Boy, we sure got prepayments wrong!
Originally published 3/16/2011 © 2021 Olson Research Associates, Inc.
The next time your regulator or auditor says you must use better prepayment data instead of bank estimates - “instead use something like Bloomberg,” consider this:
About a year ago I attended the FDIC’s Symposium on Interest Rate Risk Management in Arlington, VA. During the session entitled, “IRR Management Challenges” we heard from the Chief Investment Officer from Bank of America. Much of what he had to say was geared toward mega-bank IRR management. But there was one quote that really caught my attention, “Boy, we sure got prepayments wrong!” Now I know such an error can have large ramifications for the financial sector as a whole (B of A is rather big). But I started to wonder how this error could impact the way community banks model. It turns out if you use Bloomberg Dealer Prepayment Forecasts it could have a big impact.
Here are two snapshots of prepayment data. One is data from 07/30/2009 and the other is data from one year later 08/02/2010.
Apparently BOA did get them wrong. They made some pretty significant adjustments one year to the next: 2009 base PSA was 849, 2010 base PSA was 423. Their estimate was cut in half. Other firms changed their estimates as well, but most were not as dramatic. Here’s the interesting thing however, the Median Prepayment forecast actually went up. In July 2009 it was 690, in August 2010 it was up to 726. If BOA got it wrong – what does that say about the other firms’ estimates?
Here’s another interesting observation – take a look at the forecasts for SAL (Solomon Bros.) and GS (Goldman Sachs). They didn’t change at all, not even slightly. The numbers are exactly the same. I don’t know about you, but I have some serious data integrity questions when I see something like this.
I know I’m not going to change the ways of regulators and auditors with this one simple post. My goal is just to raise awareness. Quite often the data you get from these systems may be just as much an estimate as yours (i.e. SWAG).
And for those of you that insist on data worshiping at the house of Bloomberg…beware.