Abstract

Transparency is a critical ingredient for any statistics. It enables the understanding of their compilation and methodology, and, ultimately, ensures reproducibility and trust. Transparency becomes even more crucial for historical statistics and long series, where changing concepts and observation units across time complicate comparability. In this regard, understanding the context of the data, through extensive information on the collection, compilation and methodological choices, is an imperative to avoid misuse and misinformation. Drawing upon the insights from the new BIS central bank total assets data set, this paper advocates for an "adaptive and multidisciplinary statistical practice", encouraging the integration of historical and communication perspectives in the compilation of long series. Background: using the Central bank total assets as an example, we will shed light on the following three aspects: 1) The degree to which transparency, through extensive qualitative information, is key for any historical statistics, particularly to enable users to understand their context and manufacturing; 2) For compilers, this translates into a number of lessons and practical considerations, particularly as regards the generation, dissemination and communication of long series; 3) This also calls for an “adaptive and multidisciplinary statistical practice”, notably from the historical and communication perspectives