Panel Database and its Statistical Analysis on the Network

如果未观察到的个体影响和干扰相关，则合并横截面和时间序列数据上的最小二乘将产生不一致的估计值。众所周知，工具变量估计器为这些重复横截面产生一致估计器。最重要的一类工具是基于时间函数的，因为固定效应是时间不变的，因此必须与时间函数无关。微观数据通常包含关于出生年份等外生时不变特征的信息。这个工具变量估计量的特例是计算年龄队列细胞的组平均值，并将其用于与队列虚拟的回归。这种方法被称为“伪面板数据”。我们从1984年、1989年和1994年进行的家庭收入和支出调查中构建了伪面板数据。我们把户主的年龄和居住地点作为时间不变变量。我们汇总了家庭…的样本S按年龄和地点较多，其中1984年首领年龄在20岁至70岁之间，47个县被视为地点。因此，我们得到了以下队列。然后，我们估计了消费和流动性约束的模型。在该模型中，使用伪面板数据推导并检验了存在借款约束时的消费含义。估计的结果支持这样的假设，即无法以未来的劳动收入为抵押借款会影响消费支出。我们开发了一个公司数据库，其中包括日本2000多家公司40年来的财务数据。日本产业银行提供原始数据。该数据库建立在大阪大学经济学院的WWW服务器上。网址为http://panel.econ.osaka-u.ac.jp/micro/.利用这些数据，我们通过估计投资函数，检验了不确定性与日本制造业企业固定投资之间的关系。较少

Least squares on the pooled cross-section and time series data yields inconsistent estimator if unobserved individual effects and disturbances are correlated. It is well known that an instrumental variable estimator yields a consistent estimator for these repeated cross section. The most important class of instruments is based on functions of time, because the fixed effect is time-invariant and hence must be uncorrelated with functions of time. Micro data usually contain information on exogenous time invariant characteristic such as the year of birth. The special case of this instrumental variable estimator is to compute group means of age-cohort cells, and use them in regression with cohort dummy. This method is called "Pseudo Panel Data".We constructed pseudo panel data from the Family Income and Expenditure Survey conducted in 1984, 1989 and 1994. We consider the age of household heads and the residential location as the time-invariant variables. We aggregate the sample of household … More s by ages and locations, in which age of head was between 20 and 70 years old in 1984, and the 47 prefectures are treated as locations. Consequently, we have the following cohorts.Then, we estimated the model of consumption and liquidity constraints. In this model, implications for consumption in the presence borrowing constraints are derived and tested using the pseudo panel data. The estimated results support, the hypothesis that an inability to borrow against future labor income affects the consumption expenditures.We developed a database of corporate firms, which includes financial data of more than 2000 companies in Japan over 40 years. The Japan Development Bank supplies original data. The database is constructed on the WWW server at the Faculty of Economics, Osaka University. The URL address is http://panel.econ.osaka-u.ac.jp/micro/. Using these data, we examine the relationship between uncertainty and corporate fixed investment for Japanese manufacturing industries by estimating the investment function. Less