Yesterday I read a fascinating paper, The Panic of 2007, by Gary Gorton (via Tyler Cowen) that describes the factors that contributed to the problems in the financial sector. I learned a bunch of interesting things about subprime mortgages, RMBS (the securities used to fund subprime mortgages), and the more arcane financial structures (CDO's and CDS's) that have contributed to this crisis. The following is a list of the interesting things that I didn't previously know:
- Subprime mortgages short term loans - This was a little surprising because I had always thought about mortgages as long term investments that could possibly expire early i.e. refinancing or prepaying principle on a 3o year fixed rate mortgage. It seems that when banks made subprime mortgages, they intended force the borrower to refinance after two or three years. Page 12 starts an interesing discussion of subprime mortgage design. In short the banks intended to make a mortgage loan that would need to be refinanced in just a few years, which they assumed would be possible due appreciation on the home value. Gorton argues that this is tantamount to holding a call option on home prices.
- Subprime backed Residential Mortgage Backed Securities (RMBS) were meant to be short term investments - Again I had usually thought about these bonds as long term investments that might expire early and hence carried prepayment risk. Pages 32 and 33 show a comparison of two subprime backed RMBSs, one issued in 2005 and another issued in 2006. The first thing to strike me about these two deals is that the 2005 deal has already seen prepayments of $836 m on the original $1.2 B of mortgages (70%). In fact the vast majority of the original highest rated (AAA) bonds had been paid off. I would not have thought that less than two years after issuance the so much of this debt would be paid off.
- The problems with RMBS may have more to do with increased duration than market illiquidity - In contrast to the 2005 deal, the bonds associated with the 2006 deal are not seeing anywhere near the amount of prepayments - $518 m on the original $1.3B (40%). So investors and banks are now apparently stuck with bonds on their books that they originally expected to prepay. So it may not be so much the lack of trading as the lack of prepayment that is hurting the credit markets.
- The highest rated bonds (AAA) are still highly rated - My initial impression of the subprime mortgage crisis was that the underlying mortgages were performing so poorly that even the highest rated bonds were losing their AAA ratings. Looking at these two deals, that does not seem to be the case. Only the A4 tranche of the 2006 deal has been downgraded and even that bond still gets a AAA from Moody's. I also took a brief look at an subprime RMBS index mentioned in the paper, the ABX.HE index. Just browsing through the various AAA rated indices shows that many of the original AAA rated bonds still held their rating and many of the ones that didn't still held investment grade ratings. The real carnage appears to be in the mezzanine tranches.
- CDO complexity exposed - I guess that I pretty much understood what was going on with RMBS's being restructured into CDO's but this paper really illustrates how complex the payoff structure is. With multiple checks and tests per layer trying to model the cash flow from these investments is basically impossible. Of course if your expectation is that housing prices will keep rising and that you will get your money back in just a few years it may make sense to invest in something that you do not really understand. I guess that is the thought process.