Wednesday, December 9, 2009

The Gas Pedal: Rise of Structured Finance (Part 1-Basics)

This section becomes unavoidably semi-technical, so perhaps the best way to approach it is to list the major characters up front and give a brief description of the roles that they play. There are four main three-letter terms that will be featured in today's tale: SPVs (special purpose vehicles); MBS (mortgage-backed securities); CDOs (collateralized debt obligations); and CDS (credit default swaps). There is also a mathematical theorem that plays a major role, and that is something called the "Gaussian copula."

An SPV is an accounting entity that purchases mortgages from a bank by going into the market and issuing bonds to pay for them. These bonds are called mortgage-backed securities (MBS). An SPV that purchases these securities instead of original mortgages is called a CDO; a CDO is just the next evolutionary step in the process after MBS bonds have been created. How does the CDO buy the mortgage-backed securities? It can issue bonds, of course, but frequently the CDO is funded by using short-term financing from the commercial-paper market. The CDO has now become a sort of bank---it borrows money in the market at short-term rates and then purchases longer-term MBS (often 20-30 year maturities) with them. Because of the timing mismatch, the CDO has to continually roll over its debt and take out new short-term loans again and again. The advantage is that the short-term financing is probably very cheap. The disadvantage is that the CDO has continuous financing needs and very bad things can happen if market liquidity suddenly dries up.

If a bondholder is concerned about the issuer defaulting on the bonds and not being able to pay him, he may purchase an insurance-like contract called a CDS to protect himself. Figuring out what the chances are of a bunch of individual mortgages defaulting at the same time is the job of the Gaussian copula theorem.

So let's start with a bank that wants to remove mortgages from its balance sheet. The bank creates a new entity to purchase these mortgages from the bank. The new entity, the SPV, finances the purchases by issuing bonds to third-party investors. Particularly in the wake of the Enron scandals, there are various FASB rules in place regarding the number of voting shares that a bank can legally hold in an SPV and still keep it distinctly separate from the bank's financial statements, so banks tend to have minimal control of the SPVs once they are up and running. In fact, ownership interest in the SPV will probably be dispersed among so many parties that the SPV will not show up on anyone's balance sheet; it will become part of the "shadow banking system". However, there are some circumstances that can force a bank to take an SPV back on its balance sheet, often at the worst possible time, and those circumstances were in fact met in 2008.

When an SPV is set up to purchase mortgages from the bank and issues bonds to pay for the mortgages, these bonds are now called "mortgage-backed securities". Because banks are required to set aside capital against mortgages they create, relieving the bank of these mortgages allows the bank to be "recharged" and to create more.

The cash flows that homeowners pay for their mortgages now pass through the SPV and become the coupons that the SPV's bondholders receive; the mortgages have been "securitized". The way it normally works is that (at least) three different flavors of bond---called "tranches"---are issued by the SPV, and investors can choose which one they would like based on their risk appetites. The safest bond is called the "senior tranche" because it is the first in line to receive cash flows from the mortgages. After the senior tranche gets paid, the "mezzanine" bondholders are second. The last claims on the mortgages go to the riskiest, third-level bonds, which are confusingly termed "equity" or "preferreds". The flow of mortgage cashflows down from the safest to the riskiest MBS bonds is normally called the "waterfall". As you would imagine, those who hold the riskiest tranche want a very high return for doing this. Those who hold the senior, safest tranche will receive a more modest return.

Enter the bond ratings agencies. Before the mortgage-backed securities are issued, bond raters (S&P, Fitch, and Moody's) give these securities a rating. The ratings are very important because many of the best customers for these products, such as pension funds, are only allowed to purchase "investment-grade" (AA or higher; sometimes AAA) bonds. Additionally, the MBS bonds issued by the bank-sponsored SPVs would need to compete with those guaranteed by Fannie Mae and Freddie Mac, who used/abused their privileged government status and had the vaunted AAA rating.

It is important to understand that the buyers really wanted these things to be given an investment-grade rating: a AAA-rated MBS is deemed to have the same level of safety as a bond issued by, say, the US Government or the World Bank, but may also carry an attractive 5-6% coupon. For an insurance company, that combination may prove to be irresistible. The analysts at GaveKal Research put it another way: "(as a result of the dot.com crash)...pension funds and insurance companies around the world found themselves undercapitalized. The regulators, always keen to close the barn door once the horses have fled, decided to prevent the undercapitalized institutions from buying more equities. This left pension funds and insurance companies with a pressing need: how to replace equities, the high return part of their portfolios? Since, according to the new regulations, they could only buy more bonds, they were forced, if they wanted to boost returns, to buy very low quality bonds, offering very high yields. The problem was of course that the regulators had told them that they could not buy bonds below 'investment grade'...and that, as a result for the massive demand for yield around the world, the returns on investment grade bonds were far below the returns on equities that they now had to replace...So all of a sudden, here was a new need: the low quality bond with a high rating."

A bank that was going to be sponsoring an SPV and MBS issuance could "shop" the product around to various rating agencies and see which one would give the highest rating. In order to make things more transparent, the rating agencies would share their valuation process with the banks, so that the banks could structure MBS or CDO tranches in accordance with the known requirements to get, say, a AAA rating on the senior tranche. By diverting more of the cash flows from the MBS bonds to the highest-rated tranche, the SPV could "over-collateralize" the senior tranche of a CDO and achieve a high rating from the agencies. The logic here is intuitive: if fifteen people owe $10 each to three of us, and I get exclusive access to the payments of ten of those people, I am fundamentally safer to loan money to than is my associate who only has access to two $10 payment streams.

This leads us to our next character, the credit default swap. Without getting into the details, a CDS is an insurance-like contract that a purchaser buys in order to insure against the risk of a bond issuer defaulting. These derivatives are traded over-the-counter, which means that they are privately negotiated agreements that take place off of an exchange. The traditional narrative for a CDS is that Firm A holds many bonds issued by, say, General Motors and is concerned that GM is going to default. Firm A goes to Firm B, probably an insurance company, and buys a CDS. If GM does default Firm B has to pay Firm A. Along the way, Firm A has to make quarterly premium payments to Firm B.

CDS play an important role in this story for several reasons. First off, CDS prices were widely used for CDO tranche ratings (discussed in the next section). Second, the size of this market was absolutely huge (approximately four times the size of the entire US economy). Third, because they were not considered "insurance", CDS could be written without the purchaser needing to even own the bonds of the company on which the default concerns were raised---this is a bit like taking out fire insurance on your neighbor's house and being paid if it burns down (there is a moral hazard in this). CDS thus became an important speculative vehicle as well as a hedging/risk management vehicle, but they were not being traded on an exchange and thus lacked transparency. Fourth, the regulatory opinion that CDS were not not insurance contracts allowed insurance companies like AIG and the Monoline insurers to enter into all kinds of credit default swaps, since CDS did not require the same prudential capital reserves to be set aside that traditional insurance did.

Mortgage securitization served as a very useful process. It helped banks to take mortgages off of their balance sheets, which allowed them to make more mortgages. It helped Fannie Mae and Freddie Mac to carry out their affordable housing mandate. It helped investors who had yield requirements due to future liabilities (pension funds, insurance companies) to meet standards for both bond quality and performance.

The next installment will focus on the models used to try to calculate the risks of mortgage defaults, the way that a CDO structure built of subprime components can turbo-charge losses if a "default vector" sweeps through the underlying mortgages that fund the mortgage-back securities that the CDO owns, and the attempts made to insure against these risks by using credit default swap (CDS) contracts.

No comments:

Post a Comment