Вы находитесь на странице: 1из 2

15.

450 Analytics of Financial Engineering Review – Structured Public Securities Association (PSA) Curve: A way of modeling prepayment
Products speeds

Static Yield Spread: A fixed increment to the stripped Treasury curve that Effects of prepayment, generally:
equates the bond’s discounted (one path) cash flows to its observed
price
• Bonds with different CF patterns can have different YTM • Faster prepayment  More CF from prepayment early and less CF
• Static spread absorbs all behavior we can’t model separately from prepayment later on  MBS price ↓
• Loosely attributable to the value of the prepay/call option and the • Higher CPR  higher prepayment
value of other factors • Interest payments very sensitive to CPR
o Higher CPR  less interest payments  IO value ↓
OAS: A kind of spread applied to cash flows that have been adjusted • Higher aggressiveness (i.e. sensitivity to WAC-refi)  more aggressive
already for their embedded optionality; modeled as z below prepay  call option worth more  MBS price ↓  static spread ↓ 
OAS ↓
• Possible factors that are explained by OAS: • Interest rates↓  Prepay ↑  IO ↓ in value & PO ↑ in value since
o Credit risk principal payments are accelerated
o Liquidity risk
o Systematic prepay risk (e.g. robust economy), but not Pool Factor: The amount of the pool that is not prepaid (i.e. not counting
interest-rate driven prepayment that is in option natural amortization) – starts at 1.00 and falls
o Model risk (e.g. prepayment function, changes in
origination, changes in prepayment factors, interest rate
process assumptions) Hedging MBS with amortizing swaps is only effective if the amortization of
the swaps matches the amortization and prepayment of the pool

 CF1 CF2  Extension Risk: The risk that the PO life turns out to be longer than
P = E + +  expected because of higher interest rates (and lower realized prepayment)
1 + r1 + z (1 + r1 + z )(1 + r2 + z ) 
Prepayment Risk: The IO can have negative duration (see graph below)
• CF = Interest + Principal Amortization + Prepayment of Principal
• Expectation over all modeled interest rate paths
• CF are based on optimal or predicted behavior of bond issuer on
any given path, per an empirical prepayment model

Prepayment factors include:

• Age (new mortgages have low prepayment)


• Aggressiveness: sensitivity to (WAC – refi mtg rate)
• Seasonality (summer has higher prepayment)
• Geography (coasts have higher prepayment)
• Burnout (path dependencies)

Mortgage Backed Securities

Homeowner Originator Agency Investor


∆P / P
D effective =
∆r
• Homeowner: Principal & Interest
• Originator: Principal & Interest – Servicing • Calculate ∆ P using a fancy model that accounts for optionality
• Agency: Principal & Interest – Servicing – Guarantee
• Investor’s coupon < Weighted average coupon of mortgage pool
Pr incipal paid at time t
Weighted Average Life = ∑t ×
Valuation of MBS: (Average NPV across paths)
Total Original Pr incipal

• Interest rate simulation Static Yield: YTM based on a single path (assuming either unchanged
o Calibrated to Treasury curve so that the expected value mortgage rates or forward mortgage rates)
across paths is the same as the observed values
o Arbitrage free-pricing Static Spread: Spread over Treasury strip curve for a single path (either
o Monte Carlo simulation of Treasury rates unchanged mortgage rates (static spread) or forward mortgage rates (zero
• Prepayment model  CF projections and path dependency volatility OAS))

Conditional Prepayment Rate (CPR): Fraction of mortgages left in the Notes:


pool that are expected to prepay in the coming year modeled from
historical data, conditional on economic variables. (see graph) • OAS lowest for highest coupon even though static spread is highest
(because the option is most ITM for high coupon)
Single Monthly Mortality (SMM) rate: Monthly prepayment speed • As correlation between S/T Treasuries and mortgage rates ↑  OAS↓
because of reinvestment
• Higher volatility of interest rates  Option cost ↑  Lower PV of
(1 − SMM )12 = (1 − CPR ) ⇔ SMM = 1 − (1 − CPR )1 / 12 mortgage pool (ceteris paribus) but the price is the same in the
observed market  OAS ↓
o Effect most pronounced when option at-the-money (vega)
• Duration increases as coupon falls (with volatility ↑)
CANNOT COMPARE OAS ACROSS DIFFERENT PREPAY MODELS Coupon rate ↓ ↑
Discount rate ↓ ↑
MBS Derivatives Prepayment ↑ ↓
Net effect ↓ ↓
• IO prices decline for higher coupon rates
o Generally insensitive to coupons as prepayment speeds Problem Set/Midterm Notes
offset effect of higher coupons
• POs rise with coupon increases since prepayments good for PO • Use the WAC for amortization calculation
holders
• PV CF at the Mtg Rate
• The presence of either a very large positive OAS or a very large
• ED = 0.5 x 100 x (P-100 bps – P+100 bps)
negative OAS suggests a model problem (comparability issue)
• Lower vol  lower option  higher OAS, ceteris paribus
• Higher aggressiveness  higher prepay  higher option  lower
OAS, ceteris paribus
• WAL: Average across paths of weighted average time to principal
repayment for each path
• For secondary tranches in WAL calculation, must consider WAL of
earlier tranches first
• PACs in subsequent tranches can extend the WAL of antecedent cl.
• Changing the model:
o Higher aggress.  option worth more  OAS must fall in
order to keep NPV of pool same
o Higher burnout  less people prepaying  option worth less
IO: Interest rate ↑  Prepayment slows  IO value ↑
 OAS must rise in order to keep NPV of poo
• Analyzing PACS:
• If interest rate ↑ too much, then discount rate forces value ↓, though o Talk about rapid pre-pay scenarios  shorten later cl.
o Talk about slow pre-pay scenarios  extend earlier cl.
PO: Interest rate ↑  discount at higher rate and prepay lower  PO ↓ • Talking about interest rate risks:
o Low-rate scenarios
o High-rate scenarios
Collateralized Mortgage Obligations (or REMICs): A structured finance
instrument in which payments from the underlying pass-through are o Criteria: (see table opposite)
divided into classes or tranches with different maturity characteristics • Coupon
such that they are OBS, have monthly payments and no double taxation • Prepay
• Discount
• Remember the callable bond example: (see below)
Vanilla CMO:

• Principal payments go to first or earliest tranche until it is paid off,


then to next senior tranche
• Each class receives interest based on stated coupon in every period
• Earlier tranches (fast pay and low duration)
• Late tranches (slow pay and high duration)

Other developments:

• Accrual class or Z-class: Gets zero cash flow until all other classes
retired (like a zero)  high duration and extension risk
• Residual tranches: Slow-pay tranches with a higher coupon than a
fast pay (in an upward sloping yield curve environment), effectively
interest only with only negligible capital  pick up the slack
o WAC of tranches < WAC of pool to avoid overcommitment
• Planned Amortization Class: A tranche with specific priority rules to
make it amortize within a given payment window, conditional on
CPR being realized within a pre-set range (otherwise the PAC is
infeasible and it is said to be ‘broken’); has positive convexity
o Range of CPRs depends on the relative size of the PAC
class (big PACs are more easily broken)
o Too high realized CPR  PAC not forced to take early
payments, possibly extended if not enough principal left to
pay PAC in window (C tranche suffers)
o Too low realized CPR  PAC doesn’t have enough
principal to be repaid  extended but not as long as other
classes (A tranche suffers)
o So-called Companion classes have negative convexity,
taking on risk laid off from PAC class

Inverse Floaters: A bond with a coupon that is inversely related to the


floating rate (e.g. 44.7% - 4 x LIBOR) (different for MBS because of
prepay)

• Coupon leverage must be the ratio of principal amounts in the 2 sub-


classes (4:1 in above example)

3 effects on value Int. Rates ↑ Int. Rates ↓

Вам также может понравиться