Earlier this month ESMA published its Supervision Annual Report 2019 and Work Programme 2020. These reports are always interesting to gauge ESMA’s view on the adequacy of the quality of trade data reports (and what it’s going to do about it if the data doesn’t reach the required standard).
How to measure EMIR reporting data quality
ESMA describes the pairing and matching process as “one of the key data quality processes”. EMIR is a two-sided reporting obligation, i.e. both counterparties to a trade have to report their ‘version’ of the trade. If both sides report it with the same unique trade identifier and counterparty identifiers, then it will ‘pair’. If both parties report with the same common data attributes (such as quantity and price), then it will ‘match’. Clearly, if the trades don’t pair or match, then there is a data quality problem with the report from one or both of the counterparties.
So the pairing and matching statistics are key data quality metric. Unfortunately, the pairing and matching rates quoted in the report – 59% and 29% respectively – might suggest the quality of the reports is not up to the required standard.
Is EMIR reporting data quality improving?
In its 2018 Annual Report, ESMA stated that, as of September 2018, the average pairing rate across trade repositories (TRs) was 86%. We also know from a Freedom of Information request, that the average matching rate was 40%. This suggests that the quality of trade reports has decreased substantially. However, this is where we need to consider Disraeli’s view of “lies, damned lies and statistics” and be very careful with the story they tell. The 2018 pairing figure was derived from information provided by TRs and measured the rates against trade reports. The 2019 figure in the latest report is actually derived by ESMA based on the ‘trade state reports’ at a single point in time, so we are not comparing data produced using the same methodology.
Use of the trade state report (TSR) is a very interesting choice. The TSR, as the name suggests, is the latest state of a trade, taking into consideration subsequent modifications such as partial terminations. It sounds like a sensible methodology, but it has a big issue when it comes to exchange traded derivatives (ETDs). Unlike most OTC derivatives, ETD trades are immediately netted into a position level report. So the TSR for an ETD is actually a position level report rather than a trade level report. The significance of this is that, unlike trade reports, there is no agreement yet between the counterparties on what UTI to use on a position level report. Therefore, there is a significant probability (I calculate it at 85.256%) that the pairing rates for ETDs is very low.
However, there is a more important issue related to these statistics (both pairing and matching) – do they or do they not include delegated reports? With delegated reporting, one party reports on behalf of both counterparties – ie. it would be difficult for the reporting counterparty not to report trades that both pair and match. So, for the statistics to not be potentially misleading, we need to know the pairing and matching rates for non-delegated reporting.
Another interesting statistic in the annual report is the top reason for non-matching fields. Bearing in mind that these top ten breaks are for TSRs as of 9 August 2019, some of the fields, such as Product_ID_2 haven’t existed since November 2017. It is unclear what value offering statistics against fields that can no longer be corrected, actually delivers.
What is the true state of EMIR reporting data quality?
At Kaizen, our ReportShield™ assurance services test every field in every single trade state report to ensure that they are populated in accordance with the regulatory technical standards and the ESMA Q&A. We can also test the consistency of all the trade reports submitted against the actual values of the position level report.
If you are in any doubt about how the quality of your EMIR trade and position level reports measure up, contact us.