Nutrition security

Optimal consumer subsidies and income transfers for minimum nutritional requirements: A basic model
Malnutrition in rural highland Ecuador: The importance of intrahousehold food distribution, diet composition, and nutrient requirements
Multidisciplinary capacity-strengthening for food security and nutrition policy analysis: Lessons from Malawi

Optimal consumer subsidies and income transfers for minimum nutritional requirements: A basic model

Abstract
Introduction
Income transfers
Subsidies
Subsidies combined with income transfers
Summary and conclusions
References
Annex

Dov Chernichovsky, Uri Spiegel, Uri Ben Zion, and Mark Gradstein

Dov Chernichovsky is affiliated with the Department of Health Policy and Management and the Program for Health Policy in Economies Under Stress at the Ben-Gurion University of the Negev in Beer-Sheva, Israel. Uri Spiegel is with the Interdisciplinary Department of Social Sciences at Bar-Ilan University, Israel, and the Department of Economics at the University of Pennsylvania in Philadelphia, Pennsylvania, USA. Uri Ben Zion is affiliated with the Faculty of Industrial Engineering and Management in Technion, Haifa, Israel. Mark Gradstein is with the Department of Economics at the Ben-Gurion University of the Negev.

Abstract

The supply of food is no longer a major determinant of malnutrition in the developing world. Rather, a lack of purchasing power, ignorance about nutrition, and subjective tastes or preferences prevent some households and individuals from securing adequate diets. Some households spend more on food and other consumer items than would be needed for a minimum balanced diet. Yet they remain malnourished or have nutritionally undesirable diets. Food subsidies and income transfers have been major policy options available to governments to augment household purchasing power and change consumer preferences in order to alleviate malnutrition. Those options have traditionally addressed the problem by considering one critical nutrient and one common staple. The model discussed here provides and demonstrates a solution to the question: What is the combined optimal income-transfer and subsidy programme that would meet particular nutritional requirements with the least budget expense to the government? It is argued and shown, with the aid of an initial model, that a combination of income transfers and food subsidies that consider a range of foods, rather than a single staple, and a range of nutrients, rather than a particular nutrient, may lead to cost-beneficial policies that meet wider nutritional objectives for less cost.

Introduction

The supply of food is not a major determinant of malnutrition in the developing world. Rather, it is a lack of purchasing power of some households (and nations) that prevents them from securing adequate diets. This is one of the most important conclusions of the recent World Food Summit [1]. This view has been held for more than two decades [2].

In the classical articulation of the diet problem, Stigler [3] concluded that malnutrition is more than a problem of insufficient income to purchase enough food. Indeed, many households, especially in developing economies, probably spend more on food and other consumer items than would be needed for the minimum required diet. Yet many of them remain malnourished, in part because of ignorance about nutrition and in part because of subjective tastes or preferences that may lead to nutritionally undesirable diets.

Consequently, “ignorance” and “tastes” must be considered explicitly in food policies and programmes that in most instances attempt to modify human behaviour by changing incomes and relative prices in the short term, while relevant health education takes root [4].

Price subsidies and income transfers have been major policy options available to governments to augment household purchasing power and alleviate malnutrition [5]. Both income transfers and subsidies, largely confined to a market economy, are, however, innately problematic in that some “leakage,” i.e., support to some “wrong” people and for some “wrong” commodities, is inevitable. In spite of these shortcomings, income transfers and subsidies have major attractions. Compared with the alternatives (e.g., feeding programmes), subsidies and transfers are most effective [2]. They also rely on market rather than on administrative mechanisms. This makes them appealing in developing economies where the share of the market economy is growing but administrations may still be weak.

Through presentation of a basic model, we seek to outline the key parameters involved in the answer to the question: What is the optimal combined price subsidy and income-transfer programme that would meet particular nutritional requirements with the least budget expense? This issue of optimizing and minimizing the total amount of income transfers and subsidies has become especially significant as governments try to reduce their budgets as part of economic structural adjustment efforts.

To start answering the question, we follow earlier work by Reutlinger and Selowsky [2]. Similar to that work, we focus on “market-wide” subsidies and “target-group-oriented” income transfers. We deal also with the same parameters: income and price elasticities to capture consumer behaviour, and income distribution to capture the policy environment. The model also follows empirical research looking into the determinants of household food consumption and nutrition [6]. We depart from the work of Reutlinger and Selowsky and from common policy programmes in several ways. First, we attempt to deal with optimal combinations, from a fiscal perspective, of the alternative policies rather than viewing them as mutually exclusive options. The view that pertinent policies and programmes are mutually exclusive is also evident in the concluding remarks of the review of these policies by Pinstrup-Andersen [5]. Second, we consider a vector of nutrients rather than just one or two. Third, we deal with all foods rather than just with a particular item. A “single-nutrient, single-food” approach may be outright detrimental; it may induce consumption of, say, calories at the expense of some critical vitamins that may be ignored [7].

This paper should be viewed as part of a more general effort to develop a model that would consider optimizing income transfers and subsidies from a nutritional perspective under a variety of budgetary, production, and foreign-exchange constraints [8].

Income transfers

Income transfers involve, in our case, raising household incomes to levels that secure the minimum requirements of any desired nutrient. This option must be based on knowledge of the income distribution and demand functions that indicate how households of different income levels would modify their food consumption when their income was supplemented.

When demand functions and incomes are known, it is possible to identify the level of income (I*_tj) that provides minimum consumption of nutrient A_j (where j denotes the jth nutrient) in household t. (Households may have different I*_j values if they have different demand functions. See Annex, parts 1 and 2.) Once I*_tj is established, each household with income below or at this level needs to be supplemented with transfer payments to reach the level of income that meets the minimum requirement. (The simplest way to establish I*_j would be to estimate direct income-expenditure elasticities of consumption of nutrients [e.g., ref. 6, p. 43].) When more than one nutrient is involved, the one nutrient, A_k, requiring the highest income level to meet the minimum sets the minimum income needed, I*_tk, for household t. That is, each household with the income I*_t, < I*_tk needs to be supplemented (I_t - I*_tk)- The total cost of this programme is the sum of all such supplements or transfers across households, all with different (I*_t, - I*_tk) <0.

This policy leads, however, to some “waste.” When more than one nutrient is involved, the effort to bring consumption of the “marginal component,” Ak, to the required level causes “excess” consumption of other nutrients whose minima can be met through a lower supplement or none at all. In addition, the income transfer will also lead to an increase in consumption of other goods and services unrelated to the diet. The potential “waste” associated with the income transfer occurs even when households spend the entire transfer on food.

In general, the smaller the income elasticity and the (calculated) share of expenditure on the nutrient that is deficient (at the margin), A^, the higher the marginal waste or leakage of the income transfer, because a higher income supplement is required to bring about the desired results.

The problem stated thus far can be illustrated with a simple example. Suppose we have 100 households whose income distribution is as illustrated in table 1. We further assume that the households consume three goods, X₁, X₂, and X₃, of which the first two include the nutrients A₁ and A₂, which are of policy concern. The three goods have the properties given in table 2. The minimum requirements are set at 400 units for each nutrient B₁ = 400, B₂ = 400. In this case, only the upper-income group meets all nutritional requirements before intervention. The intermediate group is deficient in A₂ but not in A₁. The poor do not meet any of the requirements; they are at 50% of the minimum requirement for A₁ and 12.5% for A₂.

TABLE 1. Income distribution

No. of households	Income (US$)
25	120
55	900
20	1,800

TABLE 2. Properties of goods and nutrients

Good Xi	Nature	Price Pi (US$)	Share in expenditure µi,	Contents in nutrients
				t1	t2
X1	Food	3	0.5	10	1
X2	Food	2	0.1	1	5
X3,	Non-food	1	0.4	0	0

TABLE 3. Consumption and nutrition levels

Income group	Income(US$)	Consumption of			Diet
		X1	X2	X3	A1	A2
1	120	20	6	48	206	50
2	900	150	45	360	1,545	315
3	1,800	300	90	720	3,090	750

TABLE 4. Consumption and “waste” levels

Income group	Level of transfer	Consumption of					“Waste” in terms of
		X1	X2	X3	A1	A2	X3	A1	A2
1	840	160	48	384	1,648	400	336	1,248	0
2	60	160	48	384	1,648	400	24	103	0
3	0	300	90	720	3,090	750	0	0	0

Given these data and the specific demand function given in the Annex, part 1, the consumption and diet levels are established as given in table 3. Based on the derived “demand function” for each nutrient, it is possible to establish that for the poor to achieve the minimum requirements of the two nutrients, the following incomes are required:

I*₁ = US$ 233
I*₂ = US$ 960,

The “waste” associated with this policy programme is illustrated in table 4. Groups 1 and 2 increase their consumption of X₃, which has no nutritional value, and of nutrient A₁ above the level that the government is interested in achieving.

Subsidies

Rather than supporting households directly through their incomes, the government can support households indirectly through subsidized food items. (In the more general case, indirect taxes, i.e., negative subsidies, may be considered to discourage nutrition-ally detrimental consumption.) The fundamental advantage of subsidies vis-à-vis income transfers is that the former might secure better household spending of the extra resources on the foods nutritionally most desirable.

It should be noted that the model does not deal with the so-called Pareto optimum. It may be easily argued, however, that subsidies to a range of commodities are likely to be less distortive than a subsidy to one particular commodity. The model also assumes, at this stage, infinite supply elasticities; that is, all food quantities can be purchased at going (international) prices.

TABLE 5. Consumption levels by income groups after subsidy

Income group	Income(US$)	Consumption of			Consumption of
		X1	X2	X3	A1	A2
1	120	32	74	48	400	403
2	900	245	563	360	3,006	3,060
3	1,800	489	2,560	750	7,450	13,289

The government, it is assumed in this model, cannot discriminate among consumers or limit the subsidy to any particular group. [“Food stamps” are a form of subsidy to specific income groups. In the case of food stamps, our discussion would refer only to the (sub-) population that is entitled to the stamps, and would consider an optimal combined (income transfer and subsidy) policy confined to this population.] Any product X_i, the price of which is P_i, may be subsidized at the level C_i so that the effective price of X_i, to the consumer is (P_i - C_i). The total subsidy S to a household is the sum of all subsidized items purchased by the household times the value of the subsidy on each item. The total food subsidy for the economy is the sum of all subsidies across households. (For a formal presentation of the arguments, see the Annex, part 3.)

The government seeks, in this case, to minimize the total budgetary outlay on the subsidy by trying to support only households with incomes below that which meets nutritional requirements. If the highest income group in need of support is identifiable, it can be argued that only this and lower groups should be subsidized through food stamps, rather than subsidizing the entire population. The costs of a food stamp programme, as compared with an economy-wide subsidy, involve the cost of administration and the possibility that stamps will be traded. These costs need to be contrasted with the “waste” discussed in this paper. Clearly, the lower the income level of that lowest group, the higher the subsidy required to meet particular nutritional requirements.

In general, the critical parameters that influence this solution are the size of the target population in comparison with the entire population that will also benefit from the subsidy, and the price and income elasticities that determine how much of the subsidy will go to improving the diet and how much to other consumption. (For a formal discussion, see the Annex, part 3.)

Following the specific example outlined above, which is based on specific demand functions and income distribution, we can calculate the optimal subsidies on goods X₁ and X₂ by finding the subsidies (C₁ and C₂, respectively) for the lowest income groups in need. It can be shown (see the Annex, part 3) that with an income of US$ 120, the minimum requirements can be met with subsidies C₁ = 1.16 and C₂ = 1.84.

TABLE 6. Levels of subsidies by income groups

Income group	Subsidy per household (US$)	No. of households	Total subsidy(US$)
1	174	25	4,350
2	1.319	55	72,545
3	5,337	20	106,740
Totals		100	183,635

The overall consumption patterns subsequent to the subsidy are as shown in table 5, and the subsidy cost per household and across income groups is shown in table 6. It is clear from this particular example that the subsidy, requiring a budget of US$ 183,635, is a more expensive policy than the income transfer, requiring a budget of US$ 24,300, because of the levels of consumption of X₂ by households of high and intermediate incomes, and the relative sizes of these groups in the population. It is noteworthy that the main share of the subsidy in this particular case goes to the highest income group. It is findings of this nature that lend support to food stamps.

Subsidies combined with income transfers

As suggested earlier, a combination of subsidies and income transfers might yield a more efficient policy programme than either policy alone. This option is discussed and illustrated here.

Suppose that I_M is the income needed to secure minimum requirements for all households under the income-transfer policy as stated above. This income (according to our example) is a superior “opening position” to the subsidies option, which is, in our case, more costly. (The case in which subsidies are superior to income transfer as an opening position does not alter the nature of the proposed solution.) This position is used for commencing an iterative, computational trial-and-error process whereby the income transfer is reduced by marginal amounts and is substituted by subsidies that retain the same minimum requirements as under the opening position.

FIG. 1. Income transfers and subsidy combinations

The nature of the proposed solution is illustrated in figure 1. Points bb and cc on the vertical and horizontal axes indicate, respectively, the transfers needed to secure the minimum requirement either through income transfers or subsidies, as outlined earlier. The line bbcc is an “iso-requirements line” or a transformation line between subsidies and income transfers that shows the combination of income transfers and subsidies that retains particular levels of nutritional requirements. The 45° line represents an iso-cost or budget line on which government outlays remain the same regardless of the policy, whether subsidy or transfers. For a slope greater than 45° on the bbcc curve, it clearly pays to reduce the transfer and increase the subsidy, as total government outlays will decrease. For example, if the segment oq is smaller than o’q as we move from bb in the direction of cc, it pays to reduce income transfers and increase subsidies; the government will save o’q’ without sacrificing nutritional requirements. The optimal solution is reached at the tangency point o where the 45° line is tangent to the iso-requirement line; at this point there is no advantage in moving towards one policy at the expense of the other.

In practice, the optimal solutions can be obtained stepwise. For each income transfer, a vector of optimal subsidies is obtained by solving a non-linear programming problem, as suggested earlier. We then calculate the total amount of government support and find the minimum budget that maintains nutritional requirements.

TABLE 7. Income and subsidy alternatives and combinations (US$)

Minimum income	Total transfer	Total subsidy	Total government expenditure
120.00	0.00	127,512.24	127,512.24
170.00	1,250.00	75,441.72	76,691.72
220.00	2,500.00	46,698.70	49,198.70
270.00	3,750.00	39,291.67	43,041.67
320.00	5,000.00	31,166.67	36,166.67
370.00	6,250.00	25,181.31	31,431.31
420.00	7,500.00	20,571.43	28,071.43
470.00	8,750.00	16,898.05	25,648.05
520.00	10,000.00	13,891.03	23,891.03
570.00	11,250.00	11,375.00	22,625.00
620.00	12,500.00	9,231.18	21,731.18
670.00	13,750.00	7,376.24	21,126.24
720.00	15,000.00	5,750.00	20,750.00
770.00	16,520.00	4,307.90	20,557.90
820.00a	17,500.00	3,016.26	20,516.26
870.00	19,750.00	1,849.14	21,599.14
920.00	21,100.00	794.20	21,894.20
960.00	24,300.00	0.00	24,300.00

^a Boldface indicates minimum budget that secures minimum nutritional requirements.

Summary and conclusions

Both income transfers and subsidies, largely confined to a market economy, are innately problematic in that some “leakage,” i.e., support to some “wrong” people and for some “wrong” commodities, is inevitable. Income transfers are relatively efficient when it is easy to identify the needy groups and the income elasticities for food for these groups are high.

Food subsidies, on the other hand, are intended to induce consumption of those items the government is interested in supporting. In this particular regard, they have an advantage over income transfers, because they are targeted to products rather than to consumers, especially when the poor are not easily identifiable. Compared with income transfers, subsidies have, however, several shortcomings. First, since a subsidy is given to the population at large, high-income households are subsidized. This problem is particularly serious when the subsidized items have high income elasticities and consequently high-income groups may benefit substantially from the subsidy. Second, subsidies carry an income effect; the household can transfer part or all of the subsidy to consumption of other non-subsidized commodities. This problem would be relatively serious if households had low price elasticities for the subsidized goods. In that case, the quantitative response to the subsidy would be relatively small, and a larger part of the subsidy would be shifted to other consumption.

Generally, in a low-income environment where, on average, the share of expenditures on food is relatively large, food subsidies can be efficient mechanisms for targeting income transfers from a general income distribution perspective. This follows because of the low price elasticities and high income elasticities for foods in such an environment. Clearly, different price and income elasticities of different food items call for programmes that emphasize the foods with the desired elasticities and critical nutrients. As income levels rise, policies that emphasize a range of foods and micronutrients in addition to calories, for example, come into play. Such policies depend even more on models such as the one presented here.

There is no clear advantage to one policy over the other. We conclude, therefore, that it may be desirable to consider a policy that combines income transfers and subsidies, taking into account income and price elasticities for a range of foods, as well as income distribution. Such a policy could achieve minimum nutritional requirements at a lower budget cost than a policy based on either subsidies or income transfers alone.

As there are numerous combinations of income transfers and subsidies on food that can achieve desired nutritional levels, it is important to find the optimal mix also from fiscal, foreign exchange, and production perspectives that have not been considered in this discussion. These considerations should be included in an extended model. Furthermore, the basic model advanced in this paper should be applied to country data sets, and operational policy norms should be followed.

This model keeps a fairly narrow, albeit critical, nutritional perspective. At the same time, with this model it is possible to assess the effect of alternative programmes on income distribution in general, depending on the specific objectives of those programmes.

References

1. Anonymous. Will the world starve? London: The Economist, November 16, 1996.

2. Reutlinger S, Selowsky M. Malnutrition and poverty: magnitude and policy options. World Bank Occasional Papers No. 23. Baltimore, Md, USA, and London: Johns Hopkins University Press, 1976.

3. Stigler JG. The cost of subsistence. J Farm Econ 1945;27:303-14.

4. Chernichovsky D, Zangwill L. Microeconomic theory of the household and nutrition programmes. Food Nutr Bull 1990;12(1):34-52.

5. Pinstrup-Andersen P. Assuring food security and adequate nutrition for the poor during periods of economic crisis and macroeconomic adjustment: policy options and experience with food subsidies and transfer programmes. Washington, DC: International Food Policy Research Institute, 1986.

6. Chernichovsky D, Meesook AO. Patterns of food and nutrition consumption in Indonesia. World Bank Working Papers Series No. 670. Washington, DC: World Bank, 1984.

7. Williamson-Gray C. Food consumption parameters for Brazil and their application to food policy. Research Report No. 32. Washington, DC: International Food Policy Research Institute, 1982.

8. Grant SM. Food subsidies in Egypt: their impact on foreign exchange and trade. Washington, DC: International Food Policy Research Institute, 1983.

9. Lancaster KJ. A new approach to consumer theory. J Pol Econ 1966;74:132-57.

Annex

Part 1. General
Part 2. Income transfers
Part 3. Subsidies
Part 4. A combined strategy

Part 1. General

The intake of nutrient j can be expressed as a linear function of the consumption of n food items X_i,:

(1)

where t_ij is the amount of nutrient A_i, in food item X_i,. It is possible to substitute for each nutrient A_j and get m inequalities for m nutrients:

(2)

where B_j is the minimum requirement of nutrient j.

If the demand function for each X_i, as suggested by Lancaster [9], is

X_i=x_i (I_j,P₁,...,P_n+1), (3)

where X_n+1 is a composite good of all non-food items, then the demand function for each nutrient
A_j is

(4)

For simplicity and the specific illustration (in the text), it is assumed that all households share the same “Cobb-Douglas”-type utility function:

X_i =µ_i (I/P_i), (5)

where X_i is the level of consumption of X_i for the price P_i, and µ_i, is the share of total household expenditure on food X_i. This function implies that all demand functions have unitary price and income elasticities and that all cross-elasticities are zero. As implied by this function and in general, the household can be influenced to change its consumption levels of X_i by changing either income I or price P_i, or both. Accordingly, the “demand function” for each nutrient A_j is

(6)

Part 2. Income transfers

According to equation (4) in general, and equation (6) in our specific case, it is possible to identify the income level I*_j that yields A_j = B_j for each nutrient. To achieve minimum requirements across all nutrients, the highest value, I*_M, of all I*_j is needed; that is,

I*_M =Max (I*_j), (7)

so that when the household’s income is I*_M, its consumption is

A_j ³ B_j (8)

for all nutrients except A_k, for which minimum income I*_k will produce an equality in relationship (8). For most nutrients, an inequality would exist in this relationship.

Let us assume that the households’ income distribution is given by the density function F(I). That is, for each level of income î, we can find the percentage of households below this level by the integral:

(9)

and the share of population between any two levels, e.g., I₁ and I₂, by

(10)

Accordingly, total government outlays (TT) for all I £ I*_M under this policy are

(11)

where L is the number of households in the economy.

Part 3. Subsidies

The total subsidy to a household with income I that benefits from n subsidized goods is

(12)

and the total food subsidy (TS) in the economy for L households is

(13)

In this case the government seeks a vector of subsidies [C_g1... C_gn, C_gi ³ 0] that will secure the minimum requirements for the highest income group in need whose income is I₀ so that

(14)

As the total budget for the subsidies depends on the specific vector of subsidies [Cg], the solution involves seeking the vector with the lowest budget (TS₀). The minimization of function (13) subject to the set of constraints represented by (14) is a nonlinear programming problem that has a solution so that

TS₀ = Min TS([Cg]). (15)

The specific demand function for each good is

X_i = (µI_a)/(P_i - C_i) = V_iI_a (16)

where V_i = µ_i/(Pi - C_i) and I_a is the average level of income in the economy. With this demand function, which is shared by all households, the minimization problem for L households becomes

(17)

This objective function (17) is minimized subject to the minimum requirements to be met by the lowest income group with an income I₀:

(18)

In our example, the constraints for the lowest in come group with an income of US$ 120 are

120 (10V₁+V₂) = 400

120 (V₁+5V₂) = 400

These yield V₁ = .2718 and V₂. = .6122, and optimal solutions for subsidies: C₁ = 1.16 and C₂ = 1.84.

Part 4. A combined strategy

Formally, let us reduce I* by dI/* so that we get I**, i.e.,

I**=I* - dI*. (19)

I** is the “new” lowest income level in the population. The gross saving to the government in transfer payments is

(20)

* Note that From the definition of V_i, it follows that V_i,P_i, - V_i,C_i, = a_i, and from the definition of the utility function.

(21)

subject to the m constraints

The solution of the problem for nutrition levels B_j, for income group I** is an optimal vector [].

The additional subsidy needed for income group I**, and as a result for the entire population benefiting, is

(23)

The first term of this equation is the additional subsidy needed for those with income I** as a result of the reduction in transfer payment. The second term in the last equation indicates the additional “leakage” in subsidies to those people whose income is above I**. For as long as GS is larger than SS, subsidies are more efficient than transfer payments, and the process continues.

Malnutrition in rural highland Ecuador: The importance of intrahousehold food distribution, diet composition, and nutrient requirements

Abstract
Introduction
Methods
Results
Discussion
Acknowledgements
References

Peter R. Berti, William R. Leonard, and Wilma J. Berti

The authors are affiliated with the Department of Human Biology and Nutritional Sciences at the University of Guelph in Ontario, Canada.

Abstract

Our objectives were to quantify the intrahousehold distribution of food in an Andean community and to relate this distribution to dietary adequacy. Dietary information was collected using the 24-hour-recall method (n = 155 in 35 households; two or more recalls per subject). We found that food was served equitably (according to energy and protein requirements), yet the risk of inadequate intakes of four micronutrients was age-related. This was largely a function of age-based differences in micronutrient requirements per unit of energy, rather than variations in composition of the diet. A simple reallotment of food to those with higher requirements will not solve this problem, since the micronutrient density of the average diet is relatively low. Targeting of nutrient-dense foods would be difficult in this and other similar developing-world communities that are accustomed to a common pot from which foods of homogeneous composition are served. Feasible alternatives include nutrition education programmes and fortification of foods (salt, sugar, and oil) with micronutrients.

Introduction

It has been argued that throughout the developing world there is a preferential allocation of food to adult men at the expense of adult women and children. This has been observed in various countries in the developing world, but it is not a universal phenomenon. Many different food distribution patterns have been observed, including biases favouring all adults [1, 2], male adults [3-5], female adults [6], or children [7], or equitable distribution [4, 6]. The relative distribution is usually measured by energy or protein intake compared with estimated needs, although some studies examine the distribution of quality foods [6] or the pattern of serving order, the serving of second helpings, etc. [5].

There are two key reasons that knowledge of intrahousehold food distribution is important. First, for aid efforts, food distribution programmes, and research involving distribution of food supplements, workers must have an understanding of how food is distributed within households. Second, recent papers mention that malnourishment exists because of inappropriate distribution of food within households in places where the food supply is apparently sufficient [4]. It is imperative that genuine cases of insufficient food supply be recognized as such and not dismissed as cases of “inappropriate distribution.”

In this paper we will examine the intrahousehold distribution of food and its consequences for the prevalence of nutrient inadequacies in a rural highland community in Ecuador. Although the nature of the distribution is specific to this community, the interrelationships that are explored between diet composition, nutrient requirements, and food distribution are relevant to nutrition studies throughout the developing world.

Methods

Study population
Dietary methodology
Analyses

Study population

The study population was a highland Ecuadorian community of subsistence farmers, located between 3,300 and 3,600 m above sea level and 110 km south of Quito, the capital of Ecuador.

Two hundred twenty-three of the approximately 500 residents were recruited from throughout the geographic and socio-economic range of the community as part of a larger study of diet and health [8]. The dietary data were collected between January and June 1994 using the 24-hour-recall method. Data were collected from each subject for 1 to 6 days (mean, 3.1). Only two subjects refused to participate, and no subjects deliberately dropped out after recruitment. Repeat interviews were not done if the subjects could not be located (because they had temporarily left the community) or if the study period ended before an appointment could be made. Unless otherwise indicated, analyses include only those subjects for whom there were 2 or more days of dietary data and who lived in a house where there were 2 or more days of data for the male head of the household (n = 155 in 35 households).

Adult subjects were paid 2,500 sucres per interview, and adolescent subjects were paid 1,000 sucres. At the time of the study, 2,500 sucres was equivalent to approximately US$ 1.15. An adult labourer could earn between 5,000 and 10,000 sucres for one day’s work. The Human Ethics Committee of the University of Guelph approved this study.

Dietary methodology

Earlier work indicated that weighed food records would not be acceptable to most families in the community but 24-hour recalls would be acceptable and, with the modifications described below, would be reasonably accurate.

To increase the accuracy of the dietary recalls, representative samples of local foods were weighed to the nearest gram, and the volumes of all bowls and cups in which each individual’s meals were served were measured to the nearest 5 ml. In this community, family members usually eat the same food (typically soup, rice, or potatoes) from a common pot. We calculated the total volume in the pot and the proportion served to each subject. Homogeneity of contents was assumed (consistent with our extensive informal observations), unless otherwise indicated.

Each subject was questioned about the quantity of food (number of bowls or cups) consumed, but the cook was asked about the ingredients of common-pot foods. In practice, much of the interviewing proceeded as a “consensus recall,” with the husband and children helping the mother to recall the ingredients of the common pot, and all family members (especially children) helping one another to recall the amounts consumed, while two of the authors of this article prompted and recorded. If a member of the household was not present at the time of the dietary recall but had been with the family for all of the previous day, and the family knew what the missing member ate (usually the same food they ate), the family’s report of the missing member’s intake was used. All subjects were carefully probed about other foods eaten outside the home.

The first author checked the dietary data for completeness and feasibility nightly, and within six months the data were entered into a database and checked again. All days of the week were sampled, but unequally. However, there was no day-of-the-week effect on nutrient intake [8], and therefore the data were not weighted by day.

Ecuadorian [9] and occasionally Latin American [10] food composition tables were used to calculate the intakes of energy, protein, iron, vitamin A, thiamine, riboflavin, and niacin from most foods. For the few foods not listed in these tables and for zinc, vitamin B₁₂, and folate, values were obtained from Canadian food tables [11]. (Using “cross-border” food composition tables may lead to incorrect estimates of nutrient intakes [12], but if the errors in composition estimates are random, there will be little effect on estimated intake [13]. Still, extra caution needs be exercised in the assessment of the zinc, vitamin B₁₂, and folate intake.)

Analyses

The energy and protein requirements of each individual were estimated using modifications of published recommendations [14]. Energy requirements for those over 10 years of age can be estimated by the formula PAL · BMR, where PAL is the physical activity level and BMR is the basal metabolic rate, estimated with regression equations using age, sex, and weight. One year before the beginning of this study, research on energy expenditure was conducted in this community using heart-rate monitoring and activity recalls. PAL was determined to be “heavy” (2.17 in males and 1.84 in females) [15]. For the analyses in this paper, the PALs recommended by the Food and Agriculture Organization/World Health Organization/United Nations University for people doing heavy agricultural work (2.10 for males and 1.82 for females) [14] were assigned to those over 18 years of age. For those 10 to 18 years old, WHO/FAO/UNU [14] recommend PALs varying from 1.52 to 1.65 (females) and from 1.60 to 1.76 (males). However, in our experience in the community, adolescents generally take on adult-like work levels (“heavy”) by 13 to 16 years of age.

Adolescent boys work full days, caring for animals, planting, weeding, harvesting, and performing other heavy chores. Adolescent girls are similarly involved in agriculture, as well as with domestic chores, such as cooking, washing, and caring for younger siblings. The PALs used for adolescents were arbitrarily increased over the published recommendations to reflect their activity levels, so that, for example, a PAL of 1.90 (rather than 1.60) was used for males 17 and 18 years old. (Note that if these changes were not made, the results supporting our conclusions would be even stronger, i.e., Q_energy for adolescents [see below] would be higher.) Energy requirements for 0- to 10-year-olds and protein requirements for all individuals were estimated on the basis of sex, age, and body weight [14].

The ratios Q_energy and Q_protein were calculated as (intake of individual ÷ requirements)/(intake of male head of the household ÷ requirements).

The micronutrient requirements were compiled from various FAO/WHO publications [17-20]. For vitamin A and zinc, FAO/WHO provide estimates of both basal and normative requirements. The basal requirements are the levels of intake required to satisfy all demonstrable functional needs. Normative requirements are set higher than basal, providing for desirable levels of storage or “adaptive capacity.”

The prevalence of inadequacy of protein, zinc, iron, vitamin A, thiamine, niacin, riboflavin, vitamin B₁₂, and folate in seven groups (males and females 2-10 years old; males and females 10-20 years old; non-pregnant, non-lactating women 20 years or more old; pregnant or lactating women; and male heads of households) was estimated using probability analysis [16]. Although probability analysis does not allow for specific comparisons within households, it is the most reasonable way of calculating dietary adequacy at the population level and comparing dietary adequacy between groups.

All statistical analyses were done with UNIX-based SAS, version 6.09 [21].

Results

Energy and protein were distributed approximately equitably within households (figs. 1 and 2), but there was a notably higher prevalence of inadequacy of zinc, vitamin A, and vitamin B₁₂ in the youngest age group (table 1). Two possible explanations for this finding are that (1) the nutrient density of the adult diet was higher than that of children, i.e., the adults were preferentially allocated micronutrient-dense foods; or (2) other nutrient requirements in relation to energy needs are greater in children. The first of these potential explanations was tested by comparing nutrient/energy ratios between age groups. The results are presented in table 2.

The children’s diets had slightly lower densities of protein, zinc, and vitamins A and B₁₂, but the differences were slight and were not large enough to be responsible for the higher rates of inadequacy in children (as shown in figs. 3-7, discussed below). Thus, there is a puzzling situation of nearly equal distribution of food but unequal distribution of inadequate intakes, suggesting that the second explanation may be true. We therefore examined the change in nutrient requirements with age. The requirements of thiamine, riboflavin, and niacin are based on energy intake, and so an identical density is required at all ages. Thus, the prevalence of inadequacy of these nutrients should be approximately equal throughout all ages. This is indeed the case, with 0% prevalence of thiamine and niacin deficiency and approximately 70% prevalence of riboflavin deficiency (table 1). Iron intake is so high that the risk of iron dietary deficiency is uniformly less than 1% for all ages.

FIG. 1. Ratio of energy intake to energy requirement for individuals divided by the ratio of energy intake to energy requirement for the male head of the household. Mean values (Q_energy ± SD) are shown for various sex and age groups: M, male; F, female; FNPNL, female, non-pregnant, non-lactating; FPL, female, pregnant or lactating; numbers after M and F are age ranges in years

FIG. 2. Ratio of protein intake to safe level of protein intake for individuals divided by the ratio of protein intake to safe level of protein intake for the male head of the household. Mean values (Q_protein ± SD) are shown for various sex and age groups: M, male; F, female; FNPNL, female, non-pregnant, non-lactating; FPL, female, pregnant or lactating; numbers after M and F are age ranges in years

TABLE 1. Percent prevalence of nutrient deficiency in diets of different subgroups of the study community^a

Subgroup	n	Protein	Fe (PA)	Fe (B)	Zn (B)	Zn (N)	Vit A (B)	Vit A (N)	Thiamine	Riboflavin	Niacin	Vit B12	Folate
M <10 yr	26	16	0	0	15	35	15	46	0	83	0	54	5
P <10 yr	23	12	0	0	15	40	23	51	0	76	0	57	5
M 10-20 yr	19	5	0	0	2	12	0	17	0	55	0	37	4
F 10-20 yr	12	30	-	7	6	22	1	14	0	70	0	52	25
NPNL	30	8	-	7	0	2	3	26	0	73	0	24	22
PL	7	9	0	14	21	55	0	14	0	60	0	31	95
MHH	35	11	0	0	0	6	8	25	0	68	0	21	17

^a. Assuming a protein correction factor of 0.85, intermediate bioavailability of iron (10%), and intermediate bioavailability of zinc (30%). Abbreviations: B, basal; F, female; M, male; MHH, male head of household; N, normative; NPNL, non-pregnant, non-lactating women over 20 years of age (“prevent anaemia” iron levels do not exist for menstruating women); PA, prevent anaemia; PL, pregnant or lactating women over 18 years of age; Vit, vitamin.

Subgroup	n	Protein(g/kcal)	Zinc (mg/kcal)	Vitamin A (RE/kcal)	Vitamin B12 (µg/kcal)	Folate (µg/kcal)
All subjects	693	1.5±0.3	0.18 ±0.07	15 ±13	0.04 ± 0.05	4,5 ±1.9
M <10 yr	117	1.3±0.3a	0.16 ±0.06a	14 ±12a	0.03 ±0.05a	4.3 ±2.2a
F <10 yr	87	1.3±0.3	0.16 ±0.06	13 ±11	0.03 ±0.05	4.3 ±2.2
M 10-20 yr	93	1.5 ±0.3b	0.19 ±0.08b	16 ±12b	0.04 ±0.04a	4.5 ±2.2a
F 10-20 yr	63	1.4 ±0.3	0.17 ±0.05	21 ±15	0.03 ±0.03	4.1 ±1.2
M >20 yr	159	1.5±0.3b	0.19 ±0.07b	15 ±13a	0.04 ±0.04a	4.5±1.9a
F >20 yr	173	1.5±0.3	0.18 ±0.08	15 ±13	0.04 ±0.05	4.5±1.9
Age effect (p)		,0001	.0002	.0002	.06	.50

a,b. The general linear model was used to test for an age effect (sexes combined). The p-value for an age effect for each nutrient is shown. Cells with unlike superscript letters within each column are unequal (p = .05, pairwise Tukey’s LSD).

Discussion

These analyses indicate that, because of the difference in relative requirements of children, adolescents, pregnant or lactating women, and other adults, it is not sufficient to distribute food from a common pot according to energy requirements, but rather certain higher-quality foods must be preferentially allocated to individuals with higher nutritional needs. This would be difficult in this and other communities in which people are accustomed to eating from a common pot. Energy intake is correlated with (and driven by) energy requirements, but such a relationship is not likely to exist for protein or any of the micronutrients [16]. Eating to satiety, therefore, may satisfy energy requirements, but if the diet is not sufficiently nutrient-dense, it will lead to nutrient deficiencies.

FIG. 3 Dietary density of protein required to meet the safe level of intake of protein, given each individual’s energy intake. The protein density in the community diet (assuming a protein correction factor of 0.85) is shown by a solid horizontal line

FIG. 4 Dietary density of zinc required to meet the safe level of intake of zinc, given each individual’s energy intake. The zinc density in the community diet (assuming a bioavailability of 30%) is shown by a solid horizontal line

FIG. 5. Dietary density of vitamin A required to meet the safe level of intake of vitamin A, given each individual’s energy intake. The vitamin A density in the community diet is shown by a solid horizontal line

FIG. 6. Dietary density of vitamin B₁₂ required to meet the safe level of intake of vitamin B₁₂ given each individual’s energy intake. The vitamin B₁₂ density in the community diet is shown by a solid horizontal line

FIG. 7. Dietary density of folate required to meet the safe level of intake of folate, given each individual’s energy intake. The folate density in the community diet is shown by a solid horizontal line

The study community and many other Andean communities are characterized by stunted, yet otherwise apparently healthy, adults. The issues addressed in this paper may partly explain this phenomenon. The diet is apparently adequate for adults, but not for children. The inadequacy of the childhood diet, along with other environmental stressors such as water quality, hygiene, cold, and hypoxia, results in almost ubiquitous severe stunting [8]. As individuals age and mature, the nutrient density requirements drop, and as adults they are able to achieve adequate nourishment.

It is recognized that stunting most often occurs in the first few years of life [22, 23]. Although non-nutritional factors surely are important in the stunting process [24], malnourishment is often the limiting factor [25]. Targeting nutrient-dense foods available in the community to the children may cause a decrease in stunting and its associated complications. It is not known if targeting would be feasible in this community, but other work in the community suggests it might be. Stansbury [26] observed that when children were ill, their mothers gave them specific, relatively nutritious foods (toasted barley flour and toasted corn) and avoided feeding them other foods (potatoes). It may well be possible to educate the mothers in other beneficial feeding patterns.

It is unlikely that targeting alone would be sufficient to end micronutrient inadequacy in this or other similarly situated communities. Even if a household is willing to adjust and target quality foods, it may not be logistically possible to do so. Folate-rich foods need to be targeted to all adults, folate-and zinc-rich foods need to be targeted to pregnant and lactating women, and zinc-, vitamin A-, and vitamin B₁₂-rich foods need to be targeted to children.

There are three alternative strategies to targeting. One strategy, advocated by Beaton [27] and others, is to increase the nutrient density of the household diet to satisfy the individual with the highest requirements. The possibility of increasing the nutrient density of the community’s diet is explored below.

For each of the 113 foods in the community diet, the nutrient density (in grams of nutrient per unit of energy) was calculated for each nutrient for which the risk of inadequate intake was high (protein, calcium,* zinc, vitamin A, riboflavin, vitamin B₁₂, and folate). The foods that had a higher density than the arbitrary cut-off of “community mean density +1 standard deviation” and that were produced in the community were identified. The single most nutrient-dense food is turnip leaves. However, the mass of leaves that would have to be eaten to cause a substantial increase in the intake of any nutrient except vitamin A would be so high that they cannot be expected to be more than a small supplement to the diet. A 10-g serving of the leaves would supply about 20% of an adult’s daily vitamin A requirement (assuming that the provitamin A in the leaves is bioavailable) [28, 29]. The problem of consuming a sufficient quantity also applies to carrots, onions, and beets (with the exception of vitamin A from carrots). In fact, the reason that they are the most nutrient-dense foods is largely a function of their low energy content rather than their high nutrient content.

* Calcium was not included in the earlier assessment of the diet because of the lack of an adequate description of the distribution of calcium requirements [17], which is required for probability analysis. However, the dietary intake of calcium is low (the daily averages are 228 mg in children, 337 mg in adolescents, and 431 mg in adults), and therefore strategies to increase calcium intake are considered here.

A second strategy is to get the children to eat more of the foods they already eat. Although we did not collect data supporting this, our informal observations suggest that most families could feed more staple foods to their children. The low, if not inadequate, energy intake of the children (approximately 80% to 85% of estimated requirements** [8]) does not appear to be due to the limited quantity of food available, but to the high bulk and low nutrient density of the children’s diets. Zinc deficiency leads to appetite suppression, but this is likely to be only partly responsible for the low intake. Boiled potatoes were the most satiating of the 38 foods tested in a recent study [33]. They are about 3.2 times more satiating than white bread and 2.3 times more satiating than boiled white rice. (Satiation was measured both by a subjective feeling of fullness and by energy intake at a meal two hours later.)

** Recent publications, however, suggest that the WHO/ FAO/UNU 1985 recommended energy intakes for children are 15% to 30% too high [30-32].

However, increasing the total amount of food consumed does not eliminate all risk of inadequacy and may increase the risk of obesity without preventing stunting [37, 38]. If children up to 10 years of age simply increased the total amount of food consumed without changing the composition of the diet, so that their intakes met the FAO/WHO/UNU requirements [14] (note that this would put their intakes in excess of some of the newer estimates of energy requirements [30-32]), the prevalence of dietary inadequacy of protein, zinc (basal levels), and folate would fall to 0%, and the prevalence of inadequacy of vitamin A (basal) would fall to about 10%. The prevalence of inadequacy of riboflavin would be unchanged (requirements are proportional to energy intake), and the prevalence of inadequacy of vitamin Biz would remain high (42%). While still posing a health problem, mild to moderate deficiencies of vitamin A, riboflavin, and vitamin B₁₂ would probably not limit statural growth.

Because of logistical difficulties and nutritional shortcomings, neither of these first two strategies is likely to be successfully and widely implemented. A third strategy for implementation throughout Ecuador is a national food-fortification programme. An important difficulty in implementing such programmes is identifying suitable foods for micronutrient fortification. The fortified foods must be bought in stores rather than produced at home (where they could not be easily and consistently fortified), and they must be consumed by almost all people almost every day [38]. A number of candidate foods exist in this community. Vegetable lard was consumed on 98% of the surveyed days, and vegetable lard, pork lard, margarine, or corn oil was consumed on 99% of the days. Refined sugar and salt were consumed on 85% and 97% of the surveyed days, respectively. Similar results were obtained in a 1986 national survey of Ecuadorian children one to five years of age in rural and urban, coastal and highland communities. Oil or lard was consumed an average of 1.3 times per day (ranging from 1.2 on the rural coast to 1.5 in the urban highlands), and sugar was consumed an average of 1.1 times per day (ranging from 1.0 on the rural coast to 1.3 in the urban highlands) [39]. The successful implementation of a national salt iodization programme in Ecuador since the 1980s is evidence that the technical capacity and political will required for fortification programmes exist in the country.

TABLE 3. Technical possibility of fortification of foods with nutrients for which Ecuadoreans are at substantial risk of deficiency

Nutrient	Food
	Salt	Sugar	Cereal	Oil	Margarine
ß-Carotene	No	No	-a	Yes	Yes
Vitamin A	-	-	-	Yes	Yes
Riboflavin	No	No	Yes	No	No
Niacin	No	No	Yes	No	-
Folic acid	No	-	Yes	-	-
Vitamin B12	No	-	Yes	-	-
Iron	Yes	Yes	Yes	-	-
Calcium	Yes	No	Yes	-	No
Iodine	Yes”	Yes	Yes	No	No
Zinc	-	-	-	-	-

Adapted from ref. 39.

a. - indicates that possibility is not known.
b. Fortification of salt with iodine is already done in Ecuador.

In this paper dietary adequacy was considered as required density versus actual density (grams of nutrients per unit of energy, even for those nutrients whose requirements are conventionally considered as a function of body weight). The results suggest that the relative distribution of nutrient inadequacies (i.e., particular age and sex groups within a population at greater risk) might be identified given only estimates of diet composition. The ability to identify rapidly the individuals at highest risk (even if the magnitude of the risk is not known) would be a very useful tool in international development and food relief programmes. It is worthy of further study.

Acknowledgements

We thank the residents of the study community for their willing participation in this study. We thank Professor George H. Beaton for his comments on an earlier draft of this paper. Two anonymous reviewers provided very helpful comments. PRB was supported by a Natural Science and Engineering Council PGS B postgraduate scholarship, an Ontario graduate scholarship, and an International Development Research Centre of Canada Young Canadian Researchers Award. WRL was supported by grants from the Natural Science and Engineering Council (OGP-0116785) and the US National Science Foundation (SBR-9106378).

References

1. Engle PL, Nieves I. Infra-household food distribution among Guatemalan families in a supplementary feeding programme: mothers’ perceptions. Food Nutr Bull 1993,14:314-22.

2. Engle PL, Nieves I. Intra-household food distribution among Guatemalan families in a supplementary feeding program: behaviour patterns. Soc Sci Med 1993; 36:1605-12.

3. Babu SC, Thirumaran S, Mohanam TC. Agricultural productivity, seasonality and gender bias in rural nutrition: empirical evidence from South India. Soc Sci Med 1993;36:1313-9.

4. Basu A, Roy SK, Mukhopadhyay B, Bharati P, Gupta R, Majumder PP. Sex bias in intrahousehold food distribution: roles of ethnicity and socioeconomic characteristics. Curr Anthropol 1986;27:536-9.

5. Gittelsohn J. Opening the box: intrahousehold food allocation in rural Nepal. Soc Sci Med 1991;33:1141-54.

6. Wheeler EF, Abdullah M. Food allocation within the family: response to fluctuating food supply and food needs. In: de Garine I, Harrison GA, eds. Coping with uncertainty in food supply. Oxford: Oxford University Press, 1988: 437-51.

7. Leonard WR. Age and sex differences in the impact of seasonal energy stress among Andean agriculturalists. Hum Ecol 1991; 19: 351-68.

8. Berti PR. Dietary adequacy and its relationship to anthropometric status in a highland Ecuadorian community. Doctoral Dissertation, University of Guelph, Guelph, Ontario, Canada, 1996.

9. Ministerio de Salud Pública del Ecuador. Chemical composition of Ecuadorian foods. (Tabla de composición química de los alimentos Ecuatorianos.) Instituto de Investigaciones Nutricionales y Medico Sociales, División de Investigaciones Operativas. Quito, Ecuador: Organización Panamericana de la Salud, 1988.

10. Woot-Tsuen WL, Flores M. Food composition tables for use in Latin American. Bethseda, Md, USA: National Institutes of Health, 1961.

11. Dubuc MB, Lahaie LC. Nutritive value of foods. 2nd ed. St-Lambert, Quebec, Canada: Société Brault-Lahaie, 1994.

12. Bressani R. Some issues and problems in the usefulness of chemical composition data across boundaries. Food Nutr Bull 1992; 14: 128-32.

13. Beaton GH. Consideration of food composition variability: What is the variance of the estimate of one-day intakes? Implications for setting priorities. In: Rand WM, Windham CT, Wyse BW, Young VR, eds. Food composition data: a user’s perspective. Food Nutr Bull 1987; (suppl 12): 194-205.

14. Food and Agriculture Organization/World Health Organization/United Nations University. Energy and protein requirements. WHO Technical Report Series No.724. Geneva: WHO, 1985.

15. Leonard WR, Katzmarzyk PT, Stephen MA, Ross AGP. Comparison of the heart rate-monitoring and factorial methods: assessment of energy expenditure in highland and coastal Ecuadoreans. Am J Clin Nutr 1995; 61: 1146-52.

16. National Research Council. Subcommittee on Criteria for Dietary Evaluation, Committee on Evaluation of Food Consumption Surveys. Food and Nutrition Board, Commission on Life Sciences. Nutrient adequacy: assessment using food consumption surveys. Washington, DC: National Academy Press, 1986.

17. Food and Agriculture Organization/World Health Organization. Calcium requirements. Report of a joint FAO/WHO expert group. WHO Technical Report Series No. 230. Geneva: WHO, 1962.

18. Food and Agriculture Organization/World Health Organization. Requirements of vitamin A, thiamin, riboflavin, and niacin. Report of a joint FAO/WHO expert group. WHO Technical Report Series. Geneva: WHO, 1965.

19. Food and Agriculture Organization/World Health Organization. Requirements of vitamin A, iron, folate and vitamin Biz. Report of a joint FAO/WHO expert consultation. Rome: WHO, 1988.

20. Food and Agriculture Organization/World Health Organization. Report on trace elements in human nutrition. Geneva: WHO, 1996.

21. SAS Institute Inc. SAS Version 6.09. Cary, NC, USA: SAS Institute, 1992.

22. Keller W. The epidemiology of stunting. In: Waterlow JC, ed. Linear growth retardation in less developed countries. Nestle Nutrition Workshop Series, Volume 14. New York: Raven Press, 1988: 17-40.

23. Martorell R, Kettel Khan L, Schroeder DG. Reversibility of stunting: epidemiological finding in children from developing countries. Eur J Clin Nutr 1994; 48: S45-57.

24. Solomons NW, Mazariegos M, Brown KH, Klasing K. The underprivileged, developing country child: environmental contamination and growth failure revisited. Nutr Rev 1993; 51: 327-32.

25. Allen LH. Nutritional influences on linear growth: a general review. Eur J Clin Nutr 1994; 48: S75-89.

26. Stansbury JP. Culture and caretaking: maternal belief, child care practice and growth status in highland Ecuador. Doctoral Dissertation, University of Kentucky, Lexington, Ky, USA: 1996.

27. Beaton GH. Fortification of food for refugee feeding. Final report to the Canadian International Development Agency. Willowdale, Canada: GHB Consulting, 1995.

28. Solomons NW, Bulux J. Plant sources of provitamin A and human nutriture. Nutr Rev 1993; 51: 199-204.

29. Solomons NW, Bulux J. Plant sources of provitamin A and human nutriture revisited: recent evidence from developing countries. Nutr Rev 1994; 52: 62-4.

30. Davies PSW. Total energy expenditure in young children. Am J Hum Biol 1996; 8: 183-8.

31. Goran MI, Poehlman ET, Johnson RK. Energy requirements across the life span: new findings based on measurement of total energy expenditure with doubly labelled water. Nutr Res 1995; 15: 115-50.

32. Torun B, Davies PSW, Livingstone MBE, Paolisso M, Sackett R, Spurr GB. Energy requirements and dietary energy recommendations for children and adolescents 1 to 18 years old. Eur J Clin Nutr 1996; 50: S37-81.

33. Holt SHA, Brand Miller JC, Petocz P, Farmakalidis E. A satiety index of common foods. Eur J Clin Nutr 1995; 49: 675-90.

34. Brown KH, Sanchez-Griñan M, Perez F, Peerson JM, Ganoza L, Stern JS. Effects of dietary energy density and feeding frequency on total daily energy intakes of recovering malnourished children. Am J Clin Nutr 1995; 62: 13-8.

35. Michaelsen KF, Jorgensen MH. Dietary fat content and energy density during infancy and childhood: the effect on energy intake and growth. Eur J Clin Nutr 1995; 49: 467-83.

36. Malcolm LA. Growth retardation in a New Guinea boarding school and its response to supplementary feeding. Br J Nutr 1970; 24: 297-305.

37. Lampl M, Johnston FE, Malcom LA. The effects of protein supplementation on the growth and skeletal maturation of New Guinean school children. Ann Hum Biol 1978; 5: 219-27.

38. Lotfi M, Mannar MGV, Merx RJHM, Naber-van den Heuvel P. Micronutrient fortification of foods: current practices, research, and opportunities. Ottawa, Canada: The Micronutrient Initiative, 1996.

39. Freire WB, Dirren H, Mora JO, Arenales P, Granda E, Breilh J, Campana A, Perez R, D’Arquea L, Molina E. Diagnóstico de la situación alimentaria, nutricional y de salud de la población menor de cinco años. Quito, Ecuador: CONAUDE, Ministerio de Salud Pública, 1988.

Multidisciplinary capacity-strengthening for food security and nutrition policy analysis: Lessons from Malawi

Abstract
Introduction
A conceptual framework for choosing the focal points for capacity-strengthening in food and nutrition policy analysis
Food security and nutrition monitoring as a method of multidisciplinary capacity-strengthening
Multidisciplinary capacity-strengthening for food and nutrition policy analysis
Lessons for capacity-strengthening in food and nutrition policy analysis
Conclusions
Acknowledgements
References

Suresh Chandra Babu

Suresh Chandra Babu is a research fellow and head of the Training and Capacity Strengthening Program at the International Food Policy Research Institute in Washington, DC.

Abstract

Lack of sufficient analytical capacity in most of the developing countries in sub-Saharan Africa has been frequently suggested as a major factor in determining the appropriateness of food and nutrition policy interventions. This paper documents an approach implemented in Malawi for the past seven years to develop multidisciplinary capacity to analyse food and nutrition policies and programmes. A conceptual framework for identifying the areas of capacity-strengthening in food and nutrition planning and policy analysis is developed. Generalizable lessons from the Malawi experience are presented. Various issues that relate to enhancing the efficiency of capacity-strengthening programmes in sub-Saharan Africa are addressed. It is argued that continuous dialogue between food and nutrition researchers and policy decision makers and between the trainers in academic institutions and donor agencies is fundamental for achieving the goals of improved capacity for food and nutrition policy analysis.

Introduction

It is well recognized that food and nutrition policies that are ill-conceived and poorly designed could have negative effects and result in worsening the welfare of the population [1]. Although some policy analytical capacity exists in most developing countries, it has not been sufficient to meet the increasing demand for it in the assessment and evaluation of various policy reforms for their impact on the food and nutrition well-being of the population. The absence of adequate analytical capacity has been suggested frequently as a major factor in determining the appropriateness of food and nutrition policy interventions [2]. Considerable efforts have been made in developing and strengthening institutions and the necessary human capacity for designing and implementing food and nutrition programmes in developing countries [3]. However, the impact of such efforts in creating a sustainable core of food and nutrition policy analysts and planners has been limited.

Several factors contribute to the low level of capacity in food and nutrition policy analysis. In the past, national development plans - of which food and nutrition was a multisectoral theme - designed for five to ten years were the most important instruments through which governments made decisions on resource allocation among various sectors. To meet these planning needs, the capacity-building approaches emphasized sectoral planning, monitoring, and evaluation of development projects, including food and nutrition interventions. However, recently governments in developing countries have focused on policy reforms as a major tool of intervention in the process of economic development [4]. Although the methods of capacity-strengthening have changed accordingly, the capacity generated by such efforts remains grossly inadequate to meet the policy analysis needs of the governments [5].

The capacity-strengthening programmes currently offered in food and nutrition policy analysis focus largely on national macro price policies. Although the role of macroeconomic policies and their influence on the food and nutrition sector has been recognized as a potential area for capacity-strengthening, such approaches continue to emphasize policy-making at the national level [6]. Among the few approaches that focus on food and nutrition policy analysis, most concentrate on national policies, such as foodstock management, food trade, and food-aid policies, to achieve the goals of national food security and place less attention on multisectoral policies, such as food and nutrition programmes that have implications at the household level [7].

The capacity-strengthening approaches in food and nutrition policy analysis in the past have been able to cover the training needs of various groups at the same training sessions. Although this has generated some capacity in understanding the overall food and nutrition policy issues, the impact on the design of specific policies and programmes has been limited [8]. Because of limited institutions and associated infrastructure for policy analysis in many developing countries, the approach to capacity-strengthening in food and nutrition policy analysis has been confined to national institutions and headquarters of ministries, such as health, planning, finance, and agriculture. The role of these institutions, however, is limited to designing policies that will have an impact at the sectoral level or that will spread across several sectors with some spill-over effects [9]. Hence, their immediate impact on improving food security and the nutritional situation at the household level has been negligible [10].

It is recognized that policies designed to address problems of food and nutrition at the national level do not guarantee the alleviation of food insecurity and malnutrition at the household level. The farming systems, cropping patterns, and resource constraints are so diverse, even within a country, that few policies designed at the national level could have similar effects on rural households [11]. Although it is important to formulate efficient national and sectoral policies, there is a need to design programmes and policies that will have a direct and immediate effect on the food security and nutritional status of the population. However, this requires creating capacity and strengthening the existing capacity for designing such interventions. It should be noted that adequate human capacity for better policy analysis is not enough to ensure successful policy design and implementation. An overall institutional arrangement that facilitates the use of existing capacity for policy analysis is also necessary [3].

This paper documents an approach implemented in Malawi for the past seven years to develop a decentralized, multidisciplinary capacity to analyse food and nutrition policies and programmes. The specific objectives of this paper are to (1) develop a conceptual framework for identifying the areas of capacity-strengthening in food and nutrition policy analysis; (2) provide an example of a decentralized, multidisciplinary capacity-strengthening approach in food and nutrition policy analysis; (3) present specific examples of policy analysis skills that could be imparted at a multidisciplinary level; and (4) outline some of the generalizable lessons for similar approaches in other sub-Saharan African countries.

A conceptual framework for choosing the focal points for capacity-strengthening in food and nutrition policy analysis

In developing institutional and human capacity for food and nutrition policy analysis, it is important to understand the process by which information from the field is converted into policy interventions through the various institutions involved. Also, the target audience for strengthening policy analytical capacity and their training needs should be clearly identified to achieve tangible results.

The institutions that are generally involved in food and nutrition policy analysis in a country include the sectoral ministries, such as the Ministry of Health, the Ministry of Agriculture, and the Ministry of Planning, as well as specialized research centres in the universities and other academic institutions. More recently, non-governmental organizations have played key roles in generating food and nutrition information during food emergencies [12]. Figure 1 presents a conceptual model that could be used to identify different methods of training and various target groups in an institution involved in food and nutrition planning and policy analysis. In order to distinguish data management and analysis systems from training needs, the activities related to data-processing and policy analysis are represented by diamonds, the personnel by circles, and training activities by rectangular boxes.

The availability of adequate data is a prerequisite for food and nutrition planning and policy analysis. In general, in developing countries, the data for policy analysis come from two sources: primary data collected through sample surveys and participatory methods by both governmental and non-governmental organizations, and secondary data published in various official documents that either fully or partly depend on the primary data collected in the field [13]. Primary data are collected by various methods: field surveys, rapid rural appraisal, participatory methods, and key informant interviews. Sample surveys, however, continue to form a major source of data for food and nutrition policy analysis in these countries. The conceptual framework presented here could be modified for any type of governmental or non-governmental institution and for any method of data collection. It should be noted that it is not a model of policy analysis but an overall framework for identifying training needs in food and nutrition policy analysis. Generally, the data are collected by field enumerators who are supervised by field officers. Field officers compile the data for processing [14]. They jointly represent the core groups for training in data-collection methods. Given the recent advances in computer technology and its use at the field level, the data compilation is generally done using existing data-entry programmes. This introduces an element of training in operating these programmes so that data collected from the field are properly entered and checked for possible errors.

FIG. 1. Conceptual framework for identifying training needs in food and nutrition policy analysis

Once the data are entered into the computer, the process by which they are converted into policy information involves two major groups with distinct roles that are often not fully recognized by the current efforts in food and nutrition policy training. As shown in figure 1, the policy analysis group includes nutritionists, economists, sociologists, policy analysts, and other specialists, such as food technologists and agronomists. In the figure this segment is shown with solid lines to distinguish it from the other components involved in the process of converting data into policy alternatives. They are generally involved in designing various food and nutrition policy alternatives and evaluating them for their potential impact on the welfare of the population. Major areas of training for this group would involve data management: manipulating data files and processing data for policy analysis using computers. On the left side of the flow diagram, the data-processing group, or what may be called a research support group, and their training needs are presented. In most of the sub-Saharan African countries, there is a distinct group of civil service employees who work exclusively on computer information-processing [15]. Although they play an important role in policy analysis, this group has been largely ignored in capacity-strengthening efforts. Within this group, three different categories of personnel are involved, depending on the nature of the data-processing: data-processing clerks, statistical clerks, and computer programmers. Their major area of training involves understanding computer software for processing data collected from the field. They are shown with broken lines in figure 1. The computer programmes for data-processing that are widely used in sub-Saharan Africa include Lotus-123, dBASE, and Statistical Package for the Social Sciences (SPSS). Although this group of support personnel is relied upon heavily for entry, manipulation, processing, and management of data, the level of their expertise to meet these needs remains low because of inadequate training and experience.* It is not uncommon in many government ministries to find adequate computer resources with necessary software but no trained computer operators. This is in spite of established posts of computer programmers and their placement in these posts.** In figure 1 the diamond boxes with dotted lines represent the areas of their interaction with policy analysts at various stages.

* Elison M, Mthindi G, Malewa V. Optional resource use in computerization planning in the Ministry of Agriculture. Paper presented at the workshop on Computerization in Malawi - Policy and Program Planning, Mangochi, Malawi, October 1989.

** Mthindi GB, Hazeltine S. Manpower needs and capacity development for computerization in Malawi. Paper presented at the workshop on Computerization in Malawi - Policy and Program Planning, Mangochi, Malawi, October 1989.

The current models of short-term capacity-strengthening concentrate mostly on training in the collection, processing, and analysis of data, the areas shown with solid lines in figure 1. The broken lines indicate areas where additional attention should be paid in future efforts to strengthen capacity. Although the above conceptual framework could be used to identify training needs in any institution involved in food and nutrition policy analysis - governmental, non-governmental, academic, and donor agencies - some modifications may be required, depending on the type of institution in which the capacity is developed.

As pointed out, although the required human capacity for food and nutrition policy analysis may exist within an institution, it is not sufficient in itself to generate meaningful policy alternatives to implement them. The necessary institutional arrangements to facilitate the use of policy information by decision makers should be in place to benefit from this human capacity [3]. In places where adequate policy analysis capacity exists, this is frequently suggested as a major constraint to policy implementation. For example, in several countries in sub-Saharan Africa, food and nutrition policy-making remains a sectoral decision. Experience from countries