I finished reading Scott Patterson’s The Quants a week or two ago and am still thinking about how the whole housing bubble collapsed and our current economic crisis began. The book is really an interesting read and I’d recommend it to anyone.
As I understand it, here are how the system worked:
Let’s say that the bank you work for owns the deeds of 1000 homes whose inhabitants are considered high risk for defaulting on their mortgage payments (maybe their annual income is very low relative to the outstanding principal, or maybe they are all people who have defaulted on mortgages before). Let’s say that each of those households is supposed to pay $1000 a month. If everyone pays their bill on time, your bank should receive $1,000,000 a month in payments.
You can think of those mortgages as being like a company that generates (up to) $1 million a month in revenue. Whoever owns those 1000 mortgages will get a fixed income of (up to) $1 million sent to them each month. So, we can create shares in that company and divide that revenue among whomever owns the shares, just like companies distribute dividend payments to whomever owns shares. Let’s say that we make 1,000,000 shares. If every household pays their bill, then each shareholder will get $1 for each share they own. These shares constitute a ‘collateralized mortgage obligation’ (CMO).
But with high-risk mortgages, it’s a safe bet that some people won’t pay their bill on time. So we won’t be able to pay $1 to each and every share. If that happens, how do we split the money that does come in among those 1 million shares? One way would be to simply divide the money equally. If only $900,000 comes in, then each share gets paid $0.90, instead of a dollar. However, that would mean that each month, every single shareholder would be uncertain about how much money they would receive for the shares they own.
So, to reduce the uncertainty for some shareholders, the shares were divided into groups called ‘tranches’. For the sake of this description, let’s say that those tranches are called ‘Gold’, ‘Silver’ and ‘Bronze’. Gold shares get paid first and Bronze shares get paid last. So, if only $900,000 comes in one month, and there equal numbers of Gold, Silver and Bronze shares, then $333 K will be used to give the Gold shares their full $1 each. The next $333 K will be used to give all the Silver share their $1 each too. But that leaves only $133,000 left for the remaining 333,000 Bronze shares. So they can only get about $0.70 each.
Thus, it seems like the Gold shares are the safest, Silver shares slightly riskier and Bronze shares the riskiest, because the Bronze shares bear the full impact if more households default. A very large portion of the households would have to default in order for the Gold shares to be affected at all. As a result, credit rating agencies often gave the top tranche AAA ratings (even though they were based on high-risk underlying mortgages).
This is important because a AAA rating significantly increases the number of prospective buyers. If mutual fund’s prospectus says that they only deal with AAA-rated securities, then they are legally obligated to do so.
Over time this idea of dividing up the rights to mortgage payments got applied to a wide range of payments: consumer credit card debt, student loan debt, commercial debt. These were called ‘collateralized debt obligations’ or CDOs.
Financial engineers began to craft CDOs from, say, the Gold tranches of other CDOs, making what was called a ‘CDO squared’. Obviously, this vehicle, too, might get a AAA rating, because it was made up of other AAA-rated securities (and because rating agencies were getting a lot of business from the people making these new securities, so they tended to favor giving their clients what they wanted to hear). In many cases, the people buying these more-exotic shares had no clue what the underlying assets were or how risky they were.
So the floodgates of financial tools opened. People could buy option on CDOs – that is, a contract that gives the right (but not the obligation) to buy a certain CDO before a certain date at a certain price. People could short sell CDOs (that is, sign a contract to agree to sell shares they didn’t own in the hope that the price would drop and they could buy them cheaper than what they had already agreed to sell them for). And so on.
In time the value of these contracts and shares greatly exceeded the values of the underlying mortgages many times over. The dollar figures were absolutely staggering. And of course, the shares don’t need to be equally apportioned among the different tranches, so for many CDOs, the top tranche was expanded to be 80% of all shares. And as well, the prices of shares all affected each other. If default rates went up, Bronze shares would be less enticing, so the price of Silver shares might rise as people sought to buy those ones instead. But if the default rates went up further, Silver shares might start to be at risk of being affected, so their price might crash as people fled to Gold shares. Thus, Gold and Silver shares were not insulated from risk, they just saw their prices spike and crash with every tiny panic that hit the markets.
And all of this was built on top of mortgages from people who are likely to be the first hit by any economic trouble.
What really strikes me about the whole system, though, is how incredibly logical the ‘tranche’ method of allocating risk seems. Rather than have only one type of share, so that they all were exposed to the same risk if defaults went up, the tranche system shifted all risk into a set of shares that people knew beforehand were the riskiest to own. That is, people and institutions could choose how much risk they were willing to take and buy the appropriate CDO (or CDO-squared, or CDO option, or option on a CDO-squared, etc.).
The people who helped create this system of debt obligations must have thought that they were geniuses (which many were) and that they were doing important work to help contain and ameliorate risk (which they were not). How much fun it must have been while it lasted.