The literature on trade liberalization in developing countries contains divergent views
regarding the impact of liberalization on employment, incomes, and poverty. While most
studies find aggregate welfare gains, they disagree over the distribution of these gains among
households. One view is that trade liberalization generates broad-based employment gains
across regions for skilled and unskilled labor, and that consequently income gains are shared
widely (Dollar and Kraay, 2001). Under these conditions trade liberalization contributes to
reductions in inequality and poverty. The alternative view is that employment and income
gains go disproportionately to the already better-off groups, with negative implications for
inequality and perhaps also for the poor (Rodrik, 2000).
Reality likely lies somewhere in between, with the outcome depending on specific
conditions in the country in question. First, the impact of liberalization depends on the level
and structure of pre-liberalization trade barriers, which determine the sectors that gain and
lose. A differential impact on sectors holds implications for the distribution of gains among
regions, skill levels, and income groups. Second, it depends on the pre-existing distribution
of assets, that is, of land, capital, and human capital. So, for example, if trade liberalization
benefits agriculture, and if land holdings are highly concentrated, then inequality could
increase. Third, the distributional impact of trade liberalization depends on the flexibility of
domestic markets, especially (but not only) for labor. Gains from trade liberalization are less
likely to be shared equally where labor markets are segmented and barriers hinder labor
mobility across sectors. Thus for specific household groups such as the poor, the impact of
trade liberalization depends critically on local market conditions and household endowments
(Winters, 2000).
These considerations are relevant to the impact of WTO entry and concomitant trade
liberalization on employment and incomes in China. In these regards the Chinese case has
some interesting features. China’s labor markets have historically (under socialism) been
inflexible and highly segmented. Domestic economic reforms have allowed greater labor
mobility, but many observers believe that substantial institutional barriers to labor movement 2
persist. Also, in China certain assets such as land and education, while not equally
distributed, are nevertheless relatively equally distributed by developing country standards.
These two features would have counterbalancing effects in that the former would tend to
cause the gains from WTO entry to be concentrated while the latter would tend to cause the
gains to be shared more broadly compared to similar liberalizations in other countries.
This paper examines the microeconomic determinants of rural employment and
incomes in China. Using survey data, we estimate income, wage, and labor supply functions
for rural households in China. Since the households derive income from agriculture and
sideline family businesses and since labor hired in such activities is fairly rare (see Bowlus
and Sicular, forthcoming), we must impute shadow wages derived from self-employment.
Together, the income, wage, shadow wage, and labor supply functions empirically describe
household income generation from employment.
Our analysis fills a gap in the literature. While the literature examining employment
and earnings in rural China is now quite substantial, most studies of China’s rural
employment analyze the determinants of occupational status, that is, whether or not
individuals participate in different types of work such as wage jobs or non-agricultural
sidelines (examples are Hare, 1994, 1999a, 1999b, Knight and Song, 1997, 1999; Michelson
and Parish, 2000, Parish, Zhe and Li, 1995, Rozelle et al.,1999, and Zhao, 1999a, 1999b).
Relatively few studies estimate rural labor supply per se, that is, hours or days worked
(examples are Knight and Song, 1997, and Yao, 1999). Even fewer studies estimate labor
supply as a function of wages, the relationship of greatest interest here.
To our knowledge the only study that estimates labor supply for rural China and
includes a measure of the wage as an explanatory variable is by Meng (2000). Like Meng, we
estimate time worked as a function of wages and other variables. Unlike Meng, and indeed
unlike the relevant literature for other developing countries (Jacoby, 1993, Skoufias, 1994,
etc.), our labor supply functions allow for the possibility that wages or shadow wages, and
labor’s response to these wages, can differ depending on the type of wage. That is, in our 3
analysis household labor supply is a function of not just one but of multiple wages. Thus it is
possible for labor supply to be more responsive to the market wage than to the agricultural
shadow wage, or vice versa. Finally, while other studies estimate total labor supply, we
estimate total labor supply and also its components, in this case labor supply to household
agricultural production, to household non-agricultural production, and to wage employment.
By estimating the components of labor supply, we obtain information about how wages and
other variables influence the composition of employment.