Indeed, even if no VA was calculate, LTAS were published for BRL. However LTAS appear to show very limited variation, and we would like to understand which methodological elements explain this. Furthermore, could you provide details on how the transition matrices used for PD and CoD are extracted and constructed ?

The methodology for calculating LTAS and PD/CoD is detailed in Sections 12 and 14 of the RFR Technical Documentation. The same methodology was applied to the case of Brazil when BRL was in our list of relevant currencies. Please refer to the “Background Material” section in the following link for the current and earlier versions of the RFR Technical Documentation: https://www.eiopa.europa.eu/tools-and-data/risk-free-interest-rate-term-structures_en

Your specific questions/comments are addressed below:

• We publish the LTAS even if we do not publish the VA for a particular currency as LTAS is an input not only in the VA calculation but also in the fundamental spread (FS) calculation as part of the calculation of the matching adjustment (MA).
• Given our methodology, it is to be expected that the LTAS remain very stable, as they are an average of daily spreads over a 30-year period. Therefore, when one month passes, only one month of historical data is removed from the average and one month of recent data is added, both of which have only a very small weight in the average. In addition, the re-construction of the missing spread data in the early part of the 30-year history with the average spread calculated from available data is a contributing factor to the observed stability, due to its smoothing effect on early history. The specifics of this calculation are available in Section 12 of the RFR Technical Documentation.

• The transition matrices data is downloaded at the beginning of each year from our data provider, S&P Global, once that data becomes available. The data is accessed through the website of CreditPro by running a report on transition matrices, separately for financial and non-financial issuers:

  Specifically, the following criteria are used for the report: 

  • Report type: Issuer
  • Calculations based on: Number of Issuers (All)
  • Frequency: Annual
  • Static Pool: All
  • Horizon: 1 year
  • Pool start dates: from xx to xx (e.g. 01/01/1996 to 01/01/2025)
  • Regions: All
  • Countries: All
  • Industries: GICS  Economic Sectors:
    • For financials: 40 - Financials
    • For non-financials: All GICS Economics Sectors except the category “40-Financials”
  • Vintages: All
  • NR treatment: Excluded
  • Display notches: No
  • Display base: Percent

We then apply a simple transformation on the extracted transition matrices, by grouping the CC/C/D ratings in a single “Default” category and treating “Default” as an absorbing state, with no possibility of transition to higher ratings.