Actionable insights — Mortgage bonds and the refinancing option

1. Track the note-rate distribution of outstanding mortgages, not just the level of rates

1. Pull the distribution of outstanding mortgages by note rate (ICE and agency pool data publish it) — both loan counts below a low threshold and the share of unpaid balance above a higher threshold.
2. Compare that distribution with where mortgage rates could realistically go over the next year or two. The share of balances that would be "in the money" to refinance under a plausible rate drop is the size of the live prepayment option.
3. Re-run it periodically: ordinary turnover (new buyers, payoffs, moves) steadily replaces old low-rate loans even when refinancing is dead, so the option wakes up without any rate move.

• The 5%+ share crossing further milestones; the gap between the outstanding weighted-average coupon and the current mortgage rate; any rate drop of 50–100 bp that pushes a large cohort in the money.

2. Decompose a yield pickup into what you are actually being paid for

1. When a "no credit risk" bond yields more than Treasurys, name the risk the spread compensates. For agency MBS it is not default — it is the borrower's option to prepay.
2. Check whether that risk was temporarily switched off in the period the trailing returns cover. Trailing outperformance earned while the option was dormant says little about the period when it is live.
3. Only count the spread as attractive if it is wide relative to the option's current value, not relative to the dormant era.

• The current-coupon MBS spread to Treasurys widening or tightening as the option's value changes; option-adjusted spread vs nominal spread where available.

3. Stress-test both rate directions — negative convexity loses either way

1. For any callable or prepayable income asset, write out two scenarios. Rates up: prepayments slow, life extends, price falls more than a comparable bullet bond. Rates down: prepayments surge, principal returns early and must be reinvested at lower yields, and price gains are capped.
2. Ask which scenario the current macro setup makes likelier, and which one the market is pricing.
3. Size the position as if the "locked-in yield" is not locked in — the realized holding period is set by borrowers, not by you.

• Extension risk if the Fed does tighten; a sharp long-end rally (e.g. on an AI-capex bust) as the contraction trigger; the MOVE index as a gauge of rate volatility, which raises the option's value.

4. Don't model prepayment speeds purely on history — check what changed in the plumbing

1. List the structural drivers of refinancing propensity and check each against its historical norm: average loan balance (bigger loans save more dollars per refi), borrower credit quality (easier to qualify), servicer mix (nonbank servicers earn on volume and solicit refis aggressively), and process friction (time and paperwork to refinance).
2. If most drivers point to faster-than-historical response, treat speed models calibrated on past cycles as underestimating the option.
3. Look for strategist language signalling consensus complacency as a contrarian flag.

• Nonbank servicer share of agency servicing; days-to-close on refinances and AI-underwriting adoption by lenders; actual monthly prepayment speeds vs model forecasts after the next rate dip.