On December 7, 2017, the Federal Reserve released three proposals that would increase the transparency of its stress test exercises, including the Dodd-Frank Act Stress Tests (“DFAST”) and Comprehensive Capital Analysis and Review (“CCAR”).  The proposals are comprised of: (1) enhancements to the Federal Reserve’s disclosures regarding its stress test models, (2) amendments to the Federal Reserve’s Policy Statement on the Scenario Design Framework, and (3) adoption of a new policy statement on the Federal Reserve’s approach to developing, implementing, and validating models.  Comments on the three proposals are due by January 22, 2018.

The Federal Reserve’s proposals represent a substantial step toward more transparency in stress testing, but would not provide for full disclosure of the agency’s models.  The preambles to the proposals suggest that the Federal Reserve is seeking to balance the benefits of additional disclosure, including increased public and market confidence in the stress test process, with concerns over the possibility that full disclosure would allow DFAST and CCAR participant firms to “game the system” by shifting their businesses to activities that appear to be advantaged under the models, which in turn could create systemic risk by leading to increased correlations among large firms’ asset holdings.  Each of the three proposals is discussed below.

Enhancements to Disclosures

The Federal Reserve would enhance disclosures regarding its stress test models by publishing:

• key equations and variables that characterize the models, such as the equations and variables underlying the models’ probability of default, loss-given-default, and exposure-at-default measurements for corporate loans;

• loss rates (mean, 25th percentile, and 75th percentile) for eight groups of loans defined by three characteristics: sector (financial or nonfinancial), security status (secured or unsecured), and rating class (investment grade or non-investment grade); and

• hypothetical portfolios of loans based on participating firms’ loan books, and loss rates on those portfolios estimated by the supervisory models, in order to illustrate the impact of the models.

The proposal includes an example of these enhanced disclosures for corporate loans.  The example shows that there would be a substantial increase in the length and granularity of the Federal Reserve’s model-related disclosures compared to past DFAST Methodology and Results documents.  The Federal Reserve expects that, if it implemented the proposed enhancements, it would publish the enhanced disclosures in the first quarter of each year, starting with selected loan portfolios in 2018.

Amendments to Policy Statement on Scenario Design Framework

The Federal Reserve’s Policy Statement on Scenario Design Framework for Stress Testing explains how the agency designs its baseline, adverse, and severely adverse scenarios used in DFAST and CCAR each year.  The proposed amendments to the Policy Statement would:

• provide notice that the Federal Reserve may include an increase in the cost of short-term wholesale funding for banking organizations in the adverse scenario (beginning in 2019 at the earliest) and severely adverse scenario (beginning in 2020 at the earliest);

• provide that, for stress tests conducted when the unemployment rate is already elevated, the severely adverse scenario may include an increase in the unemployment rate that is toward the bottom of the Federal Reserve’s 3 to 5 percent range for an increase, as a countercyclical measure; and

• establish explicit guidance for how the housing price index would change in the severely adverse scenario.

Policy Statement on Stress Test Models

The Federal Reserve’s new policy statement on developing, implementing, and validating models would include several notable positions, including that:

• where the Federal Reserve lacks data regarding the true risk of a portfolio, it will compensate by using more conservative assumptions, such as assuming a 10th percentile pre-provision net revenue rate or 90th percentile loss rate;

• because its models are forward-looking, the Federal Reserve may incorporate events or outcomes that have not yet occurred, and the macroeconomic scenario and global market shock component are in fact intended to introduce elements outside the realm of historical experience;

• the size of firms’ balance sheets will be assumed to stay the same or grow, even in a severely stressed environment, in order to “serve[] the Federal Reserve’s goal of helping to ensure that major financial firms remain sufficiently capitalized to accommodate credit demand in a severe downturn”;

• the Federal Reserve will generally phase in “highly material” changes to its models over a period of two years; and

• the Federal Reserve generally will not disclose to participant firms information related to stress testing that it does not disclose publicly.

The Federal Reserve has not previously discussed many of these positions in explicit terms.