Bank Failures and a Coherent Model for Technocratic Intervention
How the Silicon Valley Bank failure clarifies the line around what should be governed
This post on Best Of A Great Lot is a part of a series on the subject of designing a new form of governance. Each piece aims to stand alone, but fits together on the Table of Contents.
In March, Silicon Valley Bank melted down and regulators stepped in to guarantee customer deposits. In the immediate aftermath, Noah Smith’s explanation was one of the better ones. Now that the dust has started to settle, this is a perfect opportunity to talk about one of the most important questions in my project to reinvent governance: what should be the limits of societal governance and why? Are there any intellectually clean lines?
In many ways this is a deeply philosophical question. Libertarians and Socialists (to pick two mostly opposed camps) vigorously disagree about the purpose and acceptable methods of governance. The fundamental question of what we owe to society and to the marginalized among us tears us apart both because there are pragmatic reasons to see different sides of it, and because it is a reflection of conflicts among our deepest values. The dictionary authors stay on the sidelines. Merriam Webster is typical in simply declaring that governing is exercising continuous sovereign authority over and governance the act of governing.
In casual discussion, we commonly reduce governance to the process of setting rules, such as regulations and laws. It’s clear that setting society-wide rules counts as governance. But there's plenty of activities within our government that aren't just rule-setting, and conversely, there are organizations that set de facto rules but aren't part of the government.1 The Departments of Transportation set rules, but they also build roads and investigate accidents and decide that automakers will issue recalls. In states that have lots of snow, they run snowplows and invite the public to name them. (I'm a partisan of Yer a Blizzard, Harry)
The governance action after the SVB meltdown, undertaken by the Federal Reserve, the FDIC and the Treasury Department together, doesn't look like setting rules. So let's look at it through 3 of the available philosophical lenses that we might apply: Libertarian, Socialist, and Technocratic.2
In general, libertarians argue for a very limited and constrained vision of what governance contains, while socialists argue for a maximalist version to create fairness and equity. If technocratism can be said to be a philosophy, adherents argue for governance decisions to be made by experts to make society function as well as it can.
First, the facts. On Thursday, March 9th, there was a run on Silicon Valley Bank. On Friday, the FDIC showed up to take charge. The FDIC's pre-existing insurance plan only covers the assets for customers that have below $250,000; in many bank runs, this covers a lot of the customers. Silicon Valley Bank, however, is a bank that has focused on serving tech startups and other companies. Companies that have raised investment or are just larger than a dozen people typically keep a lot more than $250,000 in the bank, if only to cover payroll.
Over the weekend, the Treasury Department, FDIC and Federal Reserve worked furiously to decide what they were going to do, and then on Monday they declared they would guarantee all of the deposits of all customers, but that they would not bail out the bank (unlike some of the bailouts they performed in 2008). Shareholders of the bank lost their shares, but customers of the bank walked away unscathed. Who paid for this? The FDIC recently announced that the largest banks would pay $16B to cover the cost of preventing SVB from collapse.
Now, the arguments. A principled libertarian view that the government shouldn't have responded to SVB focuses on letting those who take on risk experience the downside of that risk. In this view, since it's a private bank, it should be allowed to fail. Customers were informed about the deposit insurance and its limits and should have made plans to deal with the risk to deposits that were above the limit.
A principled socialist view that the government shouldn't have responded to SVB focuses on fairness. The least powerful customers of SVB were already covered by the current FDIC insurance scheme, and it's unfair that the powerful and wealthy got saved by a governmental fiat. No governmental agency holds emergency meetings over the weekend to prevent laid off families from losing their homes.
Or, more briefly:
The technocratic case for the Fed stepping in to rescue SVB customers is built on an idea of the fragility of our interwoven economy. From this perspective, the government’s job is preserving the stability of the system as a whole. If the Fed hadn't stepped in to prevent SVB's collapse, we might see the whole economy melt down. Noah Smith:
If SVB’s collapse causes depositors outside tech to focus on that fact, and start to worry that rate hikes have made their own banks insolvent, it could fuel a panic that spreads well beyond the nerds in California. This is obviously a worst-case scenario, but of course we can’t rule it out.
The problem with this technocratic view is that it provides easy cover for corruption. Coincidentally, every time the system is at risk, so is the wealth of the powerful. The well-being of the less powerful never seems to need systemic protection.
This is the core of the argument by Ann Pettifor in her claim that the Federal Reserve is engaging in class war. The Federal Reserve has spent the last 6 months raising interest rates to try to prevent an overheated economy with runaway inflation (if you believe them), or attempting to keep asset values and returns high for investors, at the cost of many jobs and financial security (if you believe the critics).
The problem with this technocratic view is that it provides easy cover for corruption. Coincidentally, every time the system is at risk, so is the wealth of the powerful. The well-being of the less powerful never seems to need protection.
I have a lot of sympathy for the technocratic view. The society we live in is extremely complex, and getting more complex, and I vastly prefer it to most of the options from history. I like clean running water and the general ability to travel safely and buy food from a grocery store whenever I need to.3 The socialist impulse to fairness and the libertarian impulse to freedom are both valuable as philosophies that inspire us to improve our lot, but as workable governance systems they don't have much in the way of successful histories compared to technocracy. I would both like to see us prioritize individual freedom wherever possible and work to build more fairness into the way society treats all of us, but at the core of what government must do is keep society functioning.
The system we have is fragile to runaway memetics. The week after the SVB collapse, even with regulators standing by to calm everyone, there were several more banks that looked shaky, and plenty of individuals who worried that the Fed's response wouldn't be enough. First Republic Bank almost collapsed that week. Though it finally did collapse in May, it is almost certain that the delay prevented panic and further systemic spiralling. Absent the regulators stepping in, the story of SVB going viral on Twitter and taking out our banking system looks very plausible. World history for the last few centuries is littered with bank runs and runaway inflation and various systemically bad economic situations that cause tremendous suffering.
A clean intellectual line
Is it possible to draw a clear line — one that isn't quite as susceptible to corruption — around the kinds of situations that need to be managed via governance and the kinds that don't?
Economists often talk about two specific terms: market failures and negative externalities. A market failure is when the market generates inefficient or non-optimal outcomes, like when oligopolistic companies collude to raise prices and minimize their need to compete to provide better service. A negative externality4 is when a private actor generates costs that they can dump (externalize) onto society, such as dumping pollution into a river for free rather than paying to clean it up.
It's not clear that SVB fits cleanly as either a market failure or negative externality. If we want to broaden the category of negative externalities to include all memetics, we’re going to be talking about a lot of events. Should we consider the Gamestop incident of 2021 to have generated negative externalities because so many other stocks fluctuated wildly? All memetic responses create ripples in the economy, but not all of them are equivalent. Similarly, if we declare a bank run to be a market failure, it would seem like every bankruptcy would have to be evaluated as well.
The idea that SVB might result in the collapse of the system but the Gamestop short squeeze wouldn't is far too subjective a line. It both encourages regulators to cry wolf when their friends are involved, and encourages players like Citadel — the hedge fund that was an institutional player in the Gamestop short squeeze — to try to become big enough, intertwined enough, and risky enough that they know they can be saved if they get in trouble.
Infrastructure
Instead of drawing a line around market failures or negative externalities, a cleaner line to draw is around a concept of infrastructure. The idea that banks are infrastructure seems intuitively appealing, but hard to square with common definitions. For example, Merriam Webster defines it as
the system of public works of a country, state, or region
Banks aren’t public works like a dam or a road. They aren’t usually managed as utilities like the power company. How can we call them infrastructure in a coherent manner?
Tech companies use the term infrastructure to mean pieces of technology that can be built further upon. There are infrastructure companies, like Amazon, who produce systems like Amazon Web Services that other tech companies build their products upon. These are differentiated from "tools" in a very fuzzy way. Perhaps the simplest way to differentiate between the two is to look at what happens when a piece of infrastructure stops working: usually, the tech company that is built on top of it is panicking because their customers are panicking because the system is broken. Tools enable people to go faster, but when they break, we can shrug and go looking for a different tool. Infrastructure, however, we depend upon.
Let's take that idea and bring it back to society. Many things that we now rely upon used to be nonexistent or rare, and then through the actions of the private market they became first popular, then common, then ubiquitous. Once they become ubiquitous, it's often hard for people to remember or fully imagine that they were ever new or rare. Occasionally these ideas or technologies became regulated by the government, or even become fully publicly operated (though more often the former). The most obvious of these are the centralized delivery of electricity, clean water and sewage treatment - the things we classically view as utilities. In some districts these are managed privately, while in others they are managed municipally.
For thousands of years society operated without electricity, piped clean water or sewage disposal. If you take a tour of Pompeii, they'll tell you about how terrible the city likely smelled: sewage was simply thrown into the street. The aqueduct of Rome is famous for bringing water into the city, but only the very wealthy had water in their home. Yet today we are often seriously distressed by having the water or electricity shut off for a few days. We’ve become thoroughly accustomed to — even dependent upon — these services.
Outside of the utilities, there are plenty of other examples. Paper money was once a brand new invention, created by banks to make it easier to conduct business. Each bank had their own money, and the wise businessman considered carefully which scrip to accept. In the early days, many people would never have touched a bill. It wasn't until the 20th century that issuing paper money became the sovereign province of the state and bills became ubiquitous.
Internet access went quickly from rare to ubiquitous. In the early 1990s only a small percentage of people had access to the internet, and fewer still were hardwired to always be connected. Today 93% of people in the US use the internet, and 77% have a broadband connection at home. Another 15% connect only with a smartphone.
Academic economics has a concept called a public good and sometimes people conflate that with the idea of infrastructure. Here's a recent article describing why public transit doesn't meet the criteria of a public good in response to another article claiming we should fund public transit because it’s a public good. The definition of a public good is about something being nonrivalrous (there can only be one) and nonexcludable (you can't allow some people in but not others). The term public good is a useful one to describe a particular category of infrastructure, like sidewalks. But banks clearly don't fit into this category, so if we draw the line of governmental intervention around public goods, we would have to narrow our view dramatically beyond the modern technocratic one.
You'd think we could point to the concept of public utility as the same as infrastructure, but the legal work around utilities doesn't draw a line that's easy to follow. The law determining which businesses qualify as a public utility is complex and highly politically motivated. Companies that might otherwise qualify have learned how to avoid being categorized as a utility because they seek to avoid the regulatory burden.
This set of requirements from Ohio relies upon your understanding of the words "essential" and "unreasonably".
First, the entity should provide “an essential good or service to the general public which has a legal right to demand or receive this good or service.” Next, the entity must provide that good or service “generally and indiscriminately.” Finally, the entity must have “an obligation to provide the good or service that cannot be arbitrarily or unreasonably withdrawn.”
Why isn't internet service a public utility? It would seem trivial to argue it satisfies any reasonable set of requirements. Sadly, the US Government doesn't agree.
The position of the U.S. government — not to mention phone and cable companies — is that the internet is a free-market service, full stop. It’s not a utility.
Ajit Pai, chairman of the Federal Communications Commission, says the internet industry merits only what he calls “light-touch” regulation, which is to say hardly any regulation at all.
Note how Pai’s argument isn’t whether or not it’s infrastructure, but about how much to regulate it. It’s very likely that Pai is entirely viewing the question in terms of how much regulation he thinks will be helpful (or good for his friends, if you hold the corrupt view). Because the question of being a utility is inextricably bound up with the facts of how we regulate utilities — and regulation is fundamentally politicized — the arguments here are often really muddy. I want to step away from the question of whether and how to regulate infrastructure, and focus on defining what it is.
My working definition of infrastructure is a technology or service that society assumes will be consistently available sufficient that other aspects of society can be built on top of it.
In this view, we could set out a test for something being infrastructure if it is:
Widely or universally available
If offered competitively, it must be at least somewhat standardized, such that switching isn't unthinkable
Assumed to be the default path.
Universality
To be infrastructure, something must be generally available. This seems obvious - it's hard to depend on something that's not available wherever you might expect to need it. The field of evolutionary biology can give us an interesting take on this aspect. As species evolve, some features go from new and unusual things that are active aspects of inter-individual competition to becoming species-universal. For example, while some humans have blue eyes, some have green, and some brown, humans generally have 2 eyes.5 We don't have some members of the species with 3 and some with 1, and over time expect to see one of these win out as a side effect of survival of the fittest.6
Imagine that some individual has a mutation to their eyes which allows them telescopic vision. They can muscularly control the lenses of their eyes to zoom in on something miles away. If for some unimaginable reason their mutation depends upon the specifics of green eyes, this mutation is vastly less likely to survive. Only if they have children with another green-eyed person -- relatively rare within the population -- will the mutation have a good chance of persisting.
Meanwhile, if it only depends upon the universal aspects of human eyes, the mutation will have a good chance of succeeding in the next generation, even though the eye is an enormously complex set of structures requiring many genes. So we see adaptations appear on top of previous adaptations only when those previous ones have become species-universal -- when they've become infrastructure.
To return to the topic of SVB, let's compare now to the 19th century. Throughout the 1800s it was entirely possible to live a perfectly normal life in society without ever interacting with a bank. Today 94% of Americans have a bank account.7 Banks are sufficiently universal that businesses assume that they can depend and build upon them. It is this very dependence and assumption of dependence which led to the SVB collapse being worrisome. Businesses do not (in recent years) spend serious effort on mitigating the risk of their bank collapsing. You could argue that this is negligent of them. I don't think it is, though. The very trustworthiness of the banking system that leads businesses to neglect the risk of bank failure is the trustworthiness that allows businesses and individuals to spend their limited cognitive effort on other risks. The dependability of infrastructure lets us build up more complex structures which allows for society to serve more complex desires.
In fact, there is a whole other level of infrastructure which has been built on top of banks, which was also threatened by the SVB collapse: Payroll Service companies. These are companies which manage the payroll, withholdings and tax paperwork aspects of paying employees. Today, very few companies do this on their own. Instead, they rely on companies like ADP, the elephant in the industry, and Rippling, an upstart newcomer.
Except, oops, Rippling banked with SVB. As the news about SVB's collapse spread, one of the biggest concerns that floated around Twitter was that companies who relied upon Rippling wouldn’t make payroll. In the end, Rippling found loans and another bank they could work with and prevented catastrophe, but it highlighted the extent to which companies think of payroll services as infrastructure. The entire structure of W2 employment and the assumptions that employees make around it are baked into society at a number of levels.
Interchangeability
If it’s standardized, there’s a good chance it’s infrastructure. Payroll Service Providers and banks take some work to swap out, but the services they provide are pretty similar to each other.
Even when a service seems to be tied to the way that it's delivered a new upstart can come along and provide it differently but still interchangeably. For example, internet service is clearly infrastructure, and SpaceX's Starlink service doesn't disprove that by providing the service via satellite. Sure, the mechanism is waves, not cables, but the actual offering is connectivity.
In the 1990s, internet service wasn't standardized. Prodigy and America Online offered connectivity, but also a suite of specific tools and services which weren't available via the general internet. For a time, it was easier to talk to other AOL subscribers if you were on AOL. Today, however, internet service is pretty interchangeable.
Invisible/Assumed use/Default
What does “assumed to be the default” mean? Consider things that you do as part of becoming an adult, or starting a business, or performing some other regular activity of society. New adults get cell phones (or, these days, rely on the ones they got as teens). Everyone assumes that the houses they buy or rent will have kitchens so they can cook and feed themselves, bathrooms so they can clean themselves, and heating so they don’t freeze. Landlords don’t specify these things in advertisements to try to differentiate their homes from those who don’t offer these things. Once they did so, but not since they’ve become ubiquitous. Businesspeople find an accountant when they first start a business, and a payroll services company when they start employing people. Employers once specified a typing speed and familiarity with spreadsheets in job descriptions, but today this is simply assumed for many roles, as familiarity with computers has become ubiquitous.
The reality is that infrastructure fades into the background because we assume it’s there and forget about it. There's a delightful book called, simply, Infrastructure, which is a field guide to the structures of modern society. In over 500 pages of dense prose and pictures, this book lays out a wide variety of large, visible things that nearly no one notices on a day-to-day basis. These objects are so ubiquitous to the landscape that we can forget that they’re there.
Infrastructure is a Spectrum
With this expanded view of infrastructure, it's wrong to view it as a binary. Things aren't infrastructure or not. Instead, as technologies and services become more ubiquitous, they may, if society views them as useful to build upon, go down the path of becoming infrastructure. Many things are ubiquitous but aren't infrastructure, such as fancy restaurants, jewelry stores, or most forms of entertainment. I love a good rollercoaster but I don't depend upon it.
Infrastructure can also stop being infrastructure. Fax machines were ubiquitous in the 1980s, and could probably have been considered infrastructure. Certainly nearly every white collar business had one, and business cards often showed a person's fax number. Today they are rare because email has replaced them.
In many other cases, companies bundle infrastructure in with their offering to build a competitive moat. One can imagine the video delivery side of Netflix, Disney+, etc to be potentially useful infrastructure, but the companies are determined not to become interchangeable. Amazon is many companies in one, several of which are infrastructural - especially AWS — which is infrastructure for building tech companies — and Fulfilled By Amazon (FBA) and the Amazon Marketplace, which are both infrastructure for running ecommerce businesses.
Why should infrastructure be governed?
Companies do a perfectly adequate job of providing a wide variety of infrastructure in modern society. Why should we prefer infrastructure be managed differently than it is now?
The power of the free market as a tool of a great society is how profit motives and competition drive innovation. Society gains when individuals are incentivized to create/discover what we want to buy. The vast panoply of offerings that exist today were all created by that mechanism. This is an evolutionary ecosystem that responds to our changing circumstances and desires better than anything else we’ve devised so far. Despite the many flaws of capitalism — such as its constant incentive to externalize pollution — no other system proposed has proven to generate the kind of abundance that we live with now.
The problems with capitalistic infrastructure arise from the blending of enormous scale with the invisibility and dependence that characterizes infrastructure. At enormous scale, companies often lose their incentive to innovate. Why bother building something new that might be worth millions of dollars for the first few years when your core business is worth billions or more? And you certainly don’t want to build anything which might reduce the need for current customers to spend those billions with you. If someone else tries to innovate in your billion dollar market, you crush them through any means necessary. When you combine that incentive with the way that we become dependent upon infrastructure, you get a bad mix. Whether we’re talking about private equity firms squeezing profit out of nursing homes or hospitals or the way that internet service providers in the US have largely stagnated ever since they’ve been able to establish local duopolies that don’t have to compete, privately managed infrastructure consistently seems to fail to deliver on capitalism’s best attributes.
Additionally, privately managed infrastructure often creates have-nots within our culture. Infrastructure is bound up with the idea of cultural expectation, but private companies don’t have to provide their services in situations where it’s unprofitable, or even if they just don’t want to. Rural people struggle to get good internet access because it’s not particularly profitable to provide those services. If we consider internet access as simply another thing that the private market offers, then it makes sense that the market may not offer it in some situations. If we consider it as infrastructure that allows people to participate fully in our culture, having people who cannot access it is a significant equity issue. Sometimes the banking system decides that OnlyFans or similar isn’t something that it wants to provide its services for, and this is currently condoned as a thing that privately managed companies are allowed to do. We should not tolerate that of something we explicitly agreed was societal infrastructure.
Run By The Government
Does that mean infrastructure should be run by the government? While I will be describing a system that I think is capable of managing infrastructure in the public interest, I don’t think our current governmental systems are good at doing so. Those who are against our current market system often point to Norwegian socialism (unless they’re ignorant enough to advocate the Soviet brand). Norwegian socialism may be better than what we have, but it may well depend for its success upon specifics of its culture that are easy to miss and hard to intentionally replicate (and if you’re advocating Soviet socialism, you should read more history).
We've had a decade of the US Digital Service doing their best to bring government into the digital age, and this has provided us with plenty of evidence that the US government is just bad at technology. With tremendous, focused effort we're able to build something like Healthcare.gov that isn't a fiasco, but the majority of government services are so difficult to interact with that people have created a term for the waste they inflict on us: time tax. Government taking responsibility for more systems in our current world is a recipe for bad, decaying systems that we all grumble about, not a plan for success.
To be fair, infrastructure should be offered to everyone. To be innovative, infrastructure should not be managed by companies with an incentive to prevent innovation, but also new infrastructure should be inventable by the private market that has the incentive to do so. To be reliable, infrastructure should be offered by organizations that are deeply competent at managing it.
If you find my work interesting, please share and subscribe. It helps tremendously.
I’ll return to these organizations in another section.
Why not liberal, conservative, Democratic or Republican? I find that these terms lack the intellectual consistency of Libertarian and Socialist, so I doubt we'd even agree about what it looks like to review this situation through those lenses.
It says something about our society that I feel the need to defend this view, but I do. Perhaps someday I’ll write something more in depth in support of capitalism with all the qualifications necessary to reflect on the damage that it also does to us.
There are also positive externalities, which are when a private actor generates value that the rest of us experience. A great example of a positive externality is the story of noncompetes being illegal in California. According to legend, because noncompete clauses in employment contracts were legal in Massachusetts, but illegal in California, California's tech industry benefited from skills transfer between companies. This wasn’t immediately a benefit to the early companies (they just had to work harder because employees left more easily), but eventually led to a vibrant ecosystem of tech-company skills within California that has made it easier to find great people. Now, Californian tech companies are so strongly entrenched that they attract great people to move to California to start or join tech companies, reinforcing the cycle.
Is born with the genetic blueprint for.
Survival of the fittest being a descriptive term for whoever survives having been a combination of luckiest and best adapted for their environment, not a normative term describing which thing we like best.
Though 13% also use alternative financial services and are considered by the Fed to be "underbanked".