The results from the NES 1996 in table 5 show that most of the workforce characteristics have a
significant impact on wage dispersion and that except for costs, none the technology characteristics have
an impact. 9 Education of production workers significantly reduces the managers- production workers
wage ratio. A greater proportion of women and minorities is associated with greater wage dispersion.
These results are consistent with those found in Black and Lynch (2000) also based on the NES 1996.
Not surprisingly, establishment that have a unionized workforce have lower wage dispersion. The
wage compression effect is however not significant in the manufacturing sector. Given that the pro-
portion of unionized firms in the manufacturing sector is not insignificant (in this sample about 35%
of manufacturing firms have a unionized workforce), this absence of compression effect suggests that
spillover effects might be important.
In the next part of the table, the results show the effect of the variables describing computer
use, workplace organization and employee involvement practices. The percentage of managers using
computer does not have a significant impact on wage dispersion suggesting that it may affect both
categories of workers similarly. 10 The percentage of non-managers using a computer has a significant
positive impact on wage dispersion suggesting a differential impact on the wages of managers and
front-line workers.
This result may be surprising given the common finding on the positive impact of computer use on
wages. One would expect the wages of front-line workers to be positively affected and therefore wage
dispersion to reduce. Given that the variable on computer use includes all non-managerial workers, it
may either not directly influence wages for front-line workers or influence it positively but in a smaller
extent than the wages of managers. One may also suspect that computer use captures greater managerial
ability which translates into greater dispersion due to greater wages for managers although the fact that
the percentage of managers using computer does affect wage dispersion rules out this last conjecture.
firms that are associated
Among the workplace organization variables, neither benchmarking, nor re-engineering have an
impact on wage dispersion. Although surprising, this result is similar to Black and Lynch (2000)
and Cappelli and Carter (2000) who find positive impacts of the same magnitude on the wages of
both managers and production workers for re-engineering and no significant impact for benchmarking.
Contrary to these two studies, the number of managerial levels significantly affect wage dispersion in
the manufacturing sector. One expects that greater hierarchical levels would be associated with greater
differences in pay.
9 The information on costs relates to energy costs, raw materials, goods and services purchased in the course of doing
business. This variable is the only technology related variable that is significant and is associated with greater wage
dispersion. This result suggests that firms may pass some of their costs onto production workers reducing their wages.
10Note that the effect remains insignificant when one does not include the variable on computer use for non-managerial
workers.
18