UsableMarkets

The FT has a good article describing the amount of data the US Financial Crisis Inquiry Commission received from Goldman after a request for information about its mortgage-backed security trading activities. (Essentially a rather … cough … large amount.) While Goldman is hardly playing nice, the more serious issue highlighted in the article is how difficult it is to wade through data and data schemes from multiple, non-connected systems in order to make sense of market activity. Even within one bank systems can be disconnected.

From the article:

The US equity market is the most advanced in terms of the disclosure of information about trades and prices. However, although regulators can see which brokers placed the trades, they cannot see who placed the orders with the brokers. Special requests for this data can be made, but even then it can be difficult to add together. The same market participant may be known to different broker-dealers by different names. The data collected by different exchanges is all in slightly different formats – getting the whole picture, rather than small slices of it, is a huge task.

It is hardly controversial, then, to expect that similar fact-finding exercises for the privately-traded over-the-counter derivatives markets will be even more complicated. Yet this type of transparency is exactly what is needed.

Perhaps financial service industry regulators, in order to get a handle on this mess of data, should compel the banks to adopt a consistent, standardized set of metadata to describe the data their systems consume and produce.

One suggestion might be taxonomies from EDM Council. “The goal of the EDM Councils Semantics Repository is to standardize the terms and definitions of all reference data attributes stored in the master files of financial institutions and passed among supply chain partners.”

What a nice thought: consistently described data. This might make the regulators jobs a bit easier. And I’m pretty sure the banks wouldn’t object much, either. Who doesn’t appreciate (and find value in) understandable data?

~alex