System improvement of the second statistics mainly on industry connection and expansion of an analysis

投入产出表自1955年以来每5年编制一次，是根据列昂惕夫的投入产出分析法编制的舒适统计表。在此统计整理表的基础上，定期对生产结构进行计量分析。特别是对于开放静态模型下的投入产出表，生产结构被描述为每个生产区段的输入矩阵。认为作为该矩阵的一个元素的输入系数是Leontief型生产函数的一个参数。本文试图利用制造业普查的微观数据对投入产出表中的理论假设进行分析。在利用普查数据构建面板数据后，我们将控制生产项目（标准行业分类第六位数）的企业分组分为(1)延续、(2)新进入、(3)退出，并估算劳动力投入系数。结果很明显，在相同的工业分类代码中，不同的业务条件（即继续、进入和退出）之间的劳动投入系数存在很大差异。例如，对于1990年最大生产项目相同的公司，退出(3)的公司系数最小，延续(1)的公司系数最大。其他行业与行业分类代码的数字在系数值上有统计学显著差异。为了提高投入产出表的统计准确性，考虑到输入系数的分布，选择样本可能是可取的。劳动投入系数。

Input-Output table has been produced every five years since 1955 it is the comfiled statistical table which is based on Leontief's input-output analysis. Based on this comfiled statistical table, econometric analysis of production structure conducted regularlly. In particular, in case of input-output table following the open static model, production structure is described as input matrix in each production section. It is considered that the input coefficient that is an element of this matrix is a parameter of Leontief type production function. In this study, we tried to analyse the theoretical hypothesis in input-output table by using micro data of Manufacturing Census. After constructing the panel data using this census data, we divided the establishment groups which controlled production items (sixth digit of standard industrial classification) by, (1) continuation, (2) new entry, (3) exit, and estimate labor input coefficient. As a result it became clear that labor input coefficients differ substantially among different business conditions (i.e. continuation, entry and exit) ever within the same industrial classification code. For example, for companies whose greatest production item is same in 1990, coefficient is smallest for companies with exit (3), and the largest for companies with continuation (1). The other industry with digits of industrial classification code shows statistically significant differences, in coefficient values. To improse statistical accuracy of input-output table, it may be advisable to select samples, taking into account of distribution of input coefficients. The labor input coefficients.