Let's face it – bankers love to look at data. Whether it is lending, analyzing a customer or reviewing ALM reports, the more data available, the more comfortable a banker feels when making a decision. As a result, financial modeling has become increasingly important to the banking industry.
In 2008, regulators are being pretty clear that they will be looking at how bankers use models, what controls they have in place to limit incorrect model output from influencing management decisions and how well bankers are keeping up with upgrades. Fundamentally, financial models describe business activity, predict future trends, provide a statistically valid basis for informed decision-making, measure risk and generally provide a common basis for understanding complex issues. Modeling can take many forms, but some of the most common in banking include ALM, budgeting, credit scoring, loan pricing and securities modeling. Many banks will use spreadsheets to capture historical performance and monitor various factors within the bank.
While some may think the use of a spreadsheet does not rate as an official "model," it is important to note that the regulators consider any output from any source that directly or indirectly contributes to the preparation of an institution's financial statements to be a model. As such, regulators want banks to make sure controls are in place, given the potential seriousness of errors. Bottom line - if your bank uses spreadsheets to feed the call report, calculate ALLL or handle the budgetary process (or any of a hundred other financial or decision making processes), proper model governance is imperative.
Banks interested in passing their next examination with flying colors should consider incorporating at least the following items into their model governance process: 1) Understand a model's intent and its inherent shortcomings. If your ALM model does not take into account optionality – state it. If the model is designed to credit score consumers – state it. Using a consumer credit scoring model to handle a small business loan is a misuse of a model. Make sure all users know the purpose and shortcomings of a model. Knowing what you don't know is as important as knowing what you know. 2) Ensure change control procedures and policies are in place to reduce errors. This is one of the most common pitfalls when using a spreadsheet. Versions are tracked and changes are not documented. A record should be kept of different spreadsheet versions, what changes were made, when they were made, who made them and who approved them. Adopt a process of periodic internal and external review for model accuracy, formulas and output reporting. 3) Have written policies at the board, management and line level on how to handle model creation, version control, validation and documentation.
Because of the nature of our business with other banks, our models get constantly reviewed by internal and external auditors as well as regulators. Models such as our Loan Pricing application, Credit Stress, ALM and others undergo rigorous review and documentation in order to ensure accuracy and save our clients work dealing with model governance on their own. We can tell you, after more than 5Ys of following the above 3 rules, they have saved our bacon on multiple occasions. Errors are now caught quicker, versions are more consistent and validation is easier. While models are beautiful, they can also require a lot of maintenance over time to keep their youthful appearance.