Mortgage passthroughs

The spread duration of a mortgage-backed security is less predictable than for a
corporate bond and is not necessarily related to its effective duration. We observed
that for corporate bonds, a change in secondary market spreads affects the cash
flows to the bondholder because of the effect on the exercise of a call or put option.
Can we make the same claim for mortgage-backed securities? In other words, can
we predict whether or not a change in secondary market mortgage spreads will affect
a homeowner’s prepayment behavior, thereby altering the expected future cash flows
to the holder of a mortgage-backed security? What implications does this have for
the spread risk involved in holding these securities?
Let us consider two separate possibilities, i.e. that homeowners’ prepayment
decisions are not affected by changes in secondary spreads for MBS, and conversely,
that spread changes do affect homeowners’ prepayments. If we assume that a
homeowner’s incentive to refinance is not affected by changes in spreads, we would
expect the spread duration of a mortgage passthrough to resemble its Macaulay’s
duration, with good reason. Recall that Macaulay’s duration tells us the percentage change in a bond’s price given a change in yield, assuming no change in cash flows.
Spread duration is calculated by discounting a security’s projected cash flows using
a new OAS, which is roughly analogous to changing its yield. Therefore, if we assume
that a change in spreads has no effect on a mortgage-backed security’s expected
cash flows, its spread duration should be close to its Macaulay’s duration.3
However, it may be more appropriate to assume that a change in OAS affects not
only the discount rate used to compute the present value of expected future cash
flows from a mortgage pool, but also the refinancing incentive faced by homeowners,
thereby changing the amount and timing of the cash flows themselves. As discussed
in the next section on prepayment uncertainty, the refinancing incentive is a key
factor in prepayment modeling that can have a significant affect on the valuation
and assessment of risk of mortgage-backed securities. If we assume that changes in
spreads do affect refinancings, the spread duration of a mortgage passthrough would
be unrelated to its Macaulay’s duration, and unrelated to its effective (optionadjusted)
duration as well.
Adjustable rate mortgage pools (ARMs) are similar to FRNs in that changes in
spreads, unlike interest rate shifts, do not affect the calculation of the ARM’s coupon
rate and thus would not impact the likelihood of encountering any embedded reset
or lifetime caps. Therefore, an ARM’s spread duration may bear little resemblance to
its effective duration, which reflects the interest rate risk of the security that is
largely due to the embedded rate caps.