Macroeconomic indicators are very often revised and the size of revisions, computed comparing subsequent estimates with previous ones, allows to assess their reliability. Along with accuracy, reliability represents a dimension of the statistics quality and is considered in the twelfth of the fifteen principles of the European statistics code of practice (Eurostat, 2011; see also IMF, 2012). Stating that "European Statistics accurately and reliably portray reality", the code of practice recognises the revision analysis as a tool "...to improve statistical processes". As stressed in de Vries (2002), accuracy refers to the closeness between the estimated value and the true unknown value and is assessed (when possible) evaluating the error associated with the estimate. On the other hand, reliability refers to the closeness of the initial estimated value to the subsequent up to the final estimated value and is partially assessed comparing estimates over time, i.e. analysing the revisions. In fact reliability is only one indicator of statistical quality and is not completely captured by revision analysis. Other aspects of quality in statistics are established in the Eurostat framework and include timeliness and robustness. Especially for short-term statistics is well known the existing trade-off between timeliness and reliability. For many of these indicators for which seasonally adjusted data are produced it is also important to distinguish between revisions in raw data and revisions in the seasonally adjusted data due to the seasonal adjustment method used.
The importance of the reliability, as one of the dimensions of quality, is confirmed by the growing interest in revisions on official statistics from international organisations (Eurostat, OECD, IMF) and National Statistics Institutes (NSIs). With the aim to produce as much as possible transparent statistics NSIs put efforts describing the revision policy, providing information about past revisions, scheduling future revisions, creating real-time data bases and analysing revisions. Most of these efforts are usersoriented because users see with a certain criticism the fact of revising economic statistics. However, in some cases, revision analysis could be helpful as well for data producer detecting possible "weakness" in the estimation procedures and to suggest suitable measures to counteract them.
Several recommendations have been set by international organisations (OECD, IMF and Eurostat) and NSIs (in particular ONS). In particular:
- revisions should follow a regular and transparent schedule (publicly available);
- preliminary and/or revised data should be clearly identified;
- non schedulable revisions due to errors should be communicated as soon as possible by data producers to the general public;
- real-time databases should be built for performing revisions analysis and made public at least for the main economic indicators;
- studies and analyses of revisions should be carried out routinely, used internally to inform statistical processes and made public (particularly for short-term statistics).
To read the entire document, please access the pdf file (link under "Related Documents" on the right-hand-side of this page).
Your feedback is appreciated. Please send your remarks, suggestions for improvement, etc. to email@example.com.