Quick answer: a mortgage servicing right (MSR) is valued as the present value of expected future net servicing income from a pool of mortgages. The cash-flow engine estimates servicing fees, ancillary income and float-like benefits, then subtracts servicing costs, especially the higher costs associated with delinquent and defaulted loans. The model must also estimate when mortgages leave the servicing pool through prepayment, refinancing, payoff or default. Those expected cash flows are discounted at a rate or option-adjusted spread consistent with their risk. Because prepayment, default and rates change the expected life of the servicing stream, MSRs behave like path-dependent financial assets rather than simple fixed annuities.

An MSR is valuable only while the mortgage remains in the servicing pool and the cost of servicing stays below the income it produces.

Exact reader question and page role

This article answers: how can a bank turn a right to service mortgages into a defensible numerical value? It does not duplicate How Banks Model Mortgage Prepayment, which owns the prepayment hazard itself, or How Banks Construct Interest-Rate Hedges, which owns hedge construction. Here prepayment and hedge sensitivities are inputs and diagnostics for valuation.

1. What cash flow does an MSR represent?

A servicer performs functions such as collecting borrower payments, managing escrow accounts where applicable, remitting principal and interest to investors, maintaining records, handling delinquency and loss-mitigation workflows, and interacting with investors, guarantors and borrowers.

A simplified per-period net servicing cash flow is:

NCF_t = servicing fee_t + ancillary income_t + escrow/float value_t − servicing cost_t − other contractual/operational costs_t.

The MSR value is then approximately:

MSR value = Σ E[NCF_t × survival of servicing right to t] / (1 + k)^t.

The survival term is crucial. If the loan prepays, refinances, liquidates or otherwise leaves the serviced pool, future servicing income disappears or changes.

2. Fair value depends on expected net income, not unpaid principal balance alone

A servicing portfolio may cover billions of dollars of mortgage principal, but the MSR is worth only a fraction of that principal because the servicer receives a servicing strip rather than the whole loan cash flow.

The OCC Mortgage Banking handbook describes MSR fair value as the present value of expected net servicing income using a market-appropriate discount rate and realistic, independently supportable prepayment assumptions. See OCC Mortgage Banking.

This immediately separates three variables that are often mixed together:

mortgage balance;
servicing fee rate;
expected life of the servicing relationship.

3. Prepayment shortens the revenue stream

If a borrower refinances or prepays, future servicing fees on that loan generally disappear. Prepayment therefore acts like an early termination option held by the borrower.

For a pool with scheduled balance B_t and servicing fee s, expected fee income before costs can be sketched as:

Fee_t ≈ s × E[B_t conditional on survival].

If mortgage rates fall, refinancing incentive can rise, expected life can shorten and MSR value can fall even if current fee income has not yet changed. This is why MSR valuations can respond strongly to rate scenarios.

4. Default is different from ordinary prepayment

Both prepayment and default can end or reduce future servicing income, but default can also make servicing much more expensive before termination. Delinquent loans require additional contact, accounting, advance management, property or legal workflows and investor/guarantor processes.

A Federal Reserve FEDS Note published on 4 June 2026 emphasises this asymmetry. It finds that higher defaults reduce MSR value partly because default servicing is considerably more expensive, while prepayment reduces the duration of servicing income. See Mortgage Servicing Right Valuations Under Stress.

The same note estimates that under its studied stress scenarios, default-driven declines in MSR values could range from about 5% to 13% depending on portfolio composition and assumptions. It also estimates that a one-percentage-point increase in prepayment rate could reduce MSR values by roughly 4% in the portfolios analysed. Those are study-specific results, not universal valuation constants.

5. Pool composition matters more than one average prepayment rate

Two servicing portfolios can have the same unpaid principal balance and fee rate but very different value because their borrowers and loans differ.

coupon relative to current mortgage rates;
loan age and seasoning;
loan-to-value;
credit quality;
geography;
investor/guarantor type;
occupancy and property characteristics;
historical prepayment and delinquency behaviour.

The June 2026 Federal Reserve analysis notes that large-bank servicing books were disproportionately seasoned and had different risk characteristics from the broader agency market, which materially affected projected default-driven MSR losses. Model calibration therefore has to reflect the actual pool rather than one generic market CPR/CDR assumption.

6. Discount rate versus option-adjusted spread

A deterministic discounted-cash-flow model can discount expected servicing cash flows at a risk-adjusted rate k. More advanced market-consistent approaches can project cash flows across interest-rate paths and value them using an option-adjusted spread (OAS) over modelled discount curves.

Conceptually:

Price = expected discounted path-dependent cash flows after incorporating a spread for residual risk and market pricing.

OAS is not a magic truth parameter. Its interpretation depends on the interest-rate model, prepayment model, cash-flow assumptions and market calibration. Changing the model can change the spread required to fit the same observed market price.

7. Servicing cost should be state-dependent

A performing mortgage and a seriously delinquent mortgage do not cost the same amount to service. A useful model therefore separates states:

cost = cost(current) × P(current) + cost(delinquent) × P(delinquent) + cost(default/workout) × P(default/workout).

If the valuation assumes one low average servicing cost while the portfolio is likely to migrate into delinquency under stress, the model can materially overstate value.

This connects to How Banks Forecast Loan Delinquency and Cure.

8. Escrow and float-like benefits can add value but are not guaranteed

Depending on contractual and regulatory arrangements, servicing can create value from custodial balances, escrow-related economics or ancillary fees. The OCC notes that discount-rate analysis should consider uncertainties in future servicing costs, ancillary income and earnings associated with escrow accounts.

These components should not be treated as permanent fixed add-ons. Rate changes, legal requirements, customer behaviour and investor agreements can alter them.

9. A miniature deterministic valuation

Imagine a simplified servicing pool expected to generate net servicing cash flow of S$4 million next year, declining by 20% per year because balances run off. With discount rate 10%, the first four expected cash flows are:

Year 1: S$4.00m;
Year 2: S$3.20m;
Year 3: S$2.56m;
Year 4: S$2.05m.

The present value of those four alone is approximately:

4/1.1 + 3.2/1.1² + 2.56/1.1³ + 2.05/1.1⁴ ≈ S$9.6m.

If faster refinancing causes 35% annual runoff instead of 20%, the value falls because later servicing cash flows shrink. If a recession raises default servicing cost at the same time, value falls further even if discount rates are unchanged.

10. Interest-rate sensitivity can reverse the intuition of ordinary bonds

When rates fall, a conventional fixed-rate bond usually rises in price. An MSR can fall because lower rates increase refinancing and shorten servicing income. This gives MSRs unusual duration characteristics.

The exact sign and size depend on the portfolio and model. A seasoned pool with low coupons may respond differently from newly originated high-coupon mortgages. The valuation engine therefore needs scenario sensitivities rather than one blanket statement that “MSRs have negative duration.”

11. Hedge sensitivity is not the same as fair value

A bank can calculate how MSR value changes under rate, volatility and prepayment shocks and use those sensitivities to design hedges. But a hedge does not change the economic definition of the MSR itself.

Useful risk measures include:

value change for parallel rate shocks;
key-rate sensitivity;
prepayment sensitivity;
default sensitivity;
volatility sensitivity;
discount-spread sensitivity;
cross-effects, such as rates changing both discounting and prepayment.

A hedge built only on rate DV01 can fail if the main valuation loss comes from default servicing cost or modelled prepayment behaviour rather than a pure discount-rate move.

12. Fair-value model and market price can diverge under stress

A model can produce a present value even when the market for MSRs becomes less liquid. The Federal Reserve’s June 2026 analysis explicitly notes that widespread mortgage distress could reduce MSR market liquidity, making it difficult to realise full carrying value through a sale.

This creates a critical boundary:

model fair value ≠ guaranteed executable sale price.

Valuation governance should therefore use observable transactions and market indications where available and apply greater uncertainty when liquidity deteriorates.

13. Accounting rules can change without the economics changing instantly

MSRs sit at the intersection of economics and accounting. FASB Subtopic 860-50 governs servicing assets and liabilities in US GAAP. In 2026, FASB added a technical project on whether the value attributable to mortgage “recapture” should be included in MSR measurement, reflecting diversity in practice. See FASB — Mortgage Servicing Rights—Recapture, updated 7 July 2026.

This is a useful reminder that accounting unit-of-account choices and valuation conventions can evolve. The cash-flow economics still need to be explicit enough that a rule change can be mapped rather than hidden inside a black-box value.

14. Failure modes

Static prepayment. One CPR is used across rate scenarios and borrower cohorts.
Uniform servicing cost. Defaulted loans inherit performing-loan cost assumptions.
Pool-average blindness. Seasoning, LTV and guarantor mix are compressed into one average.
Discount-rate plug. The discount rate is adjusted until the model reaches a desired value without evidence.
OAS mystique. A calibrated spread is treated as model-independent truth.
Liquidity omission. Model value is assumed executable even in a frozen MSR market.
Hedge=value confusion. Gains on hedges are embedded in the MSR value instead of tracked separately.
One-factor stress. Rates change without prepayment, default or servicing cost responding.

15. Counterexamples and falsifiers

Claim: “Falling rates always increase the value of mortgage-related assets.” Falsifier: refinancing accelerates enough to shorten servicing income and reduce MSR value.

Claim: “Default and prepayment are equivalent because both remove loans.” Falsifier: default generates materially higher servicing cost before termination while ordinary prepayment does not.

Claim: “A model value of S$100m means the portfolio can be sold for S$100m.” Falsifier: market depth disappears under stress and executable bids fall materially below model value.

16. Diagnostics

What percentage of MSR value comes from the first three years?
Which assumption moves value most: CPR, CDR, servicing cost or discount spread?
How does sensitivity differ by coupon and vintage?
Are default servicing costs calibrated to actual operational outcomes?
Does the model reproduce recent third-party MSR transaction levels where comparable?
How much value depends on escrow/ancillary assumptions?
What happens if refinancing and default shocks occur together?
How much of the apparent hedge effectiveness disappears under a volatility shock?

17. Verification and update triggers

backtest prepayment and default forecasts by pool;
compare servicing-cost assumptions with realised costs by delinquency state;
reconcile modelled fee cash flows with actual servicing income;
benchmark discount rates/OAS to observable market transactions;
run independent challenger valuations;
revalue after material mortgage-rate moves or volatility changes;
update after servicing contract, investor or accounting changes;
increase uncertainty reserves/governance when market liquidity falls.

18. The valuation pipeline

Define the serviced pool and contractual fee structure.
Project scheduled principal balances.
Estimate prepayment and payoff paths.
Estimate delinquency/default transitions.
Assign servicing costs by state.
Project ancillary and escrow-related economics where applicable.
Generate expected net servicing cash flows.
Discount under a governed rate/OAS framework.
Run rate, prepayment, default, cost and volatility sensitivities.
Compare with market evidence.
Separate hedge P&L from asset fair value.
Backtest realised cash flow and model errors.
Recalibrate when portfolio or market structure changes.

Research anchors

The deeper mathematical lesson

An MSR is a contingent stream whose lifetime is controlled partly by borrowers and the economy. The valuation must therefore model not just how much cash arrives, but whether the right survives long enough for that cash to exist, what servicing state the loan occupies, and what market discount is appropriate for the remaining uncertainty. The strongest model is not the one with the most precise point estimate; it is the one that makes the value’s dependence on prepayment, default, cost and market liquidity visible enough to be challenged.

Educational boundary: This article explains public mortgage-banking mathematics. It is not mortgage advice, investment advice, servicing guidance for a specific borrower, or a valuation opinion on any particular MSR portfolio.

Small Group Tutorials

How Banks Value Mortgage Servicing Rights: Servicing Cash Flows, Prepayment, Default Costs, Discount Rates and Hedge Sensitivities