Prepayment uncertainty – CMOs

For certain CMO tranche types, such as IO (interest-only), PO (principal only), inverse
floaters and various ‘support’ tranches, and mortgage strips, the prepayment uncertainty
measures can attain much greater magnitude, both positive and negative,
than for passthroughs. By the same token, well-protected PACs will exhibit a lesser
degree of prepayment uncertainty than the underlying pass-through collateral. At times, the prepayment uncertainty values for CMOs may seem counterintuitive; in
other words, tranches that one would expect to be highly vulnerable to a small
change in prepayment forecasts have fairly low prepayment uncertainties, and vice
versa. This surprising result is a good reminder of the complexity of these securities.
Consider the examples.

Here we see an IO tranche with an overall prepayment uncertainty value of 7.638;
in other words, the tranche’s value would increase (decrease) by more than 7.5% if
prepayments were expected to be 10% slower (faster) than originally forecasted. This
is not surprising, given the volatile nature of IO tranches. If prepayment forecasts
are revised to be faster than originally expected, it means that the (notional) principal
balance upon which the IO’s cash flows are based is expected to pay down more
quickly, thus reducing the total interest payments to the IO holder. In contrast, the
prepayment uncertainty of the collateral pool underlying the IO tranche is a modest
0.19 – a 10% change in expected prepayment speeds would produce only a small
change in the value of the collateral.
One would expect PO tranches to exhibit fairly large prepayment uncertainty
measures as well, as POs are priced at a substantial discount to par (they are zero
coupon instruments) and a change in prepayment forecasts means the tranche
holder expects to recoup that discount either sooner or later than originally estimated.
The total prepayment uncertainty for this particular PO is –2.731; note that the
underlying collateral of this PO is a ‘re-remic’ – in other words, the collateral is a
combination of CMO tranches from other deals, which may be backed by various
types of collateral. In a re-remic, the underlying collateral may be a combination of
highly seasoned, premium mortgages of various ‘vintages’ so it is virtually impossible
to estimate the tranche’s sensitivity to prepayment model risk simply by noting that
it is a PO. The exercise of computing a prepayment uncertainty measure for CMOs
reminds us that these are complicated securities whose sensitivities to changing
market conditions bears monitoring. 229