Financial stress test scenarios improve when risks are combined, study finds
UT Austin researchers say models that mix market shocks can better identify banks’ worst-case losses than familiar crisis scenarios.
By Lucas Ferreira · Science & Environment Writer
3 min read
Financial stress test scenarios may do a better job of finding weak points in banks when they combine several kinds of market pressure at once, according to new research from the University of Texas at Austin. The work matters for regulators because stress tests help judge whether large financial firms could keep lending through a severe shock.
The paper, by Rui Gao and Stathis Tompaidis of UT Austin’s McCombs School of Business and Rohit Arora, a McCombs doctoral graduate, was published in Management Science. The researchers developed models for choosing test scenarios that expose severe losses without relying mainly on famous crisis days.
The Federal Reserve Board said in a recent stress test that large banks could lose $708 billion in a severe recession and still remain financially sound and able to lend. The UT Austin team’s research addresses a related question: how regulators should pick the hypothetical shocks used in those tests.
How do financial stress test scenarios work?
A financial stress test asks an institution to estimate how its balance sheet, profits and losses would respond under difficult market conditions. Regulators use the results to assess whether firms could withstand shocks such as recessions, market swings or other financial disruptions.
Tompaidis said the idea resembles old “proof tests” used by gunsmiths, who fired barrels with heavy loads to see whether they would fail. In finance, the load is a set of assumptions about markets, and the result is an estimate of potential damage.
Designing the tests is hard, Tompaidis said, because regulators need scenarios that are demanding and consistent without giving firms enough information to tailor portfolios around a narrow exam. Running the tests also takes time and money, so the scenarios should focus on outcomes that are most likely to reveal serious losses.
Why combine several market risks?
Many stress tests use historic market crashes or hypothetical events modeled on them. On those days, stocks, interest rates, currencies, commodities and volatility may shift together in recognizable patterns.
Gao said profit and loss can respond to risk factors in complicated ways. That means several well-known crisis days may test the same vulnerability, while missing other combinations that could hurt an institution more.
The researchers looked for scenarios that were both severe and different from one another. Tompaidis said the goal was to find days where pressures complemented each other, rather than just the most dramatic market moves.
The team tested its approach against stress tests designed by the Commodity Futures Trading Commission. Using 2,828 historical market scenarios from April 2008 through June 2019, the models selected four scenarios with complementary stress factors.
Across 1,000 simulated portfolios, the researchers’ scenario set found the single worst historical outcome about 40% of the time. Their most accurate model identified the five worst outcomes about 95% of the time, according to the university’s summary of the research.
The researchers reported that their selected scenarios captured more severe losses than the CFTC’s baseline scenarios. Tompaidis said the findings suggest regulators may be able to model financial risk more effectively by combining stressors rather than centering tests on familiar crisis episodes.
The approach could also reduce the number of stress tests needed, Tompaidis said, because a small set of complex scenarios may capture the worst cases. Gao said the researchers want to help regulators evaluate risk more accurately before the next crisis.
This story draws on original reporting from Phys.org.