Data analytics affects all areas of the bank, from better understanding the customer to addressing regulatory issues like stress testing. However, organizations face several barriers that prevent unlocking the power of predictive analytics. John Sjaastad, a senior director at SAS, outlines these barriers, and shares how bank management teams and boards can address these issues in this video.
“It’s never simply the hammer that creates a finely crafted home. The result of the work hinges on the skills and experience of the carpenter who wields the tool.
So, too, it’s not so much the powerful cognitive intelligence software, the data and analytics tools, and the data visualization techniques that are beginning to open up opportunities for audit quality and insight enhancements from a financial statement audit. The skills and experiences of the auditors and their firms that implement these technological advancements will make the difference in the months and years ahead.”
When we think of the latest in technological innovations, we inevitably focus on the tools and techniques that benefit consumers. And, while that thinking is understandable, it would be a mistake to believe there are fewer technological advancement opportunities available for banks and other businesses. The litany of technological improvements include major commercial advances in the quality of databases, analytical capabilities and artificial intelligence.
In our world, one of the most compelling possibilities is the use of cognitive technology in the audit of financial statements. Cognitive technology enables greater collaboration between humans and information systems by providing the ability to learn over time and through repetition, to communicate in natural language and analyze massive amounts of data to deliver insights more quickly. Think of the improvements possible in the quality of audits when machine learning can be applied to deliver more actionable insights to guide and focus an auditor’s work or provide feedback on our perceptions of risks to an audit committee and management team at a bank.
While still in their infancy, there is vast potential in developing cognitive intelligence capabilities, especially given the exponential increase in the volume and variety of structured and unstructured data—this is particularly welcome given the ever increasing expectations on auditors, audit committees and management teams.
A prime example of an audit-based application of cognitive technology is the ability to test a bank’s grading or rating control over its loan portfolio. KPMG has developed a bold use case and is building a prototype that will machine “read” a bank’s credit loan files and provide a reasoned judgment on our view of the appropriate loan grade. The KPMG loan grade is compared to the bank grade, with our auditors focused on evaluating the loans with the greatest probability of a difference between the KPMG and bank loan grades.
While still in the development stage, we are encouraged by how cognitive intelligence could be applied to help us improve the quality of our bank audits. Currently, auditors carefully select a sample of loans to test from a bank’s loan portfolio. The sample is selected to provide both coverage of the loan types and grades, as well as where the auditor believes there is the greatest chance of loans being graded incorrectly. Aside from only reviewing a sample of the overall portfolio, today’s audit process is intensely manual. With the prototype being developed, the auditor would be able to select all the loans in a particular portfolio (say, oil and gas) or eventually the complete population of graded loans. The potential benefits to audit quality are very exciting—there is a distinct possibility that every loan in a banks’ portfolio could be reviewed and graded, while bringing outliers to an auditor’s attention. The bulk of the audit effort would then be focused on evaluating these potential outliers.
Further, using the combination of cognitive technologies, data visualization, predictive analytics, and overall digital automation would permit a much more granular evaluation of a bank’s enormous pool of internal and external information. Consider the potential insights that could be extracted when these powerful tools are linked to sources of market indicators. Looking into the future, the possibility exists for building a loan-grading tool to focus on grading commercial mortgage real estate loans tied to a market index of credit-quality values on commercial mortgage bonds, for example.
A tool that reviews changes in the market index against changes in a bank’s portfolio of commercial mortgage real estate loans could both improve audit quality and provide valuable insights into whether the two are consistent. If they are not consistent, those working with this technology—who are freed up from the manual duties–could spend valuable time determining whether or not there is any valid explanation for the inconsistency, better assess the remaining audit risk, and pass along the findings to a bank’s management and audit committee.
And, since such a tool would not be used in a vacuum, each bank’s results and weighted average loan grade could be compared across our portfolio of clients or a select segment of similarly sized institutions.
Even though cognitive intelligence is a powerful tool, it is important to remember that it is just a tool. The real value in cognitive and artificial intelligence is in its ability to allow human beings—in this case bank auditors—the time to think about, and respond to, the results of the testing, then work with audit committees to develop innovative solutions to real-world challenges confronting the industry.