that adopted either the partial or the full system. Note that the estimated premium are greater in mag-
nitude, especially in the manufacturing case. Given that hourly wages for managers are approximated
using an average of 40 hours per week, the approximation tends to overestimate average hourly wages
for managers. The bias seems to be even stronger in the manufacturing sector. From table 6 one can see
that the differences in the wage premia are significant though the tests show lower level of confidence
than with the NES 1996.
Overall the results are similar for both years and both sectors in terms of the positively significant
link between employee involvement practices and wage dispersion. There is evidence of lower dispersion
in establishments that complement the practices with training compared to those who do not. The
evidence is significant in the non manufacturing case in the NES 1996.
To further investigate the effect of employee involvement workplace practices on wage dispersion,
the next sub-section analyzes cross-establishment variations in wage dispersion in terms of practices
intensity of use, measured by the percentage of workers operating under the given practice.
4.2 Wage Dispersion and Employee Involvement Practices Intensity
The objectives in this sub-section are twofold. The analysis first replicates the approach taken in
previous studies on workplace practices using the NES data. In these studies 8 , employee involvement
practices like problem-solving meetings, teamwork and job rotation are analyzed as percentage of non
managerial and non supervisory workers currently under a given practice. This measure allows one to
analyze the importance of intensity of use of such practices. In a first step, the analysis investigates
the impact of factors related to workforce, technology and workplace characteristics on within firm
wage dispersion. The impact of employee involvement is then analyzed after controlling for cross-
establishments variations in workforce, technology and workplace factors. A second objective is to
investigate the possibility that the effect of the practices on wage dispersion varies between firms with
low wage or high wage dispersion and present the results of an analysis based on quantile regressions.
The results on the impact of the employee involvement practices, in terms of percentage of workers,
on the log of the wage ratio are presented in tables 5 and 6 for the NES 1996 and NES 1993 respectively.
In each table, the three first columns describe three different specifications. The first column corresponds
to the base specification in which the log of the wage ratio is regressed on establishment’s workforce and
technology characteristics. The next column describes the result of the same regression where workplace
variables have been added. The third column corresponds to the previous regression and the inclusion
of the three employee involvement practices. The last two columns consider the third specification over
the sample of manufacturing and non manufacturing establishments.
8Black and Lynch (2000), Cappelli and Neumark (2001) and Cappelli and Carter (2000).
17