The Challenges of Stress Testing US Structured Finance

Stress testing US structured finance portfolios presents a unique challenge – nowhere is tail-risk analysis more critical yet more difficult to complete. As we have witnessed over the past decade, structured finance transactions tend to carry myriad risks, therefore requiring complicated analyses. In response, banks tend to separate structured finance securities from less esoteric asset classes, both organizationally and analytically. However, when a bank conducts stress testing, it must consistently apply stresses to all its positions regardless of asset class.

Looking at a structured finance portfolio as a whole can yield useful generalizations around projected performance. For example, dropping home prices are on average going to negatively affect the credit risk of RMBS tranches. However, unlike corporate bonds, for example, it is difficult to know intuitively how a change in a given macroeconomic statistic will affect a single position. Depending on deal structure, it is possible that severe economic scenarios could improve the relative performance of some tranches and cause significant losses to others. Banks cannot determine the impact on structured finance tranches without running the cash flows on the underlying properties and loans and then passing those cash flows through the deal’s waterfall. And yet, running the cash flows opens up a new set of problems, including challenges in maintaining quality data and building the underlying asset models.

Using a consistent method to stress test across asset classes implies the ability to reliably convert forecasts on a potentially large set of macroeconomic factors into performance projections on each of the bank’s positions. In the world of structured finance, this ideally means crafting projections at the underlying loan-level. The United States is one country, but each of its fifty states has unique laws and economic environments, which means granular data at the loan-level is critical. For RMBS, the state where each loan was issued has either judicial or non-judicial mortgage laws, determining how long foreclosure proceedings could last. Loan-level data can be frustratingly scarce, especially for certain structured finance asset classes like ABS, which contributes to a dearth of granular structured finance asset models.

Lower coverage for loan-level data makes it hard, if not sometimes impossible, to develop reliable account-level models in the first place. It also means that any successful stress testing model must simultaneously and consistently support alternate methodologies. As an example, consider a bank with whole loan mortgages and RMBS on its books. The whole loans may be stressed through an account-level asset model, whereas, due to weak reporting, some of the RMBS positions can only be analyzed through an aggregation model on the underlying collateral. Despite using separate models, stressed results between the whole loan and RMBS books must be consistent. Most often, missing loan-level data forces a pool-level analysis where historical performance of a given pool, its comparables, and aggregate industry and national metrics inform the projections. Mechanisms should be in place to reconcile results from the loan-level and pool-level models.

Complexities in structured finance models are not limited to the underlying assets. Many transactions include one or many swaps intended to protect against credit risk, basis risk, and so on. However, swaps themselves introduce counterparty risk. In so-called normal economic environments, counterparty risk can be overshadowed by credit risk and extension risk, as two examples, but it strongly came to the fore during the credit crisis when protective swaps failed to deliver in times of need. Indeed, counterparty risk tends to become problematic in particularly difficult economic environments, or tail-risk scenarios, which are precisely what stress testing is designed to address. Properly tracking counterparty risk within the context of structured finance securities is especially challenging given the lack of unique identifiers and standard reporting templates for derivative transactions in securitizations. Investors often need to scour performance reports and deal documents carefully to understand their counterparty exposure.

Even if a given bank has access to a model for stress testing that features consistent implementation of structured finance analysis, that bank cannot simply run the stress test once and move on. Stress testing is meant to be an ongoing process and, therefore, any competent stress testing solution must be streamlined and user-friendly. Furthermore, the platform must be extensible and diligently supported in order for the bank to keep up with the ever-changing regulatory environment. In cases where some banks hold thousands of structured finance positions, building an efficient and scalable technology infrastructure to run a variety of stress tests in a consistent and timely manner is a challenge that must be addressed.

Stress testing with a mixed portfolio that includes structured finance securities can be a daunting task. From complicated legal structures and non-standard reporting of underlying collateral to properly incorporating macroeconomic factors, some banks may struggle to convince regulators that their structured finance testing is up to the same standard as the stress testing on their more vanilla positions. This is why it is critical to leverage a platform that provides cohesion across asset classes, strong fundamental analysis, consistent assumptions and model design, and ongoing support. Consistency across all portfolio assets is imperative to stress testing best practices.