How is technology changing the way we think about risk?

Tuesday morning’s agenda at RiskMinds International in Amsterdam zoned in on best practice for risk managers and how technology could transform the work that they do. 

Keishi Hotsuki, Executive Vice President and Chief Risk Officer at Morgan Stanley has 25 years of experience in the risk management field, and shared some of his key learnings during that time.

Reconstructing risk

Using lessons learnt from the crisis, Keishi oversaw the complete overhaul and reconstruction of the risk framework at Morgan Stanley.

That included resizing elements of Morgan Stanley’s risk profile, as well as the implementation of a diversification strategy. This ramped up the revenue stream derived from wealth management, and reduced that derived from investment banking.

The switch in strategy wasn’t an easy decision, he explained. “You have to think about what you are good at. We eliminated legacy assets, we made choices around our emerging market franchise – because our history showed that it wasn’t our core strength. We concentrated our risk capital and our human capital in the areas we were strong in.”

It worked so well because of the complementary skillset between Keishi and Morgan Stanley’s CEO, James Gorman. Gorman was the “big picture strategist” while Keishi was the “numbers guy”.

That’s not to say that there weren’t challenges. “Risk officers and the C-suite don’t speak the same language. When we had critical decisions to make regarding risk appetite, or changes to the business model, the biggest challenge was how to communicate that technical language into one that the board could understand, so that they were able to make a decision,” noted Keishi.

“Communication is a much bigger success factor in our job now,” he added.

Another challenge was how to manage the flow of risk information.

“The emphasis I make to my people is that, if you start to get concerned about certain risks and you tell me, but I didn't think it was important, and something goes bad, that’s on me. But if they didn’t tell me, that’s on them.

“You have to have to have flexible mechanism for that.”

The use of technology in risk

In five years’ time Keishi reckoned the risk management field would be unrecognisable. The only problem is “we don’t know how”.

The opportunity of AI was ignored at our peril, he warned. “The opportunity is real. There are still a lot of manual processes, particularly in the regulatory landscape. The people who are afraid of those opportunities are completely wrong.”

 But plenty are embracing that very opportunity.

For instance, Finastra walked us through how they were using advanced technologies, such as big data and GPU-accelerated computing, to solve some of the challenges in the new world of risk.

Vinod Bhaskaran, Global Solution Lead, Treasury & Capital Markets at Finastra, explained how the three traditional lines of defence were ripe for disruption. Many institutions still had a risk system based on a segmented model, with multiple data sources and calculations, and separate reporting of risks. This led to flaws in the system, such as the lack of communication between front and back office:

“Front office pricing models are typically optimised for accuracy - but not for speed. But on the risk side, it’s often optimised for speed, not accuracy, in order to be able to cope with the huge number of calculations necessary,” he argued.

Finastra’s approach was to leverage a fusion pricing service. “It’s a shared valuation and analytics layer for both front office and risk. The advantage of fusion pricing is that it leverages some of the latest computing technologies, specifically GPUs, to handle the increased computational burden. You then have the best of both worlds. A pricing service that can be used enterprise-wide, and the speeds you require for risk as well,” he explained.

 In addition, their technology allows the moving of some of these risk measures to the front line in near real time. “This enables the front line to make key decisions that are risk and capital aware.”

For instance, a risk manager looking at a capital charge can see exactly where the risk is concentrated, down to the very last detail. They can pass this information on to a trader who will come up with a strategy to bring down the capital charge by executing a trade, all done in real time.

Vinod also talked about how they unleashed AI and machine learning in order to tackle the very real threat of operational risk. This involved adding an element of machine learning to existing trade infrastructure. As and when trading errors are made, the machine learns from those mistakes.

“What we are looking to achieve is same day trade validation,” said Vinod. Their proof of concept of the system, carried out in a Tier 1 bank, resulted in 50 previously unseen errors flagged, only eight of which were false positives.

 “AI is not going to replace humans, but it’s going to make them more productive,” he said.

Big data and analytics

The incredible potential of big data and analytics was also the core message of Marcus Chromik, Chief Risk Officer & Board Member at Commerzbank.

“If you combine big data, advanced analytics, digitisation, some smart people, and a budget you can achieve great things,” he stated.

Commerzbank has invested heavily in big data and advanced analytics “because risk is not only fundamental for safeguarding a bank, it’s also central to driving efficiency and strategy.”

Marcus outlined several examples of the use of big data analytics, including how a machine learning strategy had solved 83% of the intensive care cases within the bank, without human interaction. It had also analysed the bank’s exposure to the automotive industry and isolated five names – out of 12,000 – that required risk management attention. The system had also spotted fraudulent loan applications by picking out counterfeit payslips, something which had prevented 25 million fraudulent transactions from proceeding.

 Regulators have to step up

With these exciting developments, regulators had to make sure that they kept pace, warned Marcus.

“Regulators must rework their model approval process, they must understand advanced analytics and self learning programmes. New data inputs are available all the time, so they have to accept that models are going to change constantly. If you have to keep applying for model change that’s going to kill banking completely.”

Ultimately, risk was now the driver of the transformation of banks into the digital age, felt Marcus. “You can shorten the time from analysis to decision substantially,” he said.  “If you love risk management,” he concluded. “You must embrace the future.”