possibly ethnic differences in energy metabolism and/or body composition. However, currently available information on such factors does not lead to a satisfactory explanation of the observed anomalies.

With respect to the nature and quality of the two types of data used in the present analysis, it is generally acknowledged that, in particular in developing countries with high levels of subsistence food production and consumption, estimates of food availability may be subject to considerable margins of error and possibly to locality-based bias. Furthermore, it should be stressed that food availability data are not equivalent to actual food consumption data. In high-income countries, estimates of per capita dietary energy supply can be up to 50 percent higher than actual levels of per capita dietary energy intake. For most developing countries, little is known about how food availability estimates relate to actual food consumption.

With respect to anthropometry, it is important to note that available nationally representative data are mainly on children and female adults. As high levels of undernutrition in South Asia have been attributed by some authors to female deprivation (Ramal-ingaswami et al., 1996), a special concern is whether region-specific differences in nutritional status between males and females can be observed. Utilizing data from 42 studies in non-representative population samples in Asian and Sub-Saharan African countries that provide figures on prevalence of low BMI for both males and females, analysis of variance indicates that, on average, in Asia more females than males have a BMI <18.5, while the reverse is true in Sub-Saharan Africa. Thus, on the basis of data on women, adult undernutrition might be (slightly) underestimated for Africa and overestimated for Asia. Findings, however, do not solve observed discrepancies between the two types of undernutrition estimates.

In view of above findings, it appears that a number of issues need further investigation. First, an analysis comparing FAO data on dietary energy supply with food consumption data from household surveys may yield more insight in the relationships between these two types of information. Second, there remains a strong need for better information on the health consequences of malnutrition as measured by anthropom-etry and on possible differences in this respect between Asia and Sub-Saharan Africa. Results might have consequences for the interpretation of undernutrition prevalence rates in different regions in the world. Finally, the proposition that female deprivation is a major causative factor of high levels of undernutrition in South Asia needs further investigation.


DHS, Demographic and Health Services, Macro-International Inc., Calverton, MD, USA

FAO. 1999. The state of food insecurity in the world. Rome.

FAO. 2000. The state of food insecurity in the world. Rome.

FAO. 2001. The state of food insecurity in the world. Rome.

Nubé, M. 2001. Confronting dietary energy supply with anthropometry in the assessment of undernutrition prevalence at the level of countries. World Dev., 29(7): 1275-1289.

Ramalingaswami, V., Jonsson, U. & Rohde, J. 1996. The Asian enigma. In The progress of nations, pp. 10-17. New York, UNICEF (available at

Svedberg, P. 1999. 841 million undernourished? World Dev., 27(12): 2081-2098.

Malnourished and surviving in South Asia, better nourished and dying young in Africa: what can explain this puzzle?

Stephan Klasen
University of Munich
Munich, Germany

This paper examines the factors explaining the very different relationship between an-thropometric shortfall and child mortality in South Asia and Sub-Saharan Africa. While, in the former, very high rates of anthropometric shortfall coexist with comparatively lower child mortality rates, rates of anthropometric shortfall in Sub-Saharan Africa are much lower, yet under five mortality is much higher than in South Asia. Figure 1 illustrates the puzzles graphically.



Dependent variable





Low birth weight

Low birth weight

Low birth weight

Moderate and severe


35.42*** (7.11)

31.37*** (2.20)

39.23*** (7.77)

78.42*** (15.87)

East Asia + Pacifi c

2.50*** (2.39)

2.75* (1.63)

2.33** (2.20)

6.41** (2.01)

Eastern Europe + Central Asia

-3.82 (-1.18)

-4.39* (-1.41)

Near East + North Africa

-2.03** (-2.00)

1.21** (0.97)

-2.16** (-2.08)

0.43 (0.14)


-2.02 (-1.33)

0.87 (0.34)

-2.25* (-1.52)

-8.82*** (-3.23)

South Asia

15.30*** (7.18)

13.30*** (7.05)

15.00*** (7.32)

9.68*** (2.21)

Sub-Saharan Africa

-2.87*** (-2.52)

-2.15* (-1.64)

-2.80*** (-2.48)

-7.94*** (-2.49)


0.09 (0.09)

-0.41 (-0.37)

0.15 (0.15)

-3.74** (-1.83)


1.62* (1.51)

2.14** (1.80)

1.84** (1.73)

-3.85* (-1.58)


1.98 (1.40)

4.04*** (2.62)

2.15 (1.55)*

-1.69 (-0.62)


0.019*** (5.09)

0.018*** (3.14)

0.018*** (5.09)

0.023*** (3.73)


0.353 (0.92)

0.180 (0.31)

3.259*** (3.81)


-0.045** (-2.04)

0.005 (0.83)

-0.055*** (-2.89)

0.026 (0.46)


-3.14*** (-4.15)

-3.30*** (-2.19)

-3.34*** (-5.01)

-7.60 (-4.41)


0.068** (1.89)

Sanitation access

-0.140*** (-3.20)

Low birth weight

Adjusted R-squared





Omitted variable test










Dependent variable





Moderate andsevere stunting

Severe underweight

Moderate and severeunderweight

Moderate and severeunderweight


128.60*** (13.89)

28.64 (9.05)

31.17*** (2.39)

76.56*** (12.01)

East Asia + Pacifi c

6.03** (1.89)

2.70*** (3.42)

15.65*** (8.79)

15.58*** (7.70)

Eastern Europe + Central Asia

-8.63*** (-2.49)

0.32 (0.26)

-2.90* (-1.36)

Near East + North Africa

-3.53* (-1.42)

-1.41* (-1.72)

0.16 (0.01)

-3.08* (-1.58)


-11.60*** (-4.64)

-0.36 (-0.45)

-1.29 (-0.63)

-1.55 (-0.76)

South Asia

10.62*** (2.49)

10.10*** (5.80)

20.64*** (5.63)

20.68*** (5.28)

Sub-Saharan Africa

-5.08** (-1.93)

-0.34 (-.0.46)

5.94*** (2.42)

2.96* (1.64)


-1.72 (-0.96)

-0.35 (-0.54)

0.56 (0.42)

-0.20 (-0.18)


-2.20 (-1.05)

-0.76 (-1.05)

-1.14 (-0.63)

-0.62 (-0.42)


2.76 (1.19)

0.52 (0.53)

-0.19 (-0.06)

0.38 (0.23)


0.017*** (3.03)

0.007*** (2.76)

0.022*** (3.33)

0.021*** (4.23)


1.70*** (2.87)


-0.121*** (-2.66)

-0.066*** (-3.79)

-0.030 (-0.65)

-0.150*** (-3.73)


-12.19*** (-9.62)

-2.72*** (-6.29)

-2.88** (-2.06)

-6.96*** (-7.46)


Sanitation access

-0.090*** (-2.37)

Low birth weight

0.209 (1.15)

Adjusted R-squared





Omitted variable test










Left-out categories: Latin America and 1995+.

*10 percent significance; **5 percent significance; ***1 percent significance in a one-tailed test.

This puzzle is examined using the WHO Global Database on Child Growth and Malnutrition combined with UNICEF data on child mortality. I first specify models explaining the socio-economic and demographic determinants of undernutrition and then do the same for child mortality, where I include undernutrition as one of the determinants.

The analysis suggests that socio-economic and demographic determinants of undernutrition (particularly low incomes, low female education, high fertility, poor sanitation access, poor immunization coverage) have the expected effects on undernutrition, but cannot explain more than 10 percent of the difference in undernutrition between South Asia and Sub-Saharan Africa. Combined with the conceptual doubts about the appropriateness of the currently used international reference standard for undernutrition and the very high sensitivity of measured undernutrition to small changes in the cutoff points of that standard, this analysis is consistent with the notion that the unusually high rates of anthropometric shortfall in South Asia are partially due to the use of the international reference standard (the WHO-CDC reference standard derived from the experience of US children), which appears to generate misleading international comparisons of undernutrition and to relatively overestimate undernutrition in South Asia (see Table 1).

When including the prevalence of undernutrition in a multivariate model explaining under-five mortality, it has no significant independent impact on under-five mortality. When I correct the presumed bias in the undernutrition figures from South Asia using the undernutrition regressions, however, undernutrition has a sizeable and significant influence on mortality. This is consistent with the suspicion that the international reference standard indeed generates misleading figures for international comparisons while also suggesting that undernutrition does have a significant impact on child mortality.

The mortality regressions also give some indications on the reasons for the very high child mortality in Sub-Saharan Africa. There, very high rates of under-five mortality seem to be mostly due to very high fertility, high and rising HIV prevalence and a multiplicative interaction of these and other risk factors there (data not shown).

Measuring national food security: prevalence of undernutrition and an index of national food security

Krista L. Jacobs and Daniel A. Sumner
University of California
Davis, CA, USA

Vulnerability is an important aspect of food security that remains difficult to measure. Our work emphasizes the forward-looking and inherently stochastic character of the term “security”. As Barrett notes in the “Handbook of Agricultural Economics”, food security is “a dynamic problem subject to uncertainty and thus best conceptualized as an ex ante status rather than an ex post outcome” (Barrett, in press, p. 3). We use an index of national food security (INFS) that attempts to incorporate the risk of not having adequate access to food. (For more on the INFS, see Sumner, 2000.) This index relates food security to probability distributions of household incomes and food prices. Here, we use Indonesian rice supply and demand in the 1990s as an illustration to compare the INFS with other measures of food security.

FAO and USDA estimate the percentage of the population that is undernourished, a measure that captures what we refer to as the degree of adequacy of food access rather than food security. Such measures approximate how many people are undernourished at a given time. They are estimates of a realization of a distribution of possible outcomes. They do not reveal how many people were at risk of not meeting energy requirements in the past and provide no information about the degree to which food may be lacking in the future. While these measures acknowledge the role of income and income distribution in food adequacy, they do not directly consider the role of food prices.

The INFS attempts to incorporate risk into measurement of food security by using probabilistic distributions of prices and incomes, and it provides a means of illustrating how food security might vary under different policies. When more fully elaborated, it may prove useful to policy-makers. Food security is often used as a rationale for food pricing and trade policies. It is important to evaluate how such policies affect food security in an ex ante probabilistic context.

The INFS is defined as the probability that at least Fa* percent of households spend no more than S* percent of their income on a single staple food. As an operational approximation, we use the probability that a household spends no more than S* share of its total resources on a single staple as its level of food adequacy. The probability that a household meets or exceeds a threshold level of food adequacy, Fa*, is the INFS. Formally,

INFS = Pr(Pr(Sit £ S*) ³ Fa*)

where Sit is the share spent on the staple food by household i at time t. Sit = Ptfit/Iit, where Iit is household i’s income at time t, Pt is the price of food faced at time t, and fit is household i’s consumption of food at time t determined by income and price.

As an illustration, we evaluate the food security status of Indonesia in the 1990s using the INFS and compare this with the food adequacy indicators of FAO and USDA. We use expenditure data from the Indonesian Family Life Survey to measure income and share of income spent on rice. To represent the prices resulting from actual Indonesian policy, we use monthly retail prices from the Indonesia Bureau of Statistics. To represent a policy of full access to imports, we use Thai prices. To construct the INFS in this illustration, we use a food demand function ln(fit) = b0 + b1 ln(Iit)+b2 ln(Pt) with b1 = 0.5 and b2 = -0.5. Assuming that price and income are log-normally distributed, Sit is also log-normally distributed, and we construct its distribution from the income and price distributions.

We calculated the INFS using Monte Carlo simulation from distributions for log income and log price. The probability of being less than the threshold share S* was calculated as the value of the cumulative distribution function of the log share evaluated at S*. The share of cases where this probability was greater than the threshold level of food adequacy Fa* produces the INFS.

For the criteria of S* = 30 percent and Fa* = 90 percent, the INFS is approximately 0.920 under actual policy and 0.949 under a policy of full access to imports in 1993. In 1997, the probability of food adequacy being greater than 90 percent rises to 0.966 under actual policy and 0.979 under full import access. Overall, the INFS is notably higher in 1997 than in 1993 in all cases, and the policy of full access to imports corresponds to higher levels of food security.

A qualitative comparison between the INFS and the FAO percentage of the population undernourished suggests a reasonable correspondence. Between 1990-92 and 1997-99, the percentage of undernourished fell from 10 percent to 6 percent, while the INFS increases by approximately 4 percentage points from 1993 to 1997 when S* = 30 percent and Fa* = 90 percent. Were we able to discern a correspondence between undernourishment and household food adequacy, we could make a more direct comparison between the two sets of figures. For example, assuming that spending more than 30 percent of household income on rice corresponds to being undernourished, we estimate the probability that 6 percent of the population was undernourished to be 0.929.

The illustration described here imposed a number of limiting assumptions that are being relaxed in further work. For example, it is important (1) to account for the farm income gains from higher prices of the staple crop, (2) to consider when a country may influence import prices and (3) to account for differences in local food prices within a country. Additional generalization of the INFS includes expanding from a single staple to a basket of goods, allowing prices and incomes to follow distributions besides the log-normal or to be estimated non-parametrically, using alternative consumption demand functions, or permitting price and income elasticities of demand to vary over the income distribution.


Barrett, C.B. in press. Food security and food assistance programs. In B.L. Gardner & G. C. Rauss-er, eds. Handbook of agricultural economics, vol. 2, pp. 2-88. Amsterdam, Elsevier Science.

Sumner, D.A. 2002. Agricultural trade policy and food security. Q. J. Int. Agric., 39(4): 395-409.

Nutritional status assessment methodology

Abas Basuni Jahari
Center for Research and Development
in Food and Nutrition, Ministry of Health
Bogor, Indonesia


Nutrition problems in Indonesia were prevalent in the period before the monetary crisis of 1997 and are still prevalent now. The national-level data based on a series of National Socioeconomic Surveys indicated that the prevalence of underweight among under-fives is still the highest of the countries of the Association of Southeast Asian Nations. The prevalence of underweight moderately decreased in the 11-year period 1989-1999: from 36.2 percent in 1989 to 25.4 percent in 1999, at an average rate of 1 percent decline per year. However, between 1999 and 2001, the rate of decline was much less. The decrease in the prevalence of mild-to-moderate underweight (weight-for-age Z-score below -2) within the period of 1989-1999 was not followed by a similar reduction in the prevalence of severe underweight (weight-for-age Z-score below -3), which was 6.0 percent in 1989, 6.9 percent in 1992, 10.4 percent in 1995, 9.5 percent in 1998, 7.8 percent in 1999 and 7.5 percent in 2000. The problem of severe underweight was higher among children six-35 months of age. Because the reported problem of underweight was based only on the prevalence of a weight-for-age Z-score below -2, the increasing problem of severe underweight was not detected. This is what we call “the hidden problem”. The factors related to this contradictory situation are not understood clearly. Based on national economic indicators, the Indonesian GNP during the period before the economic crisis increased and reached more than US$1 000/capita/year. It was assumed that the decrease in the prevalence of moderate underweight and the increase in the prevalence of severe underweight were related to unequal distribution of wealth where the gap between poor and rich was becoming wider. The prevalence of mild-to-moderate underweight changed little from 1999 to 2001, despite efforts to reduce the effect of economic crisis on the health and nutritional status of people.

Nutritional status monitoring in Indonesia

In January 2001, the government of Indonesia put into effect an autonomous system for district-level governments. This means that most of the decision-making processes and pro- gramme planning related to the development of the district are becoming the responsibility of district-level governments. Therefore, the nutritional status information system must provide necessary information at the district level that covers all subdistricts and villages in the area.

National nutritional status assessment survey

Nutritional status information has been assessed through the integration of anthro-pometry (weight) into the National Socioeconomic Survey (Susenas) carried out by the Central Board of Statistics since the 1980s. The survey provides national and provincial figures of the nutritional status of the people as well as its determinants, but the data cannot be disaggregated to the district level. The figures on the prevalence of underweight discussed in the Introduction were derived from the results of this survey. With this change in the government system, it is important to redirect the survey to provide nutritional status information down to the district level.

Nutritional status monitoring system

The Nutritional Status Monitoring System (NSMS) was initiated in the late 1980s. Until recently, it collected information annually only on the weight, sex and age of children under five years of age. The nutritional status information (underweight) resulting from this activity can be obtained down to the subdistrict level. This is the main source of nutritional status information for decision-making and programme planning at the district level. NSMS is still the responsibility of the central government, but ideally in the future should become the responsibility of the district-level government.

Growth monitoring activity

In the late 1960s the Growth Monitoring and Promotion (GMP) activity was introduced through a nutrition activity centre (“Taman Gizi”) in the villages. By late 1980s the GMP was expanded to the entire country through Integrated Services Post (“posyandu”) by means of monthly weighing activity. On average, there are five posyandus in each village. About 350 000 posyandus are available throughout the country. However, at present only about 50 percent of the posyandus are active, with an attendance rate that varies from 30 to 80 percent. At the individual level, the monthly weighing activity provides information whether the child has gained weight or not. This individual information is then aggregated to the subdistrict and district level to provide information on the proportion of children who gained weight during the month. Information produced by the GMP is mainly used to identify targets for food supplementation and nutrition education for the mothers at the individual level, and for programme management at the subdistrict and district level. Thus far, weight data from the posyandus have not been used to provide nutritional status information in terms of the prevalence of underweight at a certain point of time, owing to the questionable quality of the weight data measured by cadres. In addition, the representativeness of nutritional status information based on posyandu data is also uncertain owing to a low attendance rate. Only weight but not height is gathered from this monitoring activity, therefore it is difficult to understand clearly whether the nutritional problems are acute, chronic or acute-chronic. However, several small-scale studies in three areas of West Java from 1998 to 2000 did collect height as well as weight, and thus three indicators of malnutrition can be studied [weight-for-age (underweight), height-for-age (stunting) and weight-for-height (wasting).] Data from these West Java surveys in Tables 1-3. The data show that the children from families of mid-to-upper economic status in the City of Bogor in 1998 (Table 1) were better off than the children in the province of West Java in 2000 (Table 2) and the children in the District of Bogor in 1999 (Table 3). The prevalence of underweight, stunting and wasting among children of the middle-upper economic status families in the City of Bogor was much lower than those in the province of West Java or in the District of Bogor. In the province of West Java and in the District of Bogor, the problem of underweight seems to be associated with stunting. This conclusion is based on the observation that the prevalence is high for underweight and stunting but low for wasting. Therefore, the nutritional problem in these areas appears to be chronic in nature, which could be associated with the problem of poverty, poor knowledge of nutrition, inappropriate caring behaviour and chronic diseases. The associated underlying factors that may vary from area to area need to be identified and addressed. Any effort to improve nutritional status of the population needs to also consider improving the related underlying factors and not solely providing supplementary feeding. The area with low prevalence of wasting may need only targeted rehabilitative supplementary feeding program.


Indicator of nutritional status

Category of nutritional status

Good (%)

Mild to moderate (%)

Severe (%)














Indicator of nutritional status

Category of nutritional status

Good (%)

Mild to moderate (%)

Severe (%)














Indicator of nutritional status

Category of nutritional status

Good (%)

Mild to moderate (%)

Severe (%)













The problem of malnutrition is not only faced by children under five years of age, but is also seen in school-age children. A small-scale study in school-age children in the District of Sukabumi, 2001 showed that 34.8 percent of children were underweight, 25.3 percent stunted and 4.7 percent wasted. Problems of stunting was also found among middle-school students in the District of Bo-gor, 1998. The prevalence of stunting among these middle-school students was 27.3 percent, while the prevalence of underweight was 7.9 percent.


It is recommended that all future nutritional status assessment and monitoring activities collect both weight and length or height data in order to provide a clear picture of the extent of nutritional problems, to use for decision-making processes and programme planning. For evaluation purposes, the presentation of nutritional status in the form of prevalence and mean Z-scores is useful for analysing the trend of nutritional status of an individual as well as in a population.

The revitalization of the posyandu, the improvement of the quality of measurement at the posyandu and the addition of taking length measurements once a year will make posyandus an important source of nutritional status information at all levels (district, subdistrict and village). Therefore, it is recommended that revitalization of the posyandus be prioritized, not only as the frontline of a nutrition programme but also as an important source of nutritional status information.

Reducing hunger or malnutrition? The case of Bangladesh

Gerard J. Gill
Overseas Development Institute
London, UK

Bangladesh has always had particularly high levels of nutritional deprivation, but the situation has improved in recent years. Successive SOFI reports indicate that the proportion undernourished fell from 37 percent (Prevalence Category 5) in 1995/97 to 34 percent (Category 4) in 1997/99 and that the number of undernourished people, which had increased by five million between 1990-92 and 1997-99, has now apparently stabilized at around 44 million despite continued population growth. This reflects growing per capita cereal production. Between the 1970s and the 1990s, the population grew by 70 percent, but cereal production more than doubled, and by 2001 the longstanding target of cereal self-sufficiency had finally been attained. These achievements owe much to steady progress in increasing land productivity, which rose by 2.2 percent per annum between 1984/85 and 1998/99, encouraged by an improved policy environment and the increasing involvement of the private sector and NGOs in agricultural development. The public sector played a vital role, particularly in the development and popularization of high-yielding cereal varieties. As a result, the real price of rice fell from Tk16-18/kg in 1985/86 to Tk12-13/kg in 1999/2000 (Government of Bangladesh, 2000), which is important given that around a quarter of the population is now urbanized and that, even in the rural areas, marginal farm households now purchase between 62 and 80 percent of their rice.

Reinforcing this, government food distribution policy has been refocused to target the nutritionally vulnerable. Before 1990, less than 40 percent of beneficiaries were poor, but by the early 1990s, the figure was over 80 percent (ibid.). Reflecting these developments, child undernutrition showed considerable improvement throughout the 1990s, with declining levels in the three main anthropometric measures: underweight, wasting and stunting (BIDS 2001).

There are grounds for concern, however, that improvements in the quantitative aspects of nutrition have been achieved at the cost of diet quality for the poor. This is already exceptionally low, with the main food group (rice) providing as much as 81 percent dietary energy supply. For many years, the main food policy thrust has been on cereal self-sufficiency, and this was reaffirmed at the 1996 WFS with a Government commitment to halve the number of undernourished people by 2015. Since undernourishment is defined in terms of energy requirements, the target can be met most easily by increasing access to energy-rich foods like cereals. While it is understandable that policy should concentrate initially on alleviating hunger as the most serious manifestation of food insecurity, increasing attention must now be paid to the quality of diet. Many policy documents speak of the need to diversify diets, but policy implementation - as manifested in such forms as the central thrust of the agricultural research system and the range of foodstuffs provided under the public food distribution system - has focused very sharply on cereals.

Traditionally, the two most important non-cereal foods for the poor in Bangladesh were fish and pulses. The poor obtain almost all of their animal protein from capture fisheries, but stock depletion has caused per capita fish consumption to fall from 11 kg in 1970 to 7.5 kg by the late 1990s (DFID, 1998). The availability of pulses fell by 25 percent between 1989/90 and 1998/99 (Government of Bangladesh, 1999), while per capita consumption in the rural areas fell by 27 percent between 1991/92 and 1995/96 (Government of Bangladesh, 1998a). The reason is crop substitution. Pulses are primarily a winter (dry-season) crop, and with the spread of irrigation, boro (winter) rice (which, under irrigation, is a considerably less risky crop to grow than pulses) can be substituted. Boro production has been rising steadily, while the area devoted to pulses fell by 17 percent between 1983/84 and 1997/98. In the mid-1980s, the most important pulse (lentil) cost about the same as rice, but by the end of the 1990s, it cost twice as much (Government of Bangladesh, 2000). Household expenditure surveys suggest that the poorest households have reacted by substituting cereals for pulses. Given both the high protein efficiency of a balanced pulses-cereals diet and the importance of pulses as a source of iron, this substitution has seriously negative implications for those with special nutritional needs, particularly children and pregnant and lactating women.

This concern is heightened by a culture of women and children eating last and often, therefore, least. Moreover, to the extent that the more nutritious (appetizing) foodstuffs are eaten first, the problem of nutritional imbalance will be exacerbated. Lactating and simultaneously pregnant mothers are highly disadvantaged in Bangladesh, suffering as much as 30 percent nutritional deficits, and Bangladesh is one of the few countries in which female life expectancy is lower than that of males (BIDS, 2001). Age and gender bias combine to make the food insecurity position of girls the worst of all. The 1999/2000 Bangladesh Demographic and Health Survey indicates that according to all three major an-thropometric indicators, female under-fives have a significantly poorer nutritional status than their male counterparts.

The Government’s focus on quantitative aspects of diet is consistent with that of the 1996 WFS, which emphasized the number of “hungry” people in the world. The SOFI reports continue to identify food insecurity implicitly with hunger by using the energy-based concept of undernourishment as the principal measure of food insecurity. This may have encouraged policy-makers to view increased cereal output as the most rapid way of tackling the problem, when a more comprehensive approach to food security is needed. The statement in the SOFI 2000 report that there is little evidence of pro-male bias in food consumption remains a discrepancy that should be corrected, since it most emphatically does not apply in Bangladesh (or in many other parts of South Asia). In fact, the Government explicitly accepts that there is pro-male bias in food distribution (Government of Bangladesh, 1998b, p. 167), and it is a disservice to some of the most food insecure people to deny that this is the case.

At present, five methods are used to measure different aspects of hunger and malnutrition: the FAO approach to measuring the prevalence of undernourishment, household expenditure/consumption surveys, individual food intake surveys, anthropometric surveys and a range of qualitative techniques. This brief review indicates that other measures, such as analysis of production and price statistics, can provide important additional means of monitoring food insecurity and vulnerability at little additional cost. Failure to use such measures perhaps indicates a deep-seated problem of communication failure between specialists in agriculture and nutrition, and this reduces the scope for identifying remedial measures. Mapping food insecurity and vulnerability is of value only if it becomes an input in dialogue and action to improve policy formulation and implementation. This in turn requires a concerted deployment of all available resources.


BIDS. 2001. Bangladesh human development report. Dhaka, Bangladesh Institute of Development Studies for UNDP.

DFID. 1998. Support for university fisheries education and research: project memorandum. Dhaka, Department for International Development.

Government of Bangladesh. 1998a. Household expenditure survey 1995-96. Dhaka.

Government of Bangladesh. 1998b. The fifth five year plan 1997-2002. Dhaka.

Government of Bangladesh. 1999. 1998 statistical yearbook of Bangladesh. Dhaka.

Government of Bangladesh. 2000. Comprehensive food security policy for Bangladesh, final draft. Dhaka.

Previous Page Top of Page Next Page