Benchmarking and Reconciliation of Time Series: An Applied Bayesian Method

Abstract

The present article features a hierarchical Bayes method applied to solving problems of benchmarking and contemporaneous reconciliation across time series. This method enables the use of high frequency series to be either approximations or one or several related indicators. This method may be applied when facing flow or index disaggregation problems. The authors compare their results to classical procedures (viz., Denton univariate and Rossi multivariate methods) through the use of indicators. This article concludes that the suggested method bestows greater importance on the low frequency series profile, consequently providing smoother solutions than its counterparts.