Retained by Davis Polk
In China Development Industrial Bank v. Morgan Stanley & Co. Incorporated, et al. (N.Y. Sup. No. 650957/10), plaintiff China Development Industrial Bank (“CDIB”) entered into a $275 million credit default swap with Morgan Stanley on the super senior tranche of the STACK 2006-1 Collateralized Debt Obligation (“CDO”) (“STACK”). CDIB brought claims against Morgan Stanley for common law fraud, fraudulent inducement, and fraudulent concealment.
CDIB also alleged that Morgan Stanley had unique access to nonpublic information that led it to conclude that investments tied to the U.S. subprime residential real estate mortgage bonds would lead to extensive losses, leading it to transfer the risk associated with its allegedly “bad investment” in U.S. subprime mortgage bonds to other investors, such as CDIB, through the swap. CDIB further alleged that the CDO was “built upon fraudulent credit ratings” and that Morgan Stanley created and structured the CDO with knowledge that it was a “toxic and unsafe” investment.
Vega expert Dr. Ethan Cohen-Cole was retained by Morgan Stanley, through its counsel Davis Polk & Wardwell LLP, to respond to the reports submitted by Plaintiff’s experts. More specifically, he provided opinions regarding:
- The risk disclosed to CDIB in the offering documents and other disclosures
- Whether the difference in loan-level characteristics alleged by plaintiff's expert for the loans underlying the RMBS in the CDO would have materially affected the risk of the investment
- The house price appreciation and income overstatement analyses put forth in an opposing expert report
- The rating analysis performed in another opposing expert report.
Risk Disclosed in CDO Offering Materials
CDIB and Morgan Stanley entered into the swap agreement more than eight months after the STACK CDO closed in July 2006. Prior to purchase, CDIB was provided with the offering memorandum, a memo describing the deal structure. See Figure: STACK CDO Structure for the structure details outlined in the memorandum.
Figure: STACK CDO Structure
CDIB also received initial marketing materials, a break-even analysis, a list of the then-current collateral in the portfolio, and draft letters from Moody’s and S&P about the expected ratings of the swap.
Dr. Cohen-Cole discussed that a reasonable investor could analyze the expected performance of the collateral contained within the CDO at that time and would understand that the collateral could change. The offering memorandum disclosed that a significant portion of the collateral assets held by STACK could be subordinate tranches of securities, and therefore would have a higher risk of loss as a result of delinquencies or losses on the underlying assets. While an investor could expect some protection against losses within the underlying collateral, it was understood within the market that senior investments in CDOs were not risk-free.
Materiality of Loan-Level and HPA-Related Allegations
To test the materiality of the loan-level allegations, Dr. Cohen-Cole used a model similar to models used by RMBS investors to measure the impact of the opposing expert's loan-level allegations on a reasonable investor's expectation of risk. The model considered the risk of default of the loans collateralizing the assets within the STACK CDO and evaluated the materiality of opposing expert's allegations with respect to the risk disclosed to Plaintiff.
Dr. Cohen-Cole ran three tests. First, he tested whether the loan characteristics alleged by the opposing expert, assumed to be true, would have resulted in levels of risk above those expected by a reasonable investor. Second, he re-ran the same test, but incorporated House Price Appreciation (“HPA”). Finally, Dr. Cohen-Cole tested whether the loan characteristics alleged by the opposing expert would have led to greater risk than would have been expected for a managed CDO based on the offering memorandum disclosures. Dr. Cohen-Cole concluded that even if one accepted the allegations in the opposing expert report as true, the expected risk of the loans collateralizing the RMBS assets in STACK fell below the maximum risk an investor would have expected. Therefore, it would not have had a material impact on investors’ decisions regarding whether to invest in STACK.
Dr. Cohen-Cole also noted that the STACK CDO had no restrictions on the loan characteristics that could underlie the RMBS within its portfolio. He discussed the fact that the STACK was a managed CDO and the collateral manager was permitted to make changes to the collateral, which meant that the loan characteristics of the underlying RMBS collateral could change at any time.
Dr. Cohen-Cole finally revealed other flaws in the expert’s conclusions regarding HPA. For example, the opposing expert incorrectly used extrapolated data regarding the relationship between default losses and HPA. Specifically, the opposing expert failed to use a reliable method of extrapolation based on the actual historical relationship between HPA, collateral loss, and bond loss and instead simply drew a line to represent data when the HPA dropped below 2 percent. The expert did not confirm that this pattern was supported by the data and instead assumed a linear relationship.
By analyzing historical data, Dr. Cohen-Cole found that the data strongly contradicted the linear extrapolation performed by the opposing expert. Using similar publicly available data from the same date range, he reproduced the graph with four-year cumulative collateral loss observations for local HPA values below two percent added. The data points are dramatically different from the opposing expert’s extrapolated points.
Income Overstatement Analysis
Dr. Cohen-Cole evaluated whether the opposing expert correctly concluded that Morgan Stanley had unique knowledge of income overstatement. He found that the analysis in the opposing expert report did not provide any evidence of such knowledge, let alone any evidence of income overstatement.
The opposing expert opined that borrower income may have been overstated for loans underlying STACK. The expert compared average income data collected from mortgage applications, as reported according to the Home Mortgage Disclosure Act (“HMDA”), with average income reported by the Bureau of Economic Analysis (“BEA”) and the Internal Revenue Service (“IRS”).
The expert claimed that from 1996-2001, HMDA and BEA data showed largely similar rates of growth. To test this claim, Dr. Cohen-Cole performed a linear regression analysis with HMDA growth rate as the dependent variable and BEA income growth as the explanatory variable with data from 1996-2001. Had the relationship been one-to-one, the constant term in the regression would have been zero and the slope would have been one. Instead, Dr. Cohen-Cole found that the constant term was greater than zero and the slope was less than one.
Ultimately, the analysis showed that only .04 percent of the loans underlying the RMBS in STACK came from counties where income growth was above the upper bound of the confidence interval. This is in sharp contrast to the opposing expert’s claim, which used HMDA and BEA data to make income overstatement allegations for 64.6 percent of loans in STACK. Therefore, even if one compared HMDA and BEA income growth rates, the opposing expert report provided no evidence of income overstatement.
CDO Rating Analysis
Dr. Cohen-Cole performed three rating analyses to determine whether notching the collateral in STACK consistent with industry practice would affect the ratings of STACK’s super senior swap.
First, Dr. Cohen-Cole performed a rating analysis to test whether notching the tranches using the stress scenario proposed by Moody’s, would affect the rating of STACK’s super senior swap. To do so, he notched each tranche rated Baa3 and classified as subprime according to Moody’s definition by the maximum number of notches used in the stress test, five notches. He then ran the S&P CDO rating model—which was the same model used by the opposing expert. Dr. Cohen-Cole found that, when notching the subprime collateral in STACK, the rating for the super senior swap was unaffected.
Second, Dr. Cohen-Cole performed an analysis to test whether aggressively notching this collateral would have affected the rating of the super senior swap. To do so, he notched each of the 16 tranches held by STACK that were placed on negative watch in 2007 downward until it received a CCC rating, as this was the harshest possible future rating action announced by S&P. He found that, even notching all tranches in STACK placed on negative watch in July 2007 until they received a CCC rating, the rating of the super senior swap of STACK was unaffected.
Finally, based on a source cited in the opposing expert report, Dr. Cohen-Cole tested whether downgrading each RMBS and CDO tranche held by STACK by two notches would affect the rating of the swap. This was double the notches suggested and corresponded to the maximum number of notch downgrades proposed in July 2007, a full three months after CDIB’s purchase. When he ran the S&P CDO rating model using this notching and the opposing expert’s other inputs, the rating of CDIB’s tranche was unaffected. This analysis demonstrated that, even three months after purchase, the ratings of the swap would have been unaffected even by a notching schedule that was more aggressive than the one actually proposed by S&P.
These three analyses demonstrate that the rating of the super senior swap of STACK would have been unaffected when using notching methodologies consistent with industry practice.