Analytics for CECL: Challenge creates opportunity

Getting ahead of the new CECL requirements is a strategic initiative for all credit union executives in 2016. Being complaint will be no easy task, however. The new CECL requirements require life-of-loan loss forecasting capabilities. In most cases, this means the credit union will need to collect data, A LOT OF DATA. Unfortunately for most credit unions, collecting, storing, and analyzing data has not been a priority and most do not have the proper infrastructure in place to do so.

Credit unions need to start preparing for CECL, but what should they being doing to prepare? An article by Mainstreet Technologies (MST) spells it out nicely. Credit unions must learn, share, hoard, forecast, and model. Fortunately for the credit union industry, the first two points MST makes are easily achieved. Credit unions have many resources made available to them, and with the inherent propensity to collaborate, the industry can share that knowledge and best practices very easily. Unfortunately for credit unions, however, the last three points are difficult to achieve given the current state of the industry:

1. Hoard – Credit unions must have the infrastructure in place to effectively integrate, collect, and store all of their data. In order to have life-of-loan loss forecasting credit unions will need much more than just core data. Integration is key and obtaining a single source of truth is not only desired, it’s necessary. The first step in hoarding the data necessary for the CECL requirement is implementing a data warehouse to integrate, collect and store the data. Without a data warehouse (or Analytic Data Model), doing the analytics necessary for life-of-loan loss forecasting (CECL) will be nearly impossible.
2. Forecast – Forecasting is best achieved with predictive analytics. Predictive analytics harnesses patterns found in historical and transactional data to identify risks and opportunities. Through utilization of sophisticated statistical modeling techniques, machine learning, and data mining, predictive analytics look at past and present facts to make predictions about future events. With the newly introduced CECL requirements, being able to look into the future is now required. Once credit unions have a data warehouses in place to fully integrate all of their transactional data, they need to start using it. Predictive analytics allows credit unions to look at loan portfolios and apply statistical models to analyze the outcome of their future performance.
3. Model – Forecasting requires good modeling. Without proven models, a credit union’s predictions can only be taken with a grain of salt. The process has to be repeatable and accurate every time. Developing CECL-compliant models requires the skills of data scientist. Unfortunately for credit unions, having a data scientist on staff is not possibly. The level of sophistication required to develop the models necessary for CECL will need to be outsourced. Credit unions should look to partner with CUSOs (such as Deep Future Analytics) and other organizations that have dedicated themselves to the industry.

The new CECL requirements will definitely challenge the industry. Credit unions and the rest of the financial services industry have been slow to adopt data analytics in comparison to other sectors (e.g. – Retail). This great challenge introduces great opportunity, however. The industry will now have to get on top of their data and start leveraging it to predict the future. Data scientist and CEO of Prescient Models, Dr. Joe Breeden, stated in his recent blog, “The proposed CECL accounting rules for loan loss reserves will cause a dramatic shift in the use of data and analytics at credit unions, comparable to the changes occurring at larger banks due to CCAR and DFAST.”

The challenges that CECL brings on should be viewed as positive note for the industry. The industry is now feeling more pressure to be data-driven than ever before which will allow them to make better financial decisions and provide their members with the best possible service. The real trick in converting these challenges into opportunities is being prepared. Credit unions must start collecting, storing, and analyzing (forecasting and modeling) data. In order to do so, they will need to have partners. Partners for data integration (building out a data warehouse) and also for developing predictive modeling (data scientists). Credit unions need to come together as an industry to take on this great challenge. No credit union alone can do what’s required for CECL.