Knowing your sources: Pillar 3 reporting and the role of data lineage

Data is the foundation of all regulatory reporting.

Across the full range of regulatory reporting requirements, banks must source information from many different systems, enrich and aggregate it at multiple points, and eventually produce reports that go to regulators and to the public. Sourcing high-quality, well-understood data for reporting purposes is a challenge given the complexity of banks’ systems, and an updated set of Pillar 3 reporting requirements due at the end of 2017¹ are set to increase that challenge even further.

Among a long list of new templates that banks will need to fill out, two stand out in particular. The “EU LI1” (see footnote 1, page 49) template requires banks to provide a distribution of the carrying values of their assets and liabilities across the different regulatory frameworks used to calculate capital requirements for those items. Building on this, the “EU LI2” (see footnote 1, page 52) template requires banks to take the values for their assets and provide a detailed step-by-step breakdown to show how these are transformed into exposure figures that feed risk-weighted assets (RWA) calculations.

In essence, this means they must show the links between the book value of their business activity with counterparties and the extent to which they might suffer loss if those counterparties default. Although these separate sets of values share a common origin, the calculations required to derive them mean that the actual amounts can differ by an order of magnitude, and bear no obvious relation.

Take the derivatives portfolio of a large investment bank, for example. Positions purchased (long market value) and sold (short market value) appear on the balance sheet as assets and liabilities respectively, with values reaching hundreds of billions of dollars. However, the aggregated exposure value of these positions may be calculated as only a few tens of billions. For banks to tell the story of how these values are linked, they must first have a comprehensive understanding of what the underlying data is, where it comes from and how it is transformed throughout the calculation process in deriving the end result.

The picture becomes significantly more complex when you consider that different teams have ownership of different areas within those processes. No single team across any business function can provide a global view from the original data sources to the final output, because no-one owns the data from end-to-end. Based on my experience in regulatory reporting departments to date, I believe many have historically not had the ability to generate this view internally, let alone produce something to be published to the marketplace.

Relying on existing MI alone is not enough to create the level of detail required in the EU LI2 template. I’ve come across teams in different business functions using different points of focus and reconciliation criteria for MI reporting, which can easily allow differences to go unnoticed. Even well-aligned hand-offs between operations, risk and finance teams can drift over time without sufficient coordination in place.

I believe that the most successful and comprehensive solution to this problem combines three steps: end-to-end business process mapping; technology architecture and data flow mapping; and robust data lineage analysis.

The first step is comparatively easy, assuming that some level of process documentation is available. However, the required level of detail is quite fine: every change that can impact values – application of individual accounting treatments; product reclassifications; exemptions and eliminations – must be isolated and understood.

The second step overlays a technology architecture perspective on top of this to illustrate where different data flows and data processing elements take place. This links the conceptual business process to its concrete implementation, paving the way to validate the knowledge built up previously.

Finally, the third step involves testing that the model is accurate by demonstrating that values can be traced from source to origin. This provides empirical evidence that data is well-understood and of good quality – or not, as may be the case. Any mismatched or missing data will show up clearly through this exercise, enabling targeted investigations and changes to resolve these issues.

The immediate benefit from this exercise is the detailed information needed to fulfil this Pillar 3 regulatory reporting requirement. But it doesn’t end there. In the bigger picture, the value realised through this approach is greater insight into the organisation’s financial data and confidence in its quality. This has positive implications for ongoing operations, governance processes and other regulatory reporting. And what CFO wouldn’t sleep a little bit better as a result?

 

 

1. “Guidelines on disclosure requirements under Part Eight of Regulation (EU) No 575/2013”, EBA – http://www.eba.europa.eu/documents/10180/1696202/Final+report+on+the+Guidelines+on+disclosure+requirements+under+Part+Eight+of+Regulation+575+2013+(EBA-GL-2016-11).pdf