Loan loss provisioning methodology for community banks

Mikkalya Murray

I'm the chief credit officer for Harleysville National. We're a $2.4 billion community bank holding company with three nationally chartered banks, located outside of Philadelphia, Pennsylvania. Our branch system runs up to the New York border. We have a $1.3 billion loan portfolio. At year-end, $627 million of the portfolio was in commercial loans. Currently, we have about a $17.3 million reserve. As a frame of reference, when we talk about losses and loss rates, the bank's reported net charge-off at yearend was 21 net basis points.

We have been carefully following the July 2001 FFIEC pronouncement. After reading that material, we redrafted our policy and procedures information, and then adjusted our methodology so as to have a much more granular focus. In other words, we segmented the portfolios more discretely than we did in prior times.

When we look at our portfolio for reserve purposes, we break it down by C&I loans, commercial mortgage, commercial construction, direct and indirect consumer, revolving, credit cards, residential mortgage, auto, and equipment leasing. We break it down first by those segments of loan type. And then we segment based on our 10-grade risk rating system. We break the segments down further, with a special focus on the regulatory or criticized grade, which for us is doubtful, loss substandard, and special mention credits.

Within those categories, we then go to our loan review or risk rating criteria and look for the risk ratings on the credit. We have calculated loss ranges within those risks and segments. From that data we get a trend report that captures delinquency, nonaccruals, and charge-offs for each of the segments in each of those risk ratings over a 12-quarter period. It took us a number of years to develop that much data. You can start somewhere along that horizon and begin to gather the data.

Then, after we've segmented and captured all that data, we examine those loss rates in four different ways:

1. By the loss in our current period.

2. By the average loss in the most recent four quarters.

3. By the prior year-end loss rates.

4. By looking at the 12 quarters of rolling annualized loss rates.

We look at where we believe losses will be in the next year to 18 months. We consider which one of these loss rates most accurately predicts that rolling quarter approach. It's based on historical data; it's based on live loss rates. We may have a shrinking portfolio; we may have an expanding portfolio. The risk direction may be up or down. We adjust for what we believe to be the current risk profile of any particular segment.

Based on that calculation, we have a quantifiable number. We have an adjustment to losses after that quantification is done. In other words, we look at between 20 and 80 basis points of adjustment based on what we used to call BC201 factors or qualitative factors that tell us what's happening in the economy and what's happening with the management of our own portfolios. We fine-tune our quantitative number by 20 to 80 basis points. We then plug that into a spreadsheet with all the segments, and the result is the calculated reserve amount