Rate × duration
Pricing
A small rate change compounds across a long-lived balance and changes both payment and lender margin.
← Back to Strategy
FINANCIAL PRODUCT ANATOMY / 04
Mortgage Operating Model
Price the loan. Test affordability. Stress the collateral. Measure prepayment. Keep the lender P&L connected.
A mortgage is not a long consumer loan with a house attached. Its economics are shaped by duration, funding, collateral value, borrower affordability and the option to refinance when rates move. The same decision that lowers the monthly payment can extend exposure for decades.
Mortgage value is created when affordability for the borrower and duration-adjusted economics for the lender remain aligned.
José ÑáñezTechnology Advisor · Board Member
Published August 30, 2026Updated August 31, 202613 min read
Five forces determine mortgage economics
LTV × term × PTI
Affordability
Property value only becomes financeable when down payment, payment and borrower income fit together.
Long duration
Funding
The asset can remain on balance sheet for decades, so funding assumptions matter more than in short-duration credit.
House value ≠ cash recovery
Collateral
LTV protects only after property-value stress, cure, foreclosure and recovery costs are recognized.
Borrower option
Prepayment
When market rates fall, refinancing can remove high-yield balances before the lender captures their expected lifetime margin.
Where a mortgage actually makes — and loses — money
Mortgage profitability is the result of a long-lived balance, not a single approval. Digital improvement matters only when it changes one of these economic lines.
| Line | Economics | Type | Levers |
|---|---|---|---|
| Interest income | Average portfolio × portfolio yield | Revenue | Pricing, conversion, retention |
| Funding | Average portfolio × funding cost | Cost | Term, runoff, treasury structure |
| Credit loss | Default/severity after collateral recovery | Cost | Underwriting, early cure, collateral data |
| Origination OPEX | Applications + underwriting + valuation + closing | Cost | STP, data verification, AVM, eClosing |
| Servicing OPEX | Active contracts × assisted-service intensity | Cost | Self-service, automation, proactive servicing |
| Prepayment / refinance | Unreplaced runoff × contribution margin | Margin leakage | Retention, repricing, refinance journey |
DIGITAL OPERATING LEVERAGE
Digitalization is valuable only when it changes the economics
The model avoids a generic digital-transformation benefit. Each capability must change conversion, unit cost, credit loss or retained balance before PBT moves.
| Digital lever | Variable | Economic chain | Step |
|---|---|---|---|
| Application completion | Digital completion / recovered conversion | New mortgages → portfolio → interest and funding | 04–05 |
| Underwriting / STP | Eligible applications × STP × unit-cost reduction | Origination OPEX ↓; TAT stays diagnostic until conversion changes | 14 |
| Digital valuation | Eligible originations × automated valuation × cost reduction | Origination OPEX ↓; collateral data quality ↑ | 15 |
| eClosing | Eligible closings × digital closing × cost reduction | Origination OPEX ↓; errors/rework and funding cycle can fall | 16 |
| Digital servicing | Eligible contracts × self-service × cost reduction | Servicing OPEX ↓ | 17 |
| Early cure | Cure rate on deteriorated portfolio | Credit loss / recovery burden ↓ | 10–11 |
| Retention / refinance | Eligible unreplaced runoff × digital retention capture | Retained balance → interest − funding − risk | 18 |
External evidence supports the mechanisms, not the country assumptions: Fannie Mae identifies shorter life cycles, faster funding, lower operational errors and better data from eClosings; GAO notes that proptech can simplify homebuying and speed underwriting; FHFA documents regulated AVM use in collateral decisions. Fannie Mae · GAO · FHFA
EXECUTIVE INSTRUMENT
Change one mortgage decision. See both sides move.
The model separates observed market references from derived values and editable assumptions. LTV, term and rate change lender economics and borrower payment simultaneously; PTI by itself remains diagnostic.
Live economics
Colombia · COP
Current PBT
COP 232,960 M
Transformed PBT
COP 232,960 M
Net annual impact
COP 0 M
01
Reference country
Loads currency, references and starting assumptions.
2026-08 · BBVA Colombia publishes a 13.04% effective annual rate from August 20, 2026 for traditional non-VIS mortgages in COP, with terms from 60 to 240 months.
Effective annual COP rate
Assumptions and sources
Davivienda publishes financing of up to 70% of the property value for new or used housing.
COP 500 million property value, 9.5% funding, 0.8% annual loss, costs, prepayment and recovery are editable assumptions. The rate is a commercial reference, not a system average.
BBVA Colombia · Tasas hipotecarias · agosto 2026Davivienda · Crédito hipotecarioLive P&L
Accumulates Origination + Structure/Pricing + Collateral + Prepayment + Digital levers − OPEX. PTI is diagnostic and does not change the P&L by itself.
Origination COP 0 MStructure COP 0 MCollateral COP 0 MPrepayment COP 0 MDigital COP 0 MOPEX COP 0 MTotal COP 0 M
| P&G | Current | Transformed | Δ |
|---|---|---|---|
| Interest income | COP 1,460,480 M | COP 1,460,480 M | — |
| Fees and other income | COP 28,000 M | COP 28,000 M | — |
| Funding cost | − COP 1,064,000 M | − COP 1,064,000 M | — |
| Financial margin | COP 424,480 M | COP 424,480 M | — |
| Net credit loss | − COP 89,600 M | − COP 89,600 M | — |
| Margin after risk | COP 334,880 M | COP 334,880 M | — |
| Origination | − COP 6,720 M | − COP 6,720 M | — |
| Portfolio servicing | − COP 44,800 M | − COP 44,800 M | — |
| Collections and recovery | − COP 5,600 M | − COP 5,600 M | — |
| Other operating expense | − COP 44,800 M | − COP 44,800 M | — |
| Transformation OPEX | COP 0 M | COP 0 M | — |
| Profit before tax (PBT) | COP 232,960 M | COP 232,960 M | — |
20
Review and traceability
Accumulated scenario across five levers, affordability and investment.
Origination
COP 0 M
Structure
COP 0 M
Collateral
COP 0 M
Prepayment
COP 0 M
Digital
COP 0 M
Net impact
COP 0 M
Deterministic model through PBT. It excludes taxes, regulatory capital, ECL/IFRS 9, full amortization/prepayment curves, duration, convexity, inflation/UF and credit advice.
How to read this first version
Standard mortgage, not subsidized housing
The base model intentionally excludes VIS/VIP, FOGAES and other social-housing subsidies or guarantees. These require separate pricing, eligibility and risk logic.
Prepayment is economic runoff
The prepayment lever approximates margin lost or retained when eligible performing balances prepay and only part of the runoff is replaced. A formal ALM model would use seasoning, CPR/SMM curves and duration.
Collateral is stressed economically
The model converts deteriorated balance into implied collateral through LTV, applies a property-value haircut, cure and recovery cost, and compares the resulting economic burden.
Chile is modeled in UF
The Chile preset operates in UF and treats the observed housing rate as a real/UF rate. It does not forecast CPI or convert future installments into CLP.
THESIS
The best mortgage is not the one with the lowest payment, the lowest LTV or the highest spread in isolation. It is the structure that remains affordable, fundable and profitable across a long economic life.
José Ñáñez