Mortgages, MBS, Prepayments, and OAS (Chapter 20)

Book: Fixed Income Securities: Tools for Today’s Markets | Author: Bruce Tuckman & Angel Serrat | ISBN: 978-0-470-89169-8

Previous: Credit Default Swaps and the CDS-Bond Basis (Chapter 19, Part 2) | Next: Fitting Discount Curves with Flat Forwards (Chapter 21, Part 1)

Chapter 20 is the book’s tour of the US mortgage market. It is also the chapter that explains why your “bond math” intuition fails when homeowners can refinance whenever they feel like it.

Mortgage Loans

Fixed-rate level-payment loans dominate agency MBS. A $100,000 loan at 4% pays $477.42 monthly for 30 years. Early payments are mostly interest. Principal ramps up over time. “Interest lives off principal.”

Borrowers hold a prepayment option: pay outstanding balance anytime. When rates fall, PV of remaining payments exceeds balance, so prepayment is in-the-money. Unlike corporate calls, exercise is not efficient. Borrowers are slow, distracted, or cash-constrained. That gap is why MBS can trade above par at low rates.

Subprime ARMs with teaser rates fueled the 2007-2009 crisis. Teaser reset, home prices fell, delinquency on ARMs hit 25% by May 2008. By Sep 2010, 23% of mortgages were underwater nationally (46% in Florida).

From Loans to Pass-Throughs

Banks used to hold mortgages to maturity. Securitization changed that. Pools of loans become pass-through securities: interest, scheduled principal, and prepayments flow to investors minus servicing and guarantee fees.

Agency (FNMA, FHLMC, GNMA) vs private-label (jumbos, Alt-A, subprime). Post-crisis, new issuance was almost all agency.

Pool stats matter: WAC (weighted average coupon on loans), WAM (weighted average maturity), factor (current/original principal). FNMA 2004 vintage 3.5% pool: factor 68% by Dec 2010 vs ~90% scheduled amortization alone. Loans left the pool. Higher-coupon loans prepaid first.

TBAs, Dollar Rolls, and Market Structure

Specified pools trade on characteristics (loan size, geography, FICO). TBAs are forwards with a delivery option. Seller delivers cheapest pool matching issuer/coupon/term. High loan-balance pools often CTD, so TBA prices embed that.

Current coupon = front TBA trading just below par. Dec 2010: 4% coupon, interpolated par rate about 4.17%.

Dollar rolls finance MBS: sell Jan TBA, buy Feb TBA. Not quite repo. You may not get the same pool back. You miss the month’s cash flows. Roll value above carry reflects delivery option value.

Prepayment Modeling

Four drivers: refinancing (dominant), turnover (moving homes), defaults, curtailments (small partial prepays).

Refinancing incentive often looks like:

Incentive = (WAC - current mortgage rate) x WALS x annuity - fixed refinancing cost

Bigger loans prepay faster (fixed cost spread over more principal). CPR follows an S-curve in incentive. Parameters vary by SATO, FICO, LTV, geography, and history.

Burnout is the killer effect. FNMA 7% 1995 pool: CPR hit 40%+ when rates first hit 6-6.5% in 1998. Same rate zone in 2006-07? Much lower CPR. Fast refinancers already left. Pool is “burned out.” Models need path-dependent state. Trees fail here. Monte Carlo wins.

Turnover has seasonality and a seasoning ramp (nobody moves right after closing). Lock-in: borrowers with below-market rates stay put.

Valuation: Monte Carlo, Not Trees

One-factor trees assume cash flows depend only on current rates. Burnout depends on rate history. Path dependence kills the tree.

Monte Carlo: simulate many rate paths, project scheduled + prepayment cash flows per path, discount back, average. Shift curve for DV01.

Modules stack: benchmark rates, scheduled flows, mortgage rate model (often regressed on 10y swap), housing prices for defaults, prepayment model feeding off all of the above.

OAS = spread to short rates that makes model price equal market price. Constant OAS implies hedged return is short rate + OAS. Misspecified prepayment model means OAS mixes relative value with model error.

Figure 20.4: FNMA current-coupon TBA OAS mean-reverts around zero until 2008-09, then spikes to 100 bp cheap. Buy the dip? Only if you could fund the mark-to-market through the peak. OAS may have been pricing GSE credit fear, not model mispricing.

Price-Rate Behavior and IO/PO

At high rates, MBS acts like a slower amortizing bond (lower DV01). As rates fall, CPR rises, principal returns at par, price flattens. Negative convexity, like callable bonds, but borrowers are sloppier exercisers so price can sit above par.

IO strip: only interest. PO strip: only principal. When rates fall and prepayments surge, PO rallies (discount paid at par). IO collapses (interest base vanishes). IO can have negative duration. Useful for portfolio hedging, brutal to trade.

Empirical TBA hedge ratios (Dec 2010): 4% coupon hedged with 0.66x 10y Treasury or 1.15x 5y. Higher coupons prepay faster, shorter effective maturity, lower hedge ratio.

Who Needs What Hedge

Servicers earn 20-50 bp on outstanding balance. Prepay kills fee income. MSR risk resembles IO. Hedging with TBAs creates nasty convexity mismatch. Swaps add mortgage-swap basis risk.

Primary market lenders lock borrower rates then securitize later. Rate rise hurts pipeline value. Short TBAs is the standard hedge.

Why This Chapter Matters

MBS is where rate risk meets option risk meets behavioral finance. You cannot price it with static cash flows. You cannot hedge it with one Treasury future and sleep well. OAS is the relative-value language, but the prepayment model is the hidden bet.

Chapter 21 goes back to curves: how practitioners bootstrap discount factors from liquid instruments when exact arbitrage does not line up dates.