Three steps to effectively evaluating your pooled loans methodology in the estimation of the ALLL

2. Applying a Historical Loss Factor: Historical Loss Rates vs. Migration Analysis

Once the institution establishes the appropriate segmentation, they must decide what methodology to employ to determine a historical loss factor within each pool.  There are two primary approaches: the historical loss rate approach and the migration analysis approach.

Historical Loss Factor
Many banks employ this approach, at least partly because data collection is easier.  The historical loss rate approach primarily requires tracking charge-offs and recoveries within each segment over a defined period of time.  Within this approach, the two primary challenges are (1) determining the appropriate number of periods of data to incorporate and (2) whether to apply an arithmetic average or weighted average of loss rates.  Traditionally, institutions have used a longer time horizon that incorporates three to five years of loss data.  The drawback of this is that it may not reflect the increased losses that have been incurred in recent years.  As such, more and more institutions are now using a shorter time horizon such as a rolling four to eight quarters.  Another option to ensure that more recent loss rates are appropriately taken into account is to use a weighted average that applies greater weight to more recent loss rates.

Migration Analysis
This approach can be more effective in times of economic turbulence than the historical loss rate approach.  Under the migration analysis approach, institutions track the migration of loans from various buckets to charge off status which can give a more accurate picture of how the current portfolio would migrate to loss.  The basic methodology for migration analysis is to set up appropriate buckets to track within each segment.  As an example, within consumer loans, this could be based on delinquency (Current, 30-59 days, 60-89, and 90+ Days Past Due), and within commercial loan segments, this could include general risk level (Pass, Special Mention, Substandard, and Doubtful).  The biggest challenge in migration analysis is the data collection process; it takes a minimum of four quarters with a structured data collection process to gather enough data to use this approach for estimating loss factors.  The migration analysis approach can be more robust; however, for many institutions, the standard historical loss rate approach may still be adequate and most appropriate.

3. Evaluating Qualitative & Environmental Factors

Perhaps the biggest challenge that institutions face in the estimation of the ASC 450-20 (FAS5) reserves is the determination of adjustments to take into account qualitative and environmental factors that may impact loan losses.  These factors are inherently subjective, so institutions face scrutiny on the documentation used to justify any assumptions made.

The first judgment made in this process is to determine which qualitative factors to assess.  A good starting point is to use the nine standard qualitative adjustments cited on page 8 and 9 of the Interagency Guidelines on the ALLL.  These can be modified where appropriate depending on the levels of risk within an institution’s overall portfolio and within specific portfolio segments.  Secondly, the institution must evaluate its portfolio and segments on each of the qualitative factors and apply a quantitative estimate to each qualitative adjustment.  It is imperative that the institution documents all of these assumptions around the qualitative adjustment factors.  Additionally, where appropriate, the institution should reference trends in national, regional, and local economic data, which can be found through various sources, including the Federal Reserve Economic Database and the FDIC Regional Economic Conditions data, amongst others.

The institution can also consider employing a more quantitative approach for the estimation of the impact of these economic factors.  This can be done through a simplified correlation analysis approach.  For example, the institution can do a simple regression on their loss rates (using at least 12 periods) against an economic factor (i.e., using housing starts as correlated to loss rates in construction and development segments).  Additionally, the institution can employ back-testing to match up its historical losses with the levels of its ALLL over past periods to show the historical accuracy of the assumptions used in its methodology.  The estimation of the qualitative factors in this instance may still largely be judgmental and subjective, but a correlation analysis, peer comparisons, and historical back-testing can provide quantitative support to these judgments.

The most important thing to keep in mind when making all of the judgments used in assembling risk pools, applying historical loss factors, and making qualitative adjustments, is that the institution should maintain extensive documentation of its methodology and the justifications for each assumption used. This will allow for a consistent process going forward and make it more difficult for the institution to be criticized by regulators around their Allowance estimation.

Do you have any tips on the best methods to estimate reserve amounts of loan pools?