Over the past several years, I have witnessed competing priorities affecting the generation and delivery of data required for credit risk analytics. On one side, formalized data governance and controls have emerged with the intent of improving data consistency and accuracy. On the opposite side, stakeholders continue to request faster delivery.
"Analytics means providing stakeholders with accurate, actionable, and timely information, in an easy-to-digest format, that is relevant to the issues faced by banks"
Data governance and controls are key in the financial services industry, where inaccurate or mishandled data can have dire consequences for both the customers who entrust these institutions with their assets as well as investors, who need accurate information to make informed decisions. Let alone internal management, who make decisions based on a range of credit risk indicators. The regulatory environment and the financial accounting environment will likely continue to develop and require financial institutions to make even greater strides in data governance and controls. Increasingly, analytics functions are taking on a central role in data governance and controls.
More immediate data delivery continues to be an opportunity for improvement, especially in those cases where humans are required to make estimates or when new systems come on line. While more and more data processes are now automated, requests for more data increase. Sometimes these requests require employees to manually collect data. The human touch can slow processes and introduce new opportunities for data errors to creep in. And as new systems are put in place, integrating those systems into current data aggregation processes is time-consuming and is prone to errors.
I manage the credit analytics function at Zions Bancorporation. Credit analytics is cross functional, in that, analysts need to understand the data and tools at their disposal, statistical modeling techniques, the business of how banks make and lose money, how the economy affects banks, and the products that the bank offers. In addition to fielding ad-hoc requests from management, this group identifies emerging trends in credit risk, generates regular reports for management and other stakeholders, and helps management understand the implications of changing underwriting policies.
In my mind, analytics means providing stakeholders with accurate, actionable, and timely information, in an easy-to-digest format, that is relevant to the issues faced by banks. The analytics function at a bank needs to understand questions that stakeholders ask within the context of the banking industry, rather than within the context of IT. My high school computer science teacher used to say: “Computers will do what you tell them to do, not what you want them to do.” The analytics function needs to interpret instructions from management to make sure we answer fundamental business questions, rather than generating reports that are not meaningful or waste the time of stakeholders.
In a way, I think that analytics functions have become a victim of their own success. Since the science of using computers to make business decisions was first studied in the 1960’s, computing power and data accessibility has grown to the extent that bank management and external stakeholders expect the analytics function to be able to support a wide range of business decisions. Add to that, societal trends over the past couple of decades, where people have grown accustomed to finding any information that they want, on the internet, and the expectations on the analytics function are amazingly great. Stakeholders dream up increasingly more complex questions to ask, and the analytics function takes the lead role in answering those questions.
I think that the obstacles to answering those questions remain similar to what they have been over the past couple of decades. Many questions cannot be answered using data that is currently captured systematically and must be captured, at least temporarily, manually through brute force. And, sometimes, the data needed to answer a question is being captured on systems that do not link to databases in use. The same tools that were needed to overcome these obstacles in the past are still relevant, namely, effective project managers, individuals and teams that merge technical knowhow with business understanding.
In my opinion, there are two ways to deliver regular reports more quickly: require upstream data producers to generate information more quickly or use ‘estimates.’ Both of these solutions have obvious pitfalls. Each quarter, when I hear about other banks’ earnings releases, I am reminded that many banks report their earnings to the street without the benefit of final information from the previous quarter.
All employees bring their own set of biases and predispositions to their jobs, and I don’t think technology leaders are any different. In my mind, technology leaders in particular wield a lot of power and need to be careful to ensure that other stakeholders drive technology decisions rather than the technology organization determining the acceptable solutions.
In addition to understanding the technology that are available, technology leaders need to understand the whole range of stakeholder needs, in the context of the banking/financial sector and ensure that they do not impose their inclinations on other stakeholders. In the banking industry, there are myriad stakeholders that include all flavors of customers, regulators, company management, risk managers, external auditors, and more. Each of these stakeholders generally wants something different, and frequently, their needs do not intersect. Technology leaders need to be competent at implementing stakeholder requests, using appropriate technology, but if technology leaders fill in the blanks with their own biases, this can cause problems when stakeholders needs are not met.
Working in the credit risk analytics space, I am most excited about those solutions that provide early indicators of deterioration and allow for easy drill-through to identify the drivers of deterioration. Dashboard solutions like this have existed for quite some time, and I think that as we put these tools to better use, including the addition of more data sets, we will be able to better serve our stakeholders and continue to guard the soundness of the bank’s balance sheet.