Some of the most prominent theories of rising wage inequality emphasize changes in the supply of highly-educated workers, skill-biased technical change, changing labor market institutions, as well as variation in wages across occupations, industries, and geography.
David Card has highlighted some problems and puzzles for some of these prominent theories and has been focusing on the role of firms. His 2011 speech to the Society of Labor Economists highlights some of his thinking on these issues, and his recent paper with Pat Kline and Joerg Heining models and quantifies the importance of firms for rising wage inequality.
Roughly speaking, they show that working for a “good firm” has been quite important for wage growth and that “good workers” are increasingly sorting to good firms. By good firms and workers, they mean firms and workers with large firm and individual fixed effects in wage regressions. In other words, you can take the same guy and move him from the typical firm to a good firm and his wages will really go up. In particular, they find:
The rise in the variance of the person component of pay constitutes about 40% of the rise in the variance of wages, the rise in the establishment component constitutes about 25%, and their rising covariance explains about a third.
They then use their framework to evaluate the importance of education, occupation, and industry in wage inequality.
We find that nearly all of the increase in conventional (Mincerian) return to schooling over our sample period is attributable to changes in the association between education and establishment quality. Better-educated workers in West Germany are increasingly concentrated in high-wage establishments, while less-educated workers are increasingly concentrated at low-wage establishments. We find that rising assortativeness between workers and establishments also explains an important share (around 40%) of the growth in inequality across occupations and industries.
[we ask] whether the proximate source of changes in the structure of wages involves the portable component of wages, or the component associated with which particular employer a worker is paired. The data reveal that both components are increasingly important – the gap between consistently high and low wage workers has grown, as in traditional one-factor models of skill pricing (Juhn, Murphy, and Pierce, 1993), but the gap between good and bad jobs has also grown, suggesting that job search and matching is an increasingly high stakes process – one that appears to be increasingly mediated through the education system.
[...] These findings raise many questions regarding the nature of modern labor markets. Are the patterns uncovered here present in other developed economies? Which factors drive the matching of workers to firms? Why are there large firm cohort effects in wages? How much of the displacement effects of job loss are due to establishment effects? How do shocks to the product market filter into changes in person and firm effects? With the proliferation of modern employer-employee datasets, we are hopeful that progress can be made on these important questions.