Econometric Models and Methods for Project Evaluation

在本研究项目中，我们运用计量经济学模型对日本家庭微观数据进行了分析。首先，我们分析了日本已婚女性是否进入劳动力市场的决定。由女性劳动参与和工资方程组成的计量经济学模型被广泛使用，早在Gronau（1973）和Heckman（1976,1979）就进行了大量实证研究。众所周知，除非妇女工作，否则她们的工资是不被观察到的。因此，在模型的估计中必须考虑样本选择偏差的问题。然而，传统的估计方法，如Heckman的两步法，没有得到充分的检验，因此可能导致误导性的政策含义。在本项目中，日本已婚女性劳动参与率和工资方程分别采用传统方法和新提出的同时最大似然法，利用家庭经济研究所的“日本消费者小组调查”进行估计。其次，使用从“日本消费者面板调查”中获得的家庭数据面板，我们研究了家庭如何应对流动性约束。我们采用Dynan（1993）的计量经济学框架，估计了两类家庭的家庭消费欧拉方程；也就是说一个面临流动性限制而另一个不面临流动性限制。采用5型Tobit模型进行分析。我们发现，有流动性约束的家庭比没有流动性约束的家庭有更强的储蓄预防性动机。为了分析家庭微观数据，本项目开发了几种新的计量经济学方法和模型。这些方法和模型可以用于家庭分析以外的问题。我们还利用本项目开发的方法和模型，在相关领域做了一些应用研究工作。

In this research project, we analyzed Japanese household micro-data by econometric models. First, we analyzed the decision of the Japanese married female whether she would attend the labor market or not. Econometric models consisting of female labor participation and wage equations were widely used and many empirical researches were done as early as Gronau (1973) and Heckman (1976,1979). As widely known, wages of women are not observed unless they are working. Therefore, the problems of sample selection biases must be considered in estimations of the model. However, conventional estimation methods, such as Heckman's two step method, have not been sufficiently examined and therefore may result in misleading policy implications. In this project, the Japanese married female labor participation and wage equations were estimated by the conventional methods and by a newly proposed simultaneous maximum likelihood method using the "Japanese Panel Survey of Consumers" by the Institute of Household Economy.Secondly, using the panel of household data obtained from "Japanese Panel Survey of Consumers", we examined how households coped with liquidity constraints. We adopt the econometric framework from Dynan (1993) and estimated Euler's equations of household consumption for two types of households ; that is one was facing the liquidity constraints and other was not facing the liquidity constraints. The Type 5 Tobit model was used in the analysis. We found that the households with the liquidity constraints had stronger precautionary motives for saving than those without the liquidity constraints.For analyzing the micro-data of households, we developed several new econometric methods and models were developed in this project. These methods and models could be used problems other than the analysis of households. We also did several applied research works in the deeply related areas using the methods and models developed in this project.