Distrust in Data Keeps Risk Managers on Edge: Study

By Jill Gregorie
July 25, 2017

Only about half of risk management executives trust the accuracy of their organizations’ data, according to a June survey of 132 financial services companies conducted by the compliance software provider AxiomSL.

And while the credibility problem risk and compliance teams face — which often stems from disparate systems and poor data hygiene broadly — is hardly new, pending regulations are prompting many firms to finally take corrective action, one data management service provider notes.

“With the SEC modernization rule hitting this industry, now is the tipping point for everyone to start paying attention to the way we should be making investments in personnel and infrastructure, as well as developing a strategy and executing on it,” says Gary Casagrande, global head of investor communications and expense solutions at Confluence, a data management service provider.

The enforcement date for that rule is less than one year away, he notes. At fund shops, much of the skepticism stems from the fact that firms are running disparate data systems cobbled together after mergers and acquisitions, Gordon Elliot, AxiomSL’s COO, says.

“In going through mergers, you don’t necessarily consolidate all of your data systems as part of the integration process,” Elliot says. In fact, such discussions usually take a backseat to conversations about profit and growth, if they happen at all, he says.

As a result, fund shops are left with systems “that do the same thing but work with different reference data,” he says. Moreover, depending on the age of each system, they may also capture information spanning different lengths of time, he adds.

As a result, large institutions especially tend to find that they have “a lot of points of failure” in their data reporting, he says.

But even firms that have grown organically sometimes maintain separate systems for various trading instruments or asset classes, Casagrande says.

On top of reporting challenges, storing data in separate systems can make pre-trade compliance near impossible, says Terry Flynn, front office specialist and team lead at SimCorp, a Copenhagen-based investment management software provider.

Shops need to combine as many data systems as they can, says Marc Mallett, VP of product and managed services at SimCorp.

“Rather than 10 to 15 systems supporting the investment process, reduce that number down as close to one system as you possibly can, which will reduce the amount of data that needs to be moved and reconciled,” he says.

To identify where data sources can be linked, some firms create maps to show “where data comes from, where it goes and where it’s maintained,” says Arnold Wachs, principal and data management practice lead at the Hingham, Mass.-based investment operations research and consulting firm Cutter Associates.

Trimming down the number of systems on which a shop relies can foster transparency and make it simpler to validate data internally, rather than rely on custodians and fund accountants, Mallett adds.

“There’s a reliance on external parties to validate that the information they’re making investment decisions on is correct, timely and accurate when investment managers should have control of that themselves,” he says.

Before transitioning to a more streamlined system, firms should appoint a staffer to track and reconcile issues that may taint data quality, Wachs says. Those improvements will accumulate day by day, and eventually lead to a more accurate and cohesive system at large.

“The only way to effectively get rid of data issues is by getting to the root cause of them one at a time,” Wachs says.

For some firms, the transition to a unified system may be time-consuming, but it can reap benefits, says Florian Deoutteau, CEO of Dataiku, a data science platform developer.

“The difficulty is that in order to build lineage today, you have to rebuild the full workflow of data from scratch,” Douetteau says. Fund companies can build “quality checks” into the system, however, which will send automatic alerts to compliance teams whenever the process is unexpectedly disrupted, he adds.

Another way to restore confidence in a company’s IT infrastructure is by appointing a “data owner” for each segment of information — portfolio positions, trades, performance and holdings, for example — “so there’s one person that everybody knows they can go to for that data,” Wachs says.

A working group or committee can oversee the entire operation, and serve as the main point of contact if questions escalate, he adds. Some firms also appoint a chief data officer to lead the committee and help staffers abide by uniform policies and procedures.

Maintaining a “business glossary” of definitions to help employees understand what they’re doing when they work with data can also help, Wachs says.

Firms can assess the quality and accuracy of their data in a number of ways, Wachs adds. One is by conducting client surveys. Another is through systems that automatically scan sets for accuracy and completeness, he adds.

Data owners can also draft service level agreements (SLAs) containing objectives that reflect the needs of “the people they’re delivering data to,” Wachs says. At one firm he visited, the investment teams complained that the systems on which they relied for information were inefficient, despite the fact that the security data team boasted of missing only one yearly goal.

The problem was misaligned expectations. The data team’s SLA stipulated that reports would be produced within two days, whereas portfolio managers wanted them within minutes, Wachs says.

“You really have to find a situation where data owners talk to data consumers and start building that level of trust,” Wachs says.

Although addressing data systems may be arduous and resource-intensive, compliance officers should work to convince the C-suite to invest in such infrastructure and protect against risk, AxiomSL’s Elliot says.

“These aren’t exciting projects, and if you go to the CEO and say, ‘Hey, please give me $2 million or $5 million to fix our data quality,’ the CEO may not want to hear about it,” Elliot says. “But that doesn’t make it any less important.”

Originally featured in Ignites