Given the somewhat old techniques to obtain regional input ouput matrices by hybrid methods such as the popular RAS method or the implementation of location quotiens, in your opinion, what are the methods that offer a more accurate estimation of input output tables? What do you think are recent promising methods to undertake this task?
Sajad, thanks for your answer; do you have any opinion about any recent promissory method?
I think there are two recent discussions, the one between Dr. Flegg and Thomo defending their FLQi against the criticisms received in an article by Lehtonen and Tykkyläinen, I think the FLQi is still the best choice between "up-down" techniques for regionalization . On the other hand is the proposal of Asuad and Sánchez who contemplate the use of spatial correlations (Moran'I) for the estimation of sub-regional matrices and go in a "bottom-up" process, however, I think this proposal has two disadvantages when calculating weighting coefficients; a) the statistic I will always depend on the distance matrix that has been considered between two geographical spaces (cetroid distance of a polygon, road distance, and if so, the infrastructure is underestimated) and b) when using the RAS method, keep the problem of underestimating regional imports and overestimating multipliers.