Reconciling conflicting cross-border data sources for updating national accounts: The cross-entropy econometrics approach

摘要

The aim of the article is to introduce an efficient approach of combining data from various sources and to compare the results with traditional techniques used in official statistics. We used the power law-related Kullback-Leibler information divergence method, known to generalize Shannon entropy, to solve nonlinear, ill-posed inverse problems through the Bayesian philosophy. The proposed model is based on data from the most important cross-border point between Poland and Germany. Compared with traditional statistics techniques, this method produced a higher level output significance in the case of Polish balance of payments (BoP) estimation. Because of the universal character of this procedure, it can improve national accounts estimation, especially for small countries, more sensitive to cross-border processes.