|
Decision
system construction, validation and regulatory review. |
|
To contact us: |
|
Many variables that harm
the credit profile of the borrower
actually help the prepayment probability.
A high number of inquiries or “Low Documentation” may slightly
increase the probability of default but can decrease the probability of rapid
prepayment. From a fair lending
perspective, low-to-moderate income loans are commonly reputed to prepay slowly while super jumbos prepay quickly. Pricing for credit while ignoring the
impact of prepayment is a source of
disparate impact on protected groups. Prepayment models, like
their counterparts in credit, vary from time of origination as opposed to
seasoned pool evaluation. For
prepayment of seasoned pools or Portfolio Defense see the
Behavioral Prepayment
link. |
|
Management Consulting Services |
|
Mortgage Decision Technologies, Inc. |
|
Scoring loans at
origination, separating into equal thirds and observing a 50% decrease in CPR
from the median for the slow scores and a 50% increase in CPR over the median
for fast scores is quite common.
Prepayment scores rank order the likelihood of prepayment for
individual loans. Scores do not
attempt to predict the economic environment that may produce a 10 CPR or a 40
CPR environment. Assuming the
capitalization of a 40 basis point servicing strip to yield 8% over its
expected life, the servicing value of the “Slow” category in the illustration
is 1.97 points. Contrast that with the
0.90 point value of the “Fast”
category. The Modified Duration for
the groups is 4.5, 3.0, and 2.2 years
respectively. Assuming that
the lender correctly predicted an
overall duration of 3 years, the 1.07 point difference in servicing value
(1.97 less .90 = 1.07) illustrated
equates to a 36 basis point difference in yield. This compares with credit losses of 1 to 10
basis points over the past five years.
The home mortgage industry prices for credit risk and ignores the impact of predictable prepayment differences several times as large! |
