The Wide Range of Data Sources
Analyzing past delinquency and loss patterns is how you start your CECL analysis. This begins with pulling monthly performance files together and indexing each file with each account's age that month. From there you track the first occurrence of key events including write-offs based on account age.
Looking at performance at the portfolio level provides a very crude measure. Your portfolio needs to be broken down into relevant segments beginning with 'vintage' or the calendar dates when accounts were opened. Vintage analysis delivers insights into whether performance of newer accounts is getting better or worse as accounts age.
The next step is to further segment your portfolios into more detailed segments such as risk bands. The more data you are able to collect in items #2-8 below the more granular your segments can be. Each of the 8 items below are listed in priority order. The more data you collect the better off you will be down the road. But at a minimum you need items #1 and #2.
#1 Performance Data
Monthly performance data is the most crucial part of your CECL data set because you need to know if and when accounts write-off. Depending on the product you also need to know how accounts pay down since there's no risk associated with paid off loans. Segmentation of your portfolio begins here based on product characteristics and date opened.
#2 Application Data
Data captured at the time an application is received supports more detailed segmentation. Key factors here are things like risk score, income level, collateral, whether the customer lives in or outside your footprint, and the channel the account was sourced from. Other factors might be income, debt burden, and home ownership status. It's likely that you may capture different information for different products and channels. Analysis of this data might highlight how valuable each piece of data is to improved performance.
#3 Credit Bureau Profiles
Credit scores are often included in application processing files. However, significantly more data is available from the credit bureaus at the time accounts are originated. This includes the age of different tradelines, number of trades, and utilization. This data can also be linked to application data such as income to calculate debt burden and other statistics.
#4 Demographic Data
Some lenders obtain customer demographic profiles from 3rd parties. This data offers descriptive factors that may help drive further segmentation. Items here might include household size, education levels, marital status, and household asset levels.
Well Rounded Profiles and Profitability
#5 Deposit/Investment Summary
To develop a full sense of your relationship with a customer it helps to factor in the nature and depth of deposit relationships. At a minimum this provides an idea of a customer's liquidity and ability to repay their loans. This should include the type of deposits held, the amounts, and the volume of flows in and out of transaction accounts.
#6 Past Account History
Long-time customers are likely to be better credit risks than those who are new to your organization. Understanding the number of past loans, balances and payment histories can help in assessing each account's risk. The summary should include the age of the relationship. Past deposit account info also can help build out a full customer profile.
Improved Messaging and Targeting
#7 Collections History
Success in contacting delinquent customers is another potential factor in predicting future losses. Combined with delinquency history (severity, recency and frequency) you can improve short-term loss forecasts. Also, a history of account reaging could be another risk factor.
#8 Customer Service Contacts
Calls, texts, emails or letters to customer service could signal the degree of customer engagement. More engagement might mean fewer losses. Again, frequency and recency might play a role in segmentation.
The decision on how to segment your portfolios should be based on whether you can identify groups of accounts or segments that behave better or worse than average. One benefit of CECL analysis is to identify pockets of risk and opportunity. The information you glean here permits you to change strategies, tactics, and policies to grow your business effectively and improve performance.
