To: NCUA Board Members
From: Chip Filson
The industry is deeply concerned about the numbers used in taking the decisions to conserve Wescorp and US Central. I would like to explain the basis for this concern. I have attempted to describe the situation in a way that a lay person, the press, or any other interested party might understand the implications from the conclusions you reached.
The $9 Billion Credit Loss Estimate
The following is for US Central only, but we have completed the same analysis for Wescorp and would be glad to send it to you.
In our analysis we have used the entire investment portfolio, even though we understand that the focus of the PIMCO study and the primary source of the risk and loss was in the private-label mortgage backed securities ($13 billion).
The rounded numbers use US Central's Feb 28, 2009 balance sheet with the losses then calculated using the Agency's explanations as follows:
US Central's investment portfolio: $24 billion
To be insolvent in the most likely base case the following would have to have occurred:
- The entire credit enhancement of 43% on only the Non-Agency RMBS would have been lost. This amount equals $5.2 billion;
- The entire capital of US Central would be wiped out: $ 2.5 billion;
- The Insurance Fund's case-specific loss reserve, for which the industry is being billed, of approximately $1.5 bn to USC (balance of $5.0 billion to Wescorp) is then added to the above.
This means the Agency is estimating, using various models, a most likely loss on the entire portfolio of $9.2 billion. What does that number imply about future conditions for that to be a reasonable and probable outcome?
Two factors primarily determine the outcome of MBS and other ABS modeling: the rate of default (how many homes go into foreclosure) and the loss severity on each defaulted loan across the portfolio.
Are 60%+ Defaults and Loss Severities Reasonable?
To arrive at the total $9.2 billion for the entire portfolio, not just mortgage securities, the default rate would have to average 62% and the loss on every loan the same number, 62%. Any combination of these two factors that results in a $9.2 billion number is feasible, so we have just shown an average for each dimension.
The board is being asked to approve actions today based on a scenario that assumes 62% of loans go into default and there is a 62% loss rate. Is that a reasonable outlook for the actions you took?
Again, I would stress that in the Agency's discussions the focus was on the mortgage-backed securities from which statements I conclude that the default rates and loss severity numbers actually "determined" were much higher. For example, if most of the $9.2 billion projected credit losses were in the $13 billion private label MBS, does that level of defaults and severities (in the mid 70's) seem reasonable? I understand the bottom up approach, but even that can be summarized into these two numbers.
What Should the Board Do?
Therefore, I believe the question you want staff to clarify quickly is what are the average default rates and loss severities from their analysis? The question is not how does the computer work to get the numbers, but is the outcome a reasonable future assumption to take the actions you did. Is there any contemporary actual evidence from payment streams or specific bonds that would support such a scenario where 62% of loans are in default and the losses average 62% on each such default? Are market prices anywhere near the values implied by this analysis?
There are two kinds of errors that can be made in a situation like this. The first is a factual error, for example how were the credit enhancements factored in—by NCUA or PIMCO; were they assigned at the loan level? bond level? etc. The second error is in making a judgment based on the model and the likelihood of the projections occurring.
I am not sure where an error may have occurred; but I do not believe the default and loss severity are in line with anyone's view of the future.
I believe you should initiate a dialogue immediately about the appropriate way to modify your actions.
If the logic or facts above are faulty, however, I would be more than happy to learn why that is the case. This is not about access to PIMCO's model; rather, it is using the output from any model with intelligence, care and common sense. These are the numbers, along with those for WesCorp, that I think need to be published as soon as possible.
Let me close with two final points. No one is objecting to paying the bill. Everyone agrees this is a systemic crisis. However, given future uncertainties, the reasonable way to pay the bill is by aligning the actual loss discovery with the premiums. No one would object to a process of expensing real losses as they occur assuming the wind downs are done in an orderly and temperate manner.
I will be in the office until 2:30 today and then back by next Tuesday if you would like further clarification on information about the above.