The rapid diffusion of computers into the workplace coincided with increasing wage inequality. If computers are complementary to human capital, then they will have contributed to this widening by increasing the demand for skilled workers.
What seems a quite straight-forward question has not been easy to prove. However a new paper by BLS economists Cindy Zoghi and Sabrina Wulff Pabilonia may give us some answers. Their working paper Which Workers Gain Upon Adopting a Computer? uses two panels of the Canadian Workplace and Employee Survey to estimate a first-differenced model. Their results are consistent with a skill-biased explanation for wage premiums for computer users:
..controlling for individual and establishment fixed effects, we find that within a year of adopting a computer, the average worker earns a 3.6 percent higher wage than a similar worker who did not adopt a computer. Returns are even larger for managers and professionals, highly educated workers, and those with significant prior computer experience. Employees who use computer applications that require high cognitive skills earn the highest returns.
UPDATE: For a recent paper with different findings, see Computer Use and Earnings by Mary A. Silles. This paper used longitudinal data for the UK to investigate the observed correlation between computer use at work and labor market earnings.
Our findings suggest that there are no returns to computer use at work. This is evidence against the productivity interpretation of these returns and supports the view that the premium can be attributed to unobserved characteristics.
Another recent study worth reading was written by San Francisco Fed economist Robert G. Valletta last year. His paper, The Computer Evolution: Diffusion and the U.S. Wage Distribution, 1984-2003 (PDF), does not find the diffusion of computers has been a mechanism for skill-biased technological change and consequent rising wage gaps in the United States:
Analysis of data from the periodic CPS computer use supplements over the years 1984-2003 reveals that the wage increase associated with workplace computer use was relatively constant over time and across educational categories during this period, with the notable exception of an increase in the relative return to computer use for highly educated workers between 1997 and 2001. Despite this advantage in computer use for college-educated workers in recent years, quantile regressions and conditional density estimation reveal that the diffusion of computers had little overall impact on the distribution of earnings.