Robert E. B. Lucas and Sari Pekkala
IED Discussion Paper 130, March 2003

There is strong empirical evidence suggesting a positive correlation between incomes of children and those of their parents. Two explanations have been put forward for this inter-generational transmission of earnings: 1) In families with credit constraints, lack of sufficient funds, or wealth, may limit the educational attainment of children. 2) There might be a genetic transmission of unobserved abilities from parents to their children. Most studies on this subject have tried to measure the magnitude of the inter-generational transmission of earnings without attempting to disentangle these two mechanisms. Lucas and Pekkala make use of an extensive dataset from Finland to estimate the magnitude of the effect and to disentangle effects of credit constraints and the effect of inherited ability.

A persistent problem in estimating the inter-generational transmission of earnings is inconsistent estimates generated by time-dependent errors. To deal with this problem, the authors try two approaches. First, they use a between-individualsestimator by averaging earnings of children and their parents over years with non-zero earnings. Second, they adopt an instrumental-variables method. Their findings suggest that errors in measurement lead to a downward bias in the ordinary-least-squares estimates, which is corrected using a set of instruments. The estimates indicate a large elasticity of children’s wages with respect to

family income. Earnings of the family head appear to have no significant effect, providing evidence against the ability transmission story.

Since ignoring the low and zero earnings observations leads to a sample selection bias, Lucas and Pekkala then include these observations. The point estimates of the effect of family income on children’s earnings increase significantly. In Finland, children of low-income families are more likely to register as unemployed. This reinforces the correlation between family income and children’s earnings. The authors then extend the theoretical model to allow the inter-generational transmission elasticity to vary with age. They make use of the panel feature of their data to adjust the empirical specification and find that elasticity estimates rise systematically with the age of children, particularly for sons. Lucas and Pekkala conclude that the age of the children in the sample affect the estimates, providing an explanation of the sensitivity of U.S. based estimates to the age cut-off of the sons in the family.

The results of this paper suggest a low transmission from parents’ earnings to those of their children. Total family income, on the other hand, has a significant impact on the earnings of the children. The findings support the credit constraint hypothesis while rejecting a transmission of inherited abilities across generations. Furthermore, the dependence of the inter-generational transmission elasticity on age suggests that analysts should be cautious when making comparisons across nations.