Do you believe in Magic? [Fannie Mae part IA.]

This post has a couple of purposes. Firstly it is interesting in its own right – but more importantly it is key to understanding Fannie Mae’s credit profile. Consider it as Fannie Mae Part IA to go with my Fannie Mae Part I.

Mortgage Guarantee Insurance Corporation (MGIC or NYSE:MTG) is a mortgage insurance company. It was the mortgage company I hated most. I despised it. MGIC was always known as Magic – and almost whatever they said I didn’t believe. I would rather believe in witches and other black magic than almost anything the CEO of this company said.

You need a little background to know why I hated this company so much.

The changing role of mortgage insurance

There is a traditional role for mortgage insurance. The GSEs (Fannie and Freddie) were not allowed to underwrite mortgages with a loan to valuation ratio of greater than 80% without secondary mortgage insurance.

However realistically most first home buyers have never been able to put 20% down. So secondary mortgage insurance was required. This would bring the loan within Fannie Mae criteria. The traditional book of mortgage insurance companies was thus enormous piles of loans originated with 80-95 percent loan to valuation ratio – almost all loans to young couples for their first home where the total mortgage was somewhere between 100K and 200K. These loans were diversified by geography and time of origination – but not much else. They were also pretty safe.

Then a few things happened. By far the most important was that banks and the securitisation market started offering closed end seconds (CES). The idea was that you took a standard Fannie or Freddie mortgage (ie a GSE qualifying mortgage) and a second mortgage (or even two second mortgages) above that. This was the so called 80-10 product meaning an 80% GSE mortgage and a 10% second above that. There was also 80-20 product and 80-10-10 product describing different sizes of secondary mortgage.

The result was banks undercut mortgage insurers and the core product of the mortgage insurers looked obsolete. The mortgage insurers looked in a quandary – and several of the mortgage insurers had a hard time growing in the domestic market. According to the mortgage insurers association (MICA) the total mortgage insurance in force was 668 billion in December 2006 and 619 billion in December 03. That is not much growth in the most wild-ass mortgage market in the history of the USA. During that period premium rates actually fell – so revenue went nowhere to backwards.

The mortgage insurers compensated by doing other things – mostly riskier things – and that blew them apart.

• PMI went to Australia and purchased a stake in FGIC. FGIC is a bond guarantor that has been smashed.
• Radian funded a bond guarantor (Radian Guarantee – formerly Enhance Re) and blew up on it. They also invested in a fraudulent securitisation company (CBASS) I blogged about here.
• General Electric just decided it wanted to exit the space and wrapped its mortgage insurer in Genworth Financial and spun some out and sold the rest.
• Magic went subprime – and owned a stake in CBASS.

What Magic did

Magic went deep and scummy subprime. It started guaranteeing mortgages in the pools produced by the most dodgy of producers – it was particularly big into Novastar and ResCap (the GMAC mortgage business). Novastar were about as questionable as subprime mortgage providers got (and the money they owe Wachovia is a case-in-point about questionable exposures at Wachovia). ResCap by comparison were Vestal Virgins (and we know how f—ked about ResCap is). Herb Greenberg wrote more-or-less continuously about Novastar. He got a lot of flack – but he was right.

Anyway – insuring mortgages written by Novastar was a pretty quick way to the poorhouse if you asked me. Novastar also got PMI to ensure mortgages but quite quickly PMI realised what Novastar was it stopped doing business with them. Herb Greenberg also wrote about that.

Anyway Magic were the ugliest of the ugly in subprime securitisation. They insured pure crap. I couldn’t imagine how they would wind up solvent.

They then pretended it was OK by using the most absurd definition of subprime, Alt-A and prime I have ever seen. I blogged about that here. Their definition of subprime was so weak that it was possible to not be considered subprime if you had:

• A 95 percent loan to valuation in a low-end housing estate
• Where the valuation was taken by a broker-friendly appraiser at the third cash-out refinance
• Where there was no evidence that the person was able to repay principal and – to the contrary where they needed to refinance their mortgage every few years to make it current, and
• Where the customer had in the past seven years defaulted on a credit card costing the issuer money

And that was the guts of the reason why I hated Magic. It was the company with the most subprime book – but they pretended it was mostly not subprime by defining subprime in such a way that only serial credit card fraudsters seemed to meet the definition.

They also owned half of CBASS – which owned a very nasty loan servicing business (Litton). You can find out just how nasty Litton is here. I blogged about it here.

Magic were the scum of the mortgage boom.

And if you had asked me if there was any realistic proposition of magic surviving a proper crisis I would have answered no. I always said it was a $5 stock – which was just so I didn’t sound nutty saying it was a zip. [I was considered a hyper-bear in those days…]

Anyway I was wrong. Magic is sort of surviving. It is not an easy situation – but it is not completely diabolical. And sometimes I am stunned when institutions with OK management (Fifth Third) or really rich institutions (Natixis) get into trouble. But this stuns me the other way – these guys are haven’t filed the Chapter 11 they so richly deserve.

Do I have to add that they compounded these errors by repurchasing billions of dollars worth of their shares at absurd prices? No – I can thank them for it – their repurchases provided a counterparty to my short sales!

So where is Magic today?

Well – obviously they have given up doing the real dumb business. CBASS has been written off and the company has retreated to doing the traditional business of first home buyers…

I doubt I can say it any better than the Management discussion in the last annual report:

Premium Deficiency

Historically a significant portion of the mortgage insurance we provided through the bulk channel was used as a credit enhancement for mortgage loans included in home equity (or “private label”) securitizations, which are the terms the market uses to refer to securitizations sponsored by firms besides the GSEs or Ginnie Mae, such as Wall Street investment banks. We refer to the portfolios of loans we insured through the bulk channel that we knew would serve as collateral in a home equity securitization as “Wall Street bulk transactions”. During the fourth quarter of 2007, the performance of loans included in Wall Street bulk transactions deteriorated materially and this deterioration was materially worse than we experienced for loans insured through the flow channel or loans insured through the remainder of our bulk channel. Therefore, during the fourth quarter, we decided to stop writing insurance on Wall Street bulk transactions. In general, loans included in Wall Street bulk transactions had lower average FICO scores and a higher percentage of ARMs, compared to our remaining business.

In the fourth quarter of 2007, we recorded premium deficiency reserves of $1,211 million relating to Wall Street bulk transactions remaining in our insurance in force. This amount is the present value of expected future losses and expenses that exceeded the present value of expected future premium and already established loss reserves on these bulk transactions. See further discussion under “— Results of Operations —Losses— Premium Deficiency.”

C-BASS Impairment

C-BASS, a limited liability company, is an unconsolidated, less than 50%-owned joint venture investment of ours that is not controlled by us. Historically, C-BASS was principally engaged in the business of investing in the credit risk of subprime single-family residential mortgages. Beginning in February 2007 and continuing through approximately the end of March 2007, the subprime mortgage market experienced significant turmoil. After a period of relative stability that persisted during April, May and through approximately late June, market dislocations recurred and then accelerated to unprecedented levels beginning in approximately mid-July 2007. As a result of margin calls from lenders that C-BASS was unable to meet, C-BASS’s purchases of mortgages and mortgage securities and its securitization activities ceased. On July 30, 2007, we announced that we had concluded that the value of our investment in C-BASS had been materially impaired and that the amount of the impairment could be our entire investment.

In connection with the determination of our results of operations for the quarter ended September 30, 2007, we wrote down our entire equity investment in C-BASS through an impairment charge of $466 million. This impairment charge is reflected in our results of operations for 2007. For additional information about this impairment charge, see Note 8 to our consolidated financial statements. In mid-July 2007 we lent C-BASS $50 million under an unsecured credit facility. At September 30, 2007 this note was carried at face value on our consolidated balance sheet. During the fourth quarter of 2007 C-BASS incurred additional losses that caused us to reduce the carrying value of the note to zero under equity method accounting. The equity method reduction in carrying value is not necessarily indicative of a change in our view of collectability.

Well apart from not disclosing just how scummy C-BASS was all along this seems a fair disclosure. But here are the critical things for the future:

• That the excess of future claims over revenue (a premium deficiency reserve) was a bit over a billion dollars.
• That the flow business – the traditional first home business – was not showing undesirable credit characteristics as of year end.

Now, as you imagine, I never thought this company had much credibility (they continued to lie about the nature of C-BASS) and they were as close to scum as anyone I have come across in this industry. (Surprisingly the CEO has not changed.) But if the losses on the closed block of business are larger than this or the traditional business of first home buyers goes pear shaped then there is not going to be much left.

So here is the sticky question: do you believe in Magic?

The company gave us one marker that I guess you could use to test whether you believe… that marker is that they predicted paid losses this year to be just under two billion dollars.

And if you want a bull case – the margins are very much improved

Premium revenue is going up massively and they are no longer having to write business for Novastar to get that premium. Net premium written in the second quarter were 372 million compared to 321 million last year – and the total par insured has fallen from 19 billion to 14 billion.

372 million for accepting credit risk on 14 billion of mortgages is still not stunning business – it is still less than 3 percent. But it is an awful lot better than it was. However with this premium the company is still going to get into diabolical trouble if the conventional first-home owner/Fannie insured mortgage market goes pear shaped.

But current margins are not the story – the back-book and the quality of the new business is

Magic gave us a criteria by which we can judge - they told us that premium income was going to be rising and claims paid would be under 2 billion dollars this year. Surprisingly they are beating this by a very comfortable margin.

The claims paid are running at 385 million for the last quarter – about what the written premium is. The company is barely bleeding cash at all. If you had asked me whether it was possible that this – the most scuzzy of mortgage companies – could be essentially cash-flow neutral this far into a mortgage crisis this big I would have laughed and not answered. But they are essentially cash flow neutral (with a caveat that I will explain below about statutory capital). Still it is pretty hard for a company that is not bleeding cash to go insolvent. It is not impossible – but on these numbers you would have to say that you do believe in magic.

The problem with this analysis is that there is a bad reason claims paid are running below projections. It is just that the courts are jammed up, the servicers of the mortgages totally swamped and foreclosure is running much slower than anticipated.

You can see that. The delinquency on their super-scummy subprime loans is over 36 percent and rising very fast. This is serious delinquency – the company does not bother reporting 30 day delinquency (probably because its servicers do not tell it).

The company’s reserves smell very light if you adjust for this problem. They have insurance over 34 billion of bulk loans in force and they have serious delinquency on the bulk loans of is over 25 percent. They are coming in at a few billion of losses – not just the billion or so in the reserve kitty. This company looks and smells under-reserved still. And given that I probably wouldn’t buy insurance from it. Whether it is solvent or not – that is another question.

But people do buy insurance from it and in quantity – and if that insurance doesn’t blow up in their face then the company will be just fine.

Is the new business any good?

This whole thing is contingent on the new business being OK. This is a company keeping up with old losses by writing lots of new business – and to date it is working.

The new business is mostly flow business being written for Fannie Mae and Freddie Mac. It may be good business (the jury is out). But it is pretty clear why Fannie and Freddie want to support Magic. It is in the interest of Fannie and Freddie to keep this company alive. If it dies then Fannie and Freddie are on the hock for the insured business. And if Fannie and Freddie stop trusting it for insurance then the company will die.

This is a bad dynamic. Magic gets to write lots of business at today’s much improved margins at terms which might keep it alive – and it gets to do it because Fannie and Freddie refuse to face reality.

But here I think the magic is very black indeed.

Look at the delinquency rate on the flow business. It has doubled and it is running I the table above just over 6 percent. That number has me really puzzled. Indeed I needed a good lie down and think.

Why is six percent so high? Well Fannie Mae publishes the delinquency on its credit enhanced book every month – you can find it in the monthly statistics. It’s a pretty good indication of the average of the mortgage insurance industry. You can find the table in the bottom right corner of the second page of this pdf.

The Fannie Mae number is running at 3.56 percent. The Magic number is almost twice as high. Come to think of it – the Magic number is almost twice as high as the delinquencies on the 100 odd billion of Alt-A stuff at Freddie Mac which makes everyone sure that Freddie is going to fold.

This suggests to me that Magic is doing something considerably riskier than the Fannie Mae average to get the higher delinquency.

We are meant to believe that this is good business – but it doesn’t look like America. Fannie Mae’ average delinquency of under 1.5 percent looks like America. I do not get it and I have fast come to the conclusion that Magic are just unreformed and unrepentant river-boat gamblers who still operate in the risky end of what is left of the mortgage market.

It might be just the general trend that I have blogged about – that 2007 business just looks sick (see here and here). But if it is the oncoming train of 2007 business then I wouldn’t want to stand in front of it.

So in summary: I do not believe in Magic. I have covered my short – unfortunately several dollars higher than this. But something remains not quite right here.

John

Postscript: Can you go bust whilst being cash flow positive?

I asserted in my post that it was pretty difficult for Magic to go bust whilst being cash flow positive. That assertion was a gross simplification.

Magic is a holding company with considerable debt holding a regulated insurance company. When the insurance company writes premium it needs to make reserves (the so-called contingency reserve). These reserves reduce distributable capital. The losses also reserve distributable capital. The cash flow and the ability to distribute from the regulated insurance company to the holding company are not correlated.

The holding company needs to receive distribution as it has debt – and if it fails to pay that debt it will fold.

In other words it is possible for the holding company to go bust whilst the whole entity is cash flow positive. [Holders of Conseco once found that out.]

I would love it if someone has modelled the statutory accounts of this company – or could help me with that task.

John

Note: I put my MTG:NYSE short back on "last night". This is unusual for me - I originally shorted this stock above 50 and covered at 20 and 12. I put it back on below 8. For those that don't know I am in Australia - so it really was last night.

Covered bonds as a panacea – or a recipe for the FDIC to kick in more taxpayer money

I am getting jaundiced in the banks pushing covered bonds as the panacea for the housing market and the journalists who buy it. This story really got my goat because it is so one-sided.

All a covered bond is is a guaranteed securitisation.

• Currently a bank can borrow money unsecured or subordinated to buy mortgages. In which case the borrowed money carries a bank guarantee, is generally subordinate to depositors, and hits the bank’s capital ratios. In insolvency the assets are there and that helps pay depositors.
• Alternatively a bank can securitize, in which case the funding is secured, but there is no bank guarantee and hence there is no impact on the bank’s capital ratios. Ideally there is no effective guarantee at all on the assets in which case problems with those assets cannot hurt depositors.

What the investment banks want to sell is funding which is secured and carries a bank guarantee.

When a bank sells covered bonds problems with the assets hurt depositors but the assets are not available to help pay off depositors if the bank gets into trouble.

In other words it downgrades the claim of the depositors whilst not reducing the risk to deposits.

Deposits generally carry a Federal Government guarantee. Covered bonds will increase the cost of honouring those guarantees and hence increase the cost to tax payers of bank bail outs.

Paulson wants to allow this. Advice to the Senate: you will be there in 12 months. Paulson will not. Think of it that way.

J

Update

A couple of people (notably Mish) have pointed out the limitations on covered bonds will make them less useful than they might otherwise be and hence reduce the systematic risk. The Aleph Blog also made that point here.

I still think this is very dangerous. Imagine a bank in some trouble selling covered bonds. When the loans go bad they need to replace them. When the insolvency happens the bond holders wind up owning the good assets and the bad assets get left preferentially to the depositors (read the taxpayers). But the limitations which are extensive both reduce this risk and reduce the effectiveness of the bonds as a capital raising tool...

From the comments:

Paul left a comment on my “what if it was fraud” post:

I live in California and work in finance. I took on a pro-bono "client" recently who was up against the wall thanks to mortgage/housing costs that were extremely large: 90% of his gross wages - before the ARM reset. His poor decision to buy near the peak was based on speculation, but the surprising thing I discovered when I looked at the paperwork is that he did not have to commit fraud to obtain the mortgage. Countrywide originated the 0% down ARM loan with full knowledge of my client's financial status.

My client is an immigrant, and Countrywide had reps fluent in spanish who penetrated this community.

I have a retired friend who works full time with a non-profit group focusing on individuals with high debt loads. He has had over 200 "clients" in recent years who got themselves into a very similar situation to my client. Zero money down mortgages, rising house prices, and financially unsophisticated clients all contributed to the situation.

I hear variants on this story regularly. There was a fraud committed. It may not have been by the borrower – it was probably committed by Countrywide (and my guess is that Countrywide offered representations when they securitised the loan and so Countrywide will need to make good). That could be unpleasant for Bank of America.

But the issue I have is about working out what is going on in the market as a whole. Losses like the one you describe are (a) not redeemable by forbearance, (b) large and rapid. They fit the “pig through the python”.

I wish I knew how much of the market was “pig through the python” and how much was ordinary mortgage stress. If it is pig through the python there are plenty of stocks to buy right now. If the “fraud” cases are the first wave of a much broader mortgage stress then there are stocks to short right now.

The thing about the case Paul describes is that there will be no more losses from such cases in a few years. They will have all been incurred.

I would love a way to quantify this. It’s the trillion dollar question.

John

PS. Paul – thanks – can you send me an email. Much appreciated.

A correction on mortgage trends

I posted the general view that the worse your credit the faster it is improving and the better it is the faster it is deteriorating. (See this post in response to Felix Salmon.)

Again there is a gross simplication. This is true for business written in 2006 and prior. 2007 business just looks sick everywhere. I also blogged about that here where I noted that if you did 2005 and 2006 non-prime business you are seeing light at the end of the tunnel. If you did 2007 business that light is an oncoming train.

J

Jamie Dimon – it is time to pay up

Yesterday I wrote about fraud in the mortgage market.

Today I will write about a single deal – the Bear Stearns 2007-1 closed end second deal insured by Ambac.

This deal is a stinker. Ambac has provisioned for an 82 percent cumulative loss rate on this deal. I have looked at the raw securitisation data and I don’t think it will be quite that bad – but it is awful.

For this to happen 85 plus percent of the loans in the deal need to default.

That doesn’t happen to normal loans. Even in the worst of crises diversified and honestly assembled pools of loans don’t default at that rate.

It only happens when there is mass fraud involved.

Bear Stearns assembled those loans fraudulently or knew the loans were fraudulent when assembled.

But it is worse. They warranted the loans would be OK.

You will find the serving agreement here. I am not a total masochist so I haven’t read all of it… however the document contains the usual clause requiring Bear Stearns to buy back and replace loans that breached the mortgage loan purchase agreement…

Jamie Dimon didn’t create this problem. Them lying scoundrels at Bear Stearns did. They misrepresented the loans – not Jamie Dimon.

But Jamie bought this problem and it is now his.

Jamie – it is time to pay up. Hope you have provisioned for it.

John

PS. Not to be too hard on Mr Dimon I have spent some time looking at securitisation created by JPM. They perform better than average at least in the areas I have looked.

Post script: I have just listed to Ambac's conference call. I said in this post that my estimate for loss on this deal is somewhat less than Ambac's estimate. In the quarter Ambac reversed some of their reserve and the reserves look much closer to my estimate. So Jamie doesn't have to pay quite as much. But he is going to have to pay...

Felix Salmon steals my thunder

Felix has the response to my "is it fraud" post. Read it.

There was plenty of fraud - unknowable amounts. And there is an awful big pig in the python.

But there is something coming after the fraud - which is plain vanilla losses.

Now here is something that I was hoping to tease out as part of the Fannie Mae story - which is what I am working to. Its what the mortgage data says:

· If the pool of mortgages is really bad it is getting better.

· The worse it is the faster it is getting better.

The deals in the ABX – which are truly awful – are improving at a rapid rate. This is indicative of a pig in the python. The loss rate for those deals is awful but the direction is good.

· If the pool of mortgages is fairly good it is getting worse.

· The better it is the faster it is getting worse.

The loss rate might not be very bad (yet?) – but the direction is horrible.

It is not like "we are all subprime now" – just that ordinary credit is getting worse – which means what was a fraud problem is morphing into an economic problem.

There are non-prime pools with 85% defaults. There is no way that is going to morph across America. So the line “we are all subprime now” is just wrong. But the there is plenty coming and it has a different profile to what has been.

John

What if the problem was mainly fraud?

Warning: This is a post that stereotypes people to make a point. The world is more complicated than this post – and the stereotypes are precisely that. I would love complex and real examples in the comments that question my stereotypes. I am a finance junkie – and you forget when you are looking at numbers that they embody the dreams, aspirations and failures of real people. What I seek is things beyond the stereotype… so I have different ways of reading the numbers…

There are all sorts of causes of the mortgage crisis and all sorts of results. However I want to pose a question about various types of borrowers and what their loss curves are. I am really seeking anecdotal industry evidence – hopefully from people who were once mortgage brokers or similar.

Imagine a few types of stereotypical borrower:

• Jack and Jane are a typical first-home buyer. They have saved $10000 (which was hard) because Jack does contract work and sometimes his income is good and sometimes it is not-so-good. They have purchased their first home in suburbia. Its small – but property prices were high so they paid too much. Like almost all American young couples they aspire to a nicer second home. Contract work is getting thin at the moment and Jane might fall pregnant reducing the household income further. They are people of good character suffering mortgage stress.
• Margaret is a divorcee. When Patrick walked out he left her with debt against the house (but some equity) and a lifestyle she couldn’t afford. However the credit card offers came in the mail and Margaret maintained her lifestyle. Eventually she had to repay the mortgage. She did so easily by refinancing the house – and even refinanced the house three times on a stated income loan. She now has a debt that she can never afford even if she adjusts her lifestyle. And she has shown no ability to adjust. Indeed she has got fatter and sorry for herself. She smokes 40 a day. [The NYT profiled such a case.]
• Tom is a flipper. He has deliberately borrowed more than he can afford on a stated income loan. He hopes to sell the house to the next flipper. The loan would not have been granted on his real income – stable though it is – because he can’t afford it.
• Jaylin and Ebony are a lower income black couple of good character. They were victims of predatory lending. Some unscrupulous broker sold them a loan with five points up front (hidden in the fine print) and a low initial monthly and a balloon payment so they could consolidate their car loan. The car is necessary because it is how Jaylin gets to work. I am deliberately racially stereotyping here because there is plenty of evidence that predatory loans were targeted at predominantly black neighbourhoods. This couple was just defrauded by the system. They would be willing and able to repay a decent their loan on original schedule – but the balloon payment and the five points ripped off them by an unscrupulous broker puts them over the edge. They too are distressed and angry – and their willingness to work with or trust financial institutions is also low.
• Frank and Helen aspire to the upper middle class. He has a good job and she works part time and looks after the kids. The marriage is about as stable as the average American marriage – which means that when its good it is very good – but financial stress does strain the marriage (and marriage strain potentially strains the finances). They wanted to get the kids into a better school and hence purchased in a better neighbourhood. Keeping up with the Jones they purchased a better car on an incentive payment. It’s a big car (SUVs were fashionable) and the gas price and the lack of overtime is stretching Frank’s budget. But they want to keep the kids at the good school. They have fallen behind on the mortgage payments – but they make payments occasionally –and they can afford to make lesser payments without catastrophically adjusting their lifestyle.

Everyone who has followed this industry has seen borrowers that fit these stereotypes. However people are varied and complex and if you limit yourself to such stereotypes you are going to miss out on the richness of the world. The press articles tend to follow one stereotype or another - and hence also miss the richness of the world. As I am in Australia (reading bank account statements) and many of my readers are in the "real world" I seek comment.

That said – my guess is that these stereotypes have very different default profiles even in a diabolical real estate market.

On the stereotypes I have given you two of the buyers are essentially irredeemable – and three of the buyers could work the loan out with a little bit of forbearance, some modification of loan terms and a little bit of flexibility. The lenders will still take losses – but they won’t be catastrophic losses.

Think what happens if you modify the loans to Margaret, the debt ridden divorcee, or Tom the Flipper. You tell them you can keep the house if only you make 70 percent of the payment you originally promised.

• Margaret got herself into this position because there was no way that she could afford her lifestyle. She borrowed by fraudulently overstating her income. She has no hope of repaying even on modified terms. The best hope for the lender is to close her up. If she gets her act together she won’t be homeless for long. But life is about to get much worse for her.
• Tom the flipper can’t repay either (but he might be able to make 80% of scheduled payment). The problem is that Tom the flipper has no incentive to pay. He is seriously upside down [negative equity] on the loan and he only wanted to flip the house for what is now a very unlikely profit. Tom’s best strategy is to walk - and he was gaming the system in the first place - so he will walk... Again he got there through fraudulently overstating his income. The bank has the property and Tom’s lifestyle won’t change much.

Compare this to the other borrowers:

• The typical first home owners Jack and Jane are going to be very reluctant to default. They have a proven ability to save (witnessed by the deposit that they saved). They want to join the middle class. They still see homeownership as the way there. If the lender modifies their loan they will get paid most of it. Indeed you might receive all of it. It is unlikely that a well modified loan will cost more than 10% of outstandings. Jack and Jane still have a high chance of default in a recession - and there can be a tragedy underlying this - but they will fight not to default.
  
• The couple who were victims of a predatory loan are a bit more difficult. Unscrupulous brokers working for Novastar and similar companies have ripped 10 thousand plus dollars out of these people. That ten grand has gone – it has been spent. Our borrowers have paid interest at say 4 percent but accrued it at 10 percent. They now owe much more than the house is worth. They could reasonably pay the loan on the original terms – and an omniscient owner of the mortgage might restructure their loan that way. But the mortgage is now owned by a defaulted subprime pool and it is not obvious who has the power or incentive to restructure the loan. The restructure will also change the value of different classes of the mortgage securitisation. So they are going to default in my view. The loan might be able to be restructured in a perfect world – but this is not a perfect world. They will default quite rapidly when hit with increased coupon on their loan. There is an out for the lender –just the lender doesn’t know how to take it.
• My last couple will pay (almost?) all the loan and all their loan under modified terms because they want to keep their kids in that school. But they need the loan to be modified because they are at the edge of being able to pay. Still they will cut entertainment, holidays and shrink the car to stay in the house because they want their kids in that school. If there is a recession and the husband's hours get cut - well you should probably modify the loan as that is a better outcome for the bank than foreclosure.
  

Note that some of these are fast defaults and some are slow defaults. If you stop rolling loans the flipper and the divorcee default almost immediately. The victims of the predatory loan default after the low-coupon period expires. The other two loans might not default at all – but whatever happens they will default slowly. They are going to try and hang on.

My point: the fraudulent loans here (stated income, predatory) have fast defaults. The non-fraudulent loans here (typical first home buyer, middle American aspirant family) have slow defaults if they default. One group can’t or won’t hold on. The other group tries desperately to hold on and might succeed.

Or in summary: fraudulent lending leads to incredibly massive and very rapid defaults with little chance of cure. There is no point offering forbearance – it won’t work.

However once the pig (a pile of fraudulent loans) has worked through the python (the mortgage market) then the bulge (defaults) will decline just as rapidly as it rose.

Non-fraudulent but stressed loans take time and if we are seeing increasing defaults now we can expect to see them for another five years. Loan forbearance programs make sense – but they can’t stop the rise in defaults over the next three to five years.

This is where having readers around the world is really useful. To what extent is it fraud – and to what extent is it old-fashioned bad lending. Does this differ by State or even area within States?

If you look at the loss curves for companies (shooting straight up) and extend them then pretty well every institution is insolvent.

Arguing from the stereotypes – if it is fraud is the dominant problem in the mortgage industry then extending these loss curves is wrong – and the losses will drop sharply once the “pig is through the python”. Many institutions that look difficult will turn out OK.

By contrast: if the problems spread to non fraudulent loans then extending the loss curve for many years is the correct analysis – and all sorts of institutions (including the GSEs) are insolvent.

You see why I am looking for comments. This is the Sixty Four billion dollar question – or – more accurately – the trillion dollar question.

I have my views – but data on fraud is particularly hard to obtain because when you survey people they don’t volunteer that they have been fraudulent. So – as the usual data-driven guy that I am I find this one difficult to nut out.

Thanks for the help

John

Just some comments on the email I am getting

I have received a few comments on the Fannie article – which I was going to include in the parts II and III – but as the comments in the email just keep coming I will put something up now:

• The first comment is that I can’t possibly believe that they are solvent given that they are 130 times levered to a really bad housing market.
• I make no such suggestion at all. I simply have not got an answer until I do the credit analysis. However I can say that the inventory and delinquency numbers that they have right now do not indicate insolvency – for them to be insolvent a lot more bad loans have to come through the system than have already come through the system.
• The second – and more sophisticated comment is that all insurers raise rates when times are bad – but it doesn’t save many of them – so why should I spend so much time exploring this effect with Fannie?
• Well – the pricing effect does save some insurers. It only saves them if they can continue to write business – so I don’t ever talk about the pricing effect with MBI or ABK who are effectively in runoff. But I still want to work the numbers.
• That the repo inventory at Fannie is 20 percent in Michigan – even though only 3 percent of their exposure is there.
• That is a real issue which I hadn’t picked up on but should have. When you are as levered as Fannie if a single large state economically falls into the ocean then you can die. I have blogged several times about how bad the housing market is in Michigan (see here and here). It is entirely possible that Fannie loses a very much larger proportion of its inventory. However the point remains that losing 100% on the current inventory is not a problem – for Fannie to be insolvent an awful lot more loans need to pass through the inventory.

As it currently stands very few of Fannie’s losses are in California. Simple logic suggest losses are going higher – much higher. I haven’t worked out any estimate of how much higher – but please keep those emails coming.

Also – my first substantive dig didn’t point out just how exposed to catastrophe risk Fannie is. Fannie has a lot of exposure in LA. The house price in LA currently sucks – and if a major earthquake were to flatten the Inland Empire not only would it be a human tragedy – but I doubt anyone would want to rebuild a large number of the houses. What is a bad situation there for Fannie could turn suddenly diabolical. Warren Buffett once said that he thought Fannie Mae had more super-catastrophe risk than Berkshire. That is a big call by a man who should know…

On the price of hookers in Bulgaria – someone suggested that I could get my readership way up if I were to publish a price index for hookers by jurisdiction.

Well I tried…

No seriously – in the comments was the astounding observation that people in the Baltic States travel internationally to buy food. They suggested pretty well every jurisdiction – but Poland appears to be the consistent answer. This story suggests that food is more expensive in Tallinn than New York or even Finland. I can assure you that not much is cheap in Finland and that this is just another measure of an overpriced currency. Driving to Poland for big shops makes some sense…

But if you wish to confirm that Poland is cheaper than the Baltic States you can do so other ways. The immigrant population of London (who send remittances) will do that for you. Or get on the web and look at the prices of escorts. It will probably do that for you to.

On the last method of research – there is a company in India called “Escorts”. I once worked at a global fund where one staff member legitimately trolled the web for stories about Escorts in India…

How people find my blog

I use Google Analytics to track things like how people find my blog.

Sometimes it surprises me - but yesterday I had not one - but two visitors from the Google search "how much does a hooker cost in Bulgaria".

Oddly Google ranks me as the authority to answer that question. I suspect they will fix that rank soon - as I am a happily married bloke living in Australia who has never been to Bulgaria or owned or shorted a Bulgarian stock. For that reason I have preserved this page:



Well dear readers, I don't much care how you got here. Welcome anyway.



John

Picking apart Fannie Mae – PART ONE

Warning: This post in this form contains NO conclusions. I do not express a position on whether the stock is a long or short. I hope to get there by Part III - at the moment I am just conducting an intellectual exercise for me and my readers.

I have had hostile email suggesting that I couldn't possibly suggesting that the stock is a long. I am not. And I will not without a close examination. And I won't suggest its a short without a similar examination. That is the way I want this blog to work...

There has been an awful lot said about Fannie Mae and Freddie Mac of late. Almost none of it contained numbers. If they contained numbers they were just off-the-bat (ie analysis free) estimates of what a bailout might cost.

I can understand why. I had a go at picking apart Fannie Mae. The deleted post can be found here. It was one of the more embarrassing mistakes I have made. I just left a zero off a calculation and produced nonsense. Moreover I ran the calculation past one of the best known hedge fund managers in the world – and he never noticed the missing zero either. (One of my precious readers pointed it out – keep those emails coming.) My hedge fund manager friend doesn’t publish a blog – so you don’t see his mistakes… I assure you he makes them…

Picking apart the numbers is fraught with difficulty. Very few of us are used to thinking in trillions – and the rules of thumb we develop (is that a big number or a little number) become useless. I left off a zero and because the number was so big it didn’t seem too small…

This is the first part of an attempt at picking apart Fannie Mae. I started writing this without any preconception as to whether you want to be long or short. I might wind up at a conclusion – but forgive me if I sound vague or indecisive. That truly reflects my state of mind. You will find out the conclusions the way that I am finding them out – by actually running through the numbers.

Disclosure: I once knew an awful lot about Fannie Mae. I fully understood their accounting fraud before it was disclosed. Indeed I could accurately count the size of the fraud (and I was correct within a couple of billion dollars). I never understood why OFHEO with 200 staff could not (or would not) see it. I was short a lot of Fannie Mae and – whilst I made some coin – I was genuinely shocked at how little interest the market took in Fannie’s misdeeds. Having not made much money when I was right I went on to work out other ways to be right not make money – and I let my detailed knowledge of Fannie Mae lapse.

The range of outcomes to this project is enormous. It could be that

• Fannie is shockingly insolvent – and hence to be avoided like plague or
• The problems are overstated and Fannie is solvent – moreover because the competition has been removed it is likely to be far more profitable in the future than it ever has been in the past – and you should buy it because its best days are in front of it.

I am trying to start this project with no opinion.

I apologise to bond market experts who might read this post. The first bit assumes you don’t know what “negative convexity” means or how to measure it. I am going to try do this without much bond market jargon.

How Fannie Mae makes money

Fannie Mae is in two businesses.

• The guarantee business: Fannie Mae bundles up mortgages, guarantees them and sells them as “pass-through certificates”. In this business it takes only credit and fraud risk and it earns only guarantee fees.
• The portfolio business: Fannie Mae also buys mortgages for its own balance sheet and funds them using wholesale borrowings. It borrows money fairly cheaply because of a quasi-government guarantee. This allows it to make a reasonable spread by owning mortgages. However in this business it takes funding risk and interest rate risk. Because it owns the mortgages and will suffer if they default it also takes credit risk.

Now for those that are not familiar with the inner-workings of Fannie Mae I will surprise you. By far the most important business is the portfolio business. The spreads are much larger in that business and the business usually generates the bulk of Fannie’s revenue. It is also where the accounting frauds were found.

Mismanagement of the portfolio business has – over the years – cost Fannie tens of billions of dollars. The reason why Fannie has so little capital and credibility going into this cycle is that it stuffed this business up comprehensively.

The guarantee business is what is causing Fannie a lot of trouble now has – for most of Fannie’s history been trouble free. [Standard mortgages had a very low default rate until this cycle.]

This table gives from p.46 of the last annual report details Fannie’s total exposures. [Warning – head for large numbers needed as the numbers are in millions.]

year end data	2007	2006	2005	2004	2003
Mortgage portfolio	727,903	728,932	737,889	917,209	908,868
Fannie Mae MBS held by third parties	2,118,909	1,777,550	1,598,918	1,408,047	1,300,520
Other guarantees	41,588	19,747	19,152	14,825	13,168
Mortgage credit book of business	2,888,400	2,526,229	2,355,959	2,340,081	2,222,556

Since Fannie got into trouble on interest rate risk it has shrunk the portfolio business from 917 billion to 727 billion. It however sharply increased the guarantee book of business.

The credit guarantee fees are low – but are rising. The credit costs have been very low until recently (a single basis point). They were easily manageable last year. Here are the average fees and credit costs for the past five full years.

	2007	2006	2005	2004	2003
Average effective guarantee fee (in points)	23.7	22.2	22.3	21.8	21.9
Credit loss ratio (in points)	5.3	2.2	1.1	1.0	1.0

These tables are not strictly comparable because one table is year end and the other is year average. However ignoring this complication we can see that the guarantee business would have been very profitable in all years (but almost certainly loss making in 2008 and 2009).

The thing that is startling is how low the guarantee fees are. In 2003 Fannie Mae had 2.2 trillion of credit exposure. If you take 21.9 bps on this the total guarantee revenue would have been well under 5 billion. Indeed the annual report gives them as 3.4 billion – but it does not classify the profit from some credit exposure in on-balance sheet mortgages as guarantee income.

5 billion sounds a lot – but the post tax profit of Fannie Mae in 2003 on the restated accounts available to shareholders is almost 8 billion. The money wasn’t made on the guarantee business – it was made on the portfolio business.

Given that the profits historically came from the portfolio business we need to understand the portfolio business to see how it really makes its money.

The portfolio business

Fannie Mae buys mortgages. It finances them by issuing debt – “reference notes” and the like with a quasi-guarantee from the US Treasury – and makes a spread. This is old-fashioned banking business – but Fannie doesn’t take deposits.

There are a few problems with this. The first problem is that Fannie actually has to borrow the money and be able to ensure that it can always borrow the money. If Fannie can’t borrow it will fail. That almost happened recently – but since the Feds have made it clear that they stand behind Fannie it is unlikely to happen again. It’s a real risk though. I have made it clear many times (here, here) that the biggest risk faced by many banks is the ability to finance themselves. Government guarantees help a lot though.

The second problem is more continual. Fannie Mae has no idea what the duration of the mortgages it owns is. This was a very big problem indeed in the period 1999 to 2004 when there were massive waves of mortgage refinance in America. The mortgages might last ten years – or they might refinance next week. This could (and did) cause massive problems for Fannie.

The logic is as follows.

Imagine rates were to rise a lot – so that people were very reluctant to refinance and the lifespan of the mortgages were to extend to ten years. Then Fannie would want to have its financing fixed for ten years. Otherwise there could be a real problem. If the financing wasn’t fixed then the funding cost would rise and Fannie’s spread would go negative. If you are levered as many times as Fannie a large negative spread will make you insolvent quite rapidly. So when rates rose Fannie wanted to have its financing fixed.

However if rates were to fall everyone would run off to their broker (this was 2004 remember) and refinance at a lower rate. If Fannie had fixed its finance it would now have ten year finance for ten years it would now own low-yielding mortgages with old high-cost finance. The spread would be negative again. Of course it would be alright if Fannie were floating-rate financed because as rates fell the finance cost would also fall.

So Fannie had to perform the neat trick of having fixed cost financing whenever rates rose and floating rate financing whenever rates fell.

This is of course impossible.

But you can come pretty close using derivatives. What you would do is have finance of a mix of durations and buy “interest rate swaptions”. The “swaption” would give you an option to enter an interest rate swap (ie to fix or unfix your finance) at various times. That way Fannie could fix its finance when rates were rising or have floating finance when rates were falling. The swaptions were of course expensive.

Fannie could (and did) also hedge its interest rate risk by other methods including:

· issuing callable debt (which is clean but expensive and was favoured by Freddie Mac), and

· delta hedging (which is a trick of high finance that involves selling things when they fall in price and buying them back when they rise in price and hence losing money which is expensive).

You will notice that all these hedging methods are expensive and Fannie thus continuously lost money on derivatives. Derivative fair value losses at Fannie have over the past five years totalled 28 billion. (You need a lot of spread to make that up and make a profit – and Fannie made a profit in those five years.)

The frauds at Fannie were in the derivative accounting. The accounting is also a mess because Fannie accounts for the derivatives on a mark-to-market basis but accounts the spread on a yield to maturity basis. This mismatch makes the accounts incomprehensible to almost everyone (including myself most the time).

I won’t tell you how I worked out the fraud and quantified it – but Peter Eavis (a fine journalist if there ever was one) worked out a fair bit of it and put it here.

One thing that made the portfolio business hard was that the mortgage market was huge relative to the treasury market. To hedge risk Fannie needed to trade in the treasury market or buy swaptions from people who traded in the treasury market. The size of this trading moved treasury prices – and the resulting changes in interest rates were always to the detriment of Fannie. The size of Fannie and Freddie relative to the market for US Treasuries made it difficult for Fannie to hedge.

The decline of the portfolio business

Before the mortgage crisis hit the Fannie Mae portfolio business went into very sharp decline. The profitability of Fannie Mae shows this – dropping from 8 billion in 2003 to 3.5 billion in 2006 even though the book of guaranteed business continued to rise. Even a tick-up in guarantee fees could not offset the profit decline.

There are a few reasons for this. I haven’t quantified these in order of importance but I suspect (with limited evidence) that the last one is probably the single biggest factor:

• Yield curves got less steep – and so Fannie lost the option of financing at close to 1 percent and lending at 5 percent.
• Fannie shrank the “owned balance sheet” under regulatory pressure – from over 900 billion to just over 700 billion.
• Fannie reduced the amount of risk it carried – by buying more swaptions and issuing more callable debt. Lowering risk lowered profit. You can see the lowered risk because the “effective duration gap” that Fannie is running has fallen sharply over the years. [The duration gap is published in Fannie’s monthly data. A duration gap that jumps around a lot indicates that a lot of interest rate risk is being taken. At ones stage the published duration gap got over a year – which is an awful lot when you are levered as much as Fannie Mae. On a trillion dollars of mortgages a one percentage adverse movement in rates across the curve will cost you 10 billion dollars if you have a one year duration gap. The duration gap is now normally less than a month.]
• Fannie let the book run off resulting in a natural decline of spread.

The natural decline in spread as you run the book off is second nature to anyone who deals with mortgages but is not obvious to outsiders. It’s as follows:

In any pool of mortgages there are a few that will last a long time and a few that will refinance next year. If you “match” finance them you will have some finance that matures in a long time and some that matures next year.

The short dated finance is cheaper because the “yield curve” generally slopes upward. So the mortgages that don’t last very long have more spread in them than the mortgages that last a long time. As the book runs off you are left only with the low-spread business.

Anyway – however you count it Fannie Mae’s spread based portfolio business ran off – and its credit guarantee business expanded – just in time for the credit losses to hit them hard. Expanding credit risk into a crunch can’t be much fun!

A possible re-emergence of the portfolio business

Fannie’s reduced reliance on portfolio business is a pity – because if you haven’t noticed the yield curve is quite steep at the moment, interest rates are not massively volatile (reducing the cost of the hedging), and Fannie’s funding advantages are intact courtesy of Paulson et al.

Moreover Fannie can hedge far more easily now in the Treasury market because the Treasury market has grown so massively courtesy of 43s incessant government deficits.

If Fannie were to grow the spread business now it would do so at high spreads. If it were to chose to (or be allowed to) run considerable interest rate risk it could easily get the spread to 120bps (an old level). All treasuries under ten years yield under 4.2 percent. Some shorter dates yield under 2 percent. Fannie should be able to finance at 50-70bps more than Treasuries especially with explicit Treasury guarantees.

The yield on an on-the-run wholesale mortgage with a Fannie guarantee is about 6 percent. [See this graph.] Depending on where on the yield curve Fannie finances it will pick up something like one and a bit to three and a bit percentage points of spread. The “swaptions” will reduce this – but it should be possible to earn 140bps spread on 700 billion of owned mortgages if Fannie whilst hedging a fair bit – but by no means all of their interest rate risk. [They could go bust on that risk – so this is a non-trivial caveat.] 140bps annually on 700 billion is over 10 billion per annum.

Rising revenue on guarantee business

It is also entirely possible for Fannie to raise the guarantee charge on new mortgages to well over 30 bps – and the fee on the whole book will continue to rise. My guess is that the guarantee fee will rapidly go to 30bps - but 25 is a gimme. [The quasi government guarantee is worth a lot in this time of uncertainty.]

The Treasury guarantee and lack of competition will allow the guaranteed amount to go to 3 trillion. 25bps on 3 trillion is 7.5 billion. Guarantee fees were over 5 billion last year and rising rapidly – so this is not a stretch.

This looks very like 17.5 billion in revenue.

Sanity check on revenue numbers

I am back working out what the revenue numbers might be not what they are. They are considerably lower primarily because Fannie has spent the last three years shedding interest rate risk. [One of those funny things – Fannie made the same mistake as me – being petrified about interest rate risk when the right thing was to be petrified about credit risk.]

The guarantee figure estimate (7.5 billion per annum) is probably low. The last quarter the number was 1.7 billion up from just over 1 billion. At this rate of change we are very likely to get to guarantee fee income of 10 billion at some stage. Certainly the credit crisis is driving guarantee fee revenue skyward. The average effective guarantee fee is now 29.5 bps. This has risen about 50 percent since the early parts of this century.

Interest rate income is much harder to estimate. The accounting for this always sucked because you never knew what the right level of derivative expense to include was.

Net interest income was over 1600 million for the quarter – but there was over 4000 million in “fair value losses”. On that front it was not a good quarter. But the “fair value losses” are a mark-to-market concept and the interest income is a yield to maturity concept. Still if the revenue pre-derivative cost is 1.6 billion per quarter I ain’t getting to 17.5 billion per year. I am not even getting to 10 billion. Sure the 1600 is up from 1000 million quarterly as the situation gets better in the portfolio business. But unless Fannie again takes some serious interest rate risk the portfolio business is going to be stuck earning less than 6-8 billion annually.

My sanity check fails. But the possibility exists that Fannie will surreptitiously up the interest rate risk again. In which case we can get big revenue numbers. The problem is that it wasn’t very nice last time they tried it!

So let us take the low-ball numbers. The low ball numbers are revenue for guarantee business if 7.5 billion per annum, and revenue for the portfolio business of 6 billion per annum – total revenue of 13.5 billion.

Expenses other than credit losses and mark-to-market expenses

Fannie is just a big financial machine with very few staff which makes money on guarantee fees and interest rate spreads and loses it on credit and derivative hedging costs. Staff costs are close to irrelevant. Administrative expenses are about 500 million per quarter and falling. [There are half a billion of other non-interest expenses in the first quarter – but they are minorities and debt extinguishment costs and shouldn’t really concern this analysis.]

So pre-credit cost profits of Fannie are about 2 billion per annum. They will rise as credit remediation costs rise – but let’s be nasty and call it 3 billion per annum.

Pre-credit costs profits of Fannie

On this calculation the pre-tax, pre-credit cost profits of Fannie Mae will be about 10.5 billion per annum. It will be wildly volatile because the accounts of Fannie are wildly volatile due to mismatches between the portfolio and the portfolio hedges. But that – I believe – is about the underlying run-rate.

Credit costs – the problem of the moment

To date Fannie’s big problems – the ones that caused them to raise capital last time and restate – were interest rate hedging losses. Now it has credit problems. And nobody really has any good idea of how bad the situation is.

But let’s dismiss some alarmist stuff. There were plenty of alarmist pieces about how much housing stock Fannie Mae owns right now. This article suggests that Fannie and Freddie combined owned 6.9 billion of forclosed real estate – less than 5 billion of which is at Fannie Mae. They might lose 30% on it (though even that is not clear).

Well let me say at the outset that those losses are irrelevant. If they had a loss after foreclosure of 50 percent they wipe out about three months of earnings. That would be no threat.

The real estate inventory that Fannie is currently sitting on – being many thousands of houses and many thousands of families in personal tragedy – is irrelevant financially to Fannie Mae. So steer clear of the alarmist stuff. If Fannie only winds up with say 15 billion of foreclosed real estate it will skate through this crisis just fine.

What we need to really hurt is 50 billion of losses over three years – maybe 20 times Fannie’s current inventory or real-estate-owned (REO) needs to pass through Fannie’s ownership. That would wipe out three years’ profit, plus 20 billion in capital and cause all sorts of social and regulatory stress. [We could also blow up on interest rate risk!]

In summary: if the losses are less than 20 billion you will do OK buying Fannie on any of the panics. If the losses are say 50 billion the financial and regulatory stress will be such that Fannie is problematic.

If the losses are 70 billion – say seven year pre-tax earnings – then Fannie sucks as investment and the best you can hope for is a number of dilutionary capital raises. That is if the Feds don’t just confiscate it from you.

But to want to be short Fannie on the fundamentals you need the real estate owned to wind up more than ten times the current (horrific) level of inventory.

Does that happen? My instinct is no – but I would prefer examine the numbers.

I would love help modelling this – but I think it waits for Part Two.

John Hempton