Migration Analysis: A Closer Look

Chris Emery | Senior Systems Engineer | MST

Many banks are using some form of migration analysis to track and better understand the paths of loans that eventually experience loss or default. MST Senior Systems Engineer Chris Emery, who has been conducting migration analysis seminars and workshops, dove deeper into the process to discuss some of the key considerations for banks choosing to do migration analysis.

Migration analysis is a methodology to determine the expected future loss rates of given risk-stratified loan segments by tracking the historical risk characteristics of loans experiencing eventual loss or default events. It is used to look back from these events to see how the loan migrated from pass through various grades to its eventual loss or default. There are various applications and approaches to migration analysis, and a lot of room for individual bank interpretation within the framework.

In a “real-time” migration approach, the bank begins with most recent quarters, even if they do not have a full horizon of loss events. The advantage of using real-time migration, where recent loss activity is weighted equally or more than past loss activity, is that it allows the bank to react more quickly to the current loss environment. The advantage of “delayed” migration analysis, where the bank uses quarters of analysis only when they have a full horizon of loss events and past history is weighted more heavily than recent loss history, is that it provides a less volatile, more predictable analysis.

The incurred loss horizon should be the best estimate of time between the borrower’s actual loss event and the bank’s confirmation of that loss. The average horizon can vary between segments. In a migration methodology, longer look-back periods are often necessary when compared to a portfolio historical loss approach, especially when the bank is using real-time migration.

One of the biggest hurdles for banks wanting to do migration analysis is their data: not just a shortage of it, but inconsistencies within it. Another problem is loss or default events that are not tied back to the original note due to restructures, A-B notes, etc. A third issue is inconsistent risk metrics.

CECL will likely impact migration analysis approaches by extending horizons (expected loss horizon as opposed to today’s incurred loss horizon) and look-back periods, and higher reserve calculations.