A macro-micro approach to compiling statistics

IFC Working Papers No 2
May 2008

Abstract

This paper outlines a so-called macro-micro approach to compiling statistics that has been developed by De Nederlandsche Bank. Instead of trying to assess the plausibility of all individual micro data in all individual reports, this approach focuses first on the plausibility of a corresponding macro statistic. It has been implemented by the introduction of new statistical techniques, a set of drill-down tools (SIRIUS), in order to restrict checks of micro data to what is really relevant, and an increased comparison of macro-economic data from different sources. It is demonstrated that this way of data-analysis, applicable to any aggregate statistics based on large amounts of micro data, enables the compiler to raise both the quality of the statistics and the efficiency of the checking processes.

JEL Classification: C10, C40, C42, C81, C82, C88

Keywords: Data quality, plausibility checks, data control, detection of outliers, drill-down
tools