by Calculated Risk on 4/23/2008 11:40:00 PM

Here are some more Ambac comments. Note: here is the referenced Ambac Presentation

Sean Leonard, CFO: ... David, you could discuss kind of the breakout of some of the HELOC portfolio being that big portion of the portfolio is related to large bank transactions versus [investment bank generated] shelf transactions.

David Wallis, Chief Risk Officer: Yes, it is very striking. One has various hypotheses about this, and we are investigating those hypotheses. But it is very striking how concentrated, how very concentrated, some of the poor performers are, and that gives rise to all sorts of obvious questions.

I mentioned that we have diagnostic and forensic people working on some of these deals. We are beginning to see stuff back from that. The diagnostic is basically running tapes looking at delinquencies and trying to figure out given what you now know and what you knew then, would you have expected that delinquency or not? And if the answer is not, well, that is interesting.

So, in other words, you have an incredibly low FICO within a pool, and it is delinquent. Well, maybe you expected that. But if it is incredibly high and the LTV was incredibly low, then maybe you would not expect that.

So then what you do is, you take an adverse sample, so you run the tape through a program, take an adverse sample, i.e. looking for the suspicious ones, and then what you do is you go look at the files. That is a very difficult long process, but you look in the files. You look at the transcripts of servicing records, and you see what you see. And all I will say is that there is some pretty amazing stuff to see. So very concentrated adverse exposures, that is really the message here.
emphasis added

"Pretty amazing stuff to see." It's not clear if he is referring to fraud - but it sounds like it.

Q: Analyst: Can you clarify for the home equity and the Alt-A reserve strengthening, are the credit reserves there now kind of reflective of estimated lifetime losses? And if so, I guess that is my understanding of how it works per quarter. What has changed if you look at Q1 where we are versus Q4? And what could move us further down back that in terms of changes as we look forward through the year?

David Wallis, Chief Risk Officer: Sure, let me take a go at that ... the notion is that we're taking a present value of the losses or the claims that we expect to pay over the life of the deal.

Just to relate that, people always like to relate it to cumulative loss because that is the statistic that people bandy around. Let me just give you a bit more insight onto that. I think in the presentation we talked about -- in fact, it is the most egregious example, the Bear Stearns deal, where we are expecting or re-modeling at least around 82% of collateral loss. So you've got 100 people in the round that took out a closed-end second in this deal, 82 have walked out and not paid you a whole lot back, 100% severity.

Just in relation to other transactions, just to give you some more data and just be open with what we're looking at here, I mentioned that in mid prime [Alt-A], we are kind of 20 to 25% collateral loss. In HELOC they vary, the ones we have reserved from I think about a low of mid-20s to a high of just north of 50. So that is collateral loss.

In terms of what has happened and where does it go from here, probably a good thing to do is to look at the chart on page 31, and you get a pretty good sense of what has happened in the last few months. You know, basically losses have taken off in that transaction.

Sometimes you get perplexing movements. I will draw your attention to one. If you look at the First Franklin deal, which is also in the charts that I presented, you will see four months ago I think it was quite a marked flattening in delinquencies. That proved to be a false dawn because, although delinquencies -- the trajectory there has flattened, actually losses have continued to escalate. So the data is difficult. You get very odd data.

To give you a sense of how odd the data is, the remits are actually beginning to come in somewhat late because sometimes people don't believe the data that is being presented, and they send it back and say, well, that cannot be right. But, in fact, unfortunately some of it is right, and the numbers are huge.

Where can it go? Again, look at page 31, and I admit this is the extreme example. A criticism might be, well, look at the very sharp dimunition in monthly realized loss that is being projected here through the role rate methodology. And it is true, it is a fairly sharp diminution. However, it has to be because if it is not, you end up with more than 100% of collateral loss, which does not make any sense either.

So I think further discussion that Mike just had in relation to subprime and what is outstanding and what does that imply about future default and severity rates to get to a given collateral loss, we're seeing some of the same things certainly in relation to our most stressed transactions. You know, how bad can it get? 81 people in 100 walking away sounds pretty bad to me.