Uncovering the flaws in bank stress testing

Faulty tests, misleading results

Bank stress tests are critical evaluations developed and conducted by regulatory bodies to ascertain the level of resilience and robustness of financial institutions under hypothetical adverse economic scenarios. They are meant to simulate severe economic events and harsh market conditions, such as unemployment spikes or heightened market volatility, to determine if banks have sufficient capital to withstand them.

In the case of the U.S., the Fed’s annual stress test is only conducted on larger, or “systemically important,” banks with assets over $100 billion. The test conditions and hypothetical scenarios can vary somewhat. However, through the years, critics have pointed to many shortcomings and biases in the way the tests are designed and executed.

Most of those objections revolve around the idea that the scenarios are often overly optimistic and fail to capture extreme but plausible risks, which in turn results in an underestimation of potential losses. This can be extremely dangerous, as the stress test results can provide a false sense of security and lead to mass under-preparedness and complacency on the part of both the banks and the public. That makes serious banking crises and subsequent bailouts more likely, putting average households and businesses in a very precarious position.

This last stress test round in the U.S. was no different. Tomasz Piskorski, professor at Columbia University’s business school, said the Fed’s stress test “underestimated risks in the banking sector” such as those that caused Silicon Valley Bank (SVB) and First Republic to collapse last year: “If the guidelines of this year’s stress tests were based on Silicon Valley Bank’s balance sheet on Dec. 31, 2022 – three months before it went bust – I would not be surprised if it would have been deemed as ‘fine.’”

SVB’s collapse was triggered by the higher interest rate environment that forced the bank to sell some of its longer-term treasuries at a considerable loss to cover increased withdrawals. However, the latest stress test scenarios failed to properly account for similar risks going forward.

The very fact that the U.S. stress tests only involve banks that are “too big to fail” renders the results meaningless as a gauge of resilience for the entire banking sector. Case in point: This year’s stress tests showed that banks are in a position to withstand a recession and especially commercial real estate risks, the latter being a major risk to the American economy that has been widely recognized for over two years. However, this seemingly reassuring test result is entirely misleading, since it is smaller, regional U.S. banks that have by far the largest exposure to commercial real estate, and none of them underwent any tests.

Risk-weighted assets

The 2008 global financial crisis, which gave birth to early stress tests, was caused by banks investing in subprime loans that carried far higher risks than bank managers and regulators admitted at the time, or even thought possible. The Basel III set of regulations was then proposed as a global framework to ensure that this would not happen again, by forcing banks to group their assets under specific risk categories that come with different capital requirements.

As a recent Bloomberg analysis pointed out, this created another glaring problem with how the stress tests are designed:

Then there’s the question of what it means to have adequate capital. Regulators currently set minimum levels for the largest U.S. banks that vary by bank; last year’s ranged from 7 percent to 13.8 percent, which sounds robust. Yet the ratio is calculated as a percentage of the bank’s ‘risk-weighted assets,’ a measure that treats entire categories of assets as having no risk – and therefore needing no capital backing at all.

Under Basel III, U.S. government debt and securities are given a risk weight of zero percent. This seemingly arbitrary and axiomatic rule has far-reaching implications, as it clearly has a political aspect built in, incentivizing banks to lend to the government.

A general misunderstanding and misrepresentation of risk

There is an overarching theme in this debate and in all the arguments against the current stress test philosophy. At their core, stress tests seem to rest upon a misguided and dangerous misconception about the very nature of risk. Just because something is seen as stable does not mean it is safe. A whole lot of things are stable, until they are not. Nassim Taleb put it best: A turkey that is fed every day and kept healthy and comfortable enjoys a “stable” life; until Thanksgiving, that is.

Bill Blain of Windshift Capital has a more practical example:

History teaches house prices rise steadily over time. Until suddenly something happens and the housing market tumbles 50 percent. Every homeowner is then busted by negative equity, which means they can’t borrow to pay off their other loans, which means their bank goes bust on credit card losses, triggering an unravel of synthetic risk transfer trades, a collapse in confidence in markets followed by a bunch of funds admitting they were 12 times levered and are now broke, neatly collapsing the rest of the banking and insurance sector, leading to global mayhem and destruction.

 

This inability to differentiate between volatility and real risk renders stress testing highly unreliable. As mentioned above, the notion that government bonds with low volatility are “risk-free” is beyond absurd. Imagine we were talking about an over-indebted company instead of an over-indebted government that was issuing debt for which there is no collateral and which is denominated in a currency that is also issued by said company, allowing it to simply print the money it owes. In this context, this debt would probably be seen as the riskiest asset in the market. Even if it was stable for some time, it would surely be seen as a ticking time bomb, getting riskier and riskier as it grew.

This report was originally published here: https://www.gisreportsonline.com/r/banks-stress-test/