The Risk of Asymmetric Loan Prepayments
- Andrew Mitchell
- 12 hours ago
- 3 min read
One of the most influential and often under-scrutinized assumptions in interest rate risk (IRR) models is loan prepayment speed. Prepayment assumptions directly affect the timing of cash flows, reinvestment rates, earnings sensitivity, and, ultimately, the institution’s risk posture.
Yet, in many ALM frameworks, prepayment behavior is modeled simplistically, which may obscure the full earnings volatility profile.
Depending on the size and sophistication of the institution, assumptions typically fall into one of two approaches:
Static speeds based on historical or industry averages applied uniformly with no variation across rate scenarios.
Internally derived baseline speeds with linear scenario scaling, where speeds increase or decrease symmetrically as rates move. For example, if the baseline speed is 10 CPR, a Down 200bp scenario might assume 15 CPR and an Up 200bp scenario 5 CPR.
Because borrower behavior shifts significantly across rate environments, static speeds are inherently limited. Given today’s coupon distribution, cash flows are expected to slow only modestly as rates rise but accelerate sharply when rates fall. This highlights a key dynamic many IRR models fail to capture: asymmetric prepayment risk.
Fixed-rate residential mortgages provide a clear example. They represent more than 13% of loan balances in DCG’s Loans360°® cross-institution database. Most portfolios remain concentrated in loans originated in 2020-2021, when mortgage rates were below 4.00% and many borrowers were locking in rates below 3.00%. Today, approximately 61% of outstanding fixed-rate residential loans carry coupons below 4%. That embedded coupon distribution creates option-like behavior.
DCG’s Loans360° forecast projects that if rates move lower by 200bp, residential prepayment speeds may increase to about 1.5x baseline speeds. In a Down 400bp scenario, they may jump to 2.6x of baseline speeds. This ramp-up in activity is due to large cohorts of borrowers becoming positively incented to refinance down to a lower rate. Conversely, if rates move 400bp higher, speeds may only slow to about half of the baseline, as refinancing incentive erodes. The response is clearly non-linear, and heavily skewed to the downside.
In practice, this means:
Loans with coupons above 6% may respond quickly in a modest Down 100bp environment
Loans in the 3% - 4% band may not materially accelerate until rates fall 300 - 400bp
Prepayments occur in multiple waves as different coupon cohorts move “into the money”
This is negative convexity embedded within a loan portfolio that is often modeled to prepay symmetrically. While residential mortgages show the most pronounced asymmetry, similar patterns exist in CRE, Agriculture RE, and C&I portfolios. The degree of prepayment variability depends greatly on the following factors:
Coupon distribution
Remaining term and proximity to repricing
Loan size
If asymmetric prepayment risk is not properly captured, IRR results can be distorted. In rising rate scenarios, models may overstate asset sensitivity because the actual deceleration in cash flows is often less pronounced than the model assumes. In falling scenarios, models may understate earnings compression as accelerated cashflows are not fully captured. When reinvestment into lower coupons is understated, net interest income (NII) sensitivity may appear more stable than it truly is.
For institutions concentrated in fixed-rate residential or CRE loans, this distortion can be significant, particularly in Down 300 and Down 400 scenarios. If ALCO is not seeing the true magnitude of NII exposure in falling rate environments, strategic decisions around hedging, funding mix, pricing, or balance sheet restructuring could be suboptimal.
For many institutions, prepayment risk may be a skeleton in the closet. Given the current allocation of low-coupon loans that remain on balance sheets, ALCO should be asking:
How does our coupon distribution behave across a full suite of Down rate scenarios?
At what rate levels do meaningful waves of refinancing begin?
Are our Down 300 and Down 400 results reflecting true reinvestment risk, or a smoothed assumption set?
If these questions cannot be answered clearly, the institution may not be seeing its full earnings volatility profile.
To learn more about how DCG's Cross-Institution Loan Analytics can help drive strategic decision-making, click here.
© 2026 Darling Consulting Group, Inc.
