Contents - Previous - Next
This is the old United Nations University website. Visit the new site at http://unu.edu
The results as reported indicate that PROAB may have minor effects on calorie consumption but little or no effect on nutrition status and weight at birth. However, results obtained from comparative analysis may not yield reliable conclusions because other factors and effects cannot be controlled and thus the effects of PROAB cannot be isolated. As pointed out earlier, it is only through multivariate analysis that major factors and effects can be considered simultaneously and the effect of PROAB isolated with greater precision. (The equations for the multivariate analysis were given above.)
Effects on calorie acquisition
The coefficients estimated for the parameters of equation 1 are shown in table 10. Changes in household income and expenditures significantly affect calorie acquisition, but the PROAB subsidy does not affect either the slope or the intercept. Furthermore, the negative sign for the subsidy intercept effect, although not significant, indicates that target households spend less on calorie acquisition, a result already noted. But the fact that the subsidy slope effect is also not significant indicates that target households tend to spend more of their extra income on calorie acquisition. Considering only significant parameters, target and control group households show similar behaviour, and the PROAB effect, if any, is due only to increased income.
A comparison of the regressions for log income and expenditures shows better results when expenditures are used. This could be anticipated, because expenditures far exceeded income for a large number of families. (In practically all household surveys a number of families spend more than their income. Garcia and Pinstrup-Andersen, for instance, used "total expenditure as an income proxy throughout the regressions because of inherent problems in correctly measuring income" [8, p. 31] ). The calorie acquisition elasticity estimated from the income regressions (about 0.09) is probably biased downward.
The calorie acquisition elasticity estimated from the expenditures regression (0.28) is in the range of those obtained for Brazil as a whole : (0.22) and (0.08) for the lowest 30% and the highest 70% by level of income. It is also in the range of those obtained for the Philippines (0.33)  and for Thailand (0.26) and Sudan (0.30) , although higher elasticities were reported for Morocco (0.54) and Indonesia (().55) . The average increase in household incomes of ().63% estimated to come about as a consequence of the PROAB subsidy produces a meagre increase of 0.18% in calorie acquisition. Since the subsidy effect is insignificant the marginal propensity to consume (MPC) from subsidy income is equal to the MPC from all income. This result is not surprising if we consider the way the scheme operates. As regular outlets are used and households acquire most of the commodities they consume in the same place, it would be surprising if the households identified and differentiated their behaviour while acquiring the subsidized and non-subsidized commodities. Different results, as reported in several studies, are probably due to clear identification of income transfer (e.g., by means of food stamps and food-discount cards) . Also, the analysis showed that the quantities of the subsidized commodities consumed (with the exception of dry milk) did not change, thus reinforcing the conclusion that the MPC was not affected by the subsidy.
The estimator of household size is negative and significant, indicating that food consumption per AEU (adult equivalent unit) decreases as family size increases. This is a common result.
The education of the mother appears significant in the regression for income but appears insignificant when expenditures are used as a proxy. A similar result was obtained by Garcia and Pinstrup-Andersen 
The dummy variables added to control-area differences appear insignificant and cause little impact on the regression.
TABLE 10. Results of regression equations estimating calorie acquisition per adult equivalent unit per month
|Independent variable||Equation 1||Equation 2||Equation 3||Equation 4|
|Subsidy dummy (intercept)||-1,458.46||(0.14)||-5,917.67||(0.56)||2,141.18||(0.30)||-1,853.99||(0.24)|
|Subsidy dummy x Xi (slope)||363.76||(0.11)||729.13||(0.21)||0.42||(0.58)||0.41||(0.56)|
|Wife's education||2,601.15||(2.34)**||2,513.13||(2.21)**||12.25||(0.012)||- 114.29||(0.11)|
|Dummy for Mangueira||7,299.34||(0 99)||8,974.00||(1.33)|
|Dummy for Mustardinha||12,093.76||(1.77)||7,098.09||(1.13)|
|Dummy for Mini-Central||8,309.61||(0.93)||8,769.83||(1.08)|
Figures in parantheses are t ratios.
* Significant at 10 % level.
** Significant at 5 % level.
Effects on the nutrition status of children
Regression equations designed to measure the effects of PROAB on the growth of preschool-age children were specified previously.
The regression results were rather poor. This was not expected, because a similar analysis based on data collected from 726 households and 1,146 children in the Recife area found the following coefficients significant: income per capita, age of mother, age of mother as a square term, education of mother (dummy variables for educational level), a dummy for maternal employment, and age of the child . One possible reason for the poor results is that the sample for the PROAB study was taken from a low-income population and the dispersion for some variables was lower than that in a broader Recife area study.
TABLE 11. Results from regression equations estimating the effects of PROAB and other variables on the growth of preschool children
|Independent variable||Total sample||Target group|
|Mother's age squared||-0.0032||(0.429)||-0.0131||(1.526)|
|Sanitary conditions (water system)||- 0. 8763||(0.263)||4.7998||(1.985 )**|
|Sex of child||-0.0690||(0.036)||-0.8366||(0.361)|
The dependent variable is child weight as a percentage of Gomez weight-for-age standards. Figures in parentheses are t ratios.
* Significant at 10% level.
**Significant at 5% level.
Only three variables were significant in the PROAB regressions: income, the subsidy dummy, and household size (table II). Nutrition status increases with income and decreases with household size, as is expected. Thus, it is significantly affected by changes in household income. But the PROAB subsidy has a negative sign, contrary to expectations. The average increase in household income of 0.63% due to PROAB is estimated to decrease the Gomez indicator by 4.94 points.
Similar regressions were run for the target group and the control group and for the Mangueira, Mustardinha, and Mini-Central sub-areas, without the subsidy term. The results were much better. In addition to income and household size, the following variables appeared significant: birth order, sanitary conditions (presence of either potable water or sewer system), sex of child, and age of mother. The two variables that systematically appear as significant are income and sanitary conditions.
Effects on birth weight
A regression analysis was conducted to examine the factors that might affect birth weights. The results are shown in table 12. The only significant variable is prenatal care, that is, the systematic provision of health-care assistance to the mother during pregnancy. Regressions were also run for the target group, in which sanitary conditions appeared significant. This is consistent with the results obtained above. Thus, PROAB showed no positive impact on birth weight.
TABLE 12. Results from an equation relating birth weight to various explanatory variables
|Mother's age squared||-2.3258||(1.305)|
|Birth order||-103. 1809||(1.628)|
|Sanitary conditions (sewer system)|
|Sex of child||24.0536||(0.112)|
|Dummy for smoking||-340.5258||(1.378)|
|Dummy for abortion||-241.0363||(0.778)|
|Dummy for type of parturition||48.9128||(0.238)|
Figures in parentheses are t ratios.
* Significant at 5% level.
Although the r² is relatively high, the regression is considered poor because most coefficients are insignificant. The small size of the sample, the small number of studies quantitatively analysing the causes of low birth weight, and consequently the difficulties explaining weight at birth are counted among the reasons for the poor results.
PROAB had little success in improving food consumption, calorie consumption, the nutrition status of preschoolers, and birth weights. What appeared to be a low-cost programme with possible high cost-effectiveness turned out to be a transfer of income with negligible nutrition effects. No effects other than the ones from increased purchasing power were found. Moreover, these were meagre results, since consumption of the subsidized commodities did not change significantly, and the increase in purchasing power was mainly directed to consumption of non-food commodities.
Several causes are responsible for the negligible results. First and most important, the "low-income" area selected for intervention had a low percentage of households with malnourished children. Thus, even if PROAB had been effective it would not have been cost-effective because of leakages. Second, the scheme adopted for subsidizing basic foods gave great emphasis to some commodities with low income elasticities, minimizing the possible direct nutritional effects. Third, probably because 11 commodities were chosen, income transfer from the subsidy was not treated differently from other income, indicating that PROAB was not identified as a nutrition programme but rather as an income-transfer programme. Finally, some administrative problems not analysed in this report, such as vagaries in fixing subsidies, fluctuations in the regular supply of commodities, and the periodically poor quality of commodities, must have further undermined the effectiveness of the programme.
It must be mentioned that for some time PRONAN's philosophy emphasized the income-transfer effects of food and nutrition programmes as opposed to more nutrition-oriented programmes. But if the main goal was to achieve improvement in the nutrition status of preschoolers and a decrease in low birth weights, a two-step targeting procedure based on growth monitoring and follow-up care of pregnant mothers, as was adopted in Chile and Tamil Nadu, India, would have been a better choice. Given the recent data pointing to a decrease in the number of malnourished children and a decrease in the frequency of low birth weights in Brazil, even in low-income areas, a programme targeting at-risk individuals would be much more cost-effective.
Two results highlighted in the analysis were the importance of sanitary conditions as a major conditional factor affecting nutrition status, and the importance of prenatal health-care services in increasing birth weight (by almost 1 kg). These results suggest that nutrition programmes should be redirected to improving infrastructural services such as the provision of potable water and sanitary waste disposal, to developing health services that include the systematic follow-up of pregnant and lactating mothers, and to the systematic monitoring of the growth of preschoolers.
1. Berg A. Malnutrition: what can be done? Baltimore. Md, USA: Johns Hopkins University Press, 1987.
2. INAN. Subsídio pare o encontro de avaliação da execução do projeto de abastecimento de alimentos básicos em areas de Baixa Renda-PROAB. Brasilia: INAN, 1984.
3. INAN. Tabulações especiais. Brasilia: INAN, 1988 (mimeo).
4. Musgrove P. Fighting malnutrition: an evaluation of Brazilian food and nutrition programs. World Bank Discussion Paper no. 60. Washington, DC: World Bank, 1989.
5. Musgrove P. Que los pobres coman mejor: evaluación de programas destinados a mejorar el consumo alimenticio de familias pobres en el Brasil. Washington, DC: OPAS, 1987 (mimeo).
6. Garcia M, Pinstrup-Andersen P. The pilot food price subsidy scheme in the Philippines: its impact on income, food consumption, and nutritional status. Research Report no. 61. Washington, DC: International Food Policy Research Institute, 1987.
7. Galvao MRJ, Sampaio Y. Efeitos e impacto de saúde em um projeto de desenvolvimento rural integrado. Revista Econômica do Nordeste (Fortaleza) 1988; 19(4) :407-29
8. Gray CW. Food consumption parameters for Brazil and their application to food policy. Research Report no. 32. Washington, DC: International Food Policy Research Institute, 1982.
9. Alderman H. The effect of food price and income changes on the acquisition of food by low-income households. Washington, DC: International Food Policy Research Institute, 1986.
10. Coelho H de AL. Estado nutricional e condições sócio-económicas. Master's thesis, Institute of Nutrition, Federal University of Pernambuco, 1975.
Coimbra M. Dois modelos de intervenção nutricional no Brasil: PNS e PROAB. In: Castro CM, Coimbra MD. Problema alimentar no Brasil. São Paulo: Almed, 1985.
ENDEF. Consumo alimentar-antropometria: dados preliminares, Região V. Rio de Janeiro: IBGE, 1977.
Franklin DL, Harrell MW, Leonard JB. Income of donated commodities in rural Panama. Am J Agric Econ 1987; 69(1):115-22.
INAN. Políticas agrícolas orientadas pare a nutrição. Brasília: INAN, 1977.
Nunes R, Batista FM, Perez E, et al. Diagnóstico alimentar e nutricional do nordeste brasileiro. Recife: SUDENE, 1979.
PIMES. Abastecimento alimentar no nordeste urbano- grande Recife. Series Pesquisas, 5. Recife: PIMES, 1975.
Rocha JA. Baixo peso ao nascer, peso insuficiente e peso adequado em 5.940 nascidos vivos na cidade do Recife. Recife: Federal University of Pernambuco, 1985.
Sampaio Y, Campino AC. Food and nutrition interventions in Brazil. Food Nutr Bull 1991;13(3):190-201.
Zisman M. Peso ao nascer: evolução, distribuição e fatores de variação em amostras de diferentes condições socials do Recife. Recife: Federal University of Pernambuco, 1985.
Contents - Previous - Next