The Relationship between Wage Growth and Wage Levels

In the last thirty years we have witnessed large increases in the "returns to skill." These changes in the wage structure have renewed interest in increasing the skill levels of low skill workers. Attempts to do this through job training programs have been largely unsuccessful as the wage gains from these programs tend to be quite modest. In rethinking questions about subsidizing skill formation it is useful to step back and explore the issue of wage growth among low skilled workers. Despite the large amount of work in labor economics devoted towards the wage process we know surprisingly little about the mechanics of wage growth, particularly among low skilled workers. This work furthers this knowledge by exploring the link between wage growth and wage levels building on our previous work. While many different policies to raise the wages of low wage workers have been proposed, the simplest and most common is increasing labor force attachment. One of the most robust findings in labor economics is that wages increase with work experience, however very little of this work has estimated the extent of this growth among low wage workers. In part, this hole in the literature may have arisen because there are serious econometric issues behind the wage growth process involving parameter heterogeneity, endogeneity, and selection issues. We are attempting this hole and address these issues. This work extends the literature on the covariance structure of wages by focusing on low skill workers using the National Longitudinal Survey of Youth and the Panel Study of Income Dynamics. Our previous work indicates that in examining wage growth among the poor, it is extremely important to include measures of actual experience. This is problematic in that labor market experience is likely to be endogenous and related to wage levels and wage growth. We have developed a framework that allows for these relationships. It incorporates individual "fixed effects" in both wage levels and wage growth. We are estimating this model using Generalized Methods of Moments. Our results to date find little relationship between wage levels and log wage growth. After completing this, we will simulate the impact of labor force participation on future wages of low wage workers. Identification requires strong assumptions about the error structure. While we cannot completely avoid these type of assumptions, we will test the robustness of the results using a wide range of alternatives.