Fair Lending & Compliance

Does Model Risk Management (MRM) have to be such a nightmare?

Adam Kleinman
May 21, 2018

Lenders’ Model Risk Management (MRM) processes were set up to accommodate for conventional models based on logistic regression. That process worked, but I think we can all agree when I say that model verification, fair lending analysis, and model risk management documentation are pretty burdensome for these traditional models.

Machine learning (ML) — which can use thousands of variables and multiple modeling methods in its process — is likely to present a significant challenge for your organization’s established manual MRM process. Because the likelihood that a lender can manually do model verification, fair lending analysis, and model risk management documentation for their machine learning technology is almost impossible... at least not in a timely or efficient manner.

So the question rises: is there a way to adopt AI underwriting while also not putting strenuous measures on your legal department? Of course. In fact, you can actually automate your MRM framework, taking the process from nightmare to a walk in the park.

Here's some other ways that automating your MRM framework can help you achieve better results throughout your organization:

       
  1. Gets your models into production safely and quickly – It takes the typical bank about 12 months to get its credit model into production. With an automated MRM framework, you can reduce this timeline by as much 75 percent.
  2.    
  3. Improves your operational efficiency and consistency – Instead of bogging down your modeling team with compiling documentation, an automated MRM framework frees them up to focus on the modeling work. Meanwhile, your internal compliance and audit personnel will be able to review MRM documentation with a greater degree of confidence since automation standardizes the process and produces consistent results.
  4.    
  5. Accelerates your ability to keep up with external shocks – One of the lessons from the 2008 financial crisis was that traditional models were unable to keep up with changing market conditions. Lenders saw the world changing around them, but were unable to update their models to reflect the new environment. Automation of the MRM process reduces the time between model build and model deployment, which means that your organization is better positioned to react and adjust to macroeconomic conditions.
  6.    
  7. Facilitates knowledge transfer between your employees – Employee turnover forces new people to come in and work on a model they did not develop. An automated MRM process eases this transition, giving analysts and modelers insight into model build and results. This allows for a streamlined, highly efficient process that can save your company time and money.
  8.    
  9. Enhances your scalability and validation – Leveraging automation in your MRM process can facilitate scalability and repeatability. Having a multitude of modelers produce different MRM documentation is not a process that can be scaled. Automation, however, makes each MRM document, regardless of which modeling team produced it, consistent. Automation improves operational efficiency by getting models safely into production faster. And being able to verify the models’ results can help protect your organization from business losses due to human bias and error.
latest
March 28, 2024
Innovation In Lending
Looking beyond market pain points to find purpose
March 25, 2024
Women’s History Month: sisters are doin' it for themselves
February 27, 2024
Zest Cares — we mean it when we say it!