Ebook Are Income-Calorie Elasticity's Really High in Developing Countries?: Some Implications for Nutrition and Income
In the last few years there has been an intense debate on the relationship between nutrition and income. More specifically on the response of nutrition intake to rising incomes. This paper is about this relationship which has some far reaching policy implications for the developing countries on how best to reduce malnutrition. If the income elasticity is close to zero (Behrman and Deolalikar (1987)), the implication is that improvement in the income of the poor will have little impact on the extent of malnutrition. Then the developmental policies intended to improve nutrition will have to use policy instruments which attack malnutrition directly rather than relying simply on rising income. This debate was apparently triggered by the pioneering study of Behrman and Deolalikar (1987), who showed that, in the (six) ICRISAT villages of South India, the Income elasticity of calorie intake was quite low, and not significantly different from zero in statistical terms. Even among the very poor, as incomes rise households mostly purchase additional taste.
Critics on the other hand have concentrated their firepower on the finding that the income elasticity of calorie intake is low (Strauss and Thomas (1989), Ravallion (1990). Bouis and Haddad (1992), Deaton and Subramanian (1996)). Subramanian and Deaton (1996), for instance, questions the validity of the Behrman and Deolalikar (1987) initial findings, and based on the National Sample Survey data estimate the expenditure elasticity of calorie intake in rural Maharashtra. They find this to be in the range of 0.3-0.5 and in any case statistically different from zero. The debate appears to focus on the size of the calorie-income elasticity, especially at low incomes. See table 1 for a summary of various estimates from the literature.
The view that calorie-income elasticity is low is itself challenged by growing evidence that the size of the estimate is a function of the method of data collection. Low estimates are typically based on samples drawn from surveys designed to monitor nutrition. Higher calorie-income elasticity typically come from household expenditure surveys which are designed to elicit information on household consumption including food, along with other economic and social information. Not all surveys distinguish between the consumption of household members and that of visitors or hired help. Here two methodological issues, as pointed out in the debate are relevant. First, income and expenditure may be measured with random error. Second, it is possible that the error in measurement is correlated with the set of potential instruments. In general, data from household expenditure surveys are particularly susceptible to these problems.
However, some studies use instrumental variable techniques (e.g. Strauss and Thomas (1989), Bouis and Hadded (1992), Subramanian and Deaton (1996)) to get away from these errors. Using instrumental variables, Subramanian and Deaton (1996) show the calorie expenditure to lie in the range of 0.3-0.5, a range close to the conventional wisdom. And hence does not support the notion that nutrition will not increase with higher standards of living. The increase in calorie consumption with increase in income comes largely from the increase in cereals. We would therefore expect cereal consumption to grow over time. However, the empirical evidence is quite contrary to this popular claim. The official journal of the National Sample Survey Organisation (NSSO) Sarvekshana states that: "It was observed that per capita cereal consumption has registered an overall decline of about 1kg per month in rural India and about 600 gm per month in urban India between the 43rd and 50th rounds of NSS (GOI, 1996)".
"The findings of different NSS rounds from the 43rd (1987-88) to the 51st (1994-95) on quantity and value of all-India per capita consumption of important cereals . It is observed that per capita cereal consumption in rural India is declining gradually, the decline being mainly spread over wheat and jowar (GOI, 1998)".
However, the NSSO's own estimates of cereal consumption in rural areas suggest that it has fallen over time. Thus, the estimates of Subramanian and Deaton, which are also based on the same data source, but confined to one state rural Maharashtra, are seriously misleading.
The high elasticity estimate by Subramanian and Deaton (1996) is a result of applying the conversion factors to the broad groups as previously stated in the literature by Behrman and Deolalikar (1987). The total calories calculated by applying the conversion factor is higher for the top 10% than the bottom 10%. This is something, which is against the perceived evidence from the same data source.
In this paper, we relook at the evidence from India, so as Behrman and Deolalikar (1987), Subramanian and Deaton (1996) and Ullah and Roy (1998) to study this important relationship using two different sets of data. We use two data sets from the National Council of Applied Economic Research (NCAER)household surveys - HDI (1993-94) and MIMAP (1994-95) for reasons stated in the next section. Our focus is only on the rural samples but spread over all the 16 states in India, unlike that of Behrman and Deolalikar (1987) and Subramanian and Deaton (1993), which is confined to rural Maharashtra. We have 2712 rural households and are subject to similar criticism as Subramanian and Deaton (1993) of ignoring individual heterogeneity.
Our estimates of the total expenditure (income) elasticity of cereals are low (see table 1) and the cereal consumption declines with increase in income unlike that of Subramanian and Deaton (1996). This is not because the overall quality of the diet is unchanged with income but because calorie intake is a poor summary statistic for diet quality. A more comprehensive measure of diet quality or nutritional adequacy would have to include both calorie intake and the intake of other micro nutrients. We expect some of these micro nutrients to be potentially income elastic, which clearly deserves further attention.
The rest of the paper is structured as follows: section II questions the high elasticity of calorie-income relationship using examples from India. Further section III presents the definitions and data sources. Section IV presents the parametric evidence from HDI and MIMAP survey on the expenditure/income and cereal consumption relationship. We then present the nonparametric estimates of this important relationship in section V followed by policy implication and conclusion in the last section.
Posted in :
