Friday, August 5, 2011

Residential Mortgage Backed Securities: How They're Supposed To Work

Residential Mortgage Backed Securities (RMBS) are the fuel that powers our country's mortgage market. Their role in mortgage lending is not complicated, despite the fact that accurately pricing them requires in-depth knowledge of stochastic processes, matrix math, as well as a solid understanding of partial differential equations. I recently found a great article on-line by American University's Peter Chinloy that explains the how RMBS functions and the many of the assumptions that undergird them.
The long and short of it is that RMBS expands the lending capacity of financial institutions by allowing them to sell the home loans that they originate to purchasers on the secondary market to other institutions, exchanging the cash flow from the loans for cash for their balance sheets. The purchasers of these loans then package them and create investment vehicles or conduits. Investors are then offered an opportunity to participate in the cash flows that comes from the payment of these loans through the purchase of bonds issued on behalf of a conduit. These bonds are categorized as RMBS. They offer the investor exposure to the returns and activity of the housing market without the idiosyncrasies of property ownership or the capital  and labor requirements of lending. Moreover, a great deal of investment grade RMBS is guaranteed and can be insured. Add in the historically low home loan default rates with the lower capital reserve requirements of for insurance companies holding RMBS and it seems like a solid investment.
By now, the story of RMBS and its function has been told in many different places since the beginning of the economic downturn, so I harbor no delusion that what I have written thus far is not already widely understood. Chinloy's article, though written 16 years ago, offers a refreshing explanation of the purpose of each component of the mortgage lending system. It also illustrates the mathematical assumptions of lending behavior, important variables and basic RMBS pricing that can be easily used in Excel.
Chinloy describes the differences in private and agency RMBS that were once key to residential loan securitization. Originally, GNMA or agency RMBS was the outlet for borrowers that had less capital or lower credit scores than was required of conventional loans. Loans that fell under the purview of GNMA, known as FHA, VA or FmHA loans, were not only insured by FHA, VA or FmHA, but they also had default premiums priced into the mortgage payment. GNMA assumed  that the lower access to credit and/or capital typical of the borrowers of these loans would sever to lessen the frequency of prepayment through early payoff or refinance. Prepayment is generally undesirable to the bond holders of RMBS products, because it lowers the interest rate carry of each prepaid mortgage, reducing the amount of cash flow from that note and thus reducing the cash flow to the bond holder. The insurance premiums built into the loans covered prepayment due to default and the entire system was guaranteed with government credit. The added risk of the borrowers of an agency loan was therefore offset by the government's guarantee and the bond holder was generally assured that loan prepayment would be low.
Conventional mortgages, on the other hand required higher down-payments and were typically given by borrowers that had more access to capital (let's remember the borrowers give mortgages and banks give loans). Prepayment risk is generally higher via early payoffs or refinances, especially as interest rates tended downward. Fannie Mae and Freddie Mac, the largest purchasers of conventional mortgages, insured all mortgage collateral and the payments on their RMBS issues, in order to reduce this risk. The money from this insurance was both built into the price of the RMBS, but was also guaranteed by both organizations' access to a $2.5 billion credit line from the government. Principal Mortgage Insurance (PMI) was also employed in cases where the borrower's loan-to-value, LTV, exceeded 80%, making the loan more of a risk for default. PMI, however, only covered the difference between market value and mortgage value.
One of the most ironic aspects of the article is its discussion of private RMBS securitization, which it describes as a highly risky and generally below investment grade for a number of reasons. We know that this view of private securitization was largely ignored during the real estate bubble and that not only did private RMBS overtake the RMBS market, but also the resultant demand for mortgage notes led to the replacement of PMI. This practice traded insurance for increased exposure to risk from the same asset. Understanding the function of PMI, default rates and down payment requirements from the perspective of RMBS issuances, allows one to see why the once widespread practice of "piggyback loans" and "no money down" financing was a recipe for disaster. Not only did such lending practices dilute the relationship of the borrower to the property, making default much more likely, but they also eroded many of the insurance fail-safes of the private home mortgage lending system that insured the private RMBS payments to the investors in case of default. Additionally, piggyback loans, created two instances of default for a property instead of one, doubling the effect of default of each similarly financed property on the secondary market and on the related securities markets.
I could write a book about my reaction to Chinloy's article. I particularly like his analysis of mortgage payments and defaults as options, which lends borrowing behavior to derivative analysis. I also appreciate his in-depth explanation of forward and backward solving models for pricing RMBS. I am not sure which one I prefer, but I must say that I am somewhat partial to the type of analysis that the backward solving model employs. Though both types of models are useful tools, forward solving models are inherently more optimistic.  There is also a key point made on page 19 of the article regarding Fannie Mae's finding that properties with more than 10% negative equity have a high likelihood of default. Chinloy's mention of the lack of accounting for borrower liquidity in most RMBS pricing models is also noteworthy.
I do want to point out that some of the references in this article are dated. Due to mortgage acceleration clauses, new FHA, VA and FmHA mortgages are no longer assumable, therefore all of the sections about the assumablity of mortgages and selling the mortgage with the house are no longer relevant. It is also clear that Chinloy had no indication of the explosion of private and conventional loan and RMBS origination that would begin to take place just 3 or 4 years after the writing of his article and thus some of his predictions seem disconnected with what actually transpired in the real estate and securities markets. Chinloy, however, offers a cogent overview of the system of RMBS issuance that existed up until the late 1990's. It is clear that had this system been more clearly understood and followed by lenders, investment banks, investors and RMBS originators much of the calamity that recently befell our financial system could have either been predicted or avoided.