Abstract

This study evaluates the behavior of non-survey methods for estimating regional output multipliers. A Monte Carlo simulation is carried out to generate multiregional input-output tables that are assumed to be 'true'; the aggregation generates national input-output tables from which regional input coefficients are obtained using different location quotients. Then, the output multipliers estimated are compared with the 'true' multipliers through a set of statistical indicators to analyze their behavior. Unlike previous studies, three scenarios are considered that differ in the availability of historical information on the regional production vectors and the limits of the uniform probability distribution used to simulate the regional input coefficients. The results show that the SFLQ method is the best in all scenarios, although the FLQ and AFLQ methods with delta that vary by region also provide good results. Finally, it is concluded that placing the Monte Carlo simulation in a more realistic context using partial information substantially increases the precision of non-survey methods.