Finance

§

AIG - Q3 Expectations and Capital Needs

We've spent the past few days trying evaluate AIG's 2nd quarter report and use it as a basis for a guesstimate of what may be on tap for Q3. From our perspective there are four primary issues:
(1)CDS portfolio valuation
(2)Investment portfolio valuation
(3)Liquidity Needs
(4)Dividend
We note up front that our modelling methods are very crude and at best ballpark approximations as we relied heavily on the Markit ABX-HE indices as well as AIG's conference call and 10-Q presentations. The ABX-HE indices are synthetic CDS based on a pool of 20 underlying CDOs of the same credit grade issued in the prior six months. In the case of the AAA ABX-HE tranche, it is well known to be one of, if not the lowest rated AAA (it is longest in duration of the AAA's). To address this short coming, Markit has added PENAAA tranches which are one step higher up.

AIG's CDS portfolio is (notionally) concentrated in the regulatory capital relief area which for all intents has been and should remain nearly risk free. The problems have been in the multi-sector area, particularly those swaps on CDOs with subprime or alt-a collateral. AIG stopped writing these multisector CDS in the first half of 2005 and much of the portfolio was written on 2004 and earlier collateral. Thus, using the ABX-HE-2006-H1 index is, at best, only a very rough approximation. There is also the question of subprime vs alt-a collateral, though that seems to be a matter more of degree of correlation. In addition, we have made assumptions that those credits which are on watch for downgrade will only be moved down one notch this quarter.

As a test, we modelled AIG's CDS losses from Q1 and found our estimate to be about 15% too high - again, not surprising as we are using an index weighted for 2005 H2 collateral against a portfolio written against 2004/2005H1 CDOs. Adjusting for this overshoot, we come up with these figures for Q3:
We estimate the current marks as of 8/18 are $1,300B

If current price trends continue but no collateral is downgraded, we estimate unrealized losses of $2,100B

If current price levels hold and all collateral currently on watch is downgraded, losses of $6,500B

If current price trends continue and all collateral currently on watch is downgraded, losses of $7,100B

For purposes of the CDS estimates, we assumed that the downgrades were distributed by grade as in the Conf Call Supplemental pg 10.

By 9/30, assuming the collateral on watch is all downgraded, fully 60% of AIG's CDS portfolio will have been marked to A credit or lower. From a valuation perspective, the underlying collateral has been written down to near nil. At this stage it becomes pointless to try to estimate Q4 marks as there will be little 2005 exposure left and we will be guessing what, if any, of the 2003/2004 swaps remain to be hit with no easy reference valuations.

Keep in mind, the above figures are all unrealized losses, or as AIG likes to call them, economic losses. AIG does modelling for this very complex task and have estimated ultimate realized losses of $5B to $8B. Their model and assumptions have come along way since 2007 and appear to be very conservative. For example, in coming to the $5-8B figure, they assume virtually all mortgages that become delinquent will be foreclosed.

Our best guess is that the CDS will suffer on the order of $5B in marks this quarter, similar to Q2 and less than our $6.5-7.1B worst case scenarios, primarily as we expect most, but not all, of the collateral on watch to be downgraded.

The investment portfolio is a similar beast but with most of the risk in 2006 and 2007H1 RMBS. Ironically, while one half of AIG was selling protection the other was going without. We have applied a similar methodology to the RMBS portion of the investment portfolio as we used for the CDS - the various ABX-HE indices. Again, at best those indices are ball park proxies for reality.
Based on downgrades through 7/31 and assuming 2/3 of the $6.1B on watch negative are downgraded one step with 90% AAA and 10% AA:

Based on CMBS portfolio also showing same level of marks as Q2 of $400M

Based on Jumbo/Foreign MBS showing slightly worse than estimated Q2 marks of $1B

Based on HELOC and 2nd Lien already reflecting significant downgrades

We estimate (whether reflected as OTTI or unrealized loss) marks of $5,305B +/- 15% ($4,500-$6,100B)
The problem for both the investment portfolio and the CDS portfolio is not so much that the market values for any particular credit grade deteriorate but that collateral be downgarded. The drop from AAA to AA is a cliff, as is from AA to A on the older 2005 collateral. Once again though, these are only unrealized losses at this stage. In fact through July, only two tranches of the Markit indices have suffered any writedowns - 2006-BBB-H2 and 2007-BBB-H1 of about 5% in each. Realistically this will go up as homes go through the foreclosure process but considering the level of subordination in the RMBS portfolio we would only expect very small, if any losses in the Alt-A and subprime paper. The area we see as more likely to have a loss (though not large in absolute terms) would be the prime loan RMBS where subordination is only 2-3%.

But what of AIG's liquidity needs? This is currently the subject of much chatter in the market and it seems most of it from shorts. Since Q1, AIG has raised a total of $23.5B in new capital. But what have the used and what looms ahead?
AIG has posted collateral of $16.5B thru Q2 against CDS, up from $9.7B in Q1.

We expect this to rise in line with expected losses in the portfolio. Taking our 2nd worst case of $6.5B marks, expect collateral posted to climb to $23B. This would be a use of $13.3B

2a-7 puts. They have issued a total of $11.3B notional including $7.5B from a commitment made in 2005. Of this, they purchased $917M in Q2 and have taken loss reserves of $810M. The buyers of the puts have agreed to provide liquidity for up to $8.5B of which $3.2 is in use. No effect except the balance sheet enlargement.

CDS with over collateralization provisions resulting in accelerated payment: Roughly $8B notional in total of which $1.5 (6 securities) have had such defaults and $103M has been purchased at par (1 security). The majority of the $8B notional is not multisector w/subprime. Additional needs: $1.4B

Ratings Agency Downgrades. Would result in $14.5B additional collateral demands against the GIA's for a 2 notch downgrade ($13.3 for one notch, $10.5 if only by one agency). Prepayment risk of $4.6B - $5.4B on contracts subject to early termination but unlikely due to the forfeit of large economic benefits. Additional needs: $14.5B

Income from operations: $2.9B per quarter or $5.8B total which reduces need for capital.

AIG had cash from operations of $16.6B for 2008 H1

AIG has revolving credit lines, some with one year add-ons which total $9.2B


Adding that all up, AIG has already used or is expected to use about $23.4B of capital through Q3, including the potential of a ratings agency downgrade.This would leave them in the same position as the end of Q1. However, unlike Q1, the swap book has for all intents been kicked to the scrap heap. It -more-


Copyright 2004-2008 by Gedanken Experiment (previously Rant Street) except as otherwise noted. Our terms of use are as follows: Content on Gedanken Experiment may not be indexed, cached, reproduced or syndicated in any manner by any party whose purpose is to provide such an index, cache, reproduction or syndication of our content for a monetary fee or other consideration to their clients, customers or users. Content on Gedanken Experiment may not be copied, reproduced, republished, indexed, cached, uploaded, posted , transmitted, framed or distributed in any way, without the prior written permission of Gedanken Experiment, except that a) user may download, display, or print one copy of the materials on any single computer solely for user's use; b) user may briefly quote or excerpt for use in a review or criticism for purposes of illustration or comment or as part of a news report as per fair use guidelines of www.copyright.gov; c) internet search engines which provide the public the ability to search online content at no charge (free) may index and cache content from Gedanken Experiment which is not explicitly blocked by robots.txt; in case (a), (b) and (c) above user agrees to keep intact all copyright, trademark, and other proprietary notices of both Gedanken Experiment and any other third parties mentioned in our content. Modification of the materials or use of the materials for any other purpose is a violation of Gedanken Experiment's, its affiliates', or its third-party information providers' copyrights and other proprietary rights. Nothing contained herein shall be construed as conferring by implication, estoppel, or otherwise, any license or right under any copyright, patent, trademark, or other proprietary interest of Gedanken Experiment, its affiliates, or any third party, except as expressly set forth herein. 02-13-2007

Information provided by Gedanken Experiment, its predecessor Rant Street! and other sources on this Web site is believed to be accurate and reliable when placed on this site, but we cannot guarantee it is accurate or complete or current at all times. Information on this site is for informational purposes only and is not intended to provide financial, legal, accounting or tax advice and should not be relied upon in that regard.

Gedanken Experiment is not responsible in any manner for direct, indirect, special or consequential damages, however caused, arising out of your use of this Web site and/or any web browser, including any damages you may suffer if you transmit confidential or sensitive information to us or if we communicate such information to you at your request over the Internet.

Archives/Blogroll

RSS/Contact Info

XML: RSS Feed 
XML: Atom Feed 




All materials Copyright 2004-2008 Gedanken Experiment (previously Rant Sreet!) unless otherwise noted. All Rights Reserved. Detailed terms of use at bottom of page.

You may contact Gedanken Experiment by sending email to admin AT gedankenexperiment.dk