55
basin) received full compensation, it would receive 340 yuan per capita, an amount equal to
24 percent of the average household’s preprogram total per capita income in 1999 (Uchida, et
al., 2005).
The first-order conditions also imply that households that have limited liquidity and
cannot borrow will allocate more labor to farm production:
(9)
In other words, at the optimum, the household allocates land so that opportunity cost of
retiring the last unit of land (which is the marginal-value product of land from agricultural
production) equals the shadow value of compensation rate of the conservation set-aside
program. This constraint implies that the higher the shadow value of liquidity, the more the
household will allocate land to farming, ceteris paribus.
Finally, there is one more implication of the model. If the land rental market does not
exist (as has been shown to generally be the case in rural China) and the household’s liquidity
constraint is binding (which we assume is true in some households, especially for people who
live in poor, mountainous areas), then we expect that the program’s off-farm labor impact
should be affected by consumption-side characteristics in addition to production-side
characteristics. This relationship can be stated as
L = f ( Agg ; p, wω, δ, K, B, a, L ,τ ,τo ,T, z, zf ) .1
1 If only the land rental market is missing and the liquidity constraint is not binding (i.e., λB = 0 ), the model
becomes recursive and decisions on production and consumption are separable. In that case, the reduced form of