A visit to the halls of academe on a rainy spring day found the research powerhouse team of John Hull and Alan White enthusiastic and forthcoming as they discussed their recent research and its applicability to the post-meltdown financial landscape. In 2011, the two professors won the Financial Analysts Journal (FAJ) Graham and Dodd Scroll Award, and John Hull won the Risk magazine award for lifetime achievement. We took the opportunity to hear some of their many insights on securitization practices, counter-party and wrong-way risk, and setting the swap discount curve.

The Hull and White award-winning paper, published in the Sept/Oct 2010 issue of FAJ, examined the reasonableness of the ratings assigned to U.S. residential mortgage-backed securitizations (RMBS) from 1997-2007. Using two types of copula models and two criteria for deciding loss, the authors determined that the rating agencies’ results were reasonable for RMBS AAA tranches. However, this was not the case with re-securitizations. When the RMBS BBB tranches were re-packaged into asset backed securities (ABS) in the form of collateralized debt obligations (CDOs), the AAA ratings assigned to tranches of mezzanine ABS CDOs could not be justified. “Bonds on average have a recovery rate of about 40 percent,” Hull pointed out, whereas “recovery rates for mezzanine ABS CDO tranches are much lower.”

“Re-securitization was a flawed idea,” said Hull. “The problem is the loss distributions on tranches are very different than on corporate bonds, so a BBB tranche should not be treated in the same way as a BBB bond when a re-securitization is evaluated,” said Hull. “If you need diversification, then get it in the first pool,” suggests White, referring to the original securitization portfolio. Another way to achieve diversification could be through asset classes, for example, combining car loans and residential mortgages in a single securitization.

Canadian mortgage lending was considered to be more conservative in its practices and had some differences from its U.S. counterparts. The non-recourse nature of U.S. mortgages may have contributed to the spike in RMBS defaults when the underlying collateral value dropped significantly. Noting that the non-recourse provision is a “peculiar property,” White said it is not a feature of Canadian mortgages, except in Alberta.

Hull and White said the work on their 2010 paper encouraged them to write two companion papers. The first of these was presented at a Russell Sage Foundation conference in New York in April 2012, entitled Rethinking Finance. Hull and White observed ways in which investors might be misled into thinking that securitization created value. Standard & Poor’s and Fitch use a probability of loss model, which is not additive, whereas Moody’s expected loss model is additive. “We draw a parallel between this and Value-at-Risk (VaR),” said Hull. “If you put a lot of portfolios together, the expected loss adds up, whereas probability of loss doesn’t have that additive characteristic, which is where the arbitrage comes in.”

The second companion paper reviews in more detail the ratings arbitrage associated with structured products. Ratings are meant to be a measure of risk, but some investors use credit ratings as a basis for valuing products. It stands to reason that any ratings criterion should be arbitrage-free. Such was not the case for a criterion used by S&P and Fitch. Since probability of loss is non-additive, “the more you slice and dice, the larger the apparent creation of value,” said Hull.

Over the years, the research of Hull and White has been emphasizing applied studies, the issues of which are of immediate concern to the financial sector. Their work on modelling interest-rate derivative pricings in the late 1980s and their teachings have trained a generation of financial practitioners. In the 1990s, they researched the modelling of the burgeoning credit default swaps market. When asked about the reason behind their long, successful, and productive collaboration, Hull said, “We are both interested in the same topics. Getting feedback directly from practitioners, as we do, makes the research very engaging.”

Hull and White spoke enthusiastically about their two latest research projects. In a paper to be published in a forthcoming FAJ issue, they review counter-party credit risk and wrong-way risk. The expected loss from a counter-party default is measured by what is termed credit value adjustment (CVA). The usual practice is to assume that the probability of default and the exposure at default (EAD) are independent. Wrong-way risk occurs when there is positive dependence between the two so that probability of default during a future time period increases as the exposure during the time period increases. AIG and monolines, because they sold credit protection, provide examples of wrong-way risk. “We show how a relationship between probability of default and factors influencing the exposure can be developed,” said Hull.

Finally, they spoke about a forthcoming study they have conducted to investigate the discount rates that should be used for valuing swaps and other derivatives. Market practice is to use the overnight indexed swap (OIS) rate for collateralized portfolios and the LIBOR/swap rate for non-collateralized portfolios. “Usually, there’s a 10 to 20 bps spread between the two, but at the height of the crisis, the spread rose to 350 bps,” said White. The Hull and White research shows that the OIS rate should be used for all derivatives whether they are fully collateralized, or partially collateralized, or totally non-collateralized. Credit risk should be taken into account with credit risk adjustments that are now standard (i.e., credit value adjustment and debt value adjustment).

“We expect controversy with this one,” predicted Hull. The pair smiled and looked ready to take on another skirmish in the field of mathematical finance.