Will higher income improve nutrition?

Yes, according to a conventional wisdom in development economics in the '70s-80s. This was based on several findings that higher income leads to higher food consumption. One estimate (now have to find out where did I read that) on income elasticities of food expenditure was 0.8. It means that economic growth or any process that leads to higher income is good: it also leads to improved nutrition.

One major problem with this view is that increased food consumption may not lead to better nutrition intake. When their income increases, poor people may opt for a different type of diet (better taste or more variety of food) but not necessarily better nutrition. So we need to look more on the direct relations between income and nutrition. Behrman and Deolalikar 1987 found that in rural Maharashtra, India, indirect measurement of elasticity was 0.77, but the direct measure was only 0.17.

Then, there are some issues concerning choice of empirical method. Here I will particularly look at one type of nutrient, calorie. But the issue holds for nutrient in general.

Income or expenditure?

A classical issue. Ideally we would use income as it is our variable of interests. But there are many problems in collecting income data. Moreover, current income is more volatile hence a more noisy proxy for permanent income. So, in many cases, we need to rely on expenditure. In general, measured elasticity is lower when we use income.

Calorie intake or availability?

A more accurate measurement is individual calorie intake. But that will require an intensive data collection. We need specially trained diet investigators to measure individual dietary intake in their houses for a specific time. Not that it is impossible - for example they did this in India (see Behrman and Deolalikar 1987) or Bangladesh (see Pitt, Rosenzweig, Hassan 1990).

In other cases, we need to rely on the calorie availability. That is, the information on the amount of specific types of food they purchased over a specific time, and convert them to calorie intake. The main problem with this approach are wastages and leakages. What is available may not equal to what is consumed. People just don't eat up all their available food, or household might have shared or served some of their food to guests or non-household member. Another problem is more often we can only have the data in the household, not individual, level.

Another issue is estimating elasticity at a single point, for example at mean or median, will not give an accurate picture of the variations across people from different income groups. A mean or median estimation may understate the income elasticity of calorie of the poor people. A better approach is to do separate estimation for different income groups. Non- or semi-parametric approach may also work, like Gibson and Rozelle (2002) did for urban Papua New Guinea. They found he income elasticity of calorie was 0.6 for the poor, but declined rapidly for once calorie intake reached 2100 cal/day.

The bottom line: higher income may not always prevent undernutrition. We need to look more into individuals and households make decisions, including intra-household allocation of resources. Among other things, women's education could be one significant aspect in improving household demand for, and quality of, nutrition, argued Behrmand and Wolfe (1982).