Pooled data serves as a building block to building robust CECL forecasts, especially benefiting small and mid-size lenders who have limited data to use to build accurate forecasts for some portfolio segments. In addition, the pooled data can offer detailed insights through performance benchmarking.

Are you ready to jump into the deep end?  Here's a quick lesson on pooled data and the value it brings to you.

Absolutely nothing! Your data is your data. The question is simply 'how can we make your data better?' That means looking at the types and scope of your data and ways to supplement your data with other sources. One of those sources is the pooled data from other participating lenders.

Pooling of data means simply combining data from multiple sources. With CECLNow that means basing analysis on data from all lenders in order to build more accurate models. In this case we are indifferent about who the lender is; we simply want more stable and predictive models. The application of these models are adjusted for individual clients if loss performance for  the client is consistently better or worse than the pooled results. Individual clients also benefit from the pooled data via benchmark comparisons to give them more insight into their data.

One of the biggest benefits of pooled data is that it fills in thin spots in your data. Some segments in your portfolios are bound to have a limited number of accounts. Perhaps you just started marketing to a given segment or have changed key product features. One example might be auto loans. Most of your past experience might be loans with terms up to 5 years. If you  recently introduced 7 year term loans you wouldn't have many loans yet and you certainly won't have data out to maturity. This is a case where the pooled data will serve to develop an accurate CECL forecast.

Think about life insurance for a minute. If you insure 10 people and one of them dies through bad luck you will have a horrible short-term loss rate. You might close the business. But there was always a chance that at least one person would die, just a very low risk. It's also possible that no one will die which might give you a false sense of how good your underwriting is leading you to expand the business with no real basis of knowing whether your underwriting is terrific or not. Very different decisions based on very limited data.

Now if you've insured 10 million people your loss experience is much more likely to mirror death rates of the entire population. Some people will die but it's likely to represent real mortality rates. You would have a firm basis for making decisions about the future of the business. To play out our example, a primary benefit of pooled data is that you get the experience of 10 million insured people while only underwriting 10.

In sum, by blending small portfolio data together to develop  a large pool of data, each participant achieves larger numbers to develop more stable and hence, more certain results. This data is used to develop the life of loan models for CECL analysis. The larger pool of data smooths the results to a much greater degree than any one lender. A single lender may see months where there are zero losses and other months where losses are bunched together.  This can distort their results and possibly lead to bad decisions about the future direction of their business.

An incredibly valuable way to evaluate performance is to compare yourself with peers. Often this can be difficult due to  the limits on available data, its timeliness, and the relevance of the comparison. By pooling data together we are able to make targeted comparisons at the segment level. You will be able to compare approval rates at acquisition and delinquency and loss performance to identify potential changes in application processing and marketing. You can learn more about performance benchmarking here.