Contents - Previous - Next

This is the old United Nations University website. Visit the new site at

Conclusions and policy implications

The main lessons learned from this evaluation that could be of benefit to similar programmes in other developing countries are related to the areas of targeting and participation of the beneficiary population. In rural areas it is possible to target the poor through their occupation. since a significant proportion of the population in the villages is made up of marginal farmers or labourers.

The main reasons for lack of participation by potential beneficiaries in the PDS programme in different instances were lack of household income or liquidity. constrained supplies at the fair-price shop, PDS distribution of commodities (e.g., wheat and rice) that were not staple items in some regions. uncertainty in supply scheduling, and abrupt variations in the entitlement quota.

If the policy changes advocated here were adopted, the PDS could be more effective and cost-efficient in helping the poor.



Attempts at targeting subsidies to the poor in Andhra Pradesh did not meet with much success, despite efforts to reduce the number of eligible households by imposing entitlements based on income and land criteria. This can be seen in the fact that 88% of all households are included in the green-card category at a huge cost to the state government (about 30% of the budget). in addition to the cost borne by the government of India. It was also found that some of the poor (landless and marginal farmers) were left out of the scheme.

The first step in reaching the poor is to identify them. A significant proportion of the village populations (about one-third) are predominantly farm labourers or marginal farmers. These households are generally at the bottom of the income scale and do not have significant income-earning assets or skills. They form a distinct group and can be easily identified. If food grain is supplied only to this sector, the subsidy burden can be reduced. Poverty among farmers might also be tackled by other programmes to enable them to increase land productivity.


Participation in the PDS

On average, the PDS met about one-fourth of total food needs in Andhra Pradesh, less than 10% in Maharashtra, and more than one-third in Rajasthan. Several households could not obtain ration cards in Andhra Pradesh and Maharashtra villages, although the food-grain price was heavily subsidized. Most of these non-card-holders were resource-poor, belonging to backward castes, and were illiterate. and thus faced difficulties in following the official procedures for acquiring cards.

Among those who had ration cards, participation was almost 100% in the Andhra Pradesh village, where the amount of subsidy on staple foods was significant. In the tribal village of Maharashtra where geographical targeting under the ITDP was attempted, the subsidy was higher than in Andhra Pradesh. In spite of this, participation was much lower, mainly because of lack of purchasing power and constraints in supplies from the fair-price shop and because the commodities distributed (wheat and rice) were not staple food items. In another village, the participation was much lower, partly because access to the PDS was difficult, commodities were not staple food items, and there was no special subsidy as in the tribal village. In the sample village in Rajasthan, participation was quite high, although not as high as in Andhra Pradesh; however, there was no special subsidy.

In addition to the exclusion of some backward-caste households from the fair-price shops, other important reasons for lower participation were uncertainties in scheduling supplies or abrupt variation in the quotas (as reflected in the positive and significant regression coefficients of supply variables), the fair-price shops' policy of restricting supplies for short periods in order to avoid locking up capital and other resources for the whole month, lack of cash and access to credit on the part of the poor, and the supplying of less preferred cereals in areas where coarse cereal is consumed.



Commodity and price

The supply of rice and wheat in areas where these are not staple foods, as in Maharashtra villages, does not motivate the poor to consume them unless the PDS price is much lower than coarse cereal prices in the market. Furthermore, if coarse cereals were supplied, it would deter high-income groups from taking advantage of the subsidized prices in the fair-price shops.


Marketing channels

Since uncertainty and irregularity in the supply of rationed items is a major cause of lower participation, a regular supply should be ensured. Alternatively, a food-stamp scheme might avoid some of the organizational problems. Food stamps might be used in private retail stores. This could reduce the need for monitoring and other related problems in managing the elaborate network of the fair-price shop system. Furthermore, since lack of household income was also an important factor in several cases, purchase of food grains in instalments during the month should be allowed.



1. Pinstrup-Andersen P. Social and economic effects of consumer oriented food subsidies. In: Pinstrup-Andersen P, ed. Consumer oriented food subsidies. Baltimore, Md. USA: Johns Hopkins University Press, 1988.

2. Alderman H Food subsidies and the poor In: Psacharopoulos G, ed Essays on poverty, equity and growth. Elmsforth, NY, USA: Pergamon Press, 1991:172-303.

3 . Selowsky M. Target group-oriented food programmes : cost effectiveness comparisons. Am J Agric Econ 1979;61(5):988-94.

4. Bapna SL Food security through the PDS: the Indian experience In Tyagy DS. Vyas VS, eds Increasing access to food: the Asian experience New Delhi Sage Publications, 1990:99-144.

5. Lipton M. Risk m nutritional adequacy of food output: adjustment in India In: Food policy issues in low-income countries. World Bank Staff Working Paper no. 473. Washington, DC: World Bank, 1981.

6. Gupta A. Public distribution of food grains in India. CMA Monograph no. 69. Ahmedabad: Indian Institute of Management, 1977.

7. George PS. Public distribution of food grains in Kerala. Research Report no. 7. Washington, DC: International Food Policy Research Institute, 1979.

8. Kumar SK. Impact of subsidized price on consumption and nutrition in Kerala. Research Report no. 5. Washington, DC: International Food Policy Research Institute, 1979.

9. Singh VB. An evaluation of fair price shops. New Delhi: Oxford and IBH, 1973.

10. National Sample Survey Organisation (NSSO). Utilization of public distribution system, 42nd round (July 1986-June 1987). No. 362. New Delhi: NSSO, 1989.

11. Kabra KN, Ittyerah AC. The public distribution system: a report on target group orientation and the viability of retail outlets. New Delhi: Indian Institute of Public Administration, 1986.

12. Scandizzo PL, Swamy G. Benefits and costs of food distribution policies. World Bank Staff Working Paper no. 509. Washington, DC: World Bank, 1982.

13. Bapna SL. Production of coarse cereals in India: past performance and future prospects. Hyderabad: ICRISAT, 1976.

14. Bapna SL. Public distribution of food grains in India-experience and lessons. Ahmedabad: Indian Institute of Management, 1991.


Summary and concluding observations

Antonio Carlos Campino

The papers that have been assembled in this issue fall into three related categories: (1) an assessment of worldwide experiences in targeting nutrition interventions with suggestions for future programme design [1], (2) a review and evaluation of nutrition policies and programming in Brazil and Chile [2; 3], (3) in-depth evaluations of specific nutrition interventions, including a complementary feeding programme for mothers and small children in Chile [4], a school-lunch and a food-stamp programme in Brazil [5; 6], and fair-price retail shops in Brazil and India [7; 8].

None of the papers deal with general food-price subsidies, although in expenditure terms these are more significant than the food and nutrition programmes discussed here. For example, in Brazil in 1987, general price subsidies for wheat and rice amounted to US$3.39 billion [9], while expenditures in all the food and nutrition programmes amounted to US$818 million.' General subsidy programmes tend to be expensive and are not considered cost-effective in fighting malnutrition. Therefore, they are not good models to propose for the developing world when countries are struggling with the need to cut public deficits and find more cost-effective ways of providing basic services to their low-income populations.

In the following pages the programmes reviewed in the preceding papers will be characterized, their main findings will be summarized and the relevant issues identified, conclusions will be drawn, and recommendations for future evaluations of nutrition interventions will be offered.


Main findings of the overview papers

The Chilean experience indicated that efficient food and nutrition interventions could be achieved if there was a prior high-level policy decision to invest in human capital formation that included a well-functioning primary health care system with broad national coverage and a high rate of school enrolment accompanied by regular monitoring of the health and nutrition status of children. It also pointed to the importance of good targeting mechanisms at health centres, schools, and day-care centres. Given this base of a wide coverage of services and efficient targeting mechanisms, two organizational factors were important in the Chilean case: decentralization of the decisions on benefit allocation and transference to the private sector of all tasks for which it is believed to be more efficient than the public sector.

Review of the food and nutrition programmes carried out in Brazil over the past two decades indicated that improvements in health-care practices resulted in significant decreases in the percentage of low-birth-weight children and in morbidity and mortality rates, compared with the meagre results from food- and income-transfer programmes carried out in isolation from health actions [2].

The worsening socio-economic conditions in many developing countries during the 1980s called attention to the importance of high-level political commitment to health, food, and nutrition issues. This high-level commitment is cited by Pinstrup-Andersen as one of the requirements for successful nutrition interventions, the others being programme control at the periphery, decentralization of administration, and community-level participation in programme design [1]. In Brazil, however, the high-level political support that was present during the administration of President Geisel diminished substantially after 1979 [2]. The Brazilian experience contrasts with that of Chile, where high-level political support for health and nutrition programmes has persisted over a longer time period [3].

None of the four conditions for the success of nutrition interventions were apparent in Brazil after 1979. Although Chile exhibited two of the conditions for success-political commitment and decentralization-the peculiarities of its political regime allowed neither administrative control at the periphery nor a significant degree of community participation.

The main lessons from these papers are that, in order to be effective, food and nutrition programmes should be coupled with health actions, utilize a delivery mechanism with nationwide coverage, and be well targeted. Decentralization and privatization of some services can improve efficiency and programme effectiveness. However, wavering political commitment, lack of administrative control at the periphery, and lack of community participation continue to limit overall programme effectiveness, especially in Brazil.


Characterization of the programmes evaluated

The food and nutrition programmes evaluated in Chile, Brazil, and India will be characterized in terms of their beneficiaries, how they are reached, and the mechanisms used to distribute food.

Two of the programmes were targeted at pregnant and lactating mothers and small children, one at school-age children (PNAE), and two at the poorer segment of the population (PROAB and the PDS). (2) To reach these beneficiaries three main methods of targeting were used: income, outlet, and geographical location.

Targeting by income was used by the PDS in the rural and urban areas of India. Targeting by outlet was done either by the use of health-centre facilities, as in the FMP in Brazil [6] and the complementary feeding programme (CFP) in Chile, or by the use of public schools, as in the PNAE in Brazil. All of these programmes were aimed at the individual beneficiary. but there were differences according to the type of outlet used. The PNAE programme in Brazil benefited all children enrolled in public schools, regardless of their family income or socio-economic status. The programmes that distributed food through health centres used two criteria to define the vulnerable group they intended to benefit: nutrition status and family income or socio-economic status.

PROAB in Brazil targeted beneficiaries by geographic location. The vulnerable group it intended to reach were low-income families living in poor areas of the city of Recife.

The mechanisms of food delivery differed. The PDS used fair-price shops as a mechanism of delivery; this was linked to the belief that the main cause of malnutrition is lack of purchasing power associated with rising food prices. (3)

The CFP and PNAE used direct food distribution, either supplying supplementary foods to be taken home (CFP) or for on-site feeding (PNAE). Direct food distribution is compatible with the hypothesis that lack of income and insufficient nutrition education are the main determinants of undernutrition. The FMP used a food stamp that was tradable only for milk. As the stamp was valid for only one product, it did not allow for the consumer's choice, and therefore the only difference between this programme and the two mentioned above was the possibility of achieving greater efficiency by using a market mechanism rather than transforming the health centres into storage and distribution facilities.

The delivery mechanism chosen by PROAB was a market intervention by which the owners of small retail stores in the programme area could buy from a government central warehouse 11 food items at discounted prices. This enabled the retailers to sell these products at prices that were (or should be) 20% below the average prices in Recife. Apparently the underlying hypothesis was that low purchasing power was the most important limiting factor to adequate nutrition status.

Some plausible hypotheses that could have guided the formulation of the programmes have been advanced. But what is striking is the fact that, in all of the programmes, targeting criteria were formulated and mechanisms for food delivery were specified without a solid understanding of existing conditions affecting nutrition status. This could be a key factor in explaining the limited effectiveness of the programmes reviewed [1].

The constraint that all the programmes explicitly tried to remove was insufficient access to food. Pinstrup-Andersen, in his opening article, has indicated that there are other common constraints to good nutrition that programmes try to remove, such as infectious diseases, lack of knowledge, and short birth intervals [1]. None of these were addressed by the programmes reviewed. Implicitly the CFP tried to improve knowledge of nutrition when it made it mandatory that the producers of the formulas distributed by the programme should also distribute them through regular market channels. It was hoped that mothers would learn about the nutritional advantages of these products and would later buy them at commercial retail outlets.

The programmes reviewed aimed at alleviating malnutrition, but none of them addressed its fundamental causes. Also, none of the programmes attempted to modify the "economic environments within which households with malnourished members operate," which would require the formulation of specific "price, income, credit, interest rate and employment policies and policies influencing asset ownership and user rights" [1].


Main findings of the evaluations

Impact on malnutrition

A positive impact on the nutrition status of the target population was observed only in the FMP (fluid-milk programme). Even in this case a statistically significant positive impact was observed only for boys of one to two years of age. In the evaluation of PROAB, Sampaio pointed out that one of the reasons for the lack of measurable nutritional impacts was the lack of an adequate measure of family income [7].


Impact on calorie and protein availability

Pinstrup-Andersen has indicated that one reason "for observing little or no effect [of the nutrition programmes] is that the amounts of foods made available are so small that the effects cannot be measured" [1]. The evaluations of the CFP. PNAE, PROAB, and FMP addressed the question of the quantitative impact of the programmes by expressing the food distributed in terms of calories and protein and by comparing their value with some measure of the family's consumption without the programme.

The results regarding the impact of the programmes on calorie and protein availability for the beneficiaries and their families are not clear. The general conclusion is that, although the programmes are an important source of calories and protein at both the individual (CFP, PNAE) and household levels (FMP), this does not guarantee that participation in the programmes increases intake, since there is always the possibility that the food distributed is substituted for food normally bought by the beneficiaries. In just one case, PROAB, a relation between increase in intake and participation in the programme was verified.

In the cases of the PNAE and FMP, it was not possible to specify a control group, and therefore it was not possible to answer the question: Is participation in the programme responsible for an increase in calorie (and protein) availability? It was possible to observe that the programmes were an important source of calorie and protein availability at the household level (FMP) or at the beneficiary level (PNAE).

In the cases of the CFP and PROAB, it was possible to specify control groups, but the answers obtained on the association between programme participation and increase in calorie (and protein) availability are somewhat different.

In the evaluation of PROAB, it was possible to determine a clear impact of participation in the programme on calorie and protein intake at the household level. Sampaio was able to identify a control group and to compare the calorie and protein adequacy of both groups, concluding that the incidence of caloric inadequacy among the families that received the benefits of PROAB was smaller than in the control group.

Regarding the CFP in Chile, Muchnik and Vial [4] report that "the provision of free milk cereal. . . reduces household expenditure on food, allowing them to achieve a higher level of satisfaction through the consumption of other goods or through diet diversification. This does not, however, imply a higher consumption of calories." Therefore, in this case the positive impact of the programme was not to be seen in terms of calorie consumption but in terms of diet diversification.


Impact on food expenditures

In his evaluation of PROAB, Sampaio was able to compare the food expenditure of the beneficiary families with that of a control group. He concluded that "total food expenditures decreased with the subsidy" [7]. He advanced as a possible explanation the idea that there was an adjustment in consumer behaviour over time, since the programme "allowed the household to acquire the same bundle of goods at a lower price," thus freeing extra income to be spent on other non-food items. This behaviour on the part of the beneficiaries led Sampaio to draw important conclusions about targeting, which will be explored later.

The evaluation of the FMP employed a different treatment to assess the importance of the programme in terms of food expenditures. As it was not possible to obtain a control group, the objective was to assess the importance of the donated food in terms of the total value of the food available at the household level. Using the prices that the sampled families paid in the market for the same food items that were donated by the FMP and by the supplementary feeding programme (SFP), a value was imputed to these food items. Adding this value to the family's food expenditure, Campino came to the total value of the food available at the family level [6]. The result obtained indicated that the milk donated through the FMP corresponded to 16.3% of the total value of the food available to the household. The SFP provided 7.7% of the total value of food available.


Impact on schooling

The school-lunch programmes have objectives that fit into health and education policies as well. The health component of these programmes is intended to improve the nutrition status of schoolchildren and the educational objective is to improve school attendance and school performance.

The only school-lunch programme evaluated was the PNAE in Brazil. Dall'Acqua reports that 60% of the parents in sample households reported that the existence of the programme played an important role in their decision to enrol their children in school. This proportion was higher for the low-income population (71%) and decreased with income level [5]. Therefore, although it was not possible to evaluate the impact of the programme on school performance, it seems clear that it had an important impact on school enrolment.

Contents - Previous - Next