How Cloud Computing is evolving alongside Big Data, Analytics, and AI in Financial Services.
New technology like Artificial Intelligence (AI), Cloud Computing, big data, and prescriptive analytics are changing the way the Financial Services sector does business.
With evolving tech comes both new opportunities as well as different risks, and companies within the space must innovate and embrace new ideas as shifting business conditions and changing consumer preferences dictate new norms.
To find out more about how the industry is adopting new technology I sat down for an interview with Dr Richard Harmon, Managing Director of Financial Services at Cloudera. In this interview, Dr Harmon looks at how companies in the space are adapting to a very different business landscape and the role that data, analytics and machine learning have to play in helping make the industry more innovative, secure and agile.
Good morning Dr Harmon, thank you for joining us. To start us off, As we’re seeing more digital transformation and a push towards embracing new technology within the sector, how fast is the adoption of cloud computing happening and what does this mean for the sector today?
This is a great question since in the current environment much depends upon what will be the “new normal” when we finally come out of this pandemic environment. My expectation is that while some aspects of a bank’s business digital transformation – particularly from a customer experience perspective – will be accelerated, I think other areas will slow down slightly as businesses try to understand what the new normal will look like. If we look beyond the current environment, then the future is very exciting since the cloud gives financial service firms new capabilities in terms of scaling up and down, not being locked into fixed infrastructure, and having a third-party cloud service provider to enable agility and deliver immediate cost savings. This is critical for businesses from an operational efficiency perspective and to improve customer experience.
Banks will soon begin to move more of their mission-critical applications into the cloud and I can see a likely acceleration of this in customer-facing, financial crime, and risk areas. There is a realization within the industry that there are benefits for businesses to become cloud-enabled, but there is a danger for them in thinking of the move as a simple lift and shift.
What they really need to think about, when considering to migrate legacy environments, is the opportunity around how to enhance cloud computing and the opportunity to reconfigure capabilities to support future innovation and business growth.
Because of the current pandemic environment and the uncertainty of the “new normal”, adopting dynamic infrastructure capabilities becomes highly attractive to financial service firms.
For instance, the agility in being able to accelerate the restructuring or the development of machine learning models allows banks to support newer systems that can adapt to changing environments much more easily. This ability to accelerate the adoption of machine learning capabilities into new business areas becomes very attractive for the next generation of mission-critical systems in the cloud.
The value of having a hybrid cloud capability and not having to fully redevelop everything in the cloud is also critical for banks in their cloud journey. We have already seen some of the innovation benefits of this across many of our customers with sophisticated machine learning models coupled with real-time streaming data flows driving the adoption of a new generation of chatbots and robo-advisors. A carefully designed hybrid cloud strategy with a corresponding comprehensive data strategy supports the ability to accelerate one’s cloud migration strategy while derisking the whole transformation process.
As financial services firms embrace cloud computing more, what are some of the considerations and challenges of this approach to be managed?
As the Financial Services Industry accelerates the migration to the cloud, regulators are starting to express explicit concerns around the expanded use of third party outsourcing arrangements – in particular with respect to cloud computing and operational resiliency. A key concern involves the operational risk of a bank being locked into a single cloud service provider or having a concentration of critical financial applications running on a single cloud service provider even when a bank has adopted a multi-cloud strategy.
The complexity of having data and applications residing in a hybrid, multi-cloud environment has created the need to develop the next generation enterprise data cloud platform. The purpose of an enterprise data cloud platform is to have the ability to store data and run applications in any environment with a single control pane to securely manage and move data and applications to another cloud or on-premise environment. The enterprise data cloud provides this type of agility but also provides consistent data governance and security across any environment.
The agility story in Financial Services is also around how technology is rapidly evolving. This is being heavily driven by the global open source community. This is also helping to accelerate digital transformation programs across the banking sector with a significant impact on digital risk transformation. The latest innovations in digital risk support the ability to be more automated in the way banks manage risk and how quickly they can respond if there is a sudden shift in the market or some external shock. This is a core requirement from a global regulation called BCBS-239. This provides the required element of regulatory responsiveness supporting automation and on-demand recalculation of risk exposures across any dimension. When you move this requirement into the cloud, it can significantly complicate a banks’ ability to be fully BCBS-239 compliant. So the cloud is not without its own hurdles and challenges.
Are there any new risks that you foresee as a result of the industry’s adoption of cloud computing?
One of the most important “hidden” risk topics for the industry is cloud concentration risk. While there are many aspects of this, the most visible risk concern is the degree to which banks have certain core systems concentrated in one cloud service provider. This exposes the firm to what I call “firm-specific” cloud concentration risk – should something result in a disruption of the cloud service, then this core system’s ability to function is severely impacted.
Regulators have an additional type of cloud concentration risk to manage. This is from a systemic risk perspective. Systemic risk can arise from several areas but one simple example highlights their concerns even if all banks have implemented a hybrid multi-cloud environment. Suppose that the five top financial institution payment systems are all in a single cloud. Should this cloud experience a disruption, this has the potential to trigger spillover or contagion effects into the rest of the market. At the moment regulators don’t have full visibility on what applications are running on which cloud service provider but the European regulators are mandating a cloud registry – probably by the end of 2021. This is a first critical step in being able to analyze and monitor systemic cloud concentration risks.
The other risk worth noting is the potential for algorithms, as they become fully automated with learning capabilities, to create situations that impact the wider discussion around the ethics of machine learning in AI. My very niche example refers to some very interesting academic literature that demonstrates circumstances where automated machine learning pricing algorithms are designed to learn to optimize returns across a dynamic environment. If, for instance, an algorithm lowers prices it can potentially capture market share in the short term but when competitors – also using automated pricing algorithms – start matching this price, this can quickly eliminate the competitive advantage. What a few of these studies show, is that if the algorithm then increases the price, competitor algorithms can start to match this price increase as they adjust their strategies in order to maximize their returns. This can then result in collusive outcomes where the resulting market price is above the ideal competitive market price.
This is just a small example within the much wider discussion around the ethics of machine learning models. It is well documented that machine learning models will often mimic inherent human biases that are embedded in the data they train on. This highlights a fascinating global discussion on how to identify and mitigate these potential hidden risks as we try to move towards a fairer society.
For more insights on the potential for Cloud Concentration Risk, read our whitepaper – Cloud Concentration RIsk II: What has changed in the past two years