PrefBlog

Fitch Ratings has released a Special Report: Subprime Mortgage-Related Losses – A Moving Target which endorses the relatively high loss estimates of Greenlaw et al. and the IMF and contradicts the Bank of England Estimate.

Fitch, bless ’em, explicitly states that their calculation is restricted to the USD 1.4-trillion-odd of securitized subprime mortgage assets – one problem in drawing up comparable figures is determining whether the asset universes are comparable. Europe also went nuts over real-estate, particularly Spain and the UK!

Given the size of the subprime market, estimated to have originated as much as USD1.4trn of loans in the last three years (2005: USD625bn; 2006: USD600bn; and 2007: USD179bn), the poor underwriting standards deployed in originating these loans and the deteriorating economic environment, Fitch estimates total losses for the market to be in the region of USD400bn. Alternative methods of calculating the potential losses using index prices suggest potentially higher losses up to a high of USD550bn.

This compares with Greenlaw et al. (USD 400-billion); the IMF (USD 565-billion) [“Aggregate losses are on the order of $565 billion for U.S. residential loans (nonprime and prime) and securities and $240 billion on commercial real estate securities.”]; and the BoE (USD 170-billion by credit analysis vs. USD 381-billion by market price vs USD USD 317-billion by model-estimated credit-component of market price).

Of great interest to investors will be Fitch’s related conclusion that, notwithstanding that their estimate is on the low side of their mark-to-market range:

• Approximately 50% of total subprime mortgage related losses, totalling USD200bn, are estimated to reside within the banking sector. The balance 50% of losses is distributed among financial guarantors, insurance companies, asset managers and hedge funds. To the extent that institutions have effectively
  hedged their exposures with financially sound counterparties, these loss estimates may be over‐estimated. However, in the absence of detailed disclosures, it is difficult to get an accurate estimate of net losses.

• Reported losses by banks at USD165.3bn indicate that over 80% of losses stemming from ABS‐CDO and subprime RMBS exposures have been disclosed.
• As a significant proportion of the losses have been disclosed, further ratings action arising from ABS‐CDO and subprime RMBS exposures is likely to be minimal.

It’s a good paper, with a fair amount of detail provided regarding how they calculated their numbers.

Update: As noted on March 11, Fitch is very proud of how tough they are on subprime:

The full Bloomberg story explains the Fitch discrepency a little better:

“We have built in 20 percent more home price declines from the end of ’07,” said Glenn Costello, managing director for residential mortgage-backed securities at Fitch. “When you build in that much home price decline, I feel good when I pick up the paper and I see that home prices are only down another 3 percent. My ratings are still good.”

This won’t be much of a review, but I have come across a rather provocatively abstracted paper: Understanding the Subprime Mortgage Crisis, by Yuliya Demyanyk and Otto van Hemert, both of the Federal Reserve Board, dated February 29, 2008:

[abstract] Using loan-level data, we analyze the quality of subprime mortgage loans by adjusting their performance for differences in borrower characteristics, loan characteristics, and house price appreciation since origination. We find that the quality of loans deteriorated for six consecutive years before the crisis and that securitizers were, to some extent, aware of it. We provide evidence that the rise and fall of the subprime mortgage market follows a classic lending boom-bust scenario, in which unsustainable growth leads to the collapse of the market. Problems could have been detected long before the crisis, but they were masked by high house price appreciation between 2003 and 2005.

[Extract from conclusion] The decline in loan quality has been monotonic, but not equally spread among different types of borrowers. Over time, high-LTV borrowers became increasingly risky (their adjusted performance worsened more) compared to low-LTV borrowers. Securitizers seem to have been aware of this particular pattern in the relative riskiness of borrowers: We show that over time mortgage rates became more sensitive to the LTV ratio of borrowers. In 2001, for example, the premium paid by a high LTV borrower was close to zero. In contrast, in 2006 a borrower with a one standard deviation above-average LTV ratio paid a 30 basis point premium compared to an average LTV borrower.

…

In many respects, the subprime market experienced a classic lending boom bust scenario with rapid market growth, loosening underwriting standards, deteriorating loan performance, and decreasing risk premiums. Argentina in 1980, Chile in 1982, Sweden, Norway, and Finland in 1992, Mexico in 1994, Thailand, Indonesia, and Korea in 1997 all experienced the culmination of a boom-bust scenario, albeit in different economic settings.

Were problems in the subprime mortgage market apparent before the actual crisis showed signs in 2007? Our answer is yes, at least by the end of 2005. Using the data available only at the end of 2005, we show that the monotonic degradation of the subprime market was already apparent. Loan quality had been worsening for five consecutive years at that point. Rapid appreciation in housing prices masked the deterioration in the subprime mortgage market and thus the true riskiness of subprime mortgage loans. When housing prices stopped climbing, the risk in the market became apparent.

 

In the comments to my post Is the US Banking System Really Insolvent? Prof. Menzie Chin brought to my attention a wonderful paper: Leveraged Losses: Lessons from the Mortgage Market Meltdown.

This paper has also been highlighted on Econbrowser under the title Tabulating the Credit Crunch’s Effects: One Educated Guess.

The source document is in several parts – to do justice to it, I will be be posting reviews of each section.

The previous post in this series Economic Effects of Subprime, Part I: Loss Estimates, I had a look at the authors’ methodology of estimating loss. In this post, I’ll review their Section 3.4: Allocating the Losses.

Section 3.4’s main contribution to the the debate is “Exhibit 3.7: Home Mortgage Exposures of US Leveraged Institutions”, which uses unspecified Federal Reserve data to estimate that roughly 50% of all subprime exposure is held by US-based “Leveraged Institutions” – a defined term that includes Commercial Banks, Savings Institutions, Credit Unions, Brokers & Dealers, and the GSEs.

If we assume that the first three of those categories comprise all FDIC-insured institutions, then the numbers add up for RMBS exposure, more or less, anyway. The FDIC Quarterly Report on US Banks for 4Q07 has been previously discussed; the figure shown in Table II-A for “Mortgage-backed securities” is slightly over 1,236-billion, which is fairly close to the sum of the relevant categories in Exhibit 3.7 which is being examined.

So that part’s OK, but the purpose of the exercise is to determine the sub-prime exposure, not the total exposure; although there may well be losses on non-subprime paper, I think it’s pretty much agreed that these losses will be much lower, as a proportion of principal, than the losses on prime paper.

When we look at, for instance, Citigroup’s data on directly held mortgages (page 11 of the PDF), we find that the overwhelming majority of mortgages directly held are prime. Citigroup’s provides a vintage analysis of their $37.3-billion “Sub-prime Related Direct Exposures in Securities and Banking” on Schedule B of their Quarterly press release, but include the unfortunate caveat that:

Securities and banking also has trading positions, both long and short, in U.S. sub-prime residential mortgage-backed securities (RMBS) and related products, including ABS CDOs, that are not included in these figures. The exposure from these positions is actively managed and hedged, although the effectiveness of the hedging products used may vary with material changes in market condit

They are rather coy about the proportion of agency vs. non-agency RMBS held in their 2006 Annual Report, but state the total as comprising:

Mortgage-backed securities, principally obligations of U.S. Federal agencies

I don’t buy Exhibit 3.7 as evidence that US Leveraged institutions have exposure to half of the sub-prime losses. The quality of the banks’ (and bank-equivalents’, and GSE) exposure is going to be higher than average, tilted towards Agencies and AAA tranches of subprime; while “Brokers & Dealers” might – possibly – have a higher than average exposure to the mezzanine tranches, as might hedge funds, the focus is – or at least should be – on the banking system itself.

So where did it go? The Ashcraft paper, discussed in a dedicated post pointed out that the pension fund examined had all of its mortgage exposure in non-agency RMBS – I observed at that time that it was probably all AAA tranches at that. I note a Watson Wyatt press release stating:

January 30, 2007- Global institutional pension fund assets in the 11 major markets have more than doubled* during the past ten years and now total US$23,200 billion 

and another release from the same firm:

October 3, 2007 – Total assets managed by the world’s largest 500 fund managers grew by 19% in 2006 to US$63.7 trillion according to the Pensions & Investments / Watson Wyatt World 500 ranking.

I suggest that these pools of capital (one will be almost entirely included in the other, by the way!) will be a fertile hunting ground for sub-prime exposure.

Exhibit 3.8 of the paper purports to support an estimate of 50% of losses being borne by the US leveraged sector, but the source of this table is a Goldman Sachs report with no reported methodology. I will note that the table estimates exposure of $57-billion for “Mutual and Pension Funds”; using the Watson Wyatt estimate of $23,200-billion for pension funds alone, this would imply that the average pension fund (taken from the 11 major markets) has exposure of about 0.25% of assets. Given 6.5% exposure in the fund in Ashcraft’s paper, this estimate seems a little low.

In conclusion … the evidence presented that half the sub-prime losses will be borne by the US leveraged sector is unconvincing. It should also be noted that the “bottom-up” estimate of Goldman Sachs includes 17% of total exposure in US Hedge Funds to reach this 50% total. A loss is a loss is a loss, and hedge fund losses will have some effect on the overall economy, but it seems to me that the transmission of such an effect to the economy will be greatly muted relative to the effect of such losses by banks. Hedge funds can be wiped out without much affecting the price of eggs.

Update, 2008-3-12: The source document is admiringly quoted in a John Dizard piece in the Financial Times, republished by Naked Capitalism:

Since the estimates were drawn up more than 15 minutes ago, they’re already out of date, but they’re not a bad place to start. The group estimates that the losses on mortgage paper will ultimately total about $400bn, with about half of that being incurred by “leveraged US institutions”. They go on to estimate that new equity raised so far from investors such as the sovereign wealth funds is of the order of $100bn.
…
It does not, therefore, take much of a leap in imagination to suggest that the US banks need to raise well over $100bn in new Tier One capital, and perhaps more than $200bn. They also need to do it quickly, so as to avoid that spiralling destruction of capital.

In the comments to my post Is the US Banking System Really Insolvent? Prof. Menzie Chin brought to my attention a wonderful paper: Leveraged Losses: Lessons from the Mortgage Market Meltdown.

This paper has also been highlighted on Econbrowser under the title Tabulating the Credit Crunch’s Effects: One Educated Guess.

The source document is in several parts – to do justice to it, I will be be posting reviews of each section. In this post, I will examine Part 3: Estimating Mortgage Credit Losses.

The first method of estimation is described thus:

The mechanics of these estimates is best explained by focusing on the $243 billion baseline estimate produced by the global bank analysts at Goldman Sachs. Their model simply extrapolates the performance – defaults, loss severities, and total loss rates – of each “vintage” (origination year) of subprime and other mortgage loans, based on its own history as well as the typical progression pattern through time. For example, suppose that the cumulative default rate on the 2006 subprime vintage is 3% at the end of 2007. Suppose further that the 2004 vintage showed a cumulative default rate of 1% after 1 year and 4% after 3 years, i.e. a fourfold increase over 2 additional years. Their procedure is to use the data on the 2004 vintage to extrapolate the cumulative default rate on the 2006 vintage. In this scenario, the default rate on the 2006 vintage would be 12% by the year 2009.

This methodology is, of course, complete nonsense. While I am sure that it is possible to determine a factor that correlates time from origination with cumulative loss experience, it is totally unacceptable to consider this the sole factor. As the authors state in the introduction to section 3.2 regarding adjustments to this baseline forecast:

Although the modeling strategy described above seems quite logical, it does not account for the possibility of a structural break that might result from falling home prices. In particular, because the detailed mortgage performance data required to build these types of models are available only back to the mid-1990s, there are no observations on how defaults and losses on a particular vintage change through time when home prices start to fall.

It would seem much more logical to consider – at least! – a three-factor model, which would incorporate the effect of negative equity on default rates and some measure of income … in other words, the good old “asset coverage” and “income coverage” tests that will be so familar to PrefBlog’s readers. The authors do not do this, nor do they attempt to do this – they simply increase the subprime default rates by one third and assume that non-subprime [you can’t call this “prime”, because of the “jumbo” and “Alt-A” netherworlds] defaults rise to one-half of the historical peak to arrive at an estimate of $400-billion total losses.

While the authors admit that these assumptions are extremely arbitrary, I will go a bit further and say that they are so arbitrary that their inclusion detracts from the credibility of the paper.

The second method sets up a grid analyzing total sub-prime issuance of $1,402-billion into cells determined by tranche rating and vintage. Each cell is then multiplied by the price of the ABX contract corresponding to that cell to calculate a loss estimate. The authors present their data as Exhibits 3.2, 3.3 and 3.4; a highly abbreviated summary of the data is:

The authors note:

There are many caveats that come with these estimates. We know that trading is thin in the underlying loan pools. More importantly, the ABX prices probably include a risk premium that is necessary to induce investors to bear mortgage credit risk in the current mortgage credit crisis. It may therefore overstate the market’s true expectation of future losses, although the size of this overstatement is difficult to gauge. Nonetheless, it is interesting to us that the range of losses from this exercise is not too different from the one obtained using method one calculations.

I will go so far as to say that the risk premium “probably” included in ABX prices is probably dominant. Let’s have a look at some analysis that at least purports to be an analysis, rather than an academic exercise in applying the Efficient Market Hypothesis. From Accrued Interest‘s post S&P on the monolines: No problem. Why?:

I will note that the Fitch stress test of RMBS, noted in their recent report on insurance companies, allows for a 5% loss on AAA, 30% on AA and 100% on everything else. 

It is quite apparent that – regardless of the loss of information inherent in my presentation of highly compressed versions of the authors’ calculations – that there are huge differences between “price” and “value”, where price is defined by reference to the ABX indices and value is defined by S&P’s cumulative default projections.

I will not make an impassioned defense of S&P here, nor will I repeat the concerns about the ABX indices that I raised in response to Prof. James Hamiltion’s Econbrowser post “Mortgage Securitization“. I will, however, point out that it seems rather intellectually dishonest not to include any “bottom up” analyses that might, possibly, give rise to smaller numbers in this review article. If the authors don’t place any credence in the estimates – that’s fine, let them say so. But completely ignoring such estimates detracts further from the paper’s credibility.

The paper’s third method of estimation is much more robust. The authors examine prior experience in three states (California (1991-1997), Texas (1986-1989) and Massachussets (1990-1993)) that experienced sharp declines in housing prices and arrive at potential foreclosure rates in the current crisis by comparison:

Hence, we conclude from our analysis that a housing downturn that resembled the three regional busts, with a 10%-15% peak-to-trough home price fall, could triple the national foreclosure rate over the next few years. This would imply a rise from 0.4% in mid-2006 to 1.2% in 2008 or 2009. Once home prices recover, the foreclosure rate might gradually fall back toward 0.4%.

Unfortunately, the authors spoil a good start by making an aggressive assumption in the course of their calculation:

These assumptions imply cumulative “excess” foreclosures of 13.5% of the currently outstanding stock of mortgages over the next few years. [Note: The calculation is that the foreclosure rate exceeds its baseline level by an average of 0.48 percentage points per quarter for a 7-year period, which implies cumulative excess foreclosures of 13.5%.]

A seven year period? A severe nation-wide recession lasting seven years sounds more like a depression to me. Additionally, the authors arrive at this figure by tripling the mid-2006 foreclosure figure of 0.4%, when the tripling in the states’ data was achieved from base rates of about 0.2%.

To be fair, though, I will note the recent Mortgage Bankers’ Associate press release:

The delinquency rate does not include loans in the process of foreclosure.  The percentage of loans in the foreclosure process was 2.04 percent of all loans outstanding at the end of the fourth quarter, an increase of 35 basis points from the third quarter of 2007 and 85 basis points from one year ago.

The rate of loans entering the foreclosure process was 0.83 percent on a seasonally adjusted basis, five basis points higher than the previous quarter and up 29 basis points from one year ago.

The total delinquency rate is the highest in the MBA survey since 1985.  The rate of foreclosure starts and the percent of loans in the process of foreclosure are at the highest levels ever.

The increase in foreclosure starts was due to increases for both prime and subprime loans.  From the previous quarter, prime fixed rate loan foreclosure starts remained unchanged at 0.22 percent, but prime ARM foreclosure starts increased four basis points to 1.06 percent. Subprime fixed foreclosure starts increased 14 basis points to 1.52 percent and subprime ARM foreclosure starts increased 57 basis points to 5.29 percent. FHA foreclosure starts decreased 4 basis points to 0.91 percent and VA foreclosure starts remained unchanged at 0.39.

Their figure of 13.5% cumulative foreclosures is then applied to $11-trillion of total mortgage debt to arrive at foreclosure starts of $1.5-trillion. These foreclosure starts turn into actual repossessions at a rate of 55-60%, and the authors claim average loss severity of 50%, to arrive at total losses (over seven years, remember) of $400-billion.

Quite frankly, their phrasing of the justification of the 55-60% repossession strikes me as a little suspicious:

However, the percentage of all foreclosure starts that turn into repossessions – measured by the number of Real Estate Owned (REO) notices divided by the lagged number of Notices of Default (NoD) – has recently risen to over 50% according to Data Quick, Inc., a real estate information company.

It’s only recently risen to over 50%? What is it normally? What was it in the data that has been presented for the three states that give rise to the “tripling” statistic? There has been a lot of analysis to the effect that a lot of sub-prime mortgages didn’t even make their first payment – which is taken as a warning flag of intentional fraud. Has this been accounted for?

Quite frankly, there are too many unanswered questions here for me to take the loss estimates seriously. I will stress: I am not taking a position on what the actual level of subprime losses will be. I am, however, pleading desperately for a credible estimate.

I will note that Larry Summers thinks $400-billion total is optomistic.

Update, 2008-3-8: There is a laudatory article in the Economist:

The study begins by estimating the size of mortgage-related losses using three different methods.
…
Each method involves some heroic assumptions.
…
Strikingly, however, all three approaches yield similar results: that mortgage-credit losses are likely to be around $400 billion.

Not only are the assumptions heroic, they are not supported by evidence or argument. There is no indication in the paper that they are anything other than “plugs” … a factor used to ensure you get the result you want. It is therefore not terribly impressive that all three approaches yield similar results … whatever happened to critical thinking?

Update, 2008-3-23: Here are PIMCO’s views on the matter, as of January 2008:

In order to measure the distance to a resolution of the problem, we need to estimate the extent of the valuation losses in subprime loans and related products. There is no clear answer, but data from the International Monetary Fund (IMF) and the Organization for Economic Co-operation and Development (OECD) suggest subprime losses of $300 billion. Roughly speaking, this is equivalent to 15% (40% default rate times 40% loss rate) of the $2 trillion in outstanding subprime home mortgages including Alt-A loans. However, judging from losses announced recently by financial institutions, we believe that actual overall subprime losses come to nearly double this figure or approximately $500 billion. This is because in addition to losses on subprime loans themselves, there were also steep valuation losses on other securitized products that make up nearly half the total. Financial institutions have an exposure of about 40%, so we believe that their latent losses amount to $200-250 billion.

Let us now examine the disparity between these potential losses and the actual losses (including valuation losses) posted by financial institutions. Losses declared by the major banks as of the end of last year came to around $100 billion, roughly 40-50% of the estimated latent loss. (Figure 3)

Update, 2008-3-25: Goldman Sachs is estimating $460-billion to the “leveraged sector”:

Wall Street banks, brokerages and hedge funds may report $460 billion in credit losses from the collapse of the subprime mortgage market, or almost four times the amount already disclosed, according to Goldman Sachs Group Inc. Profits will continue to wane, other analysts said.

“There is light at the end of the tunnel, but it is still rather dim,” Goldman analysts including New York-based Andrew Tilton said in a note to investors today. They estimated that residential mortgage losses will account for half the total, and commercial mortgages as much as 20 percent.

Note that Goldman Sachs influenced the conclusions of the paper reviewed in this post, so this is not a fully independent estimate.

A post today in Econbrowswer (which in turn was tipped by Calculated Risk) alerted me to a new Fed study: Understanding the Securitization of Subprime Mortgage Credit, which on first examination looks excellent.

I will admit that I have not thoroughly read the paper – I dare say it probably also took the authors more than an afternoon to research and write it – but I have had a look at some of the things that matter to me.

The authors identify seven frictions involved in the securitization process; number six – and a suggestion for mitigation – is:

Frictions between the asset manager and investor: Principal-agent [2.1.6]

• The investor provides the funding for the MBS purchase but is typically not financially sophisticated enough to formulate an investment strategy, conduct due diligence on potential investments, and find the best price for trades. This service is provided by an asset manager (agent) who may not invest sufficient effort on behalf of the investor (principal).
• Resolution: investment mandates and the evaluation of manager performance relative to a peer group or benchmark

This is one of the five that they highlight as causing the subprime crisis; they note:

Friction #6: Existing investment mandates do not adequately distinguish between structured and corporate ratings. Asset managers had an incentive to reach for yield by purchasing structured debt issues with the same credit rating but higher coupons as corporate debt issues.

[footnote] The fact that the market demands a higher yield for similarly rated structured products than for straight corporate bonds ought to provide a clue to the potential of higher risk.

I’m a bit confused by their footnote. Yes, there is an incentive to reach for yield; but the fact that “the market demands a higher yield for similarly rated structured products” implies that there are some managers who do not reach for yield – for one reason or another. If there weren’t, then yields for instruments of the same rating and term would be identical.

It is very tempting to think of the markets as being homogeneous – the market says this and the market says that. While there is some reason to believe that “the market” is a good predictor – whether of investment returns or election results (except in the New Hampshire primary!) – one must always remember that the market is heterogeneous and some players are better than others.

I will also note an eighth friction that is not mentioned by the authors; this is index-inclusion friction. I do not believe that this was a factor in the subprime fiasco; as the authors note:

Note that the Lehman Aggregate Index has a weight of less than one percent on non-agency MBS.

However, this kind of thing is indeed an issue. Canadian bond indices, for instance, include the banks’ Innovative Tier 1 Capital – and these things simply aren’t bonds! However – they’re included in the index. So, to the extent that a manager exercises his discretion and does not include them in a bond portfolio, he is mis-matching his portfolio. I noted early last year that quality spreads between tiers of bank paper were awfully skinny … but there are still spreads!

If I’m matched against the index and do not hold Tier 1 paper, I’m giving up yield due to my fear and – in 99 years out of 100 – I will underperform the average idiot who sees the chance to buy Royal Bank paper at a spread to other Royal Bank paper and buys the Tier 1 crap to pick up and extra 15bp. This is something James Hamilton of Econbrowser continually – and with good reason – harps on:

But who would be so foolish to have invested hundreds of billions of dollars in extra risky assets with negative expected returns? The logical answer would appear to be– someone who did not understand that they were accepting this risk.

Which brings us to the highly interesting Table 34 in section 6.1 of the report, which shows that:

the share of non-agency MBS in the total fixed-income portfolio increased from 12% (245/2022) in 2005 to 34% (740/2179) in 2006. In other words, the pension fund almost tripled its exposure to non-agency MBS. Further, note that this increase in exposure to risky MBS was at the expense of exposure to MBS backed by full faith and credit of the United States government, or an agency or instrumentality thereof, which dropped from $489.6 million to $58.9 million.

I’ll note that I don’t understand this “full faith and credit” stuff. Agency MBS are not explicitly guaranteed by the US treasury – which, again, James Hamilton has harped on:

Frame and Scott (2007) report that U.S. depository institutions face a 4% capital-to-assets requirement for mortgages held outright but only a 1.6% requirement for AA-rated mortgage-backed securities, which seems to me to reflect the (in my opinion mistaken) assumption that cross-sectional heterogeneity is currently the principal source of risk for mortgage repayment. Perhaps it’s also awkward for the Fed to declare that agency debt is riskier than Treasury debt and yet treat the two as equivalent for so many purposes.

Technically, I suppose, the authors may justify their “full faith and credit” stance with an insistence that they are talking about the full faith and credit of the GSEs … but somehow, I have the feeling that IF Fannie and Freddie go bust and IF they are not bailed out by Congress and IF large losses are experienced by investors in this paper, then the I-told-ya-so crowd will be the first to cast aspersions at portfolio managers who paid treasury prices for agency paper.

The portfolio managers discussed by Ashcraft & Schuermann made a decision that non-agency MBS were better, on a risk-reward basis, than agency MBS. It is very easy to say that the fact that this has not turned out very well so far proves that the portfolio managers were naive – but this is the sort of assessment that active portfolio managers are called upon to make as a matter of routine. Rather than focussing on this particular instance where … er … things don’t seem to have turned out so well, it would be much better to examine the portfolio management performance over a long history of such decisions and determine the manager’s skill from these data.

In fact, the authors do look at managers, fees and performance:

In 2006, the fund’s assets were 100% managed by external investment managers. The fixed income group is comprised of eight asset managers who collectively have over $2.2 trillion in assets under management (AUM). They are (with AUM in parentheses):

• JPMorgan Investment Advisors, Inc. ($1.1 trillion, 2006)
• Lehman Brothers Asset Management ($225 billion, 2006)
• Bridgewater Associates ($165 billion, 2006)
• Loomis Sayles & Company, LP ($115 billion, 2006)
• MacKay Shields LLC ($40 billion, 2006)
• Prima Capital Advisors, LLC ($1.8 billion, 2006)
• Quadrant Real Estate Advisors LLC ($2.7 billion, 2006)
• Western Asset Management ($598 billion, 2007)

The 2005 performance audit of this fund suggested that investment managers in the core fixed income portfolio are compensated 16.3 basis points. The fund paid these investment managers approximately $1.304 million in 2006 in order to manage an $800 million portfolio of investment-grade fixed-income securities. While the 2006 financial statement reports that these managers out-performed the benchmark index by 26 basis points (= 459 – 433), this was accomplished in part through a significant reallocation of the portfolio from relatively safe to relatively risk non-agency mortgage-backed securities. One might note that after adjusting for the compensation of asset managers, this aggressive strategy netted the pension fund only 10 basis points of extra yield relative to the benchmark index, for about $2.1 million.

Look at all those name-brand asset managers with 12-figure AUMs! One wonders what the long-term performance of these managers is; and particularly, how this performance was achieved. I should also point out that I know for a fact that working for a name-brand company is not particularly well correlated with investment management skill – quite the opposite, in my experience.

Update: Oh, and another thing! The researchers gleefully imply the portfolio managers in question are dumb yumps who ignorantly reached for yield, but do not provide a lot of details. According to Table 34, all – every single dollar’s worth – of the non-agency MBS was rated A- or better; no further details are given. Given the nature of the market, the nature of the managers and the nature of the investor, I will bet a nickel that the lion’s share (if not all) of this paper is comprised of AAA tranches.

These have taken a price thumping in the past year, but credit quality – as far as I can tell – of the AAA tranches remains pretty good. Maybe not AAA, but not junk, either. It will be most interesting to learn what the ultimate return on this paper is, if held to maturity.

I’ll tell you a little story, for instance. I’m involved with a real-money account that holds about $750,000 worth of paper issued by a subsidiary of, and ultimately guaranteed by, a major US-based financial firm. I tried to sell it … couldn’t get a bid.

Couldn’t. Even. Get. A. Damned. Bid.

So … for now, anyway, the client’s going to hold it. It’s still good quality paper. More liquid paper with the same guarantor trades around 6%.

What price should this paper be marked at? Am I an idiot for recommending its purchase last year? If it matures at par, do I then become a genius?

Update, 2008-01-12: Out of the kindness of my heart, I’m going to suggest another topic for a Master’s thesis. Here’s what I want you to do … get bond prices for an enormous variety of corporates (the index prepararers would be a good place to get these data) and slice it up into tranches based on yield and term. So now you’ve got a variety of market-based yield groups … say number 1 is “4.5-5.5 year term, 5.00% to 5.25% yield” and so on.

Now what I want you to do is follow each tranche to maturity and determine its total return over the period.

The questions are: (i) Do ex-ante yields predict ex-post returns?

(ii) Do these results tie in with other work that seeks to analyze bond yields in terms of risk-free rate, liquidity and default risk? The Bank of England has published some of the results of such work, but (a) they’re more interested in junk credits, and (b) I haven’t seen their source data, or (I am ashamed to admit) thoroughly read their source documents.

(iii) How risky is liquidity risk? This is the interesting part. I contend that most investors, particularly pension funds, are way more liquid than they have to be, in large part due to the way in which actuarial work is done, with all the liabilities being discounted at the long bond yield. To move the critical point to an extreme, consider a bond with a thirty-year term that cannot be sold (don’t laugh! The Canada Pension Plan used to be invested in these things!). Isn’t such a bond simply an equity with a poor return?

(iv) If investment mandates call for a given proportion of bonds, should there be elaboration of the liquidity requirement? Let us make some assumptions about the portfolio discussed above: (a) the non-agency paper, due to credit enhancements, is perfectly good from a credit standpoint, and (b) the ex-ante liquidity premium will be captured, 100% on maturity, and (c) the paper, ex-post, is way less liquid than the ex-ante assumption. In such a case, the investment returns of the portfolio will exceed the benchmark for the period of the investment. However, there will have been more interim risk. Is there a good way to describe this?

Update, 2008-2-7: Discussion with a reader has clarified a better way of expressing part of the above (2008-1-12 Update) idea: use the Moody’s Implied Ratings to compute GINI coefficients at industry-standard time-horizons. Also, check the volatility of the these ratings. Moody’s may have already done this … I really don’t know.

Anyway, I’ll bet a nickel that Moody’s assigned ratings are better than Moody’s implied ratings under these two metrics.

Update, 2008-2-8: Further discussion and thought! The idea that yield spreads are well-correlated with credit risk has been examined by the Cleveland Fed, with the idea that bank supervision would be improved if every bank had at least one sub-debt issue that would trade in the market, providing information to the Fed. According to the Cleveland Fed, this is not yet a reliable measure.

On the other hand, one would (well … could!) expect faster and more predictable refinancing for subprime-ARMs relative to “normal” prime mortgages, with everybody trying to refinance as soon as the teaser rate expired. This would lead to more predictable cash flows and less negative convexity for these instruments, which should imply a lower required yield, which would be REALLY hard to pick apart from the credit risk.