Guide

Mortgage rates explained

Harbor Development’s 2024 land-bank model used the federal funds rate plus a fixed 250 basis-point spread to forecast buyer affordability. When the 30-year fixed mortgage rate peaked near 7.8% while the funds target sat near 5.5%, their starts forecast overshot actual housing starts by 22%. The miss was not Census noise — it was a category error. Mortgage rates are not overnight policy rates with a label. They are long-duration, prepayment-sensitive liabilities priced off Treasury yields, MBS spreads, and lender balance-sheet capacity.

This guide explains how the U.S. 30-year fixed benchmark is measured, the primary-secondary spread pipeline, why mortgage rates lag and lead the Fed at different points in the cycle, ARM and points mechanics, affordability math, the Harbor Development affordability sleeve refactor, a technique decision table versus fed-funds-only housing models, pitfalls, and an investor checklist.

What the headline mortgage rate measures

The most-cited U.S. benchmark is Freddie Mac’s Primary Mortgage Market Survey (PMMS) — a weekly average of conforming 30-year and 15-year fixed rates offered to prime borrowers with 20% down. Lenders report their best execution rate; Freddie publishes a simple average. It is a primary market rate: what a qualified borrower sees on a rate sheet before closing costs.

Other benchmarks matter for different questions:

Optimal Blue or Black Knight OBMMI — daily loan-level indices used by traders; more timely than weekly PMMS.
MBA Weekly Applications Survey — contract rates on newly originated loans; useful for application-volume context.
Secondary market current-coupon MBS yield — what investors earn on newly issued Ginnie/Fannie/Freddie pools; the wholesale price that drives primary rates.

When macro investors say “mortgage rates,” they usually mean the 30-year fixed PMMS or its daily cousins. The 15-year fixed trades at a spread below 30-year; adjustable-rate mortgages (ARMs) index off SOFR or Treasury maturities plus a margin.

How mortgage rates are built: Treasury, MBS, primary-secondary spread

A useful decomposition for the 30-year fixed:

Primary mortgage rate ≈ 10-year Treasury yield + MBS option-adjusted spread + primary-secondary spread + lender margin

The 10-year Treasury is the anchor because 30-year mortgages have average lives much shorter than 30 years — borrowers prepay when rates fall or they move. Duration is typically 6–8 years in a stable rate environment, which tracks the 7- to 10-year Treasury sector more closely than the 30-year bond.

The MBS OAS compensates investors for prepayment uncertainty, credit guarantee fees, and liquidity. When quantitative tightening shrinks the Fed’s MBS holdings, spreads can widen even if the 10-year yield is flat — lifting mortgage rates without a policy rate change.

The primary-secondary spread is the gap between the average rate lenders charge borrowers and the yield on newly issued MBS pools they sell. It reflects origination costs, servicing value, pipeline hedging, and capacity constraints. During the 2020–2021 refi boom, the spread compressed as volume scaled. In 2022–2023, as volume collapsed and lenders shed staff, the spread widened — mortgage rates rose more than Treasury yields alone implied.

Fed policy transmission: when mortgages follow and when they do not

Monetary policy moves mortgage rates through several channels, not one dial:

Expectations channel — FOMC guidance reprices the 2- to 10-year Treasury curve; mortgage rates often move on announcement days even before the funds rate changes.
Balance sheet channel — Fed MBS purchases in QE compress MBS spreads; QT lets spreads re-widen. This is independent of the funds target.
Bank funding channel — deposit betas and warehouse line costs affect lender margins, especially for non-bank originators.
Risk appetite — During stress, MBS spreads widen like corporate credit; mortgages can cheapen when Treasuries rally on flight-to-quality.

Historical rule of thumb: the 30-year fixed tracks the 10-year Treasury with a spread of roughly 150–200 basis points in calm markets. That relationship is not stable. Primary-secondary spreads of 250+ basis points have occurred when origination capacity was strained. Do not hard-code a single spread in housing models.

Affordability: payment math and the lock-in effect

Affordability is payment-driven. A $400,000 loan at 6.5% carries a principal-and-interest payment near $2,528/month; at 7.5%, the same loan costs $2,797 — roughly 11% more per month. Median home prices and down-payment norms amplify small rate moves into large payment shocks.

Two second-order effects matter for macro:

Lock-in — homeowners with sub-4% coupons avoid selling, shrinking existing-home inventory. Listings fall even when demand is stable.
ARM reset cliff — borrowers who took 5/1 or 7/1 ARMs near cycle lows face payment jumps at reset, stressing household budgets and delinquency rates.

Pair mortgage rates with personal income and consumer credit trends — rate levels alone do not tell you whether households can absorb the payment.

Points, APR, and product mix traps

Quoted rates assume discount points, credit score tiers, and loan-to-value bands. A “no-point” rate is higher than a rate with one discount point paid upfront. APR attempts to annualize fees into the cost of credit but can mislead when borrowers sell or refi within a few years.

Product mix shifts the aggregate economy:

Conforming vs jumbo — Jumbo spreads widen when bank balance sheets are constrained.
FHA/VA — Ginnie Mae MBS carry different prepayment and spread dynamics than Fannie/Freddie conventional pools.
ARM share — Rises when fixed rates are high; increases future reset risk in household sector models.

Harbor Development affordability sleeve refactor

Harbor Development’s v1 land-bank trigger fired on fed funds plus 250 bp. Refactor after the 2023 forecast miss:

Daily mortgage monitor — Optimal Blue 30-year conforming index instead of weekly PMMS lag.
Decomposed spread model — 10-year Treasury (live) + current-coupon MBS OAS (Bloomberg/Fannie) + primary-secondary spread (MBA series). Alert when spread widens >50 bp in 30 days independent of Treasuries.
Payment-to-income ratio — median home price in Harbor markets, 20% down, 30-year fixed, divided by BEA per-capita disposable income. Trigger land-bank pause when ratio exceeds Harbor’s 28% historical ceiling.
Starts confirmation — require two consecutive months of single-family starts momentum before accelerating lot purchases; rates lead starts by 1–3 quarters.
Lock-in proxy — track existing-home months’ supply and mortgage stock distribution; high lock-in suppresses resale inventory even if rates ease modestly.
QT overlay — when Fed MBS runoff exceeds $30B/month, add +25 bp stress to spread assumption in affordability scenarios.

Land-bank acquisition timing error fell from 22% to 9% on 12-month forward starts versus the v1 model. The lesson: mortgage rates are a stack, not a single spread to policy.

Technique decision table

Your situation	Prefer	Avoid
Forecasting housing demand	30-year fixed payment-to-income + starts momentum	Fed funds target alone
Rate shock attribution	Decompose Treasury vs MBS OAS vs primary-secondary	Blaming only the Fed for mortgage moves
MBS portfolio hedging	Current-coupon spread, refi incentive, duration	10-year Treasury hedge ratio held constant
Homebuilder equity read	Mortgage rate + cancellation surveys + orders	Single-month starts without rate context
Policy transmission study	QE/QT MBS leg plus curve expectations	Funds rate pass-through assumed at 1:1
Household stress test	ARM reset cohort + delinquency trends	Fixed-rate stock only at today's coupon

Common pitfalls

Equating mortgage rate to fed funds. The spread is variable and can widen sharply at cycle turns.
Using stale PMMS for trading. Weekly averages lag daily moves by days; use daily indices for timing.
Ignoring primary-secondary blowouts. 2022–2023 showed lender capacity matters as much as Treasuries.
30-year Treasury substitution. Mortgage duration is closer to 7–10 years; the 30-year bond is a poor hedge proxy.
Assuming instant housing response. Starts and new home sales lag rate changes by quarters.
Lock-in blind spots. Modest rate declines may not unlock existing-home supply if borrowers are deeply in-the-money on coupons.
Points confusion. Comparing rates without normalizing discount points distorts cross-period charts.

Investor and operator checklist

Track daily 30-year conforming rate alongside 10-year Treasury yield.
Monitor current-coupon MBS OAS and primary-secondary spread weekly.
Decompose rate moves into Treasury, MBS, and origination components.
Calculate payment-to-income for your target markets using median price and income.
Pair mortgage levels with single-family housing starts and new home sales.
Watch Fed MBS runoff pace during QT for spread widening risk.
Follow MBA application index for leading demand signal.
Map ARM origination share and upcoming reset cohorts for credit risk.
Compare jumbo vs conforming spreads for bank balance-sheet stress.
Include mortgage rate on FOMC-day scenario grids, not just the funds target.
Stress-test affordability at +100 bp and +200 bp from current levels.
Archive spread decomposition when forecasting homebuilder or REIT exposure.

Key takeaways

The 30-year fixed mortgage rate is a primary-market borrower quote — not the federal funds rate and not the 30-year Treasury yield.
Decompose into 10-year Treasury + MBS spreads + primary-secondary spread; each leg moves independently through the cycle.
QT can raise mortgage rates via MBS spread widening even when the policy rate is on hold.
Affordability is payment-driven; lock-in effects can suppress existing-home supply after rates peak.
Harbor Development cut land-bank timing error from 22% to 9% by replacing fed-funds-plus-spread with a decomposed rate and payment-to-income model.