The effects of data collection methods on calorie–expenditure elasticity estimates: a study from the Dominican Republic

Abstract

Using data from a nationally representative household survey in the Dominican Republic, we find that the estimates of calorie–expenditure elasticity vary widely depending upon the form of calorie variable used. Estimated elasticity is lower for calories purchased than for calories prepared for consumption in the home. The divergence between the estimates increases at higher income levels. There is a downward bias in the estimates obtained from purchase data due to some missing information on infrequent bulk purchases and greater memory lapse in recalling purchases. We conclude that surveys undertaken to estimate calorie–income elasticity should measure calories from consumption data. If purchase data are used, the period of recall should be flexible and defined by the respondent for each commodity so that infrequent purchases are captured. Survey methods should be standardized if elasticity estimates are to be compared across populations.