A novel method for estimating multiregional input-output tables using data at different aggregation levels

Abstract

Estimating MRIO tables is often hindered by limited access to regional data. The paper presents a novel method for estimating interregional trade matrices based on a gravity-RAS approach using survey and non-survey data at different aggregation levels. The new aggregate-disaggregate-aggregate RAS method combines estimation of à priori matrices using aggregated survey data with RAS balancing using disaggregated non-survey data for multiple commodities. The paper uses data from the Swedish Commodity Flow Survey to showcase the method's potential to improve estimations of multiregional trade matrices, highlighting trade-offs between aggregation bias and sampling errors. The performance of the method is evaluated using Monte Carlo simulation in an approach that simulates both trade matrices comprised of multiple commodities and a data sampling process for collecting CFS data. Simulation results indicate that RAS balancing at a disaggregated level can significantly improve model accuracy compared to both aggregated and disaggregated methods, highlighting the effectiveness of disaggregate-level RAS balancing. The method is demonstrated using a case study based on Swedish Commodity Flow Survey data, which also illustrates common challenges in MRIO construction under real-world data constraints.