Killing the Element of Surprise in the Mortgage Market

Windows In Financial Services - June 2007

by Desmond MacRae

Harold Geneen, the famous CEO if ITT of the 1960s, believed that "ninety-nine percent of all surprises in business are negative." If something unexpected happened, he assumed that "...management had not anticipated or planned properly."

Michael Bykhovsky, CEO and president of Applied Financial Technology (AFT) headquartered in San Francisco with offices in Boston and New York agrees. "If the recent rise in defaults of sub-prime mortgages were unexpected by the companies that granted them, or by CMO (collateralized mortgage obligations) investors, someone was clearly not doing his job," he says.

Bykhovsky and the mortgage lenders and traders who use his technology have a theory about mortgages: "Everything in the mortgage market is fairly predictable - a function of interest rates, the HPI (home price appreciation index) or the HAI (housing affordability index)," Bykhovsky says.

The head of trading for the mortgage division of a well-established foreign bank in the US agrees. His firm began working with AFT because of Bykhovsky's experience with prepayment modeling. "Like most companies that deal in mortgages, we are very interested in prepayments," he says. "The prepayment models that AFT provides us are important to the success of our mortgage business which contributes significantly to earnings."

The trader notes his firm uses AFT to compare and check its own modeling efforts. "We outsource partly because it is cost efficient, but we also do our own modeling," the trader says. "By hiring AFT, we can compare our work with that of one of the leading firms in quantitative modeling."

Ample Opportunities

Predictability should be good news for mortgage issuers, investors and traders for three reasons. One is that the US mortgage market is now $10.2 trillion. Bykovsky estimates that some two-thirds of that total are mortgages for one to four-family homes. Compare this with 2006 US GDP of $12.9 trillion (on a purchasing power basis) for 2006. The US mortgage market offers many opportunities to make money.

Another is that some 70 percent of all on to four-family home mortgages have been CMOed, that is, gathered into collateralized mortgage obligations. There are financial instruments that allocate cash-flows and defaults from the underlying mortgages among different classes or tranches according to their respective priorities.

The largest issuers are the Federal Home Loan Mortgage Corporation (Freddie Mac), the Federal National Mortgage Association (Fannie Mae) and the Government National Mortgage Association (GNMA). There are many private issuers as well. Their CMOs are known as Private Label or Whole Loan CMOs.

CMO bonds generally have greater certainty in their expected maturities and principal pay-down windows than the underlying mortgages that prepay and amortize regularly over 30 years. Prioritization of credit exposure also allows mortgage investors the flexibility to decide how much credit risk they want to take. In effect, CMOs "regularize" mortgages into discrete securities similar to bonds. Thus, they can be traded easily. The third reason is that Bykhovsky has data and methods for demonstrating that the ups and downs of this market are predictable.

Two Questions

Mortgage debtors can make regular payments, prepay, or default. If they make regular payments, all is well for mortgage holders. Thus, only two questions need be asked.

What is the connection between a protection of interest rates, a projection of the home price appreciation rate, and the resulting prepayment response? And, what is the connection between a projection of interest rates, a projection of the home price appreciation rate and resulting default and delinquencies?

"Success with mortgages is about valuing, pricing, and controlling the outcomes you get from answering these two questions," Bykhovsky says. "With these answers, you can value any mortgage instruments," he adds, "and predict these value changes in the macroeconomic variables, that is the interest rates and the home price index."

Background

Originally from Russia, Bykhovsky graduated with a degree in physics from Columbia University. He worked on his Ph.D. at the European Organization for Nuclear Research (CERN), the world's largest particle physics center near Geneva.

"Later, I started doing some simulations of the US Superconducting Super Collider near Waxahachie, Texas," Bykhovsky recalls.

Fearing that it would not be completed (Congress canceled it in 1993), he went to Wall Street in 1988 where he did simulations in finance for several firms including money managers.

"In early 1987, if you worked past 7 p.m. they served dinner with caviar. By the time I got there, they served dinner without it," he recalls. "Things have gone downhill since."

Bykhovsky developed MBS (mortgage-backed securities) analytical systems at a broker/dealer and several money managers because there were none of sufficient quality commercially available. In 1996, he went out on his own to start AFT.

Prepayment Conditions

Borrowers will prepay when it is profitable. This can be when interest rates decline enough to refinance a mortgage at a more favorable rate, or when they sell their properties.

While interest rates are not predictable, AFT has shown that declines in rates will trigger rises in prepayments very predictably. Surprisingly, geography plays a fairly important role in refinancing behavior. Because of that, AFT calculates the geographic factors in metropolitan statistical areas (MSAs).

This is because mortgage holders' behavior differs according to where they live.

"There is a misconception that California loans are prepaid faster," Bykhovsky explains. "They aren't."

Comparing similar $300,000 mortgages in California and Wisconsin shows that when rates fall, Wisconsin mortgages will be refinanced much faster. This is because in Wisconsin, $300,000 is a large mortgage, but in California it is a small one. People higher up on the socioeconomic scale are more aware of falling interest rates, and tend to prepay faster.

Knowing refinancing responsiveness to interest rate incentives is important for investors because this knowledge will tell them when their current CMOs might be "called."

"We have calculated geographic factors at the state level, controlling for housing price geographic factors at the state level, controlling for housing price appreciation differences and loan characteristics," Bykhovsky says. AFT found that Wisconsin is the fastest state for prepayments followed by Illinois and Michigan. The slowest are Texas and New York because of certain tax issues. Despite popular assumptions, California is somewhere in the middle.

Predicting Defaults

AFT uses a home price appreciation index, sometimes referred to as HPA or HPI, which is also a factor in assessing different prepayment speeds and defaults.

Borrowers will default when property values fall or when they can no longer pay because of negative changes in their incomes. However, borrowers do not default as readily as they prepay, because by defaulting they lose places to live.

When HPA is rising, excessively aggressive originations and other mistakes that borrowers, lenders or appraisers make get washed out. Mistakes become more apparent when HPA falls. As the market turns down, housing inventories start to build, but HPA can lag. Appraisers look at recent comparable sales, but they don't necessarily take into account that house owners aren't selling at high levels anymore. Houses can be on the market for six months or eight months, so one doesn't know the HPA until houses begin to move at lower levels.

Bykhovsky compares this to a Wiley Coyote cartoon. "Wiley runs off the cliff, and for a while he hovers, pedaling in mid-air, then he crashes," Bykhovsky says. "Well, that's the housing market."

AFT has found that if borrowers' loans are close to or above market value of their houses, "...a fraction of people are going to stick the bank with the loan. Once the combined loan-to-value ratio (LTV) is above 0.95, the delinquency/default rates take off," Bykhovsky explains. "HPA is the driver for default-related behaviors," he continues. "In a low to negative HPA environment, which is what is happening now in many MSAs, and given loan rates resets experienced now by many borrowers, it is entirely expected that the default/delinquency rates have become what they are now. The recent increases in delinquencies are totally consistent with the pas experience, given these macroeconomic changes," Bykhovsky concludes. "This is particularly true with subprime mortgages."

Another important factor in mortgages' propensity to default is the housing affordability index (HAI) that was in place when the loan was granted. When HAI is high, the average buyer has to stretch to afford a house. Loans originated during high HAI have a greater propensity to become delinquent or default, which is now the case with loans originated in 2006 (when HAI was very height).

AFT is Microsoft-based and uses SQL Server 2000 for its database. "We can integrate our analytical tools on clients' machines, or ship data through other machines like Bloomberg for example," Bykhovsky says, "but Windows is the dominant operating environment." AFT can also create proprietary tools to customize models for specific clients as well.

Products can be divided into three broad categories. There are prepayment models for fixed, adjustable, prime and sub-prime mortgages, home equity loans and home equity lines of credit, manufactured housing and other similar loan types. There are default risk models that can be applied to risks in each of these categories. Finally, there are scoring and valuation models that track increases or decreases in prepayments or defaults as well.

With more than 100 clients including banks, hedge funds, investment houses and research groups, Michael Bykhovsky says Applied Financial Technology's clients have to pay only a fraction of what they would pay to a single trader. Because the two keys to profits and losses are prepayments and defaults, mortgage investors and traders should be happy to know that both are predictable.