That would be worrying — except that Sproule’s methodology is all messed up. Apparently he arrived at his CPR number by looking at the change in number of loans in the trusts last month and annualizing it, and calling it the result prepayment speed. There’s tiny little problem with this technique: something like 60% of the change in the loan count in a trust is due to defaults, not prepayments. Sproule’s numbers are meaningless.
Anyway, here’s what he did (using the 2004-2 trust as an example):
Beginning loan count: 60,995
Monthly change in loans: -490
Times 12 months: 5,880
Divided by beginning loan amount: an “implied CPR” of 9.64%
Here’s how the numbers really work:
Other Adjustments (cancellations, consolidations and other): $1.47 million
Coincident Losses (change in cumulative claim payments made): $2.38 million ($21.6 million less $19.3 million)
Accordingly, the total loan count decline would be split between Other Adjustments (38%) and Coincident Losses (62%).
Are prepays and defaults something that investors ought to be vigilant about? Of course! But this particular piece of analysis simply doesn’t provide much help on that score. At all.
Disclosure: Author's fund is long FMD
