M2M

ABX

M2M

E&Y’s summary of Statement 157

http://www.aicpa.org/caq/download/WP_Measurements_of_FV_in_Illiquid_Markets.pdf

http://www.cii.org/UserFiles/file/resource%20center/correspondence/2008/CII%20Fair%20Value%20Paper%20(final)%20%20071108.pdf

http://norris.blogs.nytimes.com/2008/09/30/fair-value-follies-2/

What’s all this stuff worth?

http://business.timesonline.co.uk/tol/business/industry_sectors/banking_and_finance/article4425782.ece

http://www.washingtonpost.com/wp-dyn/content/article/2008/08/25/AR2008082502490.html

http://www.nytimes.com/2008/09/25/business/25value.html?_r=1&th&emc=th&oref=slogin (citi, 61 cents)

** http://www.citigroup.com/citi/fin/data/p081016a.pdf?ieNocache=129 (page 16 and 17, citi Q3 end; abcp subprime 57 cents, alt A 63-67 cents)

BoA

http://library.corporate-ir.net/library/71/715/71595/items/309870/BAC3Q08Presentation-Final.pdf (page 16)

Citi

Q3

We started the quarter with total sub-prime exposure of $22.5 billion, as shown at the bottom of the first column. There have been several changes in the methodology for valuing our Super Senior exposures that are worth mentioning. The discounted cash flow methodology remains generally consistent with prior quarters and was described in detail in the first quarter earnings call. As we have previously said, the model is continually subject to refinements and enhancements.

The home price appreciation assumption that we are using in our valuation methodology for this quarter has changed from the last quarter and now reflects a cumulative price decline from peak-to-trough of 32%. The assumption reflects price declines of 16% and 10% respectively for 2008 and 2009 with the remainder of the 32% decline having occurred before the end of 2007.

The projected 32% decline peak-to-trough is based both on home affordability as well as other factors, such as the large overhang of homes in foreclosure, which has further depressed prices.

We have also changed the index on which we base our home price appreciation projections from the S&P Case-Shiller Index to the loan performance index. We made this change because the loan performance provides more comprehensive data, although the loan performance and the S&P Case-Shiller national HPIs have tracked each other closely in the past.

In addition, we have updated our mortgage default model to incorporate mortgage performance data from the first half of 2008, a period of sharp home price declines and high levels of mortgage foreclosures.

As I described on the last earnings call, our valuation methodology uses a discount margin that is calibrated to the underlying instruments, such as the ABS indices and other verified cash bond marks. To determine the discount margin, we apply our mortgage default rate to the bonds underlying the ABS indices and other referenced cash bonds and solve for the discount margin that produces the market prices of those instruments.

Using this methodology, the impact of the decrease of the home price appreciation projection, from a negative 23% to a negative 32%, results in a decrease in discount margins. Taken together, these two factors directionally offset one another.

We also changed the way we value the high-grade and mezzanine positions for the quarter, from model valuation to trader prices based on the underlying assets of each high-grade and mezzanine ABS CDO. Unlike the ABCP and CDO-squared positions, the high-grade and mezzanine positions are now largely hedged through ABX and bond short positions, which of necessity are trader priced.

We believe it makes sense for there to be symmetry in the way our long positions and our short positions are valued.

Additionally, there were a number of liquidations of high-grade and mezzanine positions during the third quarter and these were at prices close to the value of trader prices. The liquidation proceeds, in total, were also above the June 30 carrying amount of the positions liquidated, contributing a substantial portion of the profit for the high-grade and mezzanine positions in the third quarter.

We intend to use trader prices to value this portion of the portfolio going forward, so long as it remains largely hedged.

With respect to monolines, the credit value adjustment for the quarter was $919.0 million, as is shown on the bottom of the slide. At quarter end, our credit value adjustment balance was $4.6 billion and the market value direct exposure to clients declined to $6.6 billion to $8.0 billion in the second quarter.

I will turn now to Slide 16. Slide 16 provides vintage and rating data for each of the exposure types. We showed you this table last quarter. Let me highlight just a couple of changes. First, in the ABCP category the percentage of AAA to AA has declined from 71% last quarter to 62% this quarter, primarily driven by downgrades of the 2005 vintages.

Second, in the high-grade category, the percentage of 2006 vintages in the portfolio substantially declined. This was due primarily to liquidations of the positions during the quarter. As a result, the overall exposure had declined as well but the mix has shifted towards the higher concentration of 2004 and 2005 vintages. All of this has resulted in a change in mark from 27% last quarter to 41% this quarter.

Finally, there are no material changes in the mezzanine exposure vintage and rating mix.

I will turn now to Slide 17. It provides more disclosure on our Alt-A exposure, which stands at $13.6 billion at the end of this quarter. Of that $13.6 billion $10.2 billion is held as available-for-sale securities in which we recorded a $580.0 million impairment charge in the quarter. The remaining $3.4 billion is held in a mark to market portfolio where the marks for the quarter were $573.0 million, net of hedges.

Moving to the box at the bottom of the page, we show you the market value of the portfolio relative to the face value and also provide vintage and ratings information.

A few points of importance. First, in the AFS portfolio approximately 1/3 of the portfolio is 2005 and earlier vintages and ¾ is rated AA to AAA. That portfolio is marked at $0.67 on the dollar.

The trading portfolio is comprised of 11% 2005 and earlier vintages. 69% is rated AA to AAA. That portfolio is marked at $0.63 on the dollar. The trading portfolio marks that I have just mentioned do not include any residual or IO positions.

Including these positions, the combined marks of the trading Alt-A portfolio would have been marked at approximately 15%, or $0.15 on the dollar.

Glenn Schorr - UBS

Could you help us with what some of the larger components of the unrealized loss position on the balance sheet. And such a big change this quarter, maybe just point us in the larger buckets.

Gary L. Crittenden

I would be happy to. There was a change, obviously, in the quarter that took place in OCI. If you go to the supplement, under the balance sheet section of the supplement, you can get a good sense for what that looks like. So quarter-over-quarter it went up about $6.0 billion. So if you look at the split of that, it was in two separate categories. The two separate categories are the AFS book and then the marks that we take as a result of deteriorating currencies in countries outside the U.S. where we have significant investments.

For example, if we have an investment in Brazil, that investment is held in local currency terms, we hedge that position, obviously. When those currencies deteriorate, that has a negative impact on our OCI calculation. And that accounts for roughly half of the value.

The other half, as I said, was related primarily to spread widening in AFS securities. And the single largest component of that was in the Alt-A category and I took you through some detail on the Alt-A category in the earlier presentation.

Glenn Schorr - UBS

And I would assume that both go toward standard tests over time for a permanent impairment and that will just probably not play out over time.

Gary L. Crittenden

We have a very comprehensive process that we go through that reviews this each quarter. What we do is we look at all our positions in the book, we compare those positions against where they are currently trading. If there was a trading analog to these securities. We make estimates about the realizability of the position that we are carrying. Unless we are absolutely convinced that for some reason we can hold these securities to value and recognize the amount that we are carrying them for, we then take an impairment as we need to.

And in this quarter you saw, I think, something over $500.0 million in impairments that we took against AFS securities for exactly that purpose.

So for those, obviously, have an other than temporary impairment are recognized as appropriate. And in the case of the FX impact, or the $3.0 million or so that was related, those kind of impairments would only be realized if we actually sold our position in a particular market.

Q3 10Q

Subprime-Related Direct Exposures in CDOs
The Company accounts for its CDO super senior subprime
direct exposures and the underlying securities on a fair-value
basis with all changes in fair value recorded in earnings.
Citigroup’s CDO super senior subprime direct exposures are
not subject to valuation based on observable transactions.
Accordingly, the fair value of these exposures is based on
management’s best estimates based on facts and
circumstances as of the date of these consolidated financial
statements.
Citigroup’s CDO super senior subprime direct exposures
are Level 3 assets and are subject to valuation based on
significant unobservable inputs. Fair value of these
exposures (other than high grade and mezzanine as described
below) is based on estimates of future cash flows from the
CITIGROUP – 2008 THIRD QUARTER 10-Q
127
mortgage loans underlying the assets of the ABS CDOs. To
determine the performance of the underlying mortgage loan
portfolios, the Company estimates the prepayments, defaults
and loss severities based on a number of macroeconomic
factors, including housing price changes, unemployment
rates, interest rates, and borrower and loan attributes, such as
age, credit scores, documentation status, loan-to-value (LTV)
ratios, and debt-to-income (DTI) ratios. The model is
calibrated using available mortgage loan information
including historical loan performance. In addition, the
methodology estimates the impact of geographic
concentration of mortgages, and the impact of reported fraud
in the origination of subprime mortgages. An appropriate
discount rate is then applied to the cash flows generated for
each ABCP and CDO-squared tranche, in order to estimate its
current fair value.
When necessary, the valuation methodology used by
Citigroup is refined and the inputs used for the purposes of
estimation are modified, in part, to reflect ongoing market
developments. More specifically, the inputs of home price
appreciation (HPA) assumptions and delinquency data were
updated during the quarter along with discount rates that are
based upon a weighted average combination of implied
spreads from single name ABS bond prices and ABX indices,
as well as CLO spreads.
As was the case in the second quarter of 2008, the third
quarter housing-price changes were estimated using a
forward-looking projection. However, for third quarter 2008,
this projection incorporates the Loan Performance Index,
whereas in second quarter 2008, it incorporated the S&P Case
Shiller Index. This change was made because the Loan
Performance Index provided more comprehensive geographic
data. In addition, the Company’s mortgage default model has
been updated for mortgage performance data from the first
half of 2008, a period of sharp home price declines and high
levels of mortgage foreclosures.
The valuation as of September 30, 2008 assumes a
cumulative decline in U.S. housing prices from peak to
trough of 32%. This rate assumes declines of 16% and 10%
in 2008 and 2009, respectively, the remainder of the 32%
decline having already occurred before the end of 2007. The
valuation methodology as of June 30, 2008 assumed a
cumulative decline in U.S. housing prices from peak to
trough of 23%, with assumed declines of 12% and 3% in
2008 and 2009, respectively.
In addition, during the second and third quarters of 2008,
the discount rates were based on a weighted average
combination of the implied spreads from single name ABS
bond prices, ABX indices and CLO spreads, depending on
vintage and asset types. To determine the discount margin,
the Company applies the mortgage default model to the
bonds underlying the ABX indices and other referenced cash
bonds and solves for the discount margin that produces the
market prices of those instruments. Using this methodology,
the impact of the decrease of the home price appreciation
projection from -23% to -32% resulted in a decrease in the
discount margins incorporated in the valuation model.
For the third quarter of 2008, the valuation of the high
grade and mezzanine ABS CDO positions was changed from
model valuation to trader prices based on the underlying
assets of each high grade and mezzanine ABS CDO. Unlike
the ABCP and CDO-squared positions, the high grade and
mezzanine positions are now largely hedged through the
ABX and bond short positions, which are by necessity, trader
priced. Thus, this change brings closer symmetry in the way
these long and short positions are valued by the Company.
Citigroup intends to use trader marks to value this portion of
the portfolio going forward so long as it remains largely
hedged.
The primary drivers that currently impact the super
senior valuations are the discount rates used to calculate the
present value of projected cash flows and projected mortgage
loan performance.
Given the above, the Company’s CDO super senior
subprime direct exposures were classified in Level 3 of the
fair-value hierarchy.
For most of the lending and structuring direct subprime
exposures (excluding super seniors), fair value is determined
utilizing observable transactions where available, other
market data for similar assets in markets that are not active
and other internal valuation techniques.

Alt-A Mortgage Securities
The Company reports Alt-A mortgage securities in Trading
account assets and available-for-sale Investments. In both
cases the securities are recorded at fair value with changes in
fair value reported in current earnings and OCI, respectively.
For these purposes, Alt-A mortgage securities are non-agency
residential mortgage-backed securities (RMBS) where: (1)
the underlying collateral has weighted average FICO scores
between 680 and 720 or, (2) for instances where FICO scores
are greater than 720, RMBS have 30% or less of the
underlying collateral composed of full documentation loans.
Similar to the valuation methodologies used for other
trading securities and trading loans, the Company generally
determines the fair value of Alt-A mortgage securities
utilizing internal valuation techniques. Fair value estimates
from internal valuation techniques are verified, where
possible, to prices obtained from independent vendors.
Vendors compile prices from various sources. Where
available, the Company may also make use of quoted prices
for recent trading activity in securities with the same or
similar characteristics to that being valued.
The internal valuation techniques used for Alt-A
mortgage securities, as with other mortgage exposures,
consider estimated housing price changes, unemployment
rates, interest rates, and borrower attributes. They also
consider prepayment rates as well as other market indicators.
Alt-A mortgage securities that are valued using these
methods are generally classified as Level 2. However, Alt-A
mortgage securities backed by Alt-A mortgages of lower
quality or more recent vintages are mostly classified in Level
3 due to the reduced liquidity that exists for such positions,
which reduces the reliability of prices available from
independent sources.
Commercial Real Estate Exposure
Citigroup reports a number of different exposures linked to
commercial real estate at fair value with changes in fair value
reported in earnings, including securities, loans and
investments in entities that hold commercial real estate loans
or commercial real estate directly. The Company also reports
securities backed by commercial real estate as available-forsale
investments, which are carried at fair value with changes
in fair value reported in OCI.
Similar to the valuation methodologies used for other
trading securities and trading loans, the Company generally
determines the fair value of securities and loans linked to
commercial real estate utilizing internal valuation techniques.
Fair value estimates from internal valuation techniques are
verified, where possible, to prices obtained from independent
vendors. Vendors compile prices from various sources.
Where available, the Company may also make use of quoted
prices for recent trading activity in securities or loans with the
same or similar characteristics to that being valued. Securities
and loans linked to commercial real estate valued using these
methodologies are generally classified as Level 3 as a result
of the reduced liquidity currently in the market for such
exposures.
The fair value of investments in entities that hold
commercial real estate loans or commercial real estate
directly is determined using a similar methodology to that
used for other non-public investments in real estate held by
S&B business. The Company uses an established process for
determining the fair value of such securities, using commonly
accepted valuation techniques, including the use of earnings
multiples based on comparable public securities, industry
specific non-earnings-based multiples and discounted cash
flow models. In determining the fair value of such
investments, the Company also considers events such as a
proposed sale of the investee company, initial public
offerings, equity issuances, or other observable transactions.
Such investments are generally classified in Level 3 of the
fair value hierarchy.

Q1

As in prior periods we used a proprietary model to calculate vectors for conditional payment, default rates and loss severity. A key input for this model is projected home price appreciation or HPA. We used the outputs to calculate the projected cash flows for the collateral assets which we then run through the CDO distribution waterfall for each transaction. Finally we discount the cash flow by a discount margin that reflects factors including a liquidity discount and the uncertainty associated with structured investments. The methodology that we used has been refined and the inputs have been modified to reflect current conditions.

The two principle refinements and modifications this quarter are the use of a more direct method of calculating projected HPA and a more refined method for calculating the discount rate. First, in the last quarter the projected HPA was based on a series of factors including projected national HPA and differences between subprime and other sectors of the mortgage market. This quarter the HPA is based on a forward looking projection of the S&P Case-Shiller Home Price Index that embodies greater subprime representation. While this change allows a more direct projection of HPA without requiring additional separate adjustments for subprime, it has a relatively small impact on our valuations.

The HPA used in our valuation methodology this quarter reflects a cumulative price decline from peak to trough of 20%, 9% which we saw through the end of 2007. Our assumptions reflect a remaining price decline of 8% and 3% respectively for 2008 and 2009. As in the past, an adjustment was made for the geographic concentration of the relevant mortgage pools. Second our methodology for calculating the discount rate was refined this quarter. In the past year we used observable CLO spreads and applied a liquidity discount to those spreads to arrive at our discount rate.

This quarter we have used a weighted average combination of the implied spreads from single named ABS bond prices and ABX indices and CLO spreads depending on vintage and asset types. This refinement was made in part in response to the combination of continuing rating agency downgrades of RMBS and ABS CDOs and a lack of CLO spreads at the resulting rating levels. Many analysts had been using the ABX as a rough proxy for the value of super seniors, while we have considered the ABX, we believe that care has to be used in applying it to determine the value of super seniors for several reasons, one of which is that it does not contain earlier vintages that make up a substantial portion of our portfolio.

William Tanona – Goldman Sachs

Good morning guys I just got one question. On the alt A portfolio, obviously you guys have $22 billion there and looking at the write down of $1 billion, when you compare that to kind of other industry participants who have taken things down to $0.70 on the dollar, it just seems like it’s a little light.

I don’t know whether or not we’re just looking at that incorrectly of if there are hedges involved or is there something that we should be looking at differently or comparing that $1 billion to the $6 billion in trading as opposed to the available for sale. Just trying to get a sense as to why that write down was so light relative to a lot of your peers.

Gary Crittenden

It’s the last point that you made. So it’s the split between, if you look on the chart, we split it out into available for sale versus trading. Obviously we do have some hedged positions against the trading book. But it’s the available for sale versus the trading. So we took the $1 billion in write down against the trading book essentially. I mean $120 million of the write down was associated with the 15.4, the large, the top right hand side of slide number 19. $120 million of the write down was associated with the 15.4; the remainder was associated with the 6.7.

William Tanona – Goldman Sachs

Okay, thank you.

Freddic Mac as a window of the impact of m2m on financial institutions.

Freddie has held-to-maturity, available-for-sale and trading securities. Trading securities and available-for-sale securities are m2m. Held-to-maturity and available-for-sale securities are tested for impairment. In 2008 Q2, Freddie recorded a 1.0 bn impairment to its available-for-sale securities, mostly mortgage-related ABS (MBS etc). This provides insight over what FMV has been for the impaired assets. However, one must note that other assets are likely marked down less when m2m because impaired assets are likely those that have incurred huge losses.

Freddie Mac:

http://www.economist.com/finance/displaystory.cfm?story_id=11751139

Both companies make a distinction between losses on trading assets (which they take as a hit against profits) and on “available-for-sale” securities which they hold for the longer term and disregard, if they think the losses are temporary. At the end of 2007, according to OFHEO, Fannie had pre-tax losses of this type of $4.8 billion; Freddie’s amounted to $15 billion.

Q1 call

Paul Miller – FBR Capital Markets

There’s a headline out there talking about Freddie Mac level three assets of $157 billion and I don’t see that in any of your releases. I was just wondering, is that true and where did that number come from? And is that related at all to the markup of the trading securities on page 4 of the $1.2 billion gain?

Buddy Piszel

No it’s not. We made a determination in the first quarter that given how wide, the pricing we were getting on the ABS portfolio, that it no longer made sense to leave that into level two. And so we essentially moved the entire ABS portfolio into level three. We are still using the mean price that we’re getting from the pricing services and the dealers, so we’re not using a model price, but that’s what all of that is. It has nothing to do with the trading portfolio.

Q1 Earnings Release Conference Call: No it’s not. We made a determination in the first quarter that given how wide, the pricing we were getting on the ABS portfolio, that it no longer made sense to leave that into level two. And so we essentially moved the entire ABS portfolio into level three. We are still using the mean price that we’re getting from the pricing services and the dealers, so we’re not using a model price, but that’s what all of that is. It has nothing to do with the trading portfolio.

Q4 2007

Patty Cook

Next, fair value, fair value declined significantly during the fourth quarter, about $17 billion after tax bringing the year to date decline to approximately $25 billion after tax. This markdown is split about evenly between the guaranty business and the investment business. I will address both. While the credit quality of our retained portfolio remains very high with 57% in agency [curities] and 33% in non agency securities of which 96% is currently triple A rated, spreads widened dramatically. This resulted in a pretax mark to market loss of $14 billion during the quarter. Almost $10 billion of which was associated with our ABS portfolio which sustained a 200 basis point widening of spreads.

During 2007, the ABS portfolio widened from a spread of 28 basis points to 335 basis points. However, given the structure of these securities and the resulting subordination, we do not believe any significant realized losses are likely and therefore have not recorded any impairments. Similarly, for the agency and [CMBF] components of the portfolio, we do not expect to realize these OAF losses. Since we do not expect to realize any meaningful losses, the reversal of these marks will over time emerge as higher course spread income and fair value. During 2007, the overall spread on the retained portfolio therefore increased from 27 basis points to 104.

In addition, the wider spreads are also creating expanded purchase opportunities in the agency mortgage securities market. Let me just remind you that all of these prices that we use for the portfolio are reflective of third party marks obtained from pricing services on the street. I want to address the ABS portfolio in a little more detail. The non agency ABSs were critical to our effort to meet our affordable lending objectives and allowed us to invest in non prime markets with substantial credit enhancements.

Despite the continued deterioration of the housing market and increases in non prime delinquencies, we remained comfortable with our risk position on these assets. In order to provide a better understanding of how we come to this view, we have posted an extensive write up of the ABS portfolio on the website with this release. Remember, all of the bonds held in this portfolio are plain pass through securities and contain no CDO exposure. A couple of quick points, look at slide 8, the circled items show the results of three stress test scenarios we run on the subprime portfolio.

The results show that a 50% default 50% severity experience, our projected losses on that position are less than $1 million. We even included two more onerous scenarios, not because we think they are likely but to show that even in these extreme cases, anticipated losses are far below the severe market estimates we have seen. Let’s move to the decline in fair value of the guaranty business. The GC business was affected by declining house prices which resulted in the current loan to value ratio of our portfolio increasing as delinquency rates increased and mortgage related credit spreads widened further. The combination of these factors led to an expectation of higher expected default costs and significantly higher risk premiums.

The result is a decline in the fair value of the GC business of about $9 billion for the quarter and $13 billion for the year after tax. About half of the increase is a function of higher expected default costs. The other is the result of market illiquidity and higher risk premium which we capture in our third party mark to market process. Said another way, if we were relying solely on our own models to price our GO, our mark to market change for the year would have been about half as much.

Remember in the fourth quarter of 2005, we moved to third party pricing for this portfolio and as a result, are picking up the risk premiums in the same manner as we do in the retained portfolio. If our expectations for default costs are close to what we ultimately realize, the portion of the mark to market loss associated with wider spreads will come back to fair value. This volatility in fair value raises the question of its usefulness as an earnings measure.

We believe fair value is a useful measure for evaluating the business over longer time periods. It is not as useful over short time periods when spread volatility can dwarf the economic spread being earned in both businesses. This is particularly true for Freddie Mac as we have both the intent and capability to hold to maturity. As a result, we believe the additional non GAAP measure of adjusted operating income will help in evaluating the business.

Q3 2007

Robert Lacoursiere - Banc Of America Securities.

Yeah. I just wondered if you could just help me understand: how you come up with the valuations for the LIA loans? There is obviously not a market. So, what are you using like a level to approach and where do you get the inputs from?

Patti Cook

No, actually we do get LIA mark from the street. And you’re right and part of that is probably reflected in the price. It is in a liquid market. It certainly not trading as many of the other securities that we are on trade but we do go to the street for any independent mark.

Robert Lacoursiere - Banc of America Securities

But you never actually trade? You don’t actually sell these things? Are you just asking them to quote your figure, right?

Patti Cook

Right. But you are on to a good point Robert, because I think that same thing is in evidence when we priced the overall GO. If you think about it our mortgages that we’re guarantying in our GSE business really don’t trade in securitize in a AAA senior sub sort of structure. So, we’ve been there, the fact that we take that structure, we go to the market, we ask for a price on something that really doesn’t trade. I think supports the notion that the uncertainty and the credit risk premium that’s embedded in those marks is likely to overshoot in a cautious credit environment.

Robert Lacoursiere - Banc of America Securities

That is precise, I am just wondering why you have to rely on those quotes when there is no market, why can’t you do a Level III approach and put your own assumptions like other institutions do?

Patti Cook

This is — you want to take it Buddy?

Buddy Piszel

We have been there in the past and the feedback that we’ve received from our auditors is the market is a more reliable source of pricing and accordingly even in 10 markets, that should be our first line of defense whether it contradicts or models or not. Since we’ve made this change, we’ve only look to the market wherever possible, and we’ve always been able to get market prices. So, whether we like them or not, they are out there and if they are out we are using them to measure because the accounting literally says that you’re supposed to use that what some one would pay you to take the obligation off your hands and if there is a market price out there, that is an indicator of the price that you would have to pay to have it taken off your hands.

Robert Lacoursiere - Banc of America Securities

If I could just bother you with one follow-up: if they are quoting, if these institutions are quoting you those prices, are they not obligated to reflect those valuations on their own positions on their own books?

Buddy Piszel

That’s an interesting question, but we are not going go there.

Robert Lacoursiere - Banc of America Securities

Thank you.