How to Decide Whether to Build or Buy CECL Model Software
Should Institutions Build or Buy a CECL Solution?
Regardless of whether financial institutions choose to stay in-house or outsource their CECL solution, they have four important considerations to make.
Takeaway 1
Institutions complying with CECL in 2023 must make their build or buy decision quickly.
Takeaway 2
Institutions will need to ensure their CECL solution has adequate data and the ability to easily assess different methodologies.
Takeaway 3
CECL solutions should support "reasonable and supportable" forecasts with transparency and auditability.
Time is ticking for 2023 adopters
Time is ticking for institutions that must comply with the current expected credit loss (CECL) by January 2023. SEC registrants that have already adopted CECL and other experts have repeatedly advised banks and credit unions to begin their CECL implementation immediately. Despite the warnings and the impending deadline, many financial institutions are at a standing start. During Abrigo’s recent CECL Kickstart, an event aimed at those institutions who are only now getting to work on this major accounting change, attendees were polled on their current implementation status. Overwhelmingly, most respondents (42.1%) reported that they were at a standing start. Meanwhile, less than 3% of attendees reported being finished with implementation.
Some financial institutions may be at a standstill as they try to build their own CECL model using spreadsheets or buy a solution through a third-party vendor. CECL requires more inputs, assumptions, analysis, and documentation, and institutions must decide how to efficiently – and accurately – take each piece into account. This can require an extensive amount of resources that many financial institutions may not have, making the option to automate and modernize the process significantly more attractive for busy banks and credit unions. Regardless of whether financial institutions choose to stay in-house or outsource their CECL solution, they have four important considerations:
Ensuring adequate data and selecting the right methodology
1. Data
An institution’s ability to calculate lifetime loss rates is predicated on the accuracy and availability of loan-level data. But having large amounts of data isn’t always enough. Many institutions lack the material loss data needed under CECL. Because of the important role data plays in calculating CECL, some financial institutions will find that partnering with a vendor can significantly help streamline the data-gathering process and identify and fill gaps in their data. If an institution is considering leveraging a third party to assist with data, understand what the vendor offers for data archive, data architecture, and data adequacy services and how that compares to in-house capabilities. Compare the degree of automation and flexibility available for each option.
2. Methodologies
The CECL model is not prescriptive, meaning banks and credit unions can determine the best methodology or methodologies to use based on their loan portfolio. With seven different methodologies to choose from, some financial institutions have “analysis paralysis” trying to determine the right approach for their institution.
Once an institution determines the methodology best suited for its portfolio, it should consider the costs for developing and maintaining the methodology in-house compared to a vendor’s costs. How automated and flexible is the solution? Can the institution easily calculate, evaluate, and change methodologies? Finally, when weighing the pros and cons of building or buying a CECL solution, banks and credit unions should consider the advisory assistance and support they may need and what resources will be available to them as they work through their methodologies.
Transition to CECL with ease with step-by-step guidance with Abrigo's CECL Prep Kit.
Download nowForecasting and documentation are key to a successful CECL solution
3. Forecasting and adjustments
Unlike the incurred loss model, the CECL model is forward-looking, estimating loans’ lifetime losses using reasonable and supportable forecasts. For periods beyond an established forecast, reversion to average historical experience is required. The subjective factors of CECL can make the new standard more daunting – another reason institutions may be at a “standing start.” Qualitative adjustments have played an important role in calculating the allowance under the incurred loss method, and these adjustments will continue to be significant under the CECL model. Models should allow for quick inclusion and exclusion of all observable analysis periods and provide forecasting intelligence, support, and application. Ideally, management should be able to evaluate all observed loss rates, make any documented exclusion/inclusion decisions, include forecasted conditions, and see those decisions reflected in the estimated reserve level.
4. Model risk and auditability
Documentation and support are an area that is often the most time-consuming exercise in today’s allowance process under the incurred loss model. CECL will require more inputs, assumptions, and analysis at the pool-level. Tracking, consolidating, and displaying all information necessary to review, support, and recalculate will be a critical function of any in-house or third-party solution. Consider the level of transparency and auditability offered by any vendors under consideration and take into account costs associated with model risk or time spent preparing documentation and support.
There are financial and resources costs associated with building an in-house solution and purchasing a vendor’s solution. Irrespective of the decision to build or buy, banks and credit unions should carefully consider the impact that these four points will make on their model to ensure a successful transition to the CECL standard. Remember, testing, discussing, and deciding does not happen overnight. Be sure to get your solution in place as quickly as possible to be ready for the 2023 effective date.
