Measuring macro- and micronutrient consumption in multi-purpose surveys: Evidence from a survey experiment in Tanzania

The nutrition transition in developing countries has increased interest in moving the measurement and analysis of nutritional choice beyond calories to a more complete understanfding of macro- and micronutrient consumption. To help move the literature on data collection forward we randomly assigned six different survey modules to measure food consumption across Tanzania, three using diaries and three using recall methods. These modules were chosen to reflect the variety of modules currently in use in multi-purpose household surveys collecting food consumption expenditures in some detail at a national scale. Using the percentage consumed relative to daily recommended intake of calories, protein, fats, sugars, fiber and 16micronutrients, we find that the household diaries and 14-day recall systematically underestimate nutrient consumption compared to the benchmark personal diaries. The 7-day recall performs better, on average. We also calculate minimum cost diets and cost-of-basic needs food poverty lines in each region, the prevalence and depth of food poverty according to these lines, and the cost of targeted transfers designed to eliminate food poverty. The 7-day recall provides reasonable results whereas the 14-day recall grossly overestimates food poverty and therefore raises the cost of hypothetical monetary transfers needed to alleviate this poverty.