ESMA Data Quality Report 2023: What does it say about SFTR?

ESMA published its 2023 Report on the Quality and Use of Data earlier this month. In this blog, we focus on the SFTR elements of the report and we also look at another ESMA report published the same week, the ESMA Market Report: EU Securities Financing Transactions Markets 2024.

The week of 8 April 2024 was an embarrassment of riches in the world of SFTR with not one but two reports relating to securities financing transactions regulation reporting, each highlighting serious issues with reporting data quality. 

SFTR ‘severe misreporting’ – fine?

Not once, not twice, but on three consecutive occasions, ESMA has raised the botheration of “severe misreporting” through a failure to close out outstanding securities loans through either modification of the open term field to false and addition of a maturity date/end date or booking of an early termination. This overreporting significantly distorts the regulator’s view of the SFT market and its ability to monitor and tackle genuine systemically significant issues or clear cases of market manipulation and abuse.

In the 2021 Data Quality Report, it was noted that the value of open SFTs was constantly increasing until November 2021 when an NCA took action against one counterparty that saw them book backloaded early terminations. The outstanding number of open SFT reports more than halved from c.7.5 million to <3 million by December 2021. This also resulted in a notable drop in the proportion of open SFTs made up by SLEBs from as much as c.85% down to c.65% between October 2020 and October 2021.

In the 2022 Data Quality Report, there were up to three entities in a single jurisdiction reporting ‘implausibly high values in the field 2.56 ‘loan value’ reported for the securities lending transactions’. This issue, while not fully resolved by the end of 2022 nevertheless saw total SLEB loan value fall from c.€250 trillion !!! to c.€10 trillion.

Three strikes and you’re out?

No fines, public disclosures and an apparent lack of regulatory jeopardy and sadly lessons do not appear to be learnt. Perhaps lightning does strike three times! This year’s (2023) data quality report highlighted that one entity was still overreporting during the first half of 2023. Their SLEB loan value represented 55% of the total market value until ESMA contacted the relevant NCA and oversaw the resolution plan and corrective actions. This saw their SLEB loan value market share drop to a relatively immaterial amount (based on ESMA’s chart).

Furthermore, ESMA noted that one counterparty was reporting repo principal amounts that accounted for c.15% of the overall market, reaching its peak in November 2022 at around €3 trillion. This entity was also populating the triparty agent field in error with the same LEI as the reporting counterparty. As remediation takes place, the market share of this reporting counterparty has dropped completely off the chart…

ESMA Market Report – data quality observations

In the meantime, the inaugural report on EU SFT markets (primarily focused on the repo market) was a welcome addition to ESMA’s market reports series.  

However, it’s important to highlight that issues with outliers, stale reports, missing maturity dates and implausible collateral to loan ratios meant that ESMA had to exclude up to 36% of overall collateral market value from their analysis. Neither the scale of the data exclusion, nor the implications for data quality appear to have been fully acknowledged by this report.  

The measure taken by ESMA to set thresholds for outliers based on “relevant monetary variable(s)” four standard deviations outside the median value indicate quite severe data quality issues. Setting the threshold so wide, I wonder how many values in the sample fell even two standard deviations outside the median. This measure resulted in a 4k reduction of the number of records in the sample, (0.6% of records) but a 19% or €2 trillion reduction in notional amount.

The removal of stale reports, including mismatched but paired double-sided trades and records with an event date older than one year resulted in another 97k records (7% of records) being excluded, resulting in a cumulative 25% reduction in collateral market value. The need to exclude 97k records does not bode well either for the desirability of two-sided reporting, nor having overly stringent reconciliation requirements encompassing so many fields with such limited (or non-existent) tolerances. In fact if you were to exclude two-sided records from a single source (delegated reporting and a single vendor enrichment), then the negative effects would be even more evident.

Finally, removal of term repos without maturity date and the exclusion of records with implausible collateral-to-loan ratios resulted in the loss of a further 5k records (1% of the total records) and a total of 36% reduction in overall collateral market value.

When data quality issues appear to have such an adverse effect on data analysis, measures of stability, regulatory oversight and transparency, maybe it’s time to revisit the rulebook and terms of engagement with reporting counterparties?

• Read our analysis of the MiFIR and EMIR angles of the 2023 Report on the Quality and Use of Data.
• For an independent, forensic review of your firm’s reporting data quality, please get in touch.