Mortgage market used to a segregated, regional, illiquid and low margin but simple business. Local banks, acted as originators, lend money to home buyers and hold the title of their houses as collateral, and enjoy a profit from the spread between higher mortgage rate received vs. lower interest rate paid to depositors. In the good old days, mortgage lending officers were careful to review and approve mortgages from homeowners. They evaluated financial situation of mortgage applicants carefully, such as household income, job security, total debt, net worth, monthly obligations. Banks were more stringent in their lending practice, and evaluated local and region real estate trend, adjusting to ensure enough cushion if there is an unexpected downturn. The main reason is that banks acted as both originators as well as life long holders of these mortgages. Even from time to time losses were unavoidable, they wanted to avoid losses large enough to put them into trouble and probably out of business. If losses happen, banks would cover first by their spread (profit), if not enough, then dip into their own capital and reserves. For mortgage officers, their current job and reputation for future career were also on the line if they are responsible for careless mistakes on making loans. At that time, home mortgage is definitely not a lucrative but low margin and simple business.
However, last 10 years or so, the rise of home mortgage securitization has thrown many good old practices out of door. Local banks and mortgage companies nowadays are acting more as agents, closing deals as originators, selling the loans to large investment banks and earning their fees. Their role has changed from a loan quality controller to a salesman, doing as many deals as possible in order to earn higher commissions. Then the structured product groups at large investment banks collect thousands of mortgages from around the country, put them into a pool, package and securitize them, slice and dice them into many tranches (products), then sell them to investors. The incentive profile has totally changed in this mortgage alchemy process. Who cares a couple years later the loan is delinquent? Maybe I am the originator, but it is your problem.
This has created some very innovative and aggressive practices of mortgage lending, since the key to earn fee here is volume, not loan quality. This is similar to house flipping during the real estate bubble. When housing price is going up 10% a year with no payment down, who doesn't want to buy a multimillion house which can be flipped for a quick profit? Same motivation on originators. The faster they finance and flip to sell a loan, the more money they can earn. Some day, a few smart people came up with a revolutionary idea called subprime, basically anyone with no income, no credit, no downpayment, no nothing, has no problem to buy a house almost at any price. Or even better, anyone can get a loan for 110% equity of the house appraisal value (since housing price is going up 10% a year anyway). Not only no downpayment, there is actually "free" money to be made by buying a house. This is such an ideal society better than anything Carl Marx could have ever imagined.
More profoundly and importantly, this securitization revolution has also brought many more players and capital into the mortgage market. Instead of only local and traditional banks lending money which have very limited capital base, now we have investors with large capital base coming into this casino to make large bets. They are insurance companies, hedge funds, corporate and state pension funds, endowment funds, foundations, fixed income mutual funds, foreign central banks, foreign large commercial banks, the so-called 144A "sophisticated" (now we know how sophisticated they are!) financial institutions which are "qualified" to bypass any regulatory requirement and oversight. Suddenly the pool of capital available is 100 times bigger than before, coupled with next to nothing mortgage lending requirement, how could housing price not go up? Since 1970s, the median home price/median family income stayed at 2.8 most of the time until 1999, then it has suddenly taken off all the way to almost 4 in 2005 in 6 year's time, over 3 standard deviation.
As I mentioned earlier, this used to be a low margin and simple business in the good old days. But lately before the real estate collapse, every seller in the process of flipping mortgage got a cut, with the largest cut going to the structured product group at Wall St., and suddenly it became a lucrative and highly profitable business. Where is all this "free" money and high margin coming from? I have seen many articles discussing subprime and MBS, even the very questionable lending practice, but I have read no article so far on explaining how and where all this large profit is created. At 60 minutes on Jan 27th, there were some good discussions about how Wall St. has made huge profit out of this, but again the program failed to explain how. So I want to give my quick two cents here.
I think the value "added" or rather mispricing is mainly from the following two steps of the mortgage securitization alchemy process:
1) By pooling thousands of mortgages across the country together, no one will be able to figure out the real and true aggregate value of the combined pool.
2) After slice and dice into many tranches, in order to sell them, each tranch (product) is manipulated to let it price more than it is actually worth, thus further squeezing additional profits.
The first one is based on the argument that the sum of the portfolio is larger than its parts. The second one is based on charging higher price premium by customizing products to fit customers with different risk profiles.
For the first one, I am using S&P as an analogy here. In order to obtain the fair value of the S&P500, there are two ways to do this. One way, I refer to as the correct way, is to evaluate each company separately by studying their financial statements, talking to management, attending earning calls, trade shows and industry symposium, and building NPV models to come up with the fair values of all these 500 companies, then capital weighting them to derive the fair value of S&P 500 Index. Even there are errors and biases here and there, but hopefully they offset each other and are cancelled out at the end of the day.
Another way is to build a complex macro econometric computer model, download all the financial data from Thomson or whatever, make many assumptions on macroeconomics such as economic cycle, interest rate, profit margin, earning increase, GDP growth, job market, payroll, inflation, etc. to come up with the fair value of S&P. A lot of the modeling is probably based on the historical data and trend to force model to fit the data anyway. This is what Abby Cohen at Goldman has been using to forecast S&P earning and price target every year. This is not bad at bullish years since it is basically a trend following approach, but would make bad calls during a major economic turning point when future economic behavior is out of sync from historical data in the past 20 years.
It is the same story but much worse results at the home mortgage securitization desk due to lack of transparency and due diligence as in the stock market. When structured product group dumps all the mortgage data across the country into a pool, its aggregate price is more subject to the assumptions of general real estate market, and historical data of default and payment patterns. It is a big question of whether even the original mortgage data is right due to the questionable or fraud lending practice. Additionally in the model, I bet assumptions about default rate is very low, interest rate remains low so refinancing would never be a problem. The most aggressive assumption must be that there has never been a down year for real estate in US as a whole since great depression in 1930s (By the way, this common myth is not true either if data has not been smoothed), so this trend will continue forever. Well, until 2007. Good time always ends.
In the subprime area, it could not be worse. Without getting into the fraud lending practice, there was almost no subprime 10 years ago, so there is little historical data except last few years. How could the structured product group at Wall St come up with a reasonable estimate of default rate and payment pattern? By assuming real estate goes up 10% every year forever? Or just grabbing a number in the air? This kind of distorted data and wrong assumptions will dramatically amplify the nonlinear behavior on the pricing of the mortgage pool. All these would not have happened in the good old days, when loan officer reviews each borrower one by one, doing due diligence on the quality of the house, resale value, local real estate situation, income and job security of the borrower. This new alchemy process of mortgage securitization offers no transparency, due diligence and even common sense at all. Let them just twist the computer model to come up with what they think the future should look like, resulting a higher aggregate price much larger than its parts.
Another argument used repeatedly is the benefit to own a diversified mortgage pool, thus investors need to pay a higher premium and price. A pool does offer lower risk since mortgage investors are diversified from specific local and regional real estate risk but only subject to general US real estate risk, similar to the fact that stock investors who invest in over 25 diversified stocks have diversified away most company specific risk but only subject to general market risk. However, why should investors pay a higher price for the pool? It is just like the more companies a stock investor buys, the higher price he or she has to pay for each company per share. Not sure that makes sense either.
Now let us take a look at the 2nd step about charging premium for customized products or tranches from this mortgage pool. From the magic hand of Wall St banks, there are so many tranches being created, sometimes it is stripping out embedded options, another time it is derivative on top of another derivative, called whatever squared or cubed, in order to squeeze more fees out of the pool, since the more products, the more exotic products the banks can create and sell, the more profit. Let us just look at the two most common derivatives of MBS: IO (interest only) and PO (principal only). Investors of IO vs. PO would only have rights to the interest payments vs. principal repayments paid by borrowers. IO will fall if rate goes down since more borrowers will refinancing (no more interest payment to the original loans anymore), so called negative duration. PO is totally the opposite that borrowers will pay them back so quickly when rate falls and they will have a free fall of cash, or positive duration. A small deviation of default rate, interest rate movement and especially their timing assumptions will dramatically change the price of them. What is the right price for them? Only God knows. It is basically whatever the price the computer model says, and we will buy it. Who wants and has time to dig behind the hundreds of thousands of lines of computer codes to figure that out? If you feel both IO and PO are way overpriced, that is too bad, since they are not like calls and puts of stock options which are governed by put-call parity and provide investors with arbitrage opportunity. There is no way to short IO and PO and long the original pool.
By designing many exotic and complex products like this, investors get totally lost, are unable to understand the behavior and do any due diligence on the products they are buying. As a result, they have to rely on the opinions of the rating agencies and bond insurers. However, they never realized the rating agencies and bond insurers were actually in bed with Wall St. banks, not only they were paid by the banks with their revenue growth mainly from increasingly complex mortgage derivatives, but also they were all using the same computer models and assumptions so structured product group at Wall St can twist the model to push yield to the ultimate threshold for a particular rating. No wonder mortgage derivatives can always pay higher yield than other bonds for a particular given rating. This alone guarantees that there would be no lack of yield hungry investors from insurance firms, hedge funds (using leverage by borrowing short term repos to generate double digit return), pension funds (need to reduce their pension deficit and liability), foreign banks (too sophisticated to know what they are getting into).
At the end of the day, mortgage securitization is a sales business, and pricing of mortgage products is market driven by competition. As people say, it doesn't matter what the product really is, it is how it is structured to make it look good to be sold. Now you know why the cosmetic industry has such a high profit margin by making some low lost powders and creams, since they make people look good. Home mortgage securitization makes yield hungry financial institutions look good. In order to guarantee their future profit, many investment banks bought mortgage originators to secure future sources of their fees, now they are guaranteed to secure future losses, write-offs, litigations and liabilities. The WSJ has several articles about the ongoing investigation on UBS about a Dillon Read mortgage trader being fired who was trying to assign a more true price to an inflated mortgage product. Come on, be realistic, mortgage trader, if you tell people the truth, how would you still be able to sell them?
As Warren Buffett quoted a banker saying: "Why banks have to find new ways to lose money when the old way had worked just fine?" The problem is without new "creative" ways of losing money, where are all the lucrative fees, salaries and yearend bonuses for investment banks coming from?