The Banks: What's Next?
The big news last weekend was that Credit Suisse managed to sell itself to UBS at a price less than half of Friday's close ($, FT). UBS credit default swaps promptly rose; combining a weak bank with a strong one means creating a bigger bank, but not a stronger one.
It's a weird deal, where the pricing for the equity has already been renegotiated up from UBS's initial bid. Credit Suisse's capital structure includes "Additional Tier 1 Capital" bonds, which are basically designed to provide a capital buffer without diluting equity—they're junior enough in the credit structure that they're capital-like, but, in good times, they pay a better rate than other bank obligations. (You can view a recent prospectus here.) These instruments, all $17.3bn worth, are getting zeroed. Which is no doubt an unpleasant surprise to the traders who were buying them on news of the deal just a few hours earlier ("Additional Tier 1 notes were quoted between 50 and 70 cents on the dollar in the wake of the deal announcement [i.e. after news emerged that UBS would make a bid, but before the numbers were announced], up from the mid 20s to high-40s earlier in the day..."). Don’t be surprised if we see the deal structure mutate a bit more over time.
In any case, the general policy goal in situations like this is to prevent single-company problems from becoming systemic problems. There are two basic frameworks for looking at systemic risk:
- The more literal, object-level one goes like this: if a bank fails, and deposits are unavailable for a while, depositors may struggle to pay other debts, leading to other bank failures. This kind of systemic risk is most directly applicable to investment banks with significant derivatives exposure: if a hedge fund in 2008 was long something via Goldman, and hedged this position through Lehman, then the collapse of Lehman would mean it was suddenly unhedged. Or, worse, that it didn't know if it was hedged or not, where exiting its position with Goldman might end up meaning that it kept the trade with Lehman after all. That kind of uncertainty is toxic for levered institutions since variance and uncertainty are, from a lender's perspective, almost entirely downside. This is an asset-based view of systemic risk, since the direct effect of failures is to make other institutions' assets worth less.
- The other kind of systemic risk is liability-based: a failure at one institution makes lenders (including depositors) less likely to want to work with similar-looking institutions. This is a more amorphous problem, because similarity is so hard to define: an American regional bank (SVB) and a Swiss investment bank (Credit Suisse) have very little in common except for their impaired assets and mobile depositors. An important consideration with liability-based systemic risk is that it's indifferent to the asset side, and all about liquidity. (The iShares MBS ETF is up about 3% since SVB announced their planned capital raise, so the specific trade that SVB lost so much on is looking better and better.)
Which, in one sense, makes liability-side contagion easier to solve: find the financial institutions experiencing outflows, and get so much cash that there's no net impact on their liquidity position and thus no reason for deposits to flee or for loans not to be rolled over. On the other hand, it's a harder problem, because it's dealing with fundamentally incommensurable things: money as a solution to a psychological problem.
As with other bailouts, the sums involved in the Credit Suisse deal are staggering: UBS is getting a $9.7bn guarantee against losses from acquiring Credit Suisse, and the ability to borrow up to $108bn from the Swiss central bank. The general approach to this kind of package seems to be: first, think about how much funding the entity will need right now. Second, think of how much more funding it would need if that's all it got and people continued to panic. Step three is to take whatever number would prevent that, and then double it, just to be safe.
(Which seems exaggerated, but a lot of risk management rules are the inverse of this: if a position or strategy loses more money than expected, a common default is to cut it in half and reassess. It's very hard to calibrate mistakes so they're just barely inconceivable, since whatever got them to the "inconceivable" level in the first place was a bigger deal than was previously imagined.)
As the bailout numbers get bigger, they lead to a second-order concern: what if all this credit and liquidity creation threatens trust in the entire financial system, and leads to faster inflation, or even hyperinflation? Some of the popular charts certainly look hyperinflationary, like this one:
But note that both of the previous episodes led, in the short term, to a drop in inflation: the CPI hit 5% for the first time since the early 90s in 2008, just before the crisis. Banking crises are typically deflationary, so you can use the central bank interventions in their wake as a proxy for how deflationary they are when there isn't a policy response. The Fed added $3.5tr to its balance sheet in the six years following the financial crisis, which was just enough to get inflation from the pre-crisis level of around 3% to the next decade's average of 1.6%. Post-Covid, a $2.9tr in balance sheet expansion was also just enough to keep year-over-year inflation slightly above zero in the first few months of the pandemic.
This is classic Lucas Critique stuff: on average, buildings that are being blasted with a firehose right now are significantly more likely to be on fire than the typical structure, but this does not mean we should ban fire departments as a clear fire hazard. It does mean that their presence coincides with a big, destructive mess, though. And, of course, they can make mistakes, and their presence is also an indicator that the previous situation was riskier than expected. And as the last two years have demonstrated, it is possible for interventions to overshoot: inflation would have been lower if the Fed had provided less liquidity, and if Congress had allowed less deficit spending, although a CPI up high single digits is preferable to one that's down by a similar amount.
Longer-term, what happens to the banks is still questionable. Regulators are in a tricky position:
- Eyeballing the current levels of inflation and unemployment, it's very easy to justify further rate increases. A few months ago, we'd say "The economy is fine aside from tech," and now we can say "The economy is fine aside from tech and certain banks." (Of course, it was perfectly accurate in 2007 to say that the economy looked fine aside from some unfortunate housing loans.)
- But those rate increases hurt banks' balance sheets, and they are particularly damaging to smaller banks.
- Providing liquidity to those smaller banks can undo some or all of the impact of tighter monetary policy. If a Fed facility is the sole reason a bank is holding underwater mortgage-backed securities it would otherwise sell, this is exactly economically equivalent to the Fed holding that same security itself, in terms of market impact and in terms of which institution ultimately gets to decide who gets what.
So we could end up in a situation where the first-order effect of prudent monetary policy is to do something that hurts the majority of banks, and there are second-order policy tweaks to ensure that these moves only hurt certain banks.
And that's politically palatable. The smaller a bank is, the more political pull it has relative to assets. Patrick McKenzie has written eloquently about this with respect to community banks:
Community banks are surprisingly powerful in the U.S., largely because they are very popular with their customers relative to Big Finance, they are indispensable for politically powerful local groups like landlords, real estate developers, and farmers through their commercial loan books, and they have earned an appealing narrative about financial access. It is difficult to overstate how dependent commercial real estate is on community banks in much of the country, and it is a symbiotic relationship; the investors who organize them are usually local business owners with heavy real estate interests.
And looking at slightly larger banks, one fun way to slice the data is to look at the share of political donations given by big banks, and the share of top donors to individual politicians from big banks versus smaller ones. The ten largest banks in the US represent 56% of total banking assets (based on this). Restricting ourselves to the top 25 donors to each member of the House or Senate, those top ten banks collectively represent 44% of banking industry donations to politicians. And slicing it more finely: 66 members of Congress have a top-10 bank in their list of top 25 donors, with no smaller banks on the list, but 99 have a below-top-ten bank among their top donors with no big banks.
At least when it comes to banking, Big Business is being outspent in both magnitude and breadth by (relatively) Little Business. The way to bet is, and for a long time has been, that smaller banks will get some political assistance when they need it, whether or not they could have avoided the problems that led them to need it in the first place. From SVB's perspective, it was very unfortunate that they weren't called Farmer's Bank of Santa Clara. But from the perspective of every other small bank that also took massive rates risks, it's very fortunate indeed that SVB had such techy branding.
So the US will likely remain an extreme outlier in terms of number of banks per capita. This arose for all sorts of historically contingent reasons—a regulatory light touch encouraged bank formation, state-level protectionism discouraged bank expansion, capital markets that can substitute for bank loans at scale, and banks' critical role in local economies means that they're hard to dislodge.
There is a sort of accelerationist hard-money argument that a deflationary collapse is just what the economy needs, because the leverage that makes deflation so dangerous is part of the problem. This debate is unlikely to be solved any time soon, though, and the big test of it has been military conflict between countries that have similar levels of natural resources and different tolerance for credit. European history from the early 18th century through the mid-twentieth century is partly a story about the UK punching above its weight by being able to out-borrow and thus out-spend the countries it went to war with. On the other hand, look at what eventually happened to them. On the third hand, their status as global financial hegemon was supplanted by the US, which has not exactly been allergic to borrowing money, whether at the government, household, or corporate level. ↩︎
For the record, I do not think that a bank that financed wealthy people's vineyards would be all that more politically sympathetic than a bank that financed their VC funds. In fact, The Diff's first post about SVB's balance sheet highlighted their wine loans as a potential topic for political grandstanding. ↩︎
A Word From Our Sponsors
This newsletter is brought to you by Tegus, your modern research launchpad.
Investors waste too much time looking for actionable information across disconnected sources. Tegus streamlines the information you need to move quickly with conviction and make better decisions. With access to the largest transcript library, expert calls with no markups, driveable financial models, and searchable SEC filings, you’ll be the smartest expert in the room.
Right now, readers of The Diff can trial the Tegus platform for free.
Mike at Nongaap Investing has a great look at SVB management's incentives and how they affected the company's decisions. Much of the piece is focused on some of the maneuvers they went through to obfuscate their risk profile, but one important section talks about how the bank's managers were compensated based on return on equity over a 1- to 3-year-period. This created an incentive to find sources of short-term profits in order to keep returns on equity up when rates were low. Granted, it's hard to balance between incentivizing management over a full cycle and having an "incentive" plan that ignores the short-term problems that can indicate long-term poor performance. On the other hand:
The trade-off is SVB was essentially generating “risk-unadjusted” ROE which created a riskier and more volatile return profile for the company since it required taking on more risk to generate the desired ROE over 1 to 3 years.
The FT has a good interview with Dina Srinivasan, author of "The Antitrust Case Against Facebook," as well as other papers on tech competition ($). One notable detail is that she found some of the same tricks identified in capital markets taking place in online ads:
For example, I was looking at something that Google did. Where it started to round up or round down the time stamps of trades to the nearest hour. I had no idea what it meant so I spent a lot of time researching financial markets. Then I found a case where a large broker in Chicago was prosecuted for insider trading, and the way that they were concealing insider trading was by changing the time stamps of trades, so that their customers could not find out what was going on.
Adtech and trading are both businesses where there's a continuous real-time auction with uncertain information, and they attract some of the same kinds of talent. (In fact, sometimes literally the same talent.) In both cases, competition tends to push down risk-adjusted returns over time, but intermediaries can continuously earn high returns by maintaining information asymmetries. This can be entirely prosocial—an "information asymmetry" can be any data point that is worth figuring out once but not worth the effort it would take if everyone involved had to discover it independently, and markets provide a nice pricing mechanism for converting that social good into a private return through more accurate pricing in return for higher profits. But imposing an asymmetry instead of discovering one is bad for everyone involved.
Cable TV increasingly sees higher ratings for afternoon shows rather than evening shows. When media get more competitive, this tends to show up first in whichever demographic has the least brand loyalty, and one obvious way to have little brand loyalty is to be a new consumer of some kind of media. So when media companies lose share, they typically lose share with their youngest audience first. The median age for cable news viewers is in the 60s already, and growth in afternoon viewing is a symptom of an increasing skew towards people who are at retirement age.
Training Your Replacement
Microsoft has a blog post and demo video for its 365 Copilot feature, adding AI text generation and summarization to the company's office suite. One way AI forces us to reframe companies is to ask: which businesses create the largest corpus of proprietary training data? Google has a huge amount of information on what people search for, and, thanks to Android and Chrome, what they do before and after a search. Meta knows more than any other company about digitally-mediated personal interactions (but they've naturally been slower than LinkedIn to introduce cues and suggested comments). And Microsoft has more data on intra-company and inter-company communications than anyone else. There's a persuasive argument that a quarter of GDP circa 1995 consisted of persuasion in one form or another. It's fair to say that Microsoft enables a plurality of that persuasion, at least weighted by GDP. And as tools like this roll out, their share will likely go up.
Disclosure: Long MSFT, META.
One important element of the rise of Airbnb and Uber was their approach to regulations. There are varying levels of cynicism about this, and somewhere in the middle of the spectrum is the theory that they wanted to comply with rules eventually, but knew that they'd be in a better position to negotiate if they got big enough to have more users with a vested interest in keeping them legal. That applies especially strongly to Airbnb; while running an informal hotel isn't a completely passive job, it's a lot more capital- than labor-intensive compared to running a one-person car service through Uber. So Airbnb's lobbying strategy benefited from the fact that, in any market where they were big enough to ban, there were Airbnb hosts who relied on the company for their livelihood and had enough free time to attend hearings.
TikTok is allegedly engineering something similar by paying influencers to go to DC and lobby on their behalf. Entertainment companies always have access to celebrities as a lever. (When commodities trader Marc Rich was part owner of 20th Century Fox, he would sometimes arrange for energy ministers in far-flung countries to visit the studio for an exclusive tour.) One difficulty for TikTok is that recommendation algorithms fragment fame; members of Congress are pretty unlikely to know who most of these people are. On the other hand, representatives' kids probably recognize most of the names on the list. It will be an interesting assignment for the TikTok data science team to tweak the recommendation algorithm so the children of members of Congress all know that mom and dad are not merely elected representatives, but are about to meet someone really famous.
Companies in the Diff network are actively looking for talent. A sampling of current open roles:
- A hedge fund is looking for an experienced alternative data analyst who can help incorporate novel datasets into systematic strategies (NYC).
- A VC backed company reimagining retirement wealth and building a 401k alternative is looking for product/GTM/bizops generalists. (NYC)
- A company building zero-knowledge proof-based tools to enable novel financial arrangements is looking for a senior engineer with a research bent. Ideal experience includes demonstrations of extraordinary coding and/or math ability. (NYC or San Diego preferred, remote also a possibility.)
- A company building ML-powered tools to accelerate developer productivity is looking for a mathematician. (Washington DC area)
- A profitable startup is looking for sales reps to market its AI-based services that help small companies accelerate their growth—especially people who are excited to use AI tools to accelerate some of their own work. (SF)
Even if you don't see an exact match for your skills and interests right now, we're happy to talk early so we can let you know if a good opportunity comes up.
If you’re at a company that's looking for talent, we should talk! Diff Jobs works with companies across fintech, hard tech, consumer software, enterprise software, and other areas—any company where finding unusually effective people is a top priority.