Non-surveying Construction of a 47 Interregional Input-Output Table and Calibration of SCGE Model

本研究的主要成果归纳如下：1）我们构建区域间投入产出表的方法被视为一种混合方法，在该方法中，根据现有的各种社会经济数据估计的初始贸易系数被连续调整以满足一致性条件。2）我们的非调查技术是基于平衡因子方法，但与传统的平衡因子方法不同的是，我们的方法需要求解线性方程组来找到一个平衡因子。3）使用日本区域间IO数据的小规模样本数据，成功地检验了算法的收敛性。该算法成功地应用于47个地区的数据，估计和实际的区域产品之间的误差小于百分之一，但在一些地区的某些部门，误差超过百分之十。由于这些部门的产品数量相对较少，我们可以通过 ...更多信息 4）构建的区域间投入产出表由45个部门和47个地区组成，其中要素数量多达4亿多个。为了求Leontief逆矩阵，我们使用了GAUSS软件包。利用47个区域间表进行了相互依存分析，发现了大都市圈与周边地区之间经济相互依存的有趣结果。5）提出了一个新的贸易模型，其中包含了不同地区经济距离的空间优势概念。如果可以使用区域间IO数据作为基准均衡，则空间优势是可测量的，然而，在通常的分析中很少使用这种数据。我们已经开发了降维方法，其中三维参数的空间优势转化为二维参数的援助外生给定的参数在不同区域之间的旅行时间。这种方法使我们能够校准贸易系数没有区域间的IO数据。6）我们已经开发和检查了几个校准程序对应的基准均衡数据是很容易获得的。最坏的情况可以是除了用于整个区域的IO数据之外没有数据可用的情况。即使在这种情况下，我们的方法可以产生令人满意的结果，在构建跨地区IO估计。如果能够使用一个完整的区域间IO数据，通过将Chenery-Moses模型转换为Isard模型，该方法可以重现与给定数据完全一致的估计，由此我们可以得出结论，我们的方法提供了一个理论上一致且实际上可行的SCGE模型。然而，不幸的是，我们不能使用本研究中创建的跨区域IO数据进行SCGE分析，主要是由于时间限制和需要解决大规模的非线性方程组。由于这不是一个根本问题，而只是一个技术问题，我们对此持乐观态度。少

Main results of this research ate summarized as follows :1)Our approach for constructing interregional input-output table is regarded as a hybrid method in which initial trade coefficients estimated by the existing various social-economical data are successively adjusted to satisfy consistency conditions.2)Our non-surveying technique is based on a balancing-factor method, however, it differs from the conventional balancing-factor method in that our approach should solve linear equation systems for finding one of the balancing-factors.3)Convergence of the algorithm was successfully examined with usage of small size sample data of Japanese Inter-regional IO data. Application of the algorithm to 47-region data was successfully completed with less than one percentage errors between estimated and actual regional products ; however, for some sectors in several regions, the error exceeds more than ten percentages. Since the amounts of products in such sectors are relatively small, we can expe … More ct that those errors can be reduced by sector aggregation and do not have significant effects on further analysis as well.4)Constructed interregional input-out table consists of 45 sectors and 47 regions, of which elements account for as much as more than four hundred millions. For finding the Leontief inverse matrix, we have used GAUSS software package. Interdependence analysis is conducted using the 47-interregional table, and we have found interesting results about the economical iuterdependences among metropolitan areas and surrounding regions.5)We have proposed a new trade model which includes the concept of spatial advantage in terms of economic distance for distinct regions. Spatial advantage are measurable if one can use interregional IO data as the benchmark equilibrium, however, such data are rarely in use in the usual analysis. We have developed the dimension-reduction method in which three-dimensional parameters concerned with spatial advantage is transformed to two-dimensional parameters with assistance of exogenously given parameter on travel times between distinct regions. This method allows us to calibrate trade coefficients without interregional IO data.6)We have developed and examined several calibration procedures corresponding to the benchmark equilibrium data which are easily obtainable. The worst situation may be the case in which no data except the IO data for total region is available. Even in such a case, our approach can produce satisfactory results in constructing interregional IO estimates. If one can use a complete interregional IO date, the method can reproduce completely consistent estimates with given data by using transformation technique of Chenery-Moses model to Isard type.From these results, we can conclude that our approach provides a theoretically consistent and practically feasible SCGE model. Unfortunately, however, we could not implement SCGE analysis using the interregional IO data created in this research, mainly due to time limitation and the needs of solving a large scale of nonlinear equation systems. Since this is not an essential problem and merely a technical one, we are optimistic with respect to that. Less