Identification of Superior Data and their Impact on the Statistical Reliability of the Regional Input-Output Table Using the New Mixed CHARM-RAS Method

Regional input-output tables (RIOTs) are considered strong tools for planning and policy making in the regional level, so the calculation of regional Input-Output tables and regional Input-Output Coefficients (RIOCs) have attracted much attention of most of input-output analysts. Compiling of RIOTs by using the statistical methods is costly and time consuming. So, since the 1950s regional analysts have introduced non-survey based methods like Location Quotients (LQ, such as , , , , , , and ), Commodity Balances (CB) and Cross Hauling Adjusted Regionalization Method (CHARM) for estimating Regional Input-Output Coefficients (RIOCs) and RIOTs. These alternative methods motivated lively debates regarding the reliabilities and accuracies of regional output multipliers.
For eliminating the shortcomings of non-statistical methods, researchers have introduced hybrid methods or simultaneous approaches “from up to down and down to up” and they have been leaded to use statistical methods (which are collected from real statistics, interview with researchers and also resources) for improving the accuracy of tables. The accuracy of hybrid methods is higher than non-survey methods and some economist such as Richardson (1986) and Lahr (1992) concluded that the mechanical non-survey methods are unsatisfactory; the short cuts are ingenious, but probably unacceptable, and therefore, the forward of RIOTs lies with mixed survey/non-survey and other hybrid methods.
The departure point of hybrid methods in calculation of RIOTs is using one non-survey method. Since collecting superior data is expensive and time consuming, the question that crops up for analysts is to identify the cells in which the collection of statistics on those data bases plays a greater role in improving the statistical accuracy of estimating. So, identifying cells related to superior data have been noticed by researchers and some criteria have been introduced like important coefficients and key sectors. The first one is applied for identifying individual cells and the other is applied for identifying an entire rows and columns.
The main objective of this paper is to demonstrate that the use of superior data in the hybrid methods of estimating regional input-output tables (RIOTs) can improve the statistical validity of tables as well as coefficients. Calculation of this paper is based on two kinds of data; one national and regional (Gilan province) symmetric activity by activity input-output tables and second, regional accounts for 1381.
In this paper, the most appropriate criterion for identifying superior data in the RIOTs is selected, using the new CHARM-RAS mixed method and the seven criteria namely LARGE1 (the largest cells in the intermediate deliveries matrix), LARGE2 (the largest cells in the direct input coefficients) and INVIMP (inverse important coefficients) COLSUM (column-sums of the Leontief inverse Matrix), ROWSUM (row-sums of the Leontief inverse Matrix) and COLHYP (the impact of hypothetically extracting an entire column on the whole economy) and ROWHYP (the impact of hypothetically extracting an entire row on the whole economy).
Results of this paper indicate that, first of all, regardless of the criteria used to identifying superior data, the use of these data will improve the accuracy and the statistical validity of the tables. Second, the highest and the lowest improvements in accuracy are related to LARGE1 and ROWSUM, respectively. Third, while criteria in identifying of individual cells have the higher improvement in accuracy than row and column criteria, but with respect to the very high costs of collecting individual cells, there is a trade -off between the statistical credibility and the cost of collecting individual cells. Fourth, column and row criterion in the hypothetical extraction method is more applicable compared to traditional method. Hence, column criteria have less statistical errors than row criteria. Therefore, the most appropriate criterion for identifying superior data is the COLHYP criterion.