By Timothy Boobier, Louise Johnston and Gayle Sansum
The Bank of England’s Statistics and Regulatory Data Division has presented analysis of revisions to monthly monetary aggregates and effective interest rates on an annual basis since 2009. Revisions are examined on a rolling three-year window of published data after two years have elapsed. This article updates the analysis for 2010-12 data, covering broadly the same series as considered last year.
For monetary series data, the results are similar to last year’s analysis for 2009-11 data, where revisions to series were found to be broadly immaterial. This finding holds even though the introduction of certain one-off methodological improvements had minor impacts on some series over the period under review. These related to M4, M4 lending and their respective other financial corporations (OFC) components. All M4 lending series investigated in this analysis refer to those series which have been relabelled as M4L ‘historical measures’ from May 2015. For effective rates data the results are also similar to last year’s analysis, where revisions were found to be broadly immaterial.
Revision size, bias and materiality
As in previous analyses, this article evaluates revisions according to measures of their size, tests for bias in revisions and measures of materiality. Methodology and results for this analysis are presented in full in Annexes A and B. Definitions for the series investigated can be found in Annex C.
For monetary series, both seasonally adjusted and non seasonally adjusted data are considered. In both cases, results are broadly similar to those found in the 2014 analysis. Mean revisions and mean absolute revisions are broadly unchanged overall (Annex B, Table 1). However, mean absolute revisions have fallen slightly for more than half of the seasonally adjusted series considered, compared to last year. For effective interest rates data, revisions are slightly larger than those observed in the 2014 analysis, although as in 2014 these can be considered broadly immaterial (Annex B, Table 2).
No material bias was found for any of the monetary aggregates series (Annex B, Table 3). This was also the case for effective rates series (Annex B, Table 4).
Even in the absence of bias, revisions could still be considered to be material. However, revisions to most series appear broadly immaterial (Annex B, Tables 5 and 6).
This section considers in more detail selected examples of series that demonstrate the largest revisions for the series considered in this analysis.
Chart A depicts revisions to the one-month growth rate in M4 lending to OFCs, non seasonally adjusted. Among the non seasonally adjusted data, this series had the highest mean absolute revision (0.66 percentage points), although no evidence of bias was detected for this series.
Chart A shows that revisions are generally larger before February 2012. This reflects the fact that methodological changes to the estimation of this series were implemented for February 2012 data and this impacted the estimation of data before this point. These changes consisted of improvements in the estimation of the effects of price movements in OFC securities held by MFIs, specifically to exclude MFIs’ holdings of bonds issued by their own securitisation SPVs. Therefore, the size and materiality of revisions to this series should be seen as reflecting a one-off change in the measure’s construction. This change is also the key driver of the revisions to the one-month growth rate of aggregate M4 lending (which show similar results in terms of materiality).
Chart B shows original and revised estimates for seasonally adjusted M4. In this case, the materiality of revisions can be attributed to a change in the seasonal adjustment methodology implemented in November 2013. This involved a change from temporary adjustment methods to standard methods for the OFC component of M4, which drove the materiality of revisions for both aggregate M4 and its OFC component. Moreover, combined with the change in estimation of MFIs’ holdings of OFC securities, this helps explain the magnitude of the revisions to seasonally adjusted M4 Lending and its OFC component.
For effective interest rates, the series with the largest ratios of mean square revision to variance of the underlying data (one measure of materiality used by the Bank) were time deposits from OFCs and other loans to OFCs (Annex B, Table 6). As discussed in last year’s analysis, revisions to these series were due to a methodological change to exclude the reporting of effective rates for intra-group business. Revisions to other rates series were largely immaterial.
To view all related charts, tables and Annexes, please download the full article:
For questions relating to this article please contact firstname.lastname@example.org or call +44 (0) 20 3461 5356.