AND

We propose a new approach to national accounts compilation, which also serves as a formalization of current compilation practices. When formalizing the procedure, a distinction is made between (basic) data, national accounts identities and so-called indicator ratios. The latter are ratios of or percentage relations between national accounts variables, such as the relation between output and value added. Indicator ratios are currently used in national accounts compilation practices in order to make adjustments to the basic data or to fill in missing data. The latter use is particularly relevant when basic data are scarce, which is the case not only in many developing countries, but also in developed countries when annual accounts are compiled for recent periods. The (basic) data, indicator ratios and identities together are used in a Bayesian approach to estimate the values of national accounts variables and analytical indicator ratios based thereon. The amendment of the current prac-tices consists in introducing reliability intervals of basic data and indicator ratios, which allows for the use of a much larger number of indicator ratios in the compilation and checking of national accounts data. The Bayesian compilation approach makes it possible-in contrast to current practices-to use indicator ratios both as priors and as analytical indicators. This paper proposes a new and relatively simple Bayesian approach t