Poverty and Time Preferences
Leandro S. Carvalho
This paper estimates the time preference of poor households in rural Mexico by combining ex-perimental variation with simulation-based econometric methods. I use data from PROGRESA, a program that randomly assigned communities to treatment and control groups and paid cash transfers to poor households in treatment communities. The randomization implies that differences in the consumption proles of control and treatment households are due to the program. A standard buer stock model predicts how consumption will respond to cash transfers as a function of the discount factor. I use the method of simulated moments to estimate such a model by matching simulated treatment eects on consumption to sample treatment eects. My estimates
indicate that, under a range of assumptions, households in the PROGRESA sample have very low discount factors. It is dificult to reconcile their behavior with the much higher discount factors that have been estimated for U.S. households. I conclude that either poor households are very impatient or a richer model is needed to describe the consumption behavior of poor households in developing countries.