What’s your institution’s segmentation strategy?
Sep 30, 2015
At last week’s 2015 Risk Management Summit, the impending accounting changes from the FASB was a major topic for nearly every banking and credit union professional in attendance. Final guidance on the CECL model is anticipated within the next several months, and it will impact every bank and credit union in the country, regardless of asset size. Bankers maximized the timing of the annual Summit to listen to industry experts speak on what CECL will mean for their institutions and how they can best prepare now, as well as network with their peers for best practices heading into the changes.
Commonly asked questions of both the experts and peer bankers revolved around ways to improve their institution’s ALLL process and how to make it more comprehensive.
Of course there isn’t any one answer to these questions, and the direction can vary based on an institution’s starting point.
A segmentation strategy is a great place to start to nail down an effective and efficient process to not only serve a substantial purpose for the ALLL, but also as a larger risk management tool. The ability to adequately meet ALLL, stress testing and other risk management requirements relies on sound segmentation practices.
For nearly any size bank or credit union with lending capabilities, a comprehensive loan portfolio segmentation strategy can enable a credit department to quickly identify the underlying behaviors that drive credit risk. To best understand that risk, bankers look at segments of the portfolio to monitor performance over time. Using segments, the institution can not only evaluate the causes driving a loss, but also adjust underwriting strategies, such as pricing or setting lower lending limits, within that segment to mitigate further loss.
Proper risk identification focuses on recognizing and understanding existing risks or risks that may arise from new business initiatives. Usually this identification is assessed during underwriting, and again during annual reviews. Borrowers within a segment generally exhibit similar financial characteristics such as capital sources and/or repayment sources. Identifying those commonalities allows an institution to look for bigger picture risks before diving into individual loan losses.
Given the importance segmentation has in risk identification, just how much should a bank or credit union segment ASC 450-20 (FAS 5) pools?
There isn’t really a clear-cut answer. There isn’t a “one size fits all” segmentation strategy. The number of pools is best decided by the individual institution, depending on factors such as portfolio composition, volume, homogeneity within pools or risk levels. Peer institutions can be helpful in this instance, too.
The current guidance suggests the loan portfolio should be broken down into homogenous pools based on similar attributes. FDIC call code segmentation is likely a good starting point for many institutions. The goal is for segmentation to be granular enough for the pools to show similar loan characteristics. One common example is breaking down an institution’s residential or even CRE lending segments into those that are owner-occupied versus those that are non-owner occupied.
Of course, there’s some risk to segmenting pools and making them too granular. With many smaller pools, the loan balances within each pool can become too small to easily identify trends and potential risk. Not only will that become too cumbersome, but too much granularity may sacrifice the statistical significance of the calculation.
Given the examiner scrutiny in the ALLL, and the changes to come to the calculation when the FASB releases its final CECL guidance, it is important to find a sweet spot of segmentation for each institution’s unique loan portfolio. Evaluating that segment of granular pools without sacrificing quality will save an institution from headaches surrounding their ALLL and risk management practices during exams.
For more information on segmentation strategies, watch the complimentary webinar Navigating Loan Pool Segmentation Under CECL.