In some areas, humans will always outperform machines. Creativity, that uniquely human spark required to come up with new ideas, products, setting strategy, managing incentives and inspiring employees to a mission, are and will always be, part of the realm of humans. Credit attribution, in so far as it is an exercise in risk assessment, is quickly becoming the domain of machines. What are the challenges ahead for the banking industry, and what can players, big and small, do to prevent being left behind?
The current decade is often called the industrial revolution of Data Science. After speaking with vast numbers of industry experts, practitioners, leaders and academics about the industrialization of Data Science, we have attempted to answer the question of what this all means for the lenders, big and small. We start by dividing the market into alternative lenders and traditional lenders, and examine what are the threats, opportunities, and actions towards reaping the rewards of the data science revolution.
Alternative Lenders
For alternative lenders, the journey towards taking full advantage this industrial revolution means crossing the notorious ‘valley of death’, so called because of the stage in which the investment required outweighs the short-term benefits. This investment includes discovering the market, collecting unique information and delivering impeccable user experience. New lenders face a dilemma: whether to invest in Machine Learning expertise from day one.
Machine learning, by design, needs considerable amounts of data, and this is exactly what new and alternative lenders often lack. On the flip side, it is also exactly the type of technology that will help lenders to make use of the data as it starts to arrive, and to monitor the quality, sufficiency, and usefulness of each data source. The Machine Learning expertise won’t make these great in the short-term (at least in the early vintages), and yet it is essential in the medium/long run, and to master it lenders need to start early to get a head start.
One could argue that this dichotomy (‘whether to invest in Machine Learning expertise from day one’) can be solved by adding business strategy to technical knowledge. Business strategy and a deep technical knowledge are essential to prioritize what is crucial to get right from day one and what can be postponed without affecting planned goals. Naturally this is not done overnight, but fortunately, many experts on business strategy are getting enthusiastic about the prospects of Data Science industrialization in Risk and Compliance. The advantages are clear: (1) better decision making, (2) higher productivity and (3) higher client satisfaction. The most important thing is to remember to work hard on finding your unique approach to using Machine Learning to your advantage.
Traditional lenders
For traditional lenders, the trade-offs are more clear-cut. The Data Science revolution simultaneously presents an opportunity, by offering higher yields and lower risks, and an existential threat of turning previously profitable institutions into obsolete museum pieces. We expect that within less than a decade a lender that does not use AI to their advantage will face the same difficulties as a lender who, as of 2017, has yet to discover the internet.
The industry is going through a technology revolution and it needs to be hand in hand with education and training in new methodologies, as well as with fair lending practices and credit market risk ethics. Traditional lenders need to map which areas are going to be dominated by automated processes where data science, machine learning and AI are crucial. Decision makers will, therefore, need to demonstrate outstanding leadership, select the right partners
and technologies, and steer their organizations through years filled with uncertainties and anxieties, with the pressure of a board asking for the groundbreaking results everyone was talking about five years ago. Done right, the rewards will be worth the journey.
In a nutshell, the goal is to industrialize the process to the extent where little or no human input is required and doing it in an efficient and innovative way that will make the lenders stand out among competitors, who will also be joining in the data science arms race.
Interestingly, as with so many defining moments, the primary focus should rest on people. Namely, on the teams’ ability to explore and find innovative ways to use data science, and to embrace, rather than fear, change. Then there is the data. In these early days, banks and financial institutions with a vast and organized data warehouse can get a head start. Organized, accessible and high-quality data procedures are determinant factors to achieve dominance in the market. Soon we will see many small monopolies in market niches, where competitive edges could either come from a client profiling, servicing excellence or from risk profiling and management. The axiom ‘the winner takes it all’ will become a reality for some of the key credit segments.
The challenges both traditional and new lenders share
The greatest challenge that all financial institutions that rely on credit as their main core business face are the complexity of bringing Machine Learning and AI to the risk practice. The complexity involved in recruiting talent, updating technology, dealing with new contractors, onboarding new service providers and managing the board and shareholders expectations within a violently tight timeframe, paired with the ever present increase in regulatory workload, will be the greatest challenge faced by any generation of lending professionals.
Key to meeting this challenge will be the bank’s new found ability to partner with FinTechs that can help them digitize and deal with the areas where Machine Learning and AI can overcome human ability. The first half of the decade saw these high-tech companies emerging to compete with the banks. The second half is defined by David and Goliath working together, towards reaping the Data Science revolution’s sweet rewards.