A detailed research report into the gathering of OTC and ETD data reveals issues with the work being done by central banks to clean the data, and missing or confused attributes.

A review of data being reported to Trade Repositories by the BIS shows that there is lots of data being collected by some countries, too little by others, and data quality issues for lots of reasons:

These exercises (this paper) have provided evidence of systematic misreporting. This can cause significant quality problems that hamper data aggregation and analysis. In particular, despite the requirement to include a UTI agreed in advance by the two counterparties, for a substantial fraction of the trades the opposite trading position (leg) cannot be found in the reported data. As a result, the two legs of a trade cannot be paired or reconciled. As the analysis suggests, the reasons include counterparties’ failure to report or to agree on the common UTI, counterparties’ misreporting of their own LEI – or their counterparties’ one – and the failure to properly report the closing of transactions. Even if both legs can be identified, counterparties might still disagree on the various objective characteristics of the trade, like notional values and timestamps. When discrepancies are large, statisticians might opt to drop observations or make simplifying assumptions. Alternatively, they can design procedures to identify systematic patterns of problematic trades (eg reporting entities, locations and counterparties that are more affected by the issue of unmatched trades) and correct them.

The PDF is hosted on this page and attached below.  Hat tip to Virginie O'Shea for spotting this.