Exogenous information and input-output updating: An evaluation

Abstract

Input-output coefficients' intertemporal instability, costs, and time lags involved in the construction of survey-based tables necessitate employment of nonsurvey updating techniques. Analysts, however, may want to include exogenous information in the updating process. The issue, then, is whether this inclusion ameliorates or aggravates the results. This paper attempts to assess the wisdom of incorporating exogenous information into the updating procedure. First, using the naive, RAS, and LaGrangian techniques, the 1966 table of the former Soviet Union was updated to 1972. Next, treating the top lO percent largest 1972 coefficients as exogenous estimates, the remaining coefficients were updated via the same three methods. Comparison of the results indicates that exogenous determination of the largest coefficients does not change the methods' rankings while yielding substantial improvements in the forecasts. (JEL C67) BackgroundInput-output (I-O) tables are compiled via surveys and are utilized for several years thereafter. However, I-O coefficients are intertemporaUy unstable. Thus, the results obtained through using the latest survey-based tables may be less than reliable. Attempts have been made to device shortcuts for updating the tables without actual surveys. The efforts resulted in the emergence of numerous, nonsurvey updating methods, although the proposed methods are minimum requirement techniques. Many researchers have wondered if the updated tables could be made more accurate if certain segments of the target year's table were exogenously estimated and if the updating techniques were used only to update the remaining portions. The question is, first, should additional exogenously determined information be combined in the updating process? Second, if that answer is affirmative, then to obtain the best results, what criteria should be used to select coefficients for exogenous estimation? The answers to these two questions will enhance the understanding and usefulness of I-O analysis. If this procedure improves the accuracy of the results, then better updated tables can be obtained and the important segments of tables may be identified and treated more carefully during the survey and construction phases.