To RAS or not to RAS? What is the difference in outcomes in multi-regional input--output models?

The global resource accounting model (GRAM), which is based on OECD input--output and bilateral trade data, is a multi-regional input--output model covering 53 countries and 2 regions. What differentiates GRAM from other state-of-the-art models in this field is that it does not use a matrix balancing technique, such as RAS, after the initial construction of the global intermediate coefficient and final demand matrices. Instead, it reproduces prescribed intermediate and final demand, and determines value added residually. This choice was made to alter the original data as little as possible and keep the calculations traceable. This simpler solution technique might, however, yield different results. This paper aims at identifying the difference between the current solution of GRAM and the solution of a RASed version of GRAM, thus contributing to the assessment of currently used methodologies in this research field. The short conclusion is that, even though some differences during the calculations are present, the calculated output (production) matrix does not differ substantially. The results show that larger differences are brought about by poor assumptions regarding missing or conflicting data rather than by applying or not applying a RAS procedure to the constructed global matrices.