Most existing studies on the wage determination in China’s urban labor market are based
on the assumption of exogenous sector choices and are therefore subject to estimation
errors when sector selections are endogenous. One important source for such
endogeneity is the unmeasured workers’ productive abilities, which affect both
workers’ sector choices and wage levels, but are not captured by individual data set
and therefore not included in the estimations. This study reconciles the problem by
treating sector selections endogenously in the wage determination model. Lee’s (1983)
generalized selection-correction technique (mlogit-OLS estimate) is used to correct
selection biases in a four-alternative choice set by distinguishing urban employment in
China by ownership into four sectors: government (GOV), state-owned enterprises (SOE),
urban collective enterprises (UCE), and private/individual enterprises (PIE). The
estimation results indicate that there exists unmeasured worker heterogeneity across labor
market sectors in urban China. With respect to their unmeasured productivity, workers
adversely choose the state sector (GOV and SOE), but positively select into the non-state
sector (UCE and PIE). The extents of the selectivity in the four sectors can be ranked in
a continuum as PIE, UCE, SOE, and GOV, with PIE having the largest positive selection
and GOV having the largest negative selection. The study further examines and contrasts
three conceptually distinct measurements of the pairwise sectoral wage differential: the
conditional differential, the unconditional differential, and the discrimination differential,
with the discrimination differential measuring the premium received by workers
participating in one sector versus the other due to the sectoral difference in their rewards
to workers’ observed human capital. The results suggest that the wage settings in China 2
are discriminatory against the non-state sectors, with state sector workers receiving a
substantial premium over non-state sector workers.