Time to Get Serious About CECL

In early January 2020 I wrote an article named: Tick tock, the CECL clock! The article explained the realities and fears associated with CECL. Today we are going to take a very sober look at the realities of implementing a CECL solution within your financial institution. Are you ready? Most aren’t. To paraphrase KPMG’s recent ‘CECL Survey 2019’ it states: “…more than 42% of larger institutions still have not designed a qualitative framework for CECL and this is the case for up to 70% of smaller institutions.” Ready or not the time to act is now and focus on what it will take to implement a CECL solution for your institution. At a high level, CECL will require: data collection and integration, forecasting and data science, and a build or buy decision for a CECL software solution. Regardless of the decisions you make, the cost of implementing a CECL solution on your own can be very expensive ranging in the hundreds of thousands of dollars or more. FIS has a much more affordable solution that could curb some of these costs and we’ll detail what that solution is at the end of this article.

Data collection and integration

CECL is going to require a richer level of data, analytics, and analysis than previously demanded. In order to defend CECL compliance it is going to require a data warehouse to retain data for the life of all loans associated with the financial institution. This will require people with a variety of skillsets including: data modeling, extract transformation, and load (ETL), and data harmonization. Alternative data will also have to be collected such as credit scores, residential and commercial real estate values, and economic indicators. Loan data might come from a variety of places too and it will have to be normalized to feed into a database. Some financial institutions will struggle with these challenges and will find that it is very costly. A decent data employee can cost an organization up to $57/per hour or over 100K per year. This is just one of the disciplines required for proper CECL regulation adherence.

Forecasting and data science

CECL is a statistical and mathematical challenge; data forecasting and data science are required skillsets. There are multiple statistical models that you can implement, however the new CECL regulation does not specify the methods to be used. It will require new skills that your organization may not currently have on staff. What exactly are these models? The FI will have to forecast the real estate markets, unemployment rates, and economic indices that might impact loan risk. These forecasts will change over time and it is not a one and done scenario. Data scientists and people who have the skillset to manage such algorithms are some of the most expensive resources in technology today. The average data scientist in the United States makes roughly $130K per year. Data scientists are also very challenging to find and keep on staff long term. Oracle recently stated that there will be 250,000 open data science positions by 2024 and there is a lot of job hopping. To put it plainly, data scientists are in high demand, which does not favor your organization’s CECL goals.

Do I build or buy a CECL solution?

For most small to mid-sized financial institutions developing software from scratch can be a very expensive proposition. Building software requires advanced technical skills that many small to mid-sized banks don’t have on staff. It also means that you go it alone. You have no one to call when things break or require support. When you license software from a vendor many of these costs go away, but licensing software can be expensive. Given that CECL has a deadline, FI leaders might be hesitant to build and are also put off by some of the costs of commercial CECL solutions. As we stated above the institution is going to require data experts and data scientists to run and manage the solution. When making a CECL solution decision don’t forget the other people impacted too! CECL will require other business people, AKA software users, including: underwriting, origination, loan processing, review, and modification. This means tracking changes associated with CECL related data will be critical to establishing a reliable solution. Whatever your organization decides the input and outputs of a CECL solution will also have to be ‘AUDIT’ ready.