Studies of economic assimilation, earnings dynamics, and race differences in wealth.

摘要

Chapter 1 presents new evidence on whether foreign-born workers assimilate, which we define as the degree to which the wages of foreign-born workers approach those of comparable native-born workers with additional time spent in the United States. While the existing literature relies on cross-section data, we use longitudinal data on native-born and foreign-born populations which allows us to control for fixed unobserved heterogeneity. Longitudinal analysis of immigration, however, may suffer from bias from sample attrition and outmigration to the extent that they are related to wages and wage growth. We draw on recent work by Hirano, Imbens, Ridder, and Rubin (2001) to develop an estimation strategy for use with overlapping rotating panel data which accounts for both sample attrition and outmigration. We provide identification conditions and an estimation procedure for the weighting function when both sample attrition and outmigration are present. We apply the methods using the matched Current Population Survey (CPS) for 1994 to 2004. Overall, we find little evidence that foreign-born workers assimilate. We also find that older migrants are more skilled than younger ones conditional on the year of entry. Our results suggest that analyses of immigrant wage growth based on repeated cross-section studies may be biased upward by individual heterogeneity. Controlling for this heterogeneity reverses the conventional result of economic assimilation.;Chapter 2 contributes to the literature on earnings dynamics. We incorporate uncertainty about future rental rates for human capital into an optimal life-cycle human capital investment model. We show that optimal life-cycle investment behavior implies not only individual heterogeneity in earnings slopes but also the presence of a persistent error process in earnings. Persistent errors are induced by the response of individuals in human capital investments to transitory shocks to the rental rate of human capital. We specify an econometric model of earnings dynamics that is loosely consistent with the solution to the worker's optimal investment decision and estimate the model using the National Longitudinal Survey of Youth 1979 (NLSY79). We confirm that heterogeneity in earnings slopes, permanent errors, and transitory shocks all play a significant role in earnings dynamics.;Chapter 3 uses new econometric methods to examine the large gap in wealth between whites and African-Americans. Methodologically, we close the gap between econometric theory and practice by applying the methods of Altonji and Matzkin (2005). We explore ways of implementing their LAR and SFD estimators when the number of control variables is large, as is common in real world applications. A number of studies have found that the effect of permanent non-asset income on wealth is smaller for blacks than whites. A partial explanation is that, as a result of past discrimination, the earnings of whites may be more strongly related to unobserved parental variables than the earnings of blacks. Altonji and Doraszelski (2005) investigate this possibility using regression models with sibling fixed effects to control for unobserved background variables. However, as they point out, their estimation strategy requires that the unobserved characteristics enter additively, which is unlikely. Consequently, we apply the LAR estimator. In addition, we use the SFD estimator to examine the role of differences in the wealth function and differences in the distribution of unobservables as the sources black-white differences in the link between wealth and earnings. Our results largely confirm the findings of AD and others that wealth is more sensitive to permanent earnings and to self employment status for whites than for blacks.