2014 Federal Reserve Stress Testing Research Conference
This Event Has Ended
Framework: The Capital and Loss Assessment under Stress Scenarios (CLASS) Model
The CLASS model is a top‐down capital stress testing framework that projects the effect of different macroeconomic scenarios on U.S. banking firms. The model is based on simple econometric models estimated using public data and also on assumptions about loan loss provisioning, taxes, asset growth, and other factors. We use this framework to calculate a projected industry capital gap relative to a target ratio at different points in time under a common stressful macroeconomic scenario. This estimated capital gap began rising four years before the financial crisis and peaked at the end of 2008. The gap has since fallen sharply and is now significantly below precrisis levels. In the cross-section, firms projected to be most sensitive to macroeconomic conditions have higher capital ratios, consistent with a "precautionary" view of bank capital.
capital, stress testing
G21, G17, G01
Framework: Dual Stress Testing and Run Vulnerability
U.S. supervisory stress testing of the banking system has to date focused on capital adequacy. But bank failures and systemic problems in the banking system are often due to a loss of short-term funding. In practice, liquidity and capital problems are closely related, in the sense that flights of liquidity are often triggered by losses or undercapitalization. This proposed presentation sets up a conceptual framework for conducting dual stress tests of capital and liquidity. It also presents projections using the approach for large U.S. bank holding companies.
Framework: Are Banks Forward Looking in Their Loan Loss Provisioning? Evidence from Senior Loan Officer Opinion Survey (SLOOS)
Our paper makes a fundamental contribution by studying loan loss provisioning over the credit cycle as three distinct phases. Looking at the three distinct phases of the financial crisis – the precrisis period, crisis period, and post crisis period – is important as loan loss provisioning is driven by different factors in each, in part due to extensive shifts in (or in the application of) regulatory rule. We show evidence of forward-looking loan loss provisioning by utilizing Senior Loan Officer Opinion Surveys (SLOOS) which provide useful controls for credit cycle information. Though the SLOOS dataset is a restricted sample and generalizability to a broader sample could potentially be a stretch, we control for credit cycle factors as part of an identification strategy to sort out changes in the credit market equilibrium. We contribute to the growing literature on forward-looking loan loss provisioning and early in the cycle loss recognition by incorporating a broader range of available credit information.
Loan loss provisioning, forward-looking, income smoothing, capital management, early loss recognition
Framework: Stress Testing and Portfolio Granularity
We explore the question of what is the appropriate level of portfolio granularity for a stress testing model when the design of the stress test is based on fluctuations in observable economic variables. We conduct model comparisons of loan-level and portfolio-level models of default probabilities, using as an example home equity loans and lines of credit in the Los Angeles county. We also consider hybrid approaches where we aggregate data to sub-portfolios formed by quantiles of LTV, FICO score, or DTI distributions. We evaluate model performance in terms of in- and out-of-sample forecast accuracy and ability to capture the impact of macroeconomic variables, which is important for scenario-based stress testing. Our goal is to evaluate the optimal level of aggregation to achieve both these goals.
Scenario Design: Choosing Stress Scenarios for Systemic Risk Through Dimension Reduction
This paper proposes a methodology for choosing one or a small number of stress-scenarios with the property that if the banking system is well capitalized against these scenarios it will be well capitalized in a systemic risk sense against a wide set of other plausible scenarios that could affect the banking system. The scenarios are chosen by using dimension reduction techniques that create factors (linear combinations of variables) based on their ability to explain measures of systemic risk. If the banking system is well capitalized against the movements of these factors, then it is well capitalized against a wide set of scenarios that generate systemic risk. In regulatory stress-testing practice the methodology can be used as a challenger methodology to examine whether current regulatory stress-tests miss directions of risk-taking that the method identifies as important. In addition, as the methodology is refined variations on the methodology have the potential to blended in with or replace current regulatory methods for generating stress scenarios.
Stress Testing, Financial Stability, Banking
Scenario Design: Constructing Yield Curves from Macro Scenarios of Bank Stress Tests
We describe how arbitrage-free Nelson-Siegel (AFNS) term structure models are used to generate a variety of yield curves consistent with the macroeconomic scenarios in the Federal Reserve's annual bank stress tests required under the Dodd-Frank Act.
Scenario Design: Robust Stress Testing
Thomas Tallarini Jr.
In recent years, stress testing has become an important component of financial and macroprudential regulation. Despite the general consensus that such testing has been useful in many dimensions, the techniques of stress testing are still being honed and debated. This paper contributes to this debate in proposing the use of robust forecasting analysis to identify and construct adverse scenarios that are naturally interpretable as stress tests. These scenarios emerge from a particular pessimistic twist to a benchmark forecasting model, referred to as a 'worst case distribution'. Importantly, this approach allows for the enormous Knightian uncertainty faced by regulators and market participants who need a method of identifying vulnerabilities, even while acknowledging that their models are misspecified in possibly unknown (or unknowable) ways.
We first carry out our analysis in the familiar Linear-Quadratic framework of Hansen and Sargent (2008), based on an estimated VAR for the economy and linear regressions of bank performance on the state of the economy. We note, however that the worst case so constructed features undesirable properties for our purpose in that it distorts moments that we would prefer were left undistorted. In response, we formulate a finite horizon robust forecasting problem in which the worst case distribution is required to respect certain moment conditions. In this framework, we are able to allow for rich nonlinearities in the benchmark process and more general loss functions than in the L-Q setup, thereby bring our approach closer to applied use.
Risk Modeling: Banking Sector Operational Losses and Macroeconomic Environment
Rob T. Stewart
We investigate the relationship between operational losses at large U.S. banking organizations and the macroeconomic environment. We find evidence of a negative relationship between macroeconomic growth and operational losses in two Basel II loss event type categories: Clients, Products, and Business Practices and Execution, Delivery, and Product Management. Our analysis suggests that the negative correlation of losses in these two categories with macroeconomic growth is concentrated in the tails of loss distributions. We also find evidence of the negative correlation of the tail losses in External Fraud event type category with macroeconomic growth, while this correlation is positive for smaller losses in this category. Losses in these three event type categories comprise about 93 percent of the total industry losses in our sample. We demonstrate how our findings can be used in practice for modeling operational losses conditional on macroeconomic growth.
Risk modeling; Operational risk; Basel II; Stress testing;
C22, C23, G21
Risk Modeling: From Originations to Renegotiation: A Comparison of Portfolio and Securitized CRE Loans
This paper uses unique bank supervisory data to compare portfolio and securitized commercial real estate loans. We document how the types of loans banks hold in portfolio differ substantially from the types of loans the same banks sell. Banks tend to hold loans that are "non-standard" in some observable dimension. Among the portfolio and securitized loans of similar type, we analyze loan distress and renegotiation. We find that bank loans are riskier than securitized loans and banks are more likely to extend loans in distress. Our results suggest that banks have a comparative advantage in funding risky assets with contracts that may need to be renegotiated.
Securitization, CMBS, Commercial banks, Asymmetric information, Renegotiation
G14, G21, D82, G33
Risk Modeling: The End of the Line: Behavior of Borrowing Constrained HELOC Borrowers
Many existing home equity lines of credit (HELOCs) are structured such that when they reach the end of the draw period, they convert from open-ended, non-amortizing lines of credit to closed-end, amortizing loans. Given that a large share of outstanding HELOCs will reach end of draw in the next several years, an important question is whether the increase in required payments that accompany the end of draw will affect default rates. We estimate a competing hazard model of default and payoff for HELOCs, and use it to predict outcomes in 2012 and 2013. We find that the model does a good job of predicting default for HELOCs that do not reach end of draw during the examination period, but it significantly underestimates default for HELOCs that do reach end of draw. The underestimate is greater for HELOCs originated to borrowers with lower credit scores that have high combined loan to value ratios as they approach end of draw. The underestimate for these "low quality" HELOCs is particularly large for HELOCs with balloon payments at the end of draw.
home equity, HELOC, end of draw, consumer credit, payment changes
G21, R31, D14
Risk Modeling: Stress Testing Interest Rate Risk Exposure
In the current low interest-rate environment, the possibility of a sudden increase in rates is a potentially serious threat to financial stability. As a result, analyzing interest rate risk (IRR) is critical for financial institutions and supervisory agencies. We propose a new method for generating yield curve scenarios for stress testing banks' exposure to IRR based on the Nelson-Siegel (1987) yield-curve model. We show that our method produces yield curve scenarios with a wider variety of slopes and shapes than the historical scenarios and hypothetical scenarios typically used in stress testing. We stress test the economic value of equity of several hypothetical bank balance sheets and show that our method provides more information about banks' exposure to IRR using fewer yield curve scenarios than the historical and hypothetical methods.
Bank, Interest Rate Risk, Stress Testing, Scenario Generation, Nelson-Siegel Model
E47, G21, G28