Quantitative Study of Underemployment in Agricultural Sector of Thailand Using Farm Village Development Database

以往使用泰国农村发展数据库（Farm Village Development Database）估算1986 - 2003年农业生产函数的工作，从统计学和经济学的角度来看都不令人满意。当使用Ohtsuka、Isvilanonda和Hossain等人基于1997年和1998年在中原素潘布里省的三个村（Wang Yang、Jora Kae Yai和Sa Ka Chum）进行的农户调查的微数据的替代数据时，估计的生产函数变得更加可取和可信。通过比较由估计生产函数得到的边际劳动生产率和个体农户水平上的工资率，我们发现三分之二的农户使用最优劳动投入，其余三分之一的农户使用超过最优劳动投入。这与这样一个事实形成对比，即在总体水平上，平均农户使用最优劳动力投入。2002年，Shintani也在同样的三个村庄进行了调查，并获得了很好的生产函数估计。通过比较劳动边际生产率和工资率，再次得到了关于最优和非最优劳动投入的类似结果。综上所述，我确认在个别层面上，部分农户存在就业不足，而在总体水平或平均水平上，就业不足并未被观察到。在分析中，对传统生产函数和随机前沿生产函数进行了估计。

Previous work on the estimation of agricultural production functions using Farm Village Development Database of Thailand from 1986 to 2003 was unsatisfactory from both statistical and economic points of view. When the alternative data based on micro-data from agricultural household survey in 1997 and 1998 conducted in three villages(Wang Yang, Jora Kae Yai, and Sa Ka Chum)from Suphan Bri province in Central Plain by Ohtsuka, Isvilanonda, and Hossain and others is used, estimated production functions become more preferable and credible. By comparing the marginal productivity of labor obtained from the estimated production function and the wage rate at the individual household level, we find that the two thirds of total agricultural households use optimal labor input and the remaining one third use more than optimal labor input. This contrasts the fact that, at the aggregate level, the average agricultural household uses optimal labor input. In 2002, Shintani also conducted a survey in the same three villages, and obtained good estimates of production functions. Similar results on the optimal and non-optimal labor inputs are again obtained by comparing the marginal productivity of labor and the wage rate. In conclusion, I confirm the existence of the underemployment in some parts of the agricultural households at the individual level, while the underemployment is not observed at the aggregate level or on average. In the analysis, both traditional production function and stochastic frontier production function are estimated.