Archive for March, 2009

In the beginning there were ratings agencies, and they rated corporate bonds from the very highest of credit quality (AAA) down to junk (CCC).

Now AAA means that the chances of losing money are very, very low. With each level of increased incremental risk comes a lower rating. If a corporate bond was at risk for losing just one dollar, it was rated all the way down to junk. And that was fine. Everybody knew the rules of the game.

But then investment banks asked the agencies to rate a large group of home mortgages in a pool known as a Residential Mortgage Backed Security (RMBS). The investment bank would divide the pool (the RMBS) into various tranches. The highest-rated tranche would be given a rating of AAA. Let's say that the AAA tranche was 92% of the loan pool. The AAA tranche would get the first 92% of all monies coming into the pool before the other investors were paid (again, really oversimplified, but that is the net effect). That would mean that the pool could have 16% of the home loans default and lose 50% of their value before the AAA tranche would lose even one dollar.

We all know now, though, that some of those AAA-rated tranches are in fact going to lose money. And the rating agencies are now writing down the ratings on the former AAA tranches.

I am not talking about the exotic CDOs and CDO squareds, or some of the truly toxic securitized assets which are going to zero. What I am writing about today are plain vanilla mortgages grouped together in securitized pools.

I wrote three weeks ago, "The downgrades by Moody's today of 2,446 different classes of Residential Mortgage Backed Securities will be a real blow. Moody's warned in a report last week that loss assumptions would be increased for RMBS and that downgrades could be expected. Moody's is projecting that alt-A deals originated in the second half of 2007 will experience 25.5% losses of original balance, compared to 23.9% of 1H07 deals, 22.1% for H206 deals, and 17.1% for 1H06 deals. The rating agency in May expected average losses for 2006 and 2007 vintage deals to reach 11.2% and 14.7%, respectively." (The Big Picture)

Fitch and S&P are also piling on with downgrades. Most of them see RMBS's go from AAA all the way down to junk. This has some very bad unintended consequences.

Let's say a bank has a loan portfolio of 1,000 individual mortgages valued at an average $200,000, for a total portfolio value of $200 million. The loan officers were not very good, and it turns out that 18% of the homes went into foreclosure and lost an average of 50%. That means 180 homes went into foreclosure and that the bank lost an average of $100,000 per home, or $18 million overall. The bank was charging 6% interest, so in a few years it would at least have its original investment back, although the losses would eat into capital.

To make those loans of $200 million, the bank would need at least $20 million in capital, and so would need to go raise some money or reduce its loan portfolio by selling the performing loans. The reality is that for a bank to have such a large mortgage book, it would probably be a much larger and better-capitalized bank. If it were not, it would soon be taken over by the FDIC.

Note that the remaining 82% of loans are still performing and are carried on the books at full value (again, oversimplified). There is real value in the remaining loan portfolio.

But what if the bank invested in a RMBS that was rated AAA, and 18% of the loans in the security went bad? Remember, the AAA tranche gets the first 92% of income. The loss to the RMBS is 9% of capital. The losses to the AAA tranche are only 1%. Hardly a catastrophe. Annoying, but something you can deal with. Except for some very nasty rules.

Remember, a bond is downgraded to junk if it loses even $1. Now, let's take it to the real world.

Say a bank buys a $1-million AAA portion of that large RMBS. It can use that AAA debt in its capital base, and can actually lever it up about five times, as the rules only make the bank take a 20% "haircut" on an AAA bond. But if the bond goes to CCC, the bank must now move the entire bond to its "risk-impaired" portfolio. And because most institutions cannot buy junk paper, there are very few buyers out there who will want to buy it -- mostly hedge funds and private capital. The price on that paper might easily drop to $.50 on the dollar because of the potential for a 1% loss.

The accountants, being conservative and living with new mark-to-market rules, make the bank take a $500,000 loss. This directly reduces regulatory capital by $500,000. Banks are required to have a maximum of 8% of risk-impaired assets as compared to solid capital to be considered adequately capitalized. Keeping the asset on the books means they have $1 million of risk-weighted assets. If they have to sell to get the capital required to follow the regulations, they will lose $500,000.

And they lose this on an asset that the rating agencies say might lose $1 ten years from now.

...

Here's the truth. That bond should never have been rated AAA to begin with, and it shouldn't be rated CCC today. The ratings agencies took a perfectly fine corporate bond rating system and tried to bootleg it onto a security that has an entirely different set of circumstances. A corporate bond is a bond from one company or one obligor. An RMBS might have several thousand obligors. (An obligor is a person or entity that is obligated to pay back debt.)

It was very convenient for investment banks to get the rating agencies to use the corporate bond analogies, because that meant they did not have to explain a new system. Everyone knew what AAA meant, or AA or BBB. A bond buyer in Europe or at a pension fund simply looked at the rating and hit the buy button. Easy. No need for a lot of research. Make your purchases and go to lunch.

Some simple rules changes would solve a lot of this problem. First, let's recognize that the root of this particular problem is the ratings system. If an RMBS is likely to get $.95 of its capital, then it should be valued at some number below that, but don't make them assign it 100% to their risk capital. That is like making the bank with the 1,000 home loans in its portfolio write off all of them because 18% are bad. In principle, there should be no difference.

Then, the Federal Reserve should call in the rating agencies and have a "come to Jesus" meeting. They are at the heart of the problem, and they need to fix it. They need to change their ratings system for packaged securities like RMBS's.

Let me throw out one idea (there are likely to be a lot better ones, but let's get some ideas on the table). Let's move away from using standard bond ratings for multi-obligor securities. Why not rate a bond by the percentage of capital likely to be returned? Let's call it the Impairment Factor, or I-Factor. If a bond is likely to lose 10% of its capital, then it would have an I-Factor of 10%. An I-Factor of 0% would mean the bond should see all its capital returned, and an I-Factor of 100% would mean that all the money will be lost.

Now, that tells investors something. That's a useful statistic, as opposed to "CCC." What does CCC mean? Am I going to lose $1 or $1,000 or all my money? CCC gives me no useful information if I want to buy or sell a bond. And without real transparency, you end up with a world in which a few very knowledgeable buyers can make a lot of money.

That is because there are a lot of AAA bonds that are going to zero, as in 100% loss. If you are on an institutional desk and would like to participate in getting some of the better values, unless you have a very sophisticated team with good analysis software, you simply can't take the risk.

Further, if the rating agencies do their homework to figure out what the I-Factor is, they will have all sorts of useful information that can be disclosed about the security, such as average loan balance, average loan-to-value, how many loans are at risk of default, where the loans are, and scores of other details. Armed with that information, buyers can make rational decisions.

It may be able to add to the accountability issue as well. Additionally I would say simply throw out any leverage brought about with a rating agency behind it, even with this I-Factor rating; an I-factor range is easily calculated by anyone, using objective criteria (such as percentage in default, delinquency rates, foreclosure rates etc.); given that it is, let the leverage or capital requirement be addressed by this objective calculation, which has to be released for all MBS or pooled product. Then, the rating agency doesn't matter - anyone can rate. The agencies can become providers of this objective information, with a pull-back on fees if they provide false information; plus, ANYONE should be able to provide this information, not just rating agencies, based on the public information given by the pool.

This will truly make rating agencies irrelevant, but retain rating as a method to evaluate investments. Anything outside the current realm of information gets an adverse rating of 100% - meaning put full capital on it. Anything inside gets rated objectively. No one to blame other than ourselves; and no one can hide behind the "first amendment". Oh, I like it.