This post by Dave Curran originally appeared on Thomson Reuters’ Legal Executive Institute on May 2018.
NEW YORK — The concept of “culture” within financial services organizations is undergoing a transformation, primed by management’s growing realization of the importance of culture in maintaining a successful company. Add to this the increased use of data science, predictive analytics, and behavioral science to quantify previously “soft” concepts of culture; it’s clear the idea of organizational culture is being revolutionized.
As part of that revolution, bank management is making the commitment to “harden” these softer culture issues — based mostly around how people are given incentives to do the right thing — and craft a metric-driven strategy to successfully implement sustainable changes that would, hopefully, result in behavior that drives the institution toward better business and regulatory outcomes.
In an effort to help analyze the impact that metrics and behavioral science are having on the culture debate, we convened the second panel of Thomson Reuters’ “Bank Culture Reform & Behavioral Science” event on April 9, which again was hosted by the Legal Managed Services and Regulatory Intelligence divisions of Thomson Reuters, and by Starling Trust Sciences.
I was eager to moderate this panel, and stimulate the discussion with panelists Michael Schlee, Managing Director at Goldman Sachs; Azish Filabi, Executive Director of Ethical Systems; Mark Cooke, Global Head of Operational Risk at HSBC; and Stephen Scott, CEO of Starling.
The panel began — like its predecessor panel — by acknowledging that regulators, company management, and society as a whole are coming together around the idea that better bank culture is simply good for business. “We’re seeing that the community really cares about culture as a determiner of individual conduct within the organization,” noted one panelist.
Next of course, comes the tough follow-up question: What are the best ways to improve it? And — more to the point of this panel — does data science provide a new pathway?
However, as more than one panelist pointed out, an initial wrench in the system is that the data around culture — unlike hard data around profits, margins, costs, and assets — may be difficult to quantify or even identify. “To measure things, you need data and numbers,” explained another panelist. “To measure culture — that’s too meta for me to think about!”
Other panelists and some audience members disagreed, noting that there are ways to measure observable behavior, identify patterns and even locate the rogue elements that may be exhibiting abnormal behavior. “Like with any objective, you have to identify what about culture you consider important and how you want to measure it,” another panelist said.
Indeed, one panelist explained how a firm can measure a simple behavior — comings and goings in the office by keycard swipes or computer log-ons, for example — and then build-out a good predictor of errant or out-of-the-norm data points. “It becomes a relatively easy way of identifying conduct risk and looking for abnormal behavior,” he said.
“To measure things, you need data and numbers. To measure culture — that’s too meta for me to think about!”
Before the discussion got too far down the data road, one panelist urged that ethics needed to be brought into the conversation, if indeed the idea was to ultimately create a more positive organizational culture. “You have to layer in the ethics as a base,” the panelist explained. “Because ethics is really about an individual’s relationships, both with other people and with the company. The human element has to be brought in early on.” Several panelists agreed, adding that a discussion of ethics needs to involve “finding the root causes of behavior” so that good behavior can be encouraged or “nudged” and bad behavior discouraged.
Clearly, if you can assess conditions that lead to certain behaviors, and you can change those conditions, then those behaviors can also be changed for the better. “In this way, you can optimize performance and reduce risk,” a panelist said, adding that this all must be grounded in a thesis of how humans behave.
The AI Solution
The role of artificial intelligence (AI) in this debate soon took front and center, as the panel recognized the benefit of using properly crafted algorithms to sift through the massive amounts of banks’ internal data, identify patterns, and locate anomalies. “Our firm is not alone in making a huge investment in this space,” observed one panelist.
As the panel explained, AI could be extremely useful in looking at how people operate within the organization even through simple monitoring of phone calls and emails and effectively tracing the “neural network” of how people interact with one another.
“Tracking these cultural measurements is extremely important to get to an understanding of your organization’s culture and how it’s viewed by your people,” said a panelist, adding that by measuring what people are saying and what they’re doing, AI can identify areas of concern within a culture. In organizations where this has been done, key components of a positive culture have been identified, the panelist continued, and include such components as a strong organizational fairness, the ability to speaking-up, and a management team that has demonstrated commitment to cultural issues.
One audience member then asked the natural question: “But where to start?”
The panel agreed that a firmwide analysis is an important first step, building it around the questions of infrastructure (What do you have?), analytical capabilities (What can you do?), and an idea of the desired outcome (What do you want?) Then, and only then, should you try to craft the AI lay-over to get at the necessary data.
As one panelist summed up nicely: “Very quickly, you’re going to realize it’s a long journey and a long process — but well worth it.”
