Perspective on Risk - Dec. 10, 2024

Why Banks Fail; KYC Breakdown at MS; 2024 FSOC Annual Report; and the Evolution Thereof; Deep, Deep Tail Risk; Lots More

Charity Request

If you’ve been a reader for a while, you know that I volunteer twice a week at Helping Hands Rescue Mission, my local food pantry.

This blog is of course free. If you or your company value this writing, consider throwing HHRM a few bucks. Here’s a shortcut to the donation page, or simply Zelle or Venmo them a few bucks. Be generous. Thank you.

NY Fed On “Why Banks Fail”

Liberty Street, the NY Fed’s blog, recently posted three pieces on the above topic, summarizing research published in a Staff Report Failing Banks.

• Why Do Banks Fail? Three Facts About Failing Banks (Liberty Street)

• Why Do Banks Fail? The Predictability of Bank Failures (Liberty Street)

• Why Do Banks Fail? Bank Runs Versus Solvency (Liberty Street)

This study analyzes over 5,000 U.S. bank failures from 1865 to 2023, finding that banks typically fail due to deteriorating fundamentals rather than bank runs. The authors document three key patterns:

• failing banks experience rising asset losses and declining solvency in the years before failure;

• they increasingly rely on expensive "non-core" funding sources like wholesale funding and time deposits; and

• they undergo a boom-bust cycle where rapid asset growth precedes eventual failure.

Bank failures prove highly predictable using simple accounting metrics and the study finds that fundamentals can predict both individual bank failures and waves of failures during banking crises with high accuracy.

But particularly for the modern era, this model is pretty shocking for it’s construction. The model is extremely simple and does not need to include directly any measure of asset quality - it only implicitly comes in through Net Income/Assets!

From Table 2 Panel D (Modern Era 1959-2023):

AUC scores:

• 1-year horizon: 0.951 (in-sample), 0.938 (out-of-sample)

• 3-year horizon: 0.878 (in-sample), 0.854 (out-of-sample)

• 5-year horizon: 0.816 (in-sample), 0.787 (out-of-sample)

Model Specification

P(Insolvency) = 1/(1+EXP(-Formula))

Where

Formula = β₁*[EquityToAssets] + β₂*[WholesaleFundingToAssets] + β₃*([EquityToAssets]*[WholesaleFundingToAssets]) + β4*[ThreeYearAssetGrowth] + β5*[ThreeYearGDPGrowth]

And in the Modern Era (1959-2023)

• β₁ (Net Income/Assets) = -53.03

• β₂ (Time Deposits/Deposits) = 2.18

• β₃ (Interaction term) = -354.33

• β4 (ThreeYearAssetGrowth) varies from 0.07 to -0.06 to depending on quintile (see paper)

• β5 (ThreeYearGDPGrowth) = -0.06

Maybe in a future post I will recreate this model and test it on current FFIEC data. Should be pretty easy. The hardest part will be downloading the data from the FFIEC’s API.

Breakdowns in KYC at Morgan Stanley

How Morgan Stanley Courted Dodgy Customers to Build a Wealth-Management Empire (WSJ)

The bank wooed clients from countries known for financial corruption and drug trafficking, including Venezuela. Federal investigators are looking into whether the bank allowed accounts to be used for money laundering funds gained through corruption that were connected to a former government minister there.

Documents warned the bank’s risk had increased because of its exposure to Russia and a sanctioned Russian bank—Morgan Stanley received a caution for it from the Treasury Department. In one problem account, the SEC has been probing the bank’s work with a billionaire with ties to Russia who was sanctioned by the U.K. after Russia’s invasion of Ukraine.

Overall, bottlenecks in the bank’s anti-money-laundering procedures held up the vetting of thousands of accounts, which were allowed to proceed in the meantime.

The 2023 document showed that 24%, or 46,572, of Morgan Stanley’s international wealth-management accounts were designated by the firm as “High/High+” risk for money laundering.

E*Trade, in contrast, had used a centralized data system that compiled information about a customer. The data included alerts on suspicious activities, negative developments about clients that would be in the public record, and a log of a customer’s transactions so that atypical moves could be flagged and assessed for risk.

Morgan Stanley dropped much of E*Trade’s automated vetting procedures.

By the spring of 2021, it had turned off E*Trade’s risk-based score that would alert the firm to higher risks associated with an account or when suspicious activities occurred. The capability wasn’t turned back on until early this year.

2024 FSOC Annual Report

The 2024 FSOC Annual Report has been published. I’m not going to rehash the full report. Instead here are a few tidbits that struck me as new and worth reporting.

Insurance Sector Transformation

The FSOC’s focus on the insurance sector has expanded, and my former insurance sector colleagues will want to review section 3.2.4.. The insurance sector, particularly in life insurance, is experiencing unprecedented structural changes that carry potential financial stability implications. The report highlights that life insurers have been rapidly accumulating balance sheet risks through increased investment in nontraditional assets, such as private credit, structured credit, and alternative investments. Total investments by private-equity-owned U.S. life insurers have grown by 93% over the past five years, now controlling approximately $1 trillion in investments, or about 20% of the sector's total assets under management.

The report highlights the growing linkages between insurers and private credit.

Additionally, life insurers are increasingly using nontraditional liabilities, such as greater borrowing from capital markets and Federal Home Loan Banks, with advances to life insurers reaching an all-time high of over $150 billion in 2024. Perhaps most significantly, the sector has dramatically increased its use of offshore reinsurance, particularly with Bermuda-based reinsurers that are wholly owned by the same insurance groups. Life insurance and annuity reserves transferred offshore rose to $1.2 trillion at year-end 2023, representing about 45% of total ceded reserves. These changes are raising concerns about regulatory arbitrage, reduced transparency, and increased interconnectedness within the financial system, potentially creating new channels for systemic risk transmission.

Artificial Intelligence Risks

The Council's treatment of AI risks in 2024 represents a significant evolution from previous years, moving from general concerns to specific financial stability implications.

The report identifies several key vulnerabilities introduced by AI adoption in financial services. First, the lack of explainability and high complexity of AI approaches may heighten financial instability beyond effects on individual financial actors. This "black box" nature of AI systems makes it particularly challenging to assess and manage risks at a systemic level. Second, the report highlights concentration risk, noting that multiple financial institutions using the same AI models, vendors, or datasets could lead to correlated behaviors and amplified market movements during stress periods.

The Council expresses particular concern about the potential for AI systems to create new forms of interconnectedness and herding behavior that could exacerbate market volatility. Additionally, AI's data dependency creates new operational vulnerabilities, as data poisoning or integrity attacks could simultaneously affect multiple financial institutions using similar datasets. The report also notes that AI tools could make certain types of cyber attacks more sophisticated and easier to execute, potentially increasing the frequency and effectiveness of such attacks against financial institutions. These concerns have led the Council to recommend increased monitoring and development of specific regulatory frameworks for AI use in financial services, while emphasizing the need for robust risk management practices and enhanced supervision of AI applications in critical financial functions.

Credit Union Failure

Hat tip to Bank Reg Blog for highlighting this tidbit:

The Council encourages the NCUA to continue efforts to mitigate the risk of a significant credit union failure. If the NCUA is unprepared for significant failures, including of the largest credit unions or highly interconnected credit unions, or third-party service providers, then the SIF may be unable to withstand the resulting losses. …

The Council recommends that the NCUA use its existing powers in managing the SIF to increase the reserves and normal operating level—the target equity ratio—to better safeguard against losses and adopt a countercyclical approach.

Evolution of FSOC Reports 2021-2024

Just as with Global Trends, I have looked across the reports from 2021-2024 to see how their views on the risk environment have evolved.

Banking Sector Stability

Banking sector stability concerns underwent significant transformation during this period. In 2021, the focus was primarily on fundamental metrics like capital and liquidity pressures. This shifted in 2022 to emphasize exposure to interest rate risk as monetary policy tightened. The landscape dramatically changed following the Spring 2023 bank failures, which prompted intensive regulatory scrutiny and response. By 2024, while acute crisis concerns had diminished, the focus remained on persistent funding pressures and deposit stability issues.

Commercial Real Estate

The CRE trajectory shows a dramatic escalation in concern over the four-year period. What began as a concentration risk in 2021 quickly elevated to a broader market vulnerability in 2022. By 2023, CRE had emerged as a major stress point, with particular concerns about office space occupancy and valuations. The situation reached a critical level in 2024, with the reports highlighting specific worries about the approaching maturity wall and refinancing challenges in a high-interest-rate environment.

Private Credit

Private credit markets represent a risk that grew substantially in regulatory focus over the period. The 2021 and 2022 reports made only limited mention of these markets. By 2023, they had begun to draw growing attention from regulators. In 2024, private credit had become a major focus area, with the report specifically noting $1.6 trillion in assets under management and expressing concerns about interconnectedness with traditional financial sectors, particularly insurance.

Technology

Technological risks evolved from relatively straightforward cybersecurity concerns to a complex web of interconnected threats. The 2021 report focused primarily on basic cybersecurity and third-party risks. In 2022, cloud service adoption risks were added as a major concern. The 2023 report marked a significant expansion with the introduction of AI and quantum computing threats. By 2024, the focus had sharpened on AI implementation risks and specifically called out increased cyber threats from state actors, reflecting a more sophisticated understanding of technological vulnerabilities in the financial system.

Crypto

If Tether continues its alleged current rate of Treasury purchases, it could become a significant holder of U.S. Treasuries and could present risks to the stability of the Treasury market if it experienced a run.

Digital asset and crypto markets showed an interesting evolution in how they were perceived as risks. In 2021, they were viewed primarily through the lens of operational risks to the financial system. By 2022, they had been elevated to a significant vulnerability requiring broader regulatory attention. The 2023 report maintained this high level of concern, particularly in light of market disruptions. However, by 2024, the focus had narrowed specifically to stablecoin concentration risks, with particular attention to the fact that 70% of the market was concentrated in a single issuer.

Climate

Climate-related financial risks show a clear progression from emerging concern to concrete impact. In 2021, these risks were newly emphasized as an emerging threat to financial stability. The 2022 report focused on integrating climate risks into assessment frameworks. By 2023, the focus had sharpened to specific impacts on the insurance sector. The 2024 report documented concrete disruptions in insurance markets attributable to climate risks, marking a progression from theoretical to actual impacts.

Other Observations

Artificial Intelligence risks represent a newly identified category that emerged strongly in the 2023 report and gained prominence in 2024. The initial focus was on potential threats to financial stability from AI adoption, but this evolved to include specific concerns about implementation risks and model governance.

Third-party service provider concentration risk, while not entirely new, evolved significantly over the period. Initial concerns about basic operational dependencies in 2021 expanded to encompass cloud service adoption risks by 2022, and by 2024 included complex interconnections between service providers and financial institutions.

Deep Deep Tail Risk

Global Catastrophic Risk Assessment (RAND)

This report summarizes what is known about the risks associated with six threats and hazards: artificial intelligence; asteroid and comet impacts; sudden and severe changes to Earth's climate; nuclear war; severe pandemics, whether resulting from naturally occurring events or from synthetic biology; and supervolcanoes.

Overall, global catastrophic risk has been increasing in recent years and appears likely to increase in the coming decade

• For supervolcanoes and asteroid and comet strikes, risk should remain constant or decrease in the next decade.

• For the remaining threats and hazards, the risk appears to be increasing in the next decade because of current or expected human activities.

• For artificial intelligence, the uncertainties are sufficiently large that it is difficult to determine the extent or magnitude of changes in risk with any confidence.

More

Measuring Systemic Risk

SYSTEMIC RISK MEASURES: TAKING STOCK FROM 1927 TO 2023

Three of the leading systemic risk researchers, Acharya, Brunnermeier and Pierret) complete a stock-taking exercise. Good overview if you are not familiar with the literature.

Norinchukin Market Risk

Norinchukin’s market RWAs blow up 342% in Q3 (Risk Quant)

• Norinchukin calculates its market risk RWA under the internal models approach of Basel III. This significant increase, one of the steepest recorded under the Basel III framework, reduced the bank's Common Equity Tier 1 (CET1) capital ratio by 117 basis points. Norinchukin is a rather unique bank, however the massive jump in RWA suggests Norinchukin's portfolio became significantly more sensitive to market shocks, either due to portfolio changes, rate volatility, or stricter regulatory models.

Wirecard

I know I’m probably the only person here obsessed with this case, but here are recent developments:

‘Sophisticated’ spy ring passed secrets to Russia for three years, UK court told (Guardian)

A “sophisticated” UK-based spy ring passed secrets to Russia for nearly three years and gathered information on targets across Europe, a court has heard.

Three Bulgarian nationals – Katrin Ivanova, 33, Vanya Gaberova, 30, and Tihomir Ivanov Ivanchev, 39 – allegedly carried out surveillance on individuals and places of interest to Russia.

… the defendants plotted with a Russian agent, Jan Marsalek, who was known as “Rupert Ticz” and was said to be an Austrian national, to obtain material useful to Russia.

Jan Marsalek was the Chief Operating Officer (COO) of Wirecard before fleeing to Russia.

Bank Regulation in the Second Trump Administration

BankRegBlog has a good writeup Five Thoughts About Bank Regulation in the Second Trump Administration on Basel III, a “broad support” standard at Fed Board, tailoring regulations, new bank charters and FDIC resolution. Head over there if interested.

Bank Service Providers

FDIC Looks to More Fintech Tracking After Synapse Collapse (Bloomberg)

This increased scrutiny has come after the the bankruptcy of Andreessen Horowitz-backed “banking as a service” startup Synapse Financial Technologies Inc., which offered record-keeping services that enabled fintechs to provide financial products by connecting them with FDIC-insured banks.

Synapse imploded in April, and in the months since has been embroiled in a dispute with one of its sponsors, Evolve Bank & Trust. Evolve was itself hit with an enforcement action by the Federal Reserve in June. And potentially thousands of people have been left without access to their funds.

The Federal banking agencies have the authority to “examine” bank service providers under the Bank Service Company Act. Here is a specific reference to the FDIC program: The FDIC’s Regional Service Provider Examination Program. IMO, a vastly under-utilized tool, and one that may need expansion as fintech and private credit grows in importance.

Banks and ESG

Banks and ESG

This paper responds to the growing controversy surrounding the proper role of ESG-related issues as part of the business of banking. … First, it argues that federal banking agencies incorporating ESG-type considerations into regulation and supervision is legally sound and serves legitimate policy objectives. … Second, this paper argues that two recent Supreme Court decisions call into question whether anti-ESG efforts are legally permissible.

Modeling Demand Deposits

Why is Risk Modeling of Nonmaturity Deposits So Difficult? (UponFurtherAnalysis)

Passive (Equity) Investing

Assessing the Impact of Passive Investing over Time: Higher Volatility, Reduced Liquidity, and Increased Concentration (Apollo Academy)

… this paper illustrates how passive investors, who primarily track major indices, have contributed to reduced price elasticity and market responsiveness, which, in turn, have led to amplified price movements, decreased liquidity, potential macroeconomic inefficiencies, and a disproportionate concentration of market influence in a few dominant stocks, such as the so-called “Magnificent Seven.”