An Empirical Comparison of the Mathematical Methods for the Time Series of the Supply and Use Tables Construction

Abstract

We investigate the relative effectiveness of the projection methods of Supply and Use tables in relation to Use tables. The empirical bases of the study are the Use tables of 28 countries for the period from 1995 to 2010 from WIOD project. We conduct a comparative study of three mathematical methods that have proven the most effective in constructing projection of Use tables for Spain and the Netherlands from the empirical study by Temurshoev, Webb, and Yamano (2011). In these methods, a Use table is constructed based on the benchmark table and the sums of the columns and the rows of the table under construction. The most effective of these methods is GRAS, a version of the classical RAS algorithm. The results of applying this method under the number of criteria are closer to the published tables than the results of the INSD method and Kuroda method, which are based on quadratic programming. We conclude that GRAS method is a priority in the extrapolation of Use tables for Russia. At the same time we have shown that in some cases the table cannot be balanced by GRAS method because of sig nificant changes in the structure of the table. In 80% of these cases the tables were successfully balanced by the two quadratic methods. In these cases the Kuroda method is the most effective.