Just over a third of swaps trades reported to specialist repositories under European trading rules are correctly matched, data from the regulator shows, raising concerns over the quality and completeness of records regarding the derivatives market in the EU.

Only 40% of swaps trades reported under the European Market Infrastructure Regulation (EMIR) are correctly matched up at trade repositories, data from the European Securities and Markets Authority (Esma) has revealed, following a freedom of information request by Risk.net.

This release is thought to be the first time the European supervisory agency has disclosed an official matching rate since EMIR entered into force in January 2014. Although still low, the matching rates are higher than most market participants expected. An industry source says they were told matching rates were in single digits, while the treasurer at a multinational company was told they were higher than single-digits, but not by much.

“I was pleasantly surprised at the 40% figure, because, from my discussions with trade repositories, I thought the matching rates would be lower,” says the industry source, but adds: “I don’t think we can pat ourselves on the back as regulators would remain unhappy that rates are at 40%.”

The treasurer concurs, saying the rate is “still devastating”. Low matching rates limit the usefulness of swaps data submitted to regulators for monitoring systemic risk as they cannot be sure all the information is correct. Since January 2014, EU swaps counterparties have been reporting details of deals to trade repositories under EMIR. Unlike US legislation, the EU requires both counterparties of a trade to report the swap, known as dual-sided reporting. Regulators have justified this practice as a way to verify reported information is correct. Repositories must, therefore, start by pairing up the same unique code representing a single transaction, and then 58 of the data fields considered key by Esma need to be identical or almost identical in both reports to count as a successful match.

Matching rates have been a closely guarded secret. A source who spoke to Risk.net in September 2015 said repositories had been explicitly instructed by Esma not to discuss their pairing and matching rates publicly, although Esma declined to comment on the claim at the time.

Today, sources are still wary, mainly trade repositories, to disclose matching or pairing rates. Esma’s November 2017 annual report only disclosed pairing and not matching rates – something sources suspected was because matching rates were significantly lower. According to the report, pairing rates stood at 87% at that point.

But, in response to Risk.net’s freedom of information request, Esma has disclosed that pairing and matching rates in February this year stood at 86% and 40%, respectively. These figures are averages of the rates of all reports submitted by each trade repository in February and include the rates for reports reconciled between different repositories. The need to match so many different fields is a significant contributor to failures in matching rates and is not completely necessary, says the industry source.

“There are too many fields which need to be matched to count as a successful match,” says the industry source. “There may need to be a recalibration to assess exactly what fields are systemically important to conduct matching on. If you have 40-plus fields then it is really a very high watermark for success.”

David Nowell, a senior regulatory reporting specialist at consultancy Kaizen Reporting, agrees: “Those 58 fields that need to match, they are extensive and the chances of them matching are slim.”

Many a Slip

Differences in the interpretation of data fields also lead to lower matching rates as parties to a trade do not always agree on how all of the fields should be completed.

“Despite an EMIR requirement that counterparties should agree details before reporting, a lack of accessible trade confirmation mechanisms means trade details may not be agreed before a trade is reported,” says a source at a trade repository.

The same problem, however, can also occur due to separate reports of the same trade being stored in different repositories, because they sometimes store information differently, says the treasurer at the multinational company. He says small differences such as swaps having two legs can lead to unmatched reports: some repositories treat a two-leg trade as two separate reports, while others record it in only one report.

Text fields also lead to reports being unmatched, as trade repositories express the same information differently. For example, if a trade is governed by an International Swaps and Derivatives Association master agreement, some repositories may store the information as “Isda,” while others refer to it as “Isda master agreement.” Nowell says give-up trades, whereby a broker executes a trade, but another dealer clears it, also trigger breaks in a pairing. Under EMIR, the end-user and clearing member must report the details of the trade and not the broker. However, brokers often still report the trade, which can lead to pairing errors if either the dealer or end-user uses the unique code generated by the broker or identifies the counterparty in their report as the broker.

The source at the trade repository says sometimes reporting entities still do not report details of trades, especially if they use derivatives only occasionally and so don’t have an institutional memory of which trades fall under EMIR and how they should be reported. All of these factors drive down pairing and matching rates.

Progress made

Esma’s figures do show improvement has been made since the start of EMIR, at least for pairing rates. Esma previously disclosed that pairing rates were just 55% in November 2016. But, given that rate rose to 87% in November 2017, the response to Risk.net’s freedom of information request shows this improvement has now stagnated.

The introduction of a hierarchy assigning which counterparty to a trade generates a unique transaction identifier has led to significant improvements in pairing rates. More focus from reporting institutions on remediating errors has also improved pairing and matching rates, according to market participants.

“There has been a lot of focus by institutions to carry out exception management and remediation,” says Mark Husler, chief executive of trade repository UnaVista. “An increased regulator focus on the need to improve reporting quality, completeness and reconciliations has obviously increased the focus on operational improvements.”

Husler adds that repositories have also helped improve matching rates by allowing values in data fields that have a marginal difference to count as a match.

“Where we have been able to introduce tolerance levels for matching, it has helped significantly in enabling regulators to match off the transactions and the legs of derivatives positions,” says Husler. “It means even if certain attributes are a decimal place off due to a rounding error, those reports can still match.” While Esma has not previously published matching rates, one repository estimated this to be about one-third for over-the-counter derivatives and just 3% for exchange-traded derivatives in June 2014.

“Given time and resources, everything is achievable, but, as we know, there are many compliance demands on firms and they can only go so far,” says Nowell. “Trades with third-country firms cannot be paired, and even if the two counterparties are reporting, they have got to identify each other correctly,” says Nowell. That is not as easy as it sounds, he adds, because there is scope to choose the wrong legal entity identifier if two companies have similar sounding names.

Failure to report trades led to one of the most infamous cases of sanctions being meted out under EMIR in November 2017, when the Financial Conduct Authority issued Merrill Lynch International, part of banking group Bank of America, a £34.3 million ($44.2 million) fine for failing to report 68.5 million exchange-traded derivatives. This was considered a substantial penalty for a purely technical violation.

Our sister website, Risk.net, first published this story.