Contents - Previous - Next

This is the old United Nations University website. Visit the new site at http://unu.edu

TABLE 14. Average prices per kilogram, nutrients per gram, and price per nutrient unit(calorie and protein)^a

a. Price/1000 nutrient units = price/kg divided by nutrient unit/g.
b. Computed by taking a weighted average of rice prices and rice product prices, the weights being their relative shares in the consumption of the group.
c. Computed by taking a weighted average of corn prices and corn product prices, the weights being their relative shares in the consumption of the group.
Source: FNRI [14].

Rice, corn, and cooking oil are reasonable candidates for a subsidy policy. Rice is an important component of the Filipino diet. It is a preferred cereal, especially for the lower income groups, and has desirable nutritional qualities: it is not only the major calorie source but is also a significant source of protein. Corn, on the other hand, is the cheapest calorie source in terms of nutrient units per peso; it is also consumed mostly by the low-income groups. The third commodity, cooking oil, has a high calorie density and is easily digestible, even by children, making it easy to use as a calorie supplement. In addition, increasing of) consumption not only alleviates calorie deficiency but also aids in the metabolic process by providing a vehicle for fat-soluble vitamins. A pilot food discount project is presently being undertaken in the Philippines; preliminary results from a discount scheme targeting rice and oil are encouraging.

Table 15 presents the estimated changes in nutrient consumption resulting from a general non-targeted price subsidy policy; table 16, the results of subsidies to quartiles I and II; and table 17, subsidies to quartile I only.

In all cases the calorie and protein gains were largest when rice was subsidized. Relative to rice, oil subsidies offered an insignificant calorie gain, perhaps because of lower elasticities compared to those for rice, which can be explained with reference to actual dietary patterns at the time of the study. Oil accounted for only a small share of the budget and of total calorie consumption. Moreover, the amount of oil ingested was far more limited than that of rice.

A surprising result was that of the corn price subsidy. For most income groups, a 10 percent price subsidy was estimated to yield a small, but negative, percentage decrease in calorie and protein consumption, with a few exceptions. Using inelastic supply elasticity assumptions, most of the exceptions were for the upper quartile income groups.

The unexpected results for corn can be explained as follows. Even if there were a big percentage increase in corn consumption, the decrease in the consumption of a commodity that made a greater nutrient contribution, for example rice, would more than offset the positive own price response. Another reason for this behaviour was that corn was a substitute for, rather than a complement to, other goods: a decrease in the corn price would lead to a decrease in the consumption of the commodities for which corn was a substitute.

TABLE 15. Estimated change in nutrient consumption, 10 percent consumer price subsidy, all quartiles

TABLE 16. Estimated change in nutrient consumption, 10 percent consumer price subsidy, first and second quartiles

There was also the possibility that our results were artificial effects, since corn was consumed primarily in certain regions of the Visayas and Mindanao; a corn subsidy programme would reasonably be implemented in those regions, not on a nation wide scale. More precise estimates can be obtained by adding a regional dummy variable to the estimating equation. While these results were only indicative and were very tentative, they suggested that both substitution and complementarily effects should be taken into account when formulating a subsidy programme.

Regalado's results for rice, although based on the simulation of a 10 percent increase in the price of rice, were consistent with those of this study [36]. She estimated that a 10 percent increase in the rice price would decrease calorie consumption in the strata used in her study by 8.02, 10.48, 8.50, and 7.32 percent in strata I, II, III, and IV respectively.

Examining the effects of alternative supply elasticity assumptions, we found that caloric gains were larger assuming supplies were elastic rather than inelastic for cases where subsidization led to increased total calorie consumption, regardless of the degree of targeting. This result was expected, since elastic supplies would be able to accommodate a demand shift for one commodity without prices being bid up by the increased demand. In table 16, which illustrates the case of a non-targeted or general subsidy, the gains of price subsidies were apparently counteracted by price increases: there were insignificant or negligible gains if inelastic supplies were assumed.

With regard to targeted or non-targeted programmes, the results indicated that the more precise the degree of targeting, the greater the gain for the nutritionally at risk groups and the smaller the nutritional waste, defined as increased consumption for nutritionally sufficient groups. This result was consistent with the view that general price subsidies are very cost-ineffective and imply a regressive use of fiscal resources, for the simple reason that a general price subsidy implies an institutional leakage of fiscal resources to the non-target population. Broad subsidies and the use of international trade instruments, e.g. imports with subsidies, are extremely expensive and can also have powerful disincentive effects on the agricultural sector. Targeting consumer subsidies to just the households in greatest need provides nutritional gains comparable to those of the broader subsidies without the enormous fiscal burden or the production disincentives [44] .

TABLE 17. Estimated change in nutrient consumption, 10 percent consumer price subsidy, first quartile

There are many schemes that can be used to target nutrition intervention programmer, including: food-for-work projects, food stamps distributed according to a means test or income-based participation criteria, supplementary feeding, targeted food ration programmer, or fair-price shops that use explicit discriminating devices such as identification cards, or that are self-discriminating through geographical location or types of commodities. Many of these schemes involve sizeable administrative costs to prevent leakages to the non-target population. It is possible that the cost of implementing screening mechanisms may be greater than the cost of leakages, especially if the difference in income or nutritional status between the target and the non-target population is small. Since poverty and malnutrition tend to be location-specific, it is probably easier and more cost-effective to use geographical targeting, for example to target to an entire barangay with many malnourished members rather than to specific households within barangays.

Table 16 includes additional commodities and combinations for targeting. Starchy roots, such as cassava, have often been suggested as a possible candidate for subsidization, usually in Indonesia, because they are consumed primarily by low-income groups. However, estimates using our Philippine elasticity figures indicated that the calorie gains from subsidizing starchy roots were minimal. In addition, it is doubtful whether cassava is acceptable as a staple food since most Filipinos prefer rice.

The combination with the greatest promise was that of rice and oil, which is the combination presently being tested. Whether subsidization of these commodities can be implemented on a wider scale depends on many other factors, including the willingness of producer or marketing lobbies to have these commodities subsidized and the need to set up administrative mechanisms to implement the subsidy programme and assure that programme targets are the actual beneficiaries.

Cost-effectiveness Measures of Target-group-oriented Programmes

Because of the fiscal costs of general subsidies and the concern for nutritionally at-risk members of the population, target-group-oriented programmes have been gaining support. In addition, since malnutrition in the Philippines is the result of unequal distribution of food, rather than a lack of aggregate supply, there is a need for programmes that redistribute food within the population instead of general subsidies encouraging aggregate consumption [38] .

In order to rank target-group-oriented policies, this section estimates the treasury costs of increasing calorie consumption in quartile I households through a food-budget transfer or price subsidies. A direct food-budget transfer means that the food budget would be effectively increased. The cost of achieving this increase through an income transfer would, of course, be higher because of leakages to non-food purchases. In our case, the cost of the corresponding income transfer would be equal to cost of the food-budget transfer divided by the income elasticity of food expenditure. We examined the treasury costs of increasing calorie intake at the margin and by amounts up to 28 percent (the calorie deficit of the lowest income quartile!, assuming unitary and zero supply elasticity for the calorie deficit.

The results are presented in table 17. In general, the costs of increasing calorie consumption would be greater under inelastic supplies, as expected. In addition, while the cost function for a food-budget transfer was linear, the cost per increase in calorie consumption attained with price subsidies rose with the desired calorie gain because the cost functions were quadratic. Computation of the marginal cost for a 1 percent increase in calorie consumption by quartile I households indicated that, at the margin, the most cost-effective food policy for the S = 1.0 case was a food-budget transfer to quartile I (or its corresponding income transfer), followed by a subsidy on rice and oil; a subsidy on all food; and subsidies on rice, sugar, rice and corn, oil, and roots in that order. For S = 0, the ranking generally remained the same, except that the ranks of rice/corn and sugar, and roots and oil, were interchanged.

Table 18 presents formulas for the total and marginal treasury cost per poor person per day (in 1978 pesos) of increasing the calorie consumption of quartile I. The costs are expressed as a function of the desired percentage of calorie gain for quartile I, ECU, with the parameters of the cost functions defined by the demand elasticities and supply elasticity assumptions. Table 19 presents the computed estimates of the treasury costs associated with specific values of the desired percentage increase. Due to the quadratic cost function, there was wide divergence among the costs of closing the calorie gap- an increase of 28 percent. Assuming S = 1.0, we found that the rank order from the cheapest to the most expensive intervention was as follows: the cheapest intervention, that of a food-budget transfer, assuming no leakages to non-food items, would cost 13.44 pesos per person per day, followed by an income transfer, with allowance for non-food leakages, which would cost 20.36 pesos per person per day, and then by a food subsidy, whose cost increases abruptly to 1,066.24 pesos per person per day. The magnitude of these costs at the higher levels of desired gains suggested that even target-oriented programmes are expensive, and thus care must be taken to prevent leakage through resale or arbitrage, However, the administrative costs of such a programme are also likely to be high.

While we have obtained estimates of treasury costs, these should not be used as precise estimates of the cost of intervention policies. It has been pointed out that the costs in table 19 are unrealistically high relative to the per capita income statistics in table 5. This apparent overstatement is due to two factors: (1) severe understatement of income data; and (2) approximation errors arising from the use of a quadratic cost function. To illustrate how severe income underestimation in the survey was, the food-budget share in the lowest income group was 500 percent. Although income was underestimated, we believe food expenditures were accurately measured, and consequently were used to provide the base estimates for the cost calculations, Perrin and Scobie mentioned the possibility of errors due to the non-linearity of the demand curves as the size of the subsidy increases [30]. They indicated that the model might not be reliable in predicting the equilibrium displacements from single commodity subsidies except for small values of EC₁, for example, 1 to 2 percent. However, they also expected that approximation errors due to the non-linearities of demand and supply curves would be substantially smaller for subsidies to groups of commodities Nevertheless, the results should be taken as merely indicative and as a basis for ranking nutrition intervention policies, not for estimating actual costs.

TABLE 18. Total and marginal treasury costs per person per day for increased calorie consumption in quartile I (in 1978 pesos)

In practice, when the administrative costs of implementing a target-group-oriented programme are taken into account, the above ranking, which assumed perfect targeting and no arbitrage, may not hold. Some programmes which may appear more expensive from the cost-effectiveness ranking used above may actually be cheaper if they are easier to supervise. An example would be a commodity for which a reliable distribution system already exists. It is also possible that policy-makers may decide to implement a general subsidy in a particular geographic area if the incidence of malnutrition is so high that targeting would not be effective.

Lustig suggests that households may have different subsistence floors for various items; once the floor for food consumption is met, the family may decide to use additional income to satisfy other non-food needs first [23] . The family's perceived food threshold may, therefore, be lower than the nutritional standard, and a household would experience net nutritional gains only after the other subsistence requirements are met. Malnutrition cannot be eradicated in isolation from the fulfilment of other basic needs.

The magnitude of the costs suggests that intervention policies undertaken for an extended period will be very expensive, The alternatives are to adopt short-term targeted interventions in response to shortages or disasters or to target more specifically to vulnerable members of the family. Cost is the rationale for intervention programmes targeted directly to children, who are often at a disadvantage in the intra-family food distribution. While short-run programmes may be mere palliatives in that they do not affect the mechanisms that create poverty and malnutrition in the first place, they may mean the difference between subsistence and malnutrition for the low-income groups in the short run. The issue of sustainability is yet another factor to be studied by policy-makers.

continue

Contents - Previous - Next