The 2023 Crisis Episode

In the first two quarters of 2023, two separately classified crisis events took place. The first of these was the problems of Credit Suisse, which ultimately led to the controlled takeover of the bank by UBS. This case is analyzed primarily from the perspective of the effectiveness of the regulations implemented after the 2008 crisis in the area of restructuring methods and orderly liquidation (the resolution mechanism) at a moment of crisis, for a financial institution of sig-nificant systemic importance. The Credit Suisse case, although interesting and deserving of a more in-depth description, will not be the subject of this paper. In this study, I will focus on the second crisis episode of 2023: the crisis of regional bank failures in the United States. These events, despite the small scope of their occurrence, seem to give a better picture of today's integrated crisis mechanisms.

Silicon Valley Bank (SVB), headquartered in Santa Clara, California, specialized in provid-ing private and commercial banking services for the venture capital industry. At the end of 2022, it had total assets of approximately USD 209 billion, which, after the raising of the regu-latory subject-matter thresholds in 2018 (above USD 250 billion), did not qualify the bank for the category of institutions subject to Enhanced Prudential Standards (EPS). During the peri-od of strong growth, in the years of the technology boom (2020-2021), there was an enormous inflow of deposits into the bank, which it placed in long-term treasury bonds. The significant increase in interest rates that took place in 2022, however, caused a decline in the yield of these securities relative to newly issued instruments on the market, which forced a reduction in the value of the bonds held by SVB. Initially, however, the loss in asset value was not dis-closed by the bank. Meanwhile, liquidity problems arose, and in order to raise cash, the bank had to sell the bonds, and the loss, previously invisible to the market, became real. On March 8, 2023, SVB sent two communications to the market: the sale of bonds at a loss of USD 1.8 billion, and simultaneously information about the issuance of new shares in order to raise capi-tal. Both indicated the bank's financial problems. The bank's depositors were primarily tech-nology companies and VC funds — entities that understand market signals perfectly well. By the end of the day on March 9, almost 30% (USD 40 billion) of deposits had been withdrawn, and another USD 100 billion was expected to be withdrawn the following day. On March 10, the California Department of Financial Protection and Innovation (CADFPI) closed SVB and appointed the FDIC as receiver of the bankruptcy estate. The first steps of the bank's restruc-turing process were initiated, but the pace of the deposit outflow made it impossible to find potential buyers for the institution before its failure.

The further course of events was a classic contagion problem, following the loss of confi-dence in the banking sector. Over the weekend of March 11-12, depositors began massively withdrawing funds held in accounts at another bank: Signature Bank (total assets of USD 110 billion). On Sunday, March 12, regulators, alarmed by the turn of events, preventively closed the bank. At the same time, a decision was made, with respect to both banks, to adopt a sys-temic risk determination (pursuant to the Federal Deposit Insurance Act). The adoption of such a determination for these events enabled the FDIC to extend deposit insurance protection for deposits held at both banks and to establish two bridge banks, which could be used to facil-itate the orderly liquidation of both institutions. The FDIC subsequently carried out a sales process for both banks, divesting practically all deposits and selected loan portfolios of Signa-ture Bank within a week, and all deposits and loans of SVB within two weeks. In both cases, the institutions could now fail: shareholders lost their investments, unsecured creditors bore losses, and the boards and senior management of both banks were removed. Ultimately, as stated in the supervisory document referring to the events of 2023 (FSB, 2023), all losses re-sulting from both failures were covered by the banking sector. This means that the costs of the crisis were not transferred to taxpayers (there was no need to apply bail-outs), and thus did not spill over beyond the area where the crisis was triggered.

At the moment of SVB's failure, contagion effects began to spread to other American re-gional banks, including First Republic Bank — a bank headquartered in San Francisco, with a balance sheet total of approximately USD 230 billion, focused on banking services for wealthy individual clients.

Instead, the bank's assets consisted primarily of illiquid long-term bonds, which could not pro-vide protection in such a stress situation. The NSFR, meanwhile, requires stable funding for long-term assets. SVB, however, was funding long-term bonds with short-term corporate depos-its — thus doing exactly what the NSFR was meant to prevent. Perhaps it is precisely these lowered prudential standards that lie at the root of the problems of American regional banks at the threshold of 2023. It is worth recalling, however, that the General Manager of the BIS, in his speech (Hernández de Cos, 2026), noted that the events of 2023 took place despite the ful-fillment of capital requirements, as well as, quite often, liquidity requirements.

Attention is drawn here to yet another reason, pointed to by all the reports mentioned above. FSB (2023), in its document, writes that by the end of the day on March 9, more than USD 40 billion in deposits had left the bank, and another USD 100 billion was expected to be withdrawn the following morning. It further explains that this deposit outflow was driven by social media and a concentrated network of venture capital investors and technology firms, who withdrew their deposits in a coordinated manner with unprecedented speed. The dynamics of deposit outflows are, according to the authors of the report, the main observation and chal-lenge emerging from this crisis episode. The prevalence of social media and mobile banking may mean that bank runs, once they occur, proceed more quickly. Although the underlying problems of the banks themselves may vary (e.g., poor management practices, asset quality, etc.), once a bank run gains momentum, stopping the outflow of funds becomes extremely dif-ficult.

In the Fed report (Barr, 2023), which analyzes the case of SVB, particular attention is drawn to the influence of social media on the dynamics of the deposit panic. Social media posts urged depositors to withdraw funds from the bank, and many uninsured depositors did so sim-ultaneously. The concentration of these large deposits among technology firms and individuals, meanwhile, may have contributed to the synchronized character of the deposit outflow.

A separate problem, pointed to in the reports, is the trap of technological development in the banking sector. The ease of using digital banking applications, available around the clock, enabled depositors to withdraw funds even on weekends. It was thus possible to make with-drawals on an enormous scale within just a few hours. Additionally, as is usual in such cases, a contagion effect occurred, which caused the panic to spread to other entities, not necessarily in a state of poor financial condition.

All of these factors thus influenced the scale of the events at the beginning of 2023. Occur-ring in an aggregated and mutually reinforcing manner, they constituted a wave, difficult to stop, striking at the liquidity security of banks.

Summary

The 2023 crisis episode was an event limited in its scale — it affected a few American regional banks and did not spill over into the real economy. However, it is precisely this small scope of events that may be misleading. The conclusions that emerge from the crisis are, in fact, sys-temic in significance.

First, the post-crisis financial security architecture, painstakingly built since 2008, proved insufficient — and not because the banks violated capital or liquidity requirements. Qualita-tive factors failed: risk management, organizational culture, supervision. This means that the gap in the security system is of a nature that no quantitative indicator is able to capture.

Second, the 2023 episode revealed a new mechanics of banking crises. Social media and digi-tal banking shortened the time it takes for a deposit panic to materialize from days to hours. This is a qualitative change in the dynamics of extreme events — tail risk is no longer the same risk that has been modeled up to now. The tails of distributions are fatter, and the time window for supervisory response is dramatically narrower.

Third, existing regulations do not yet respond to the challenge of changes in the dynamics of tail events. The paradigm of normality of phenomena, built into methods such as IRB, as well as the system of buffers adopted in CRR III, were constructed with a different, slower mechanics of crises in mind. Supervisory reports, published after the 2023 episode, indicate the regulators' awareness of the new mechanisms behind the spread of crisis panic. What is lack-ing, however, is an institutional response.