Where e is an i.i.d. random noise satisfying the classical properties of a white noise. If work-
place practices are associated with greater (lower) wage dispersion the estimated γ will be positive or
(negative).
In order to better approximate LF and LM , variables related to average worker characteristics
for managers and production workers will be added to capture differences in worker quality. Firm
characteristics such as size, industry and its proportion of unionized workers will be included as controls
for differences in companies usually observed in the empirical literature to affect wage outcomes and
wage dispersion. The value of capital and equipment are also added as controls, for the same reasons.
3 The Data
This section introduces the dataset used in this paper and discusses issues such as choice of variables
related to workplace practices and measurement method for the analysis. Section 3.1 describes the
two surveys used for the analysis. Section 3.2 summarizes the different workplace organization and
employee involvement variables and discusses measurement issues in analyzing the effect of adoption of
the practices (intensity of adoption versus adoption against non adoption). Since the data come from
establishment-level surveys, information on workers is limited. Section 3.3 discusses this point.
3.1 The NES Surveys
The data come from the National Employer Survey (NES), an establishment-level survey of employment
practices conducted by the Bureau of the Census for the National Center on Education Quality of the
Workforce (EQW). The survey was first administered in 1994 to a nationally representative sample of
private establishments with more than 20 employees. It was repeated again in 1997 using the same
sampling frame as in 1994, based on the Bureau of the Census SSEL file. Public sector employers,
not-for-profit institutions and corporate headquarters were excluded from the sample. The questions
apply to firm outcomes during the year preceding the survey year.
These surveys provide detailed information on employers human resource management practices
such as recruitment strategies, organization of the workplace and training investments. They also
report information on a company’s equipment and technology and on the average characteristics of
its workforce. The workforce information is available for different categories of workers: managers,
supervisors, clerks, technicians and production or front-line workers.
The questionnaires for the two surveys are not identical. The NES 1993 has a detailed section
related to training investments and whereas NES 1996 has additional questions about school-to-work
transition and community involvement. On the other hand, there is a common set of core questions on