prevalence of incoming workers in the village. If communities have even more strict
policies to encourage the employment of village resiedents, and enterprises (either private
or collective) can hire incoming labor only when no village residents are willing to take
the job, we should expect the size of the labor force to be negatively correlated with the
proportion of incoming labor.
Empirical Specification
To examine these issues, we estimate an empirical model of the demand for
incoming labor in village industry. The dependent variable in our model is the percent of
incoming labor among the village industrial workforce. We estimate separate equations
for total incoming labor as a percentage of the village’s labor force, commuters, and
migrants. The model is estimated only on the villages with at least one industrial firm,
which in our sample included 87 villages in 1988 and 130 villages in 1995 out of the 215
villages surveyed.vii
In our final labor demand specification, we regress a measure of incoming worker
demand on variables that capture the costs of hiring, constraints on local employment
decisions, and provincial dummies. Separate equations are estimated for 1995 and 1988
since a Chow test of structural change indicates firms are behaving differently over time.
In addition to the local wage, we also include a lagged dependent variable in the 1995
equation. We believe that this variable will measure, among other things, the experience
and investments that firms have made in hiring incoming workers in the past, and can
serve as a proxy for the non-wage costs of hiring incoming workers.
In addition to variables measuring the direct costs of hiring, four other variables
are included. If a village has firms producing products in the light industrial category, a
21