Applying DFAST and CCAR Scenarios Across Asset Classes
In this article, we execute the three Fed scenarios on a sample of non-agency RMBS to demonstrate how to apply one approach to both the securitized tranches and the underlying residential mortgage portfolios collateralizing the securitizations.
Calculating stressed losses on structured finance portfolios to comply with DFAST or CCAR can be challenging for many financial institutions. Consistency between the underlying asset and whole loan portfolio analyses is critical, and yet few stress testing solutions in the marketplace offer a singular solution for all asset classes.
When reviewing potential losses under the Federal Reserve (Fed) macroeconomic scenarios, it helps to have a sense of what to expect. Table 1 demonstrates the average discounted tranche loss across the three Fed scenarios through the life of the transaction. Table 2 shows the average projected loss on the underlying mortgages from the Fed’s CCAR report, as well as the results from Moody’s Analytics.
The numbers used in this exercise are based on a sample portfolio of 583 US non-agency residential mortgage backed securities (RMBS) tranches. In order to run the three Fed scenarios, this methodology leverages a top-down modeling framework, which enables macroeconomic assumptions from Moody’s Analytics economists to automatically filter down into loan-level credit model projections. These loan-level cash flows for the RMBS transaction are then allocated to the tranches based on the legal structure (the waterfall).
Table 1. Sample US RMBS Portfolio
Source: Moody's Analytics
Table 2. Underlying mortgages from the Fed’s CCAR report
Source: Moody's Analytics
Based on these figures, the findings indicate that for the Fed scenarios, the senior RMBS notes would lose, on average, about 35%, while the mezzanine notes would lose around 70%. The incremental loss between the Baseline and Severe scenarios is roughly 6% to 8% on average for each of the Senior and Mezzanine tranches, respectively. We found that the underlying mortgage pools would suffer around 7% loss on currently performing mortgages. Given that RMBS pools tend to hold primarily first-liens, this estimate is not much different than the Fed’s projections. As we would expect, a 7% loss on the underlying collateral translates to a lower loss for the senior notes relative to the mezzanine notes.
While averages can be useful for thinking about a portfolio, individual tranche results may diverge significantly. Table 3 breaks down the projected tranche losses into quartiles.
The variance in tranche loss is non-trivial. While the average senior note loss is about 35%, around a quarter of the senior notes in the sample lose more than 50% and another quarter lose less than around 15%. As expected, the mezzanine notes are even more volatile. Even under the Baseline Fed scenario, a quarter of the mezzanines lose almost 100%, while another quarter lose less than 35% – a smaller loss than the overall average for the senior notes. Additionally, for all three scenarios and for all tranche categories, the maximum loss is 100% and the minimum is 0%. Anything can happen given the specific performance of each deal.
Figure 1 reinforces the high variance by showing the distribution of projected tranche losses by seniority. Both senior and mezzanine notes have bar-belled distributions but the senior notes have a higher concentration in the lower losses.
Table 3. Sample US RMBS Portfolio
Source: Moody's Analytics
Figure 1. Distribution of projected tranche loss
Source: Moody's Analytics
Table 4 breaks down the sample portfolio results by asset class, which highlights the variance in the individual tranche results. The overall average loss for senior notes is around 35% but there’s a large dispersion based on asset class – under the Baseline Fed scenario, the average Prime tranche loss is 28% while the average Option ARM loss is 70%. It is interesting to note, too, that Subprime losses in the sample are actually among the lowest for the senior tranches. Despite having lower-quality underlying collateral, in these cases the senior notes tend to have more effective subordination due to safer deal structure to offset the riskier assets and higher current credit support.
Table 4. Sample US RMBS Portfolio
Source: Moody's Analytics
Featured Experts
Cristian deRitis
Leading economist; recognized authority and commentator on personal finance and credit, U.S. housing, economic trends and policy implications; innovator in econometric and credit modeling techniques.
Juan Manuel Licari
Dr. Juan M. Licari is a managing director at Moody's Analytics. Juan and team-members are responsible for the research and analytics that enable our quantitative solutions. The team helps our customers solve complex business problems; adding value through data and analytics.
James Partridge
Credit analytics expert helping clients understand, develop, and implement credit models for origination, monitoring, and regulatory reporting.
As Published In:
Explores how North American financial institutions can leverage stress testing regulations to add value to their business, for compliance and beyond.
Related Articles
The Challenges of Stress Testing US Structured Finance
Stress testing structured finance transactions presents unique challenges due to large and diverse portfolios of underlying assets, limitations on data availability, and the idiosyncrasies and complexities of the structures and associated risks.
Challenges of Stress Testing Structured Finance in Europe
This article addresses some of the challenges banks face in stressing their structured finance positions within the context of a larger enterprise-wide stress testing exercise.
CLO Deal Ranking - Third Quarter 2011
Moody's Analytics recently reviewed our database of CLO deal performance statistics to assess relative performance across CLOs in each of the 2003 through 2007 vintages. 453 U.S. and 194 European CLO transactions were reviewed in our study and the top five deals in each vintage were selected. The ranking is presented below. While any relative performance ranking is subjective in terms of deal inclusion, parameter selection and weights, and other considerations, the underlying data used in our study is available for readers to perform their own analysis.