B.I.S.S Research White Papers

Culture and how to fix Data

Richard Robinson

Richard Robinson

A senior business executive with more than 25 years of experience in the financial industry, with a rare perspective that spans working in operations and technology positions at global custodial banks, international brokers, investment managers as well as core industry utilities.
I recently attended the annual end of year EDMCouncil Meeting in New York, and was happy to listen in to the well informed panel discussion that included Harold Finkel, Marc Alvarez, David Goldberg, Bala Ayyar, and Gavin Kaimowitz.

At one point, Bala made a very good point in response to a question about monetization of data; that too often, generating value from data is the focus as opposed to cost savings. In other words, a penny saved is a penny earned – but that perspective is typically less ‘sexy’ than a cool webpage with multi-colored analytical charts on it.

As I nodded my head in agreement (because, of course, efficiency is the first mantra of data governance), I began to think of the challenges we have in data, data governance, standards, and the financial services world. And the challenges in agreeing about it.

Just previous to the panel, Mark McQueen was introduced, who will be running Best Practice Groups focused on a number of activities – one of which is the creation of a glossary of terms to create a common language among data professionals.

It’s a worthy endeavor (and one that I plan to participate in), but bouncing between that and Bala’s comment, something came to mind.

I’ve done this before.

The International Society of Chief Data Officers, out of MIT (and started through the efforts of the fantastic Professor Richard Wang), began an effort last year to properly define what a ‘Chief Data Officer’ is, as well as some of the terms around the roles and responsibilities of a CDO.

I’ve been part of MDDL, vocabulary discussions, ISO15022, FIBO, LEI, ISO20022, FpML, FIX, and a host of other data-definition efforts over the past 25 years.

I was struggling with forming a question for the panel around this, and by the time I formed a semi-coherent point, and waved down fellow data wonk, John Bottega, for the mic, the panel was closed as Marc Alvarez gave a typically funny and sage closing point.

But what took me so long to figure out what I wanted to say? I wanted to jump right on Bala’s point that there needs to be more focus on cost saving. A number of audience members did grab the mic and echoed that point.

But that wasn’t it, really. It’s was just the cost saving point.

I wanted to ask why we continue to define and redo definitions. And why, when everyone agrees on the basic tenants, we seem to have trouble implementing, or convincing people of good data practices. I wanted to ask why, when two parties execute a trade, it is so hard to reconstruct that to report to a regulator (or why doesn’t the matched trade from the mid office suffice for a regulator’s needs)? Why is CAT (Consolidated Audit Trail) so hard, when it should essentially just be a copy of what we already send out post execution to the exchange or our counterparties? There are privacy issues in data, but they differ across jurisdictions and even between companies. We all seem to know and agree on the basic details, but argue nuance. What is the source of rogue data and why does it continue to be rogue?

Then I realized something.

Our challenge in the data world has nothing to do with data. Our challenge is culture;

We, especially in the financial industry, do not share well.

The inability to share may be the single biggest cultural challenge facing the advancement of data governance and transparency.

This is why we have umpteen slightly different versions of a definition for a Chief Data Officer. And why it gets re-written with each new glossary. Or, for that matter, why there are dozen of definitions for ‘end of day price.’

We, as individual firms, individuals, standards organization, regulators, and industry bodies, are unable to give up control. A fear pervades the industry that something we give away may suddenly turn out to be valuable, and therefore a lost opportunity or competitive edge. Everything is potentially chargeable intellectual property, and therefore potentially revenue generating. Open data is looked at with suspicion. Some of it may be buried in capitalism, but I believe it is deeper than that. The view we individually hold is the only view – we insist on static, single versions of our own truth. Our context is the only context, and we ignore that context is really a court jester running about ruining our concise and specific plans and definitions (that individual “we” applying equally to a firm, group, organization, or industry segment).

As an industry, we find it very hard to give up control of even commoditized data (information), and we are exceptionally poor at valuing data properly in context. Language dictionaries have multiple definitions for the same word, and a host of synonyms – yet we insist on creating closed glossaries with singular definitions, tweaked to our own sensibilities, down to the most specific unassailable meaning. Regulators create their own definitions, and their own versions of what a trade and transaction are. Firms insist on creating their own versions of financial instrument classifications. Standards organizations fight with each other. You can’t simplify inherent contextual complexity, but in data that is what we insist on.

The question that every data professional should ask first, then, is how do we start to change this culture? How do we start to enable us to share what should be shared, in an open way? How do we accept the different (and quite valid) contextual views that are shared by different sub-groups? How do we recognize where it is OK for firms or organizations to have different definitions and contexts – while still coming together and sharing common concepts when, in fact, they are the same?

How do we start to agree on what we should agree on, while also agreeing where we rightly should disagree? What is wrong with agreeing and sharing multiple contextual definitions and/or standards for the same ‘thing’? How do we create open data, where it makes sense to share? How do we enable an approach to sharing data akin to a technology cloud web service?

I believe we are on the right path, even if it is a bit wobbly. Efforts like what isCDO and EDMCouncil are undertaking, the work that groups like ISITC and SIFMA and ISDA produce, and standards work from OMG and ISO all speak to pockets of efforts in the right direction. But there is a tendency to focus on ‘my’ version, to keep it closed or behind the walls lest someone else take it. There still is more drive to insist everyone only use one version of the truth – that being your version, not theirs.

Changing culture is hard. But we need to address that first. We need to find a better way to share. We need to acknowledge – not necessarily adopt – multiple points of view and context. We need to unleash the power of open shared data. If we do these things, we are sure to find that data, in the end, is easy. 

It’s just data, after all.

Richard Robinson

Richard Robinson

A senior business executive with more than 25 years of experience in the financial industry, with a rare perspective that spans working in operations and technology positions at global custodial banks, international brokers, investment managers as well as core industry utilities.