渡十娘｜特约专家为你揭开硅谷银行（SVB）爆雷内幕

做公众号里的《纽约客》

戳蓝字一键关注渡十娘

转发也是一种肯定

文字｜Neal

翻译｜Judy

编辑｜渡十娘

作者简介

Neal，从事银行监管工作三十多年，曾监管美国最大的几家银行。毕业于康奈尔大学和卡内基梅隆大学，CFA特许注册金融分析师。

本文英文和译文版权属作者和译者所有。

小编按：看完文章请你判断以下中国银行是否这次躺赢了。

2023年3月10日，硅谷银行(SVB)被宣布破产并由联邦存款保险公司（FDIC）接管。总资产为2090亿美元，总存款为1750亿美元，这是2008年以来规模最大的银行倒闭，也是史上第二大银行倒闭。毫无疑问，将会有很多事后追溯、银行失败报告及监管机构被国会召见听证。我想在这里提供一些初步看法。

硅谷银行是联邦储备系统的一家州授权成员，由加利福尼亚州和美联储监管，其中联邦存款保险公司担任后备监管机构和保险机构。我的观点完全基于公开可得的信息，以及从监管银行的职业中得出得的见解。

银行的规模并不是FDIC接管意义重大的唯一因素。通常情况下，首选途径是为陷入困境的银行寻找收购者，有时需要FDIC的协助。华盛顿互惠银行、美联银行及大多数其他倒闭的银行就是这样。这次是一次清算。这通常意味着监管机构已经放弃希望，得出结论没有人想要购买。它基本上没有特许经营的价值。

银行持续存在的两个最关键要素是其流动性和资本。媒体报道经常混淆了这两个概念，但它们是截然不同的。流动性是可用于履行责任的现金或优质投资的数额。这些责任包括日常运营，或承受压力例如银行挤兑。资本是银行的净值：其资产（如贷款和投资）减去其负债（如借款和存款）。监管机构同时关注资本相对于总资产（这称为杠杆率）和相对于风险加权资产的情况，将更高的权重分配给风险较高的资产。其中一些资产，如美国国债，风险加权为零。

我看了硅谷银行截至2022年12月31日的财务报告和统一银行业绩报告(UBPR)。财务报告包括银行的详细财务报表。UBPR是一份分析报告，可以比较同类银行的趋势和关键比率。这些报告可在FFEIC的公共网站上找到FFEIC'spublicwebsite。乍一看一些数字还不错。硅谷银行的基于风险的比率远高于15%，轻松超过最低要求。其杠杆率略低于8%，也远超要求。净收入也与同业银行相当。

那么问题是什么？有几件方面。新闻报道显示，该银行的可供出售(AFS)投资组合约20亿美元低于购入价值。这种未实现的损失在银行的GAAP财务报表中被认可，但出于监管资本的目的被加回（最大的一些银行是不允许加回）。出售AFS投资意味着硅谷银行需要承受收益和监管资本的冲击。20亿美元的打击对一家2000亿美元的银行来说会造成伤害，但未必致命。

还有另外两个更大的危险信号。第一个涉及存款类型。FDIC保险最高250,000美元的存款。此金额以下，如果银行破产你不必太担心，有些不便但没有灾难性的后果。超过这个数额的存款就没有保障了，你只是银行的另一个债权人。硅谷银行存款总额中超过250,000美元的存款占97％，平均余额超过400万美元。即使不包括大额存款的受保部分，存款余额的93％以上都是无保险的。难怪存款人在发现问题的第一时间就匆忙逃离，提取了惊人数额的420亿美元存款。

另一个危险信号，硅谷银行持有至到期证券(HTM)规模从两年前的166亿美元增长至913亿美元。当银行将证券指定为HTM时，它要有意图和能力将其保持到到期。出售HTM证券可能会产生不利的会计后果，出售某些证券可能会污染整个投资组合。但在某些情况下，如流动性紧张，不会发生污染。问题在于这些证券的公允价值仅为760亿美元，这意味着低于购入成本150亿美元。出售HTM证券将摧毁硅谷银行的资本。

将证券指定为HTM是银行为减轻其证券投资组合的利率风险而采取的一种常用方式。如果利率上升，他们不需要在监管资本或公认会计准则财务报表中认可价值下降。如果利率上升只是周期性而不是长期趋势，银行可以安然度过直到利率回落。但如果利率上升持续超过一段时间，或银行需要出售HTM证券以满足流动性需求，这种方法就行不通了。被击中之前闭上眼睛可能是一种自然反应，但没有真正的帮助。

拥有超过2000亿美元的资产使硅谷银行成为美国第18大银行18th largest bank，但它的规模并不足以使该银行受到某些监管要求的约束，这些监管要求可能可以提早发现问题。

2018年的《经济增长、监管放松和消费者保护法案》（EGRRCPA）意味着拥有少于2500亿美元资产的银行将不受流动性覆盖率（LCR），或每年资本压力测试的约束（对银行控股公司的资本压力测试称为CCAR，对银行为DFAST。）该法案是由特朗普政府和国会共和党推动，得到两党的支持，以67-31的投票结果在参议院通过。该立法的监管实施称为“调整规则”。

对硅谷银行进行年度压力测试和LCR是否能更早地发现问题？也许能也许不。CCAR下的压力测试情景通常侧重于银行面临经济衰退的风险，而不是其利率风险。2021年的严重不利情景，假设短期利率将保持在接近零的水平，而长期利率将适度上升。10年期国债利率假设达到1.5%（目前为3.5%）。2022年的情景假设类似的利率路径，而2023年的情景假设利率将会下降。就像将军们一样，监管机构有时会用过去一场战争的战略战术来打新战。

对于最大的一些银行，美联储要求至少运行一个公司自己定义的情景。这曾经是监管机构对所有资产超过500亿美元的银行的要求，且仍被认为是谨慎资本规划的重要组成部分。这样做旨在识别那些让管理层夜不能寐的情景类型。对硅谷银行来说，这可能包括利率急剧上升，接着出现银行挤兑的情景。但不能保证银行会测试遇到这种严峻情况。只要银行测试结果差于监管机构设的场景，管理层就会辩称自己的方法是“保守的”。

硅谷银行存在很大的流动性风险，但尚不完全清楚LCR会在多大程度上提示这种风险。LCR着眼于银行是否有足够的现金和其他流动资产来处理压力时期的现金流出。分子考虑了一个叫做“优质流动资产”（HQLA）的东西。HQLA背后的想法是，它们可以轻松立即转换为现金而很少或不会损失价值。好吧，就是这想法。问题是HQLA定义并没有太多关注利率风险。HQLA并不全是现金和国库券。它可以包括长期政府义务，以及（受一定限制）抵押贷款证券、股票证券和市政债券。如果利率急剧上升，这些资产的价值真的会暴跌。SVBFinancial的FRY-15Report报告显示，其AFS投资组合的94%符合HQLA定义，但该投资组合仍约20亿美元低于购入价值。

没有足够的公开可用的信息来看HTM投资组合的HQLA情况，但它似乎主要由房利美和房地美抵押贷款证券组成，算作2A级HQLA。这意味银行可以将其85%用于满足LCR要求。

在分母方面，硅谷银行的FRY-15报告其大部分存款为短期批发融资。目前尚不清楚这是否也适用于LCR，特别是如果硅谷银行可以提供证据表明它与一些大储户有日常业务运作关系，以使大储户与银行更紧密结合。没有足够的公开资料来证明这个假设。

如果硅谷银行的失败只是一次孤立的事件而不是蔓延的开始，它可能不会引发很多变化。我的看法是，资本压力测试可以提供一个良好的银行脆弱性指标，但是目前的流程过于依赖获得完全正确的场景。这可能会给硅谷银行这样的银行带来一种错误的安全感，因为它所处的场景与监管机构所选择的不同。我也相信流动性规则应该更加关注利率风险，并充分考虑流动性与资本之间的相互作用。现在仍然是事发初期，从硅谷银行的倒闭，我们还有很多东西要了解学习。

后注：

这篇文章完成于2023年3月12日，在联邦存款保险公司宣布覆盖所有存款（包括超过25万美元的限额）之前。这一政策变化可能会降低未投保存款引发未来银行倒闭的可能性，但这仍是硅谷银行倒闭的一个主要因素。

以下英文原文：

WHAT HAPPENED WITH SILICON VALLEY BANK?

On March 10, 2023, Silicon Valley Bank (SVB) was declared insolvent and taken over the by the FDIC. With total assets of $209 billion and total deposits of $175 billion, this was the largest bank to fail since 2008 and the second largest bank failure of all time. No doubt there will be multiple post-mortems, failed bank reports, and regulators hauled before Congress. In the meantime, though, I wanted to provide some initial thoughts.

SVB was a state-chartered member of the Federal Reserve System and was regulated by the State of California and the Federal Reserve, with the FDIC acting as backup regulator and insurer. I’m basing my observations entirely on publicly available information and the insights gained from a career examining banks.

The size of the bank was not the only element that made the FDIC’s takeover significant. Usually, the preferred path is to find an acquirer for the troubled bank, sometimes with FDIC assistance. That’s what happened with Washington Mutual, Wachovia, and most other failed banks. This was a liquidation. That usually means that the regulators have thrown up their hands and conclude that no one wants to buy the place. There is essentially no franchise value.

The two most critical elements of a bank’s continued existence are its liquidity and its capital. Press coverage often conflates the two concepts but they are quite distinct. Liquidity is the amount of cash or high-quality investments available to meet obligations. That can include day to day operations or under stress, such as a bank run. Capital is the bank’s net worth: its assets (like loans and investments) minus its liabilities (like borrowings and deposits). Regulators look at capital both relative to total assets (this is known as the leverage ratio) and relative to risk[1]weighted assets, which assigns a higher weight to riskier assets, with some assets, such as U.S. Treasuries, risk weighted at zero.

I looked at SVB’s Call Report and Uniform Bank Performance Report (UBPR) as of December 31, 2022. The Call Report contains the bank’s detailed financial statements and the UBPR is an analytical report that looks at trends and key ratios relative to similar banks. These reports are available on the FFEIC's public website. At first glance, some of the numbers look pretty good. SVB’s risk-based ratios are well over 15%, comfortably above minimum requirements. Its leverage ratio is just a shade under 8%, also well in excess of requirements. Net income was also in line with peer banks.

So what was the problem? A few things. News reports showed that the bank’s available for sale (AFS) investment portfolio was about $2 billion underwater. This unrealized loss is recognized in the bank’s GAAP financial statements but gets added back for regulatory capital purposes. (The very biggest banks aren’t allowed to add this back.) Selling the AFS investments meant SVB needed to take the earnings and regulatory capital hit. A $2 billion hit would hurt but wouldn’t necessarily be fatal for a $200 billion bank.

There are two other items that are even bigger red flags. The first relates to the type of deposits SVB had. The FDIC insures up to $250,000 in deposits. Up to that amount, you’re not too concerned if the bank fails. Some inconvenience but nothing catastrophic. Above that amount, there’s no guarantee. You’re just another one of the bank’s creditors. Deposits above $250K were 97% of SVB’s total deposits with an average balance of more than $4 million. Even excluding the insured portion of the large deposits, more than 93% of SVB’s deposit balance was uninsured. No wonder depositors rushed for the exits at the first sign of trouble. And rush they did with a staggering $42 billion in deposits leaving the bank.

The other red flag is that SVB had $91.3 billion in held-to-maturity (HTM) securities, up from $16.6 billion two years earlier. When a bank designates securities as HTM, it is supposed to have the intent and ability to hold them until they mature. Selling securities out of HTM can have adverse accounting consequences, where selling some securities can taint the entire portfolio. However, tainting does not apply in certain circumstances, such as a liquidity crunch. The trouble is that the fair value of these securities was only $76 billion, meaning they were $15 billion underwater. Selling the HTM securities would wipe out SVB’s capital.

Designating securities as HTM has been a popular way for banks to try to mitigate the interest rate risk of its securities portfolios. If rates go up, they don’t need to recognize the decline in value in their regulatory capital or GAAP financial statements. If a rise in rates turns out to be cyclical rather than part of a longer-term trend, the bank can presumably ride things out until rates go back down. That approach doesn’t work if the increase in rates persists over a longer period or if the bank needs to sell HTM securities to meet liquidity needs. Closing your eyes before getting hit may be a natural response but it doesn’t really help.

While having more than $200 billion in assets made SVB the 18th largest bank in the country, it wasn’t large enough to subject the bank to certain regulatory requirements that may have identified the problem earlier. The Economic Growth, Regulatory Relief, and Consumer Protection Act of 2018 (EGRRCPA) meant that banks with less than $250 billion in assets would not be subject to the Liquidity Coverage Ratio (LCR) or to annual capital stress tests, known as CCAR for bank holding companies and DFAST for banks. The bill was pushed through by the Trump Administration and Republicans in Congress but enjoyed bipartisan support, passing the Senate by 67-31. Regulatory implementation of the legislation was called the Tailoring Rule.

Would subjecting SVB to annual stress tests and LCR identified the problem earlier? Maybe yes, maybe no. Stress testing scenarios under CCAR have typically focused on a bank’s exposure to an economic downturn rather than to its interest rate risk. The 2021 Severely Adverse scenario assumed that short term rates would remain near zero with longer term rates increasing modestly. The 10-year Treasury Rate was assumed to reach 1.5% (it’s currently at 3.5%). The 2022 scenarios assume a similar rate paths and the 2023 scenario assumes that rates will decline. Like generals, regulators sometimes fight the last war.

The largest banks are required to run at least one firm-defined scenario. This was once a requirement of all banks over $50 billion in assets and is still considered an important element of prudent capital planning. These scenarios are supposed to identify the types of scenarios that keep management up at night. For SVB, that might have included a sharp rise in interest rates followed by a bank run. However, there’s no guarantee that the bank would have run that type of severe scenario and management could argue that its approach was “conservative” as long as it came out worse than the supervisory scenario.

SVB had a lot of liquidity risk but it’s not entirely clear to what extent the LCR would have flagged that risk. LCR looks at whether the bank has sufficient cash and other liquid assets to handle cash outflows during a time of stress. The numerator considers something called “high quality liquid asset” (HQLA). The idea behind HQLA is that they easily and immediately convert to cash with little or no loss of value. Well, that’s the idea. The trouble is that HQLA definitions don’t pay a lot of attention to interest rate risk. HQLAs aren’t all cash and T-bills. The definition can include longer term government obligations, and (subject to some limitations) mortgage-backed securities, equity securities, and municipal bonds. The value of these assets can really tank if rates rise sharply. The FR Y-15 Report for SVB Financial shows that 94% of its AFS portfolio met the HQLA definition, but the portfolio was still $1.8 billion underwater.

There’s not enough publicly available information on the HTM portfolio to see where it shakes out in terms of HQLA. However, it appears to primarily consist of Fannie Mae and Freddie Mac MBS, which would count a Level 2A HQLA. That means that you can count 85% of it toward meeting your LCR requirements.

On the denominator side, SVB’s FR Y-15 reports most of its deposits as short-term wholesale funding. It’s not clear whether that would also be the case with LCR, particularly if SVB could provide evidence that it had an operational relationship with some of its large depositors to tether them more closely to the bank. There’s just not enough publicly available to play out this hypothetical.

SVB’s failure may not spur a lot of changes if it remains an isolated event rather than as the start of a contagion. I have my own wish list. Capital stress testing provides a good indicator of a bank’s vulnerability, but the current process gets too hung up on getting the exact right scenario. This can provide a false sense of security for a bank such as SVB, that’s exposed to a scenario different to what the regulators chose. I also believe that liquidity rules should give more consideration to interest rate risk and take greater account of the interplay between liquidity and capital. It's very early days and we still have a lot to learn about SVB’s collapse.

P.S.

This article was written on 3/12/2023, prior to the announcement that the FDIC would cover all deposits, including those above the $250,000 limit. This policy change may make uninsured deposits less likely to drive future bank