Does Measurement Error Explain a Paradox About Household Size and Food Demand? Evidence from Variation in Household Survey Methods

Several recent papers report a puzzling pattern of food demand falling as household size rises at constant per capita expenditure, especially in poorer countries. This pattern is contrary to a widely used model of scale economics. This paper exploits within-country differences in household survey methods and interviewer practices to provide a measurement error interpretation of this puzzle. A comparison of household surveys in Cambodia and Indonesia with the results from Monte Carlo experiments suggest that food expenditure estimates from shorter, less detailed recall surveys have measurement errors that are correlated with household size. These correlated measurement errors contribute to the negative effect of household size on food demand and cause upward bias in Engel estimates of household scale economies.