Internal Migration in Mexico: A Test of the Todaro Model

Abstract

The results of an empirical test of the Todaro model--a model of migration dynamics in which migration is a function of expected income differentials instead of simply wage differentials--using 1969-70 Mexican interstate migration data from the 1970 Mexican Census are reported and compared with empirical tests using data from other countries. The process of assimilation of migrants into both the formal and informal sectors of the labor force is examined in general terms and in the specific case of Mexico City. In the Todaro model the potential migrant responds rationally to economic considerations and acts to maximize the current value of the net stream of his or her expected income. The expected urban income is determined by the urban wage rate and the probability of obtaining a job in the urban modern sector. The probability of finding such a job is inversely related to the urban open unemployment rate. The expected rural income is determined by the rural wage rate and the probability of obtaining rural employment is assumed to be 1. The Mexican regression shows that distance has a negative and statistically significant effect on the migration flows between 2 states. The Todaro model should predict positive and significant signs for the earnings of destination area and negative signs for the earnings of the origin area. The unemployment variable which is inversely related to the probability of obtaining a job should be negative for the destination area and positive for the origin area. The Mexican data fail to provide a robust support of the Todaro model given this functional representation. Coefficients for the destination state are positive and significant for Mexico and for Venezuela. The Mexican data also produce a positive and significant coefficient for the earnings of the origin state; this is contradictory to what was expected. In Mexico the unemployment rate variable is not significant for the destination state but is significant and negative for the origin state. In the Todaro framework data for Venezuela make more sense as the unemployment rate in the destination state has a negative and significant impact on migration but is not significant in the origin state. The fact that the unemployment variables fail to function according to the Todaro hypothesis prediction does not provide sufficient basis for rejecting the Todaro model because an interstate migration model may be too aggregate and may blur important intrastate differences in unemployment rates. The Todaro model can be improved by introducing the probability of obtaining work in the informal sector and earnings levels in that sector.