Data & Analytics Lending & Credit Risk Risk Ratings Portfolio Risk & CECL

Why financial institutions are rethinking 2D risk rating models

Kent Kirby

April 8, 2025

Read Time: 0 min

This article covers these key topics:

The difference between 1D and 2D risk rating models
How CECL has impacted the necessity of a dual approach
Why the LGD variable is so difficult to pinpoint

Does your risk rating framework align with your CECL needs?

More banks are rethinking the value of two-dimensional (2D) risk rating models as CECL and real-world challenges with LGD raise questions about their practicality. While 2D models offer structure and clarity, many institutions are shifting toward simpler frameworks. This blog breaks down the pros, cons, and what financial institutions should consider when evaluating their risk rating approach.

Is a 2D risk rating model still worth it?

An effective risk rating framework is probably the single most important tool a bank can use when it comes to managing credit risk. One of the ongoing debates in building these frameworks is whether to go with a one-dimensional (1D) or two-dimensional (2D) model.

For the last 20 years or so, I’ve been firmly in the 2D camp. But lately, I’ve started to reconsider. Let me explain.

Transform CECL data into stress testing insight.

LEARN MORE

What’s the difference between 1D and 2D?

A two-dimensional model—sometimes called an Expected Loss (EL) model—breaks things down into three parts:

Probability of Default (PD): How likely it is that the borrower will default.
Loss Given Default (LGD): How much we’d lose (as a percentage) if that default actually happens.
Exposure at Default (EAD): The dollar amount at risk if default occurs.

The math behind this calculation is:

EL = PD × LGD × EAD

In contrast, a one-dimensional model combines PD and LGD into a single score or rating. It also tends to factor in other elements that may (or may not) have anything to do with actual credit risk.

The big selling point of the 2D model has always been clarity. Bankers using this model can see how much risk is tied to the borrower vs. the structure of the facility. And when CECL came along, with its focus on expected loss, it seemed to align naturally with the 2D approach—at least at first.

Do CECL and 2D risk rating models align?

In my opinion, not anymore.

CECL looks at expected credit losses over the life of the loan. However, most risk rating frameworks are designed with a much shorter time horizon in mind—usually 12 to 18 months. That mismatch alone raises questions about how well the two actually fit together.

Data shows that banks and credit unions have been trending away from using 2D risk rating frameworks in recent years. In its annual loan review survey, Abrigo asks financial institutions about 1D vs. 2D modeling. Here are the results of the most recent survey:

For banks under $10B in assets, only 5% used a 2D model in 2024 (down from 15% historically).
For the $10B–$50B crowd, 2D model use dropped from 60% to 50% in 2023.
Even at institutions over $50B—where 2D adoption was nearly universal—it fell to 50% in 2024.

That’s a significant shift. And it tells me something’s changing in how banks think about risk frameworks.

While many institutions have adopted a dual risk rating system over the past couple of decades, the OCC and NCUA make it clear: it's not a regulatory requirement. A well-designed single rating system may be more than enough for smaller or less complex institutions. The key is consistency and accuracy—your risk rating approach should clearly reflect both the borrower’s ability and willingness to repay, as well as the strength of any supporting structure or collateral.

The only real argument left in favor of 2D is that it’s more precise and provides financial institutions with more visibility into risk components. And to be fair, PD is pretty straightforward. Banks and credit unions have a lot of history and data to support it. The biggest variable tends to be each institution’s risk appetite and how they define thresholds.

But LGD? That’s where it starts to fall apart.

Why LGD is so hard to pin down

Here’s where my experience—and maybe some of yours—comes into play.

LGD estimates at origination often miss the mark.

We typically assess LGD at origination based on a going-concern assumption. But when defaults actually happen, all kinds of issues come into play—deferred maintenance, collapsing collateral value, and lost revenue streams, for example. I’ll never forget a deal from the late 1980s: five commercial lots in North Dallas were appraised at $110 million in September 1987. By January 1988, they were worth $49 million. Nothing had changed physically; the market just evaporated. The original LGD estimate can quickly go out the window.

We rarely account for the true cost of collecting.

Early in my career, I worked a small $15,000 equipment loan. The loan went south; it turned out that the equipment was junk, and we lost everything. But we also spent at least $10,000 in legal fees, repossession costs, and other expenses trying to recover the funds. In the end, the real loss to the financial institution was at least $25,000—or 166% of the loan. And that doesn’t even count internal time and salaries. Most banks don’t even track those internal and risk-related costs. So, how can we calculate LGD with any confidence?

Time changes everything.

During the 2008 financial crisis, our regulators directed us to charge down certain residential lot loans. But instead of foreclosing, we kept some borrowers barely afloat—just enough to survive—because we didn’t want to manage land. Over the next few years, we recovered much of what we wrote off.

How do financial institutions factor these situations into LGD? The time value of money? What is the ongoing cost of “keeping the lights on” while you wait for recovery? It gets complicated fast.

Where do financial institutions go from here?

I still believe the structured thinking behind a 2D model has value. But I’m starting to wonder if it’s worth the complexity, especially when one of the key components—LGD—is unpredictable in real-world practice.

What matters most is finding a framework that fits your institution’s size, complexity, and strategic goals—without overcomplicating the process. Whether you stick with a 1D model or continue to refine a 2D approach, the key is to ensure it delivers meaningful insights that help you manage credit risk effectively.

If you’ve got thoughts, I’d love to hear them.

Email [email protected]. Let’s figure out what makes sense—for our institutions, our borrowers, and the future of credit risk management.

Adapt your credit analysis processes to address economic volatility.

View the webinar Talk to an advisor

Webinar

CECL mastery series: From compliance to confidence

Webinar

CECL model governance: Controls, documentation, and audit readiness

Whitepaper

A banker’s guide for CECL compliance and backtesting

About the Author

Kent Kirby

Senior Consultant, Portfolio Risk

Kent Kirby is a retired banker with over 39 years of experience in all aspects of commercial banking: lending, loan review, back-room operations, credit administration, portfolio management and analytics and credit policy. As Senior Consultant in the Portfolio Risk practice, Kirby assists institutions in the review and enhancement of commercial

Full Bio

Looking for Valuant? You are in the right place!

Looking for DiCOM? You are in the right place!

Looking for TPG Software? You are in the right place!

Why financial institutions are rethinking 2D risk rating models

This article covers these key topics:

Does your risk rating framework align with your CECL needs?

Is a 2D risk rating model still worth it?

Transform CECL data into stress testing insight.

What’s the difference between 1D and 2D?

Do CECL and 2D risk rating models align?

Why LGD is so hard to pin down

Where do financial institutions go from here?

Adapt your credit analysis processes to address economic volatility.

CECL mastery series: From compliance to confidence

CECL model governance: Controls, documentation, and audit readiness

A banker’s guide for CECL compliance and backtesting

Kent Kirby

Senior Consultant, Portfolio Risk

About Abrigo