Market Structure

Market structure is the part of finance nobody thinks about until it stops working. Every trade that clears, every price that sets, every benchmark that resets overnight depends on a stack of conventions, intermediaries, and trust assumptions that took centuries to assemble and can be compromised in weeks. When the plumbing fails, the damage is never local.

This hub is about the plumbing. The Buttonwood Agreement — twenty-four brokers agreeing in 1792 to trade only with each other at a minimum commission — is the root commit of what would become the NYSE and, eventually, the market microstructure of most listed securities on earth. Lloyd's of London began as a coffee house where ship captains compared rumours and left, three centuries later, as the market that insures spacecraft and footballers' legs on unlimited-liability syndicate capital. LIBOR was a set of daily phone calls between sixteen banks' treasury desks that somehow ended up underpinning 350 trillion dollars of contracts, right up to the point where it became obvious nobody could trust the phone calls.

The failure cases are equally instructive. The Flash Crash of 2010 revealed that the electronic market microstructure nobody had designed end-to-end could evaporate liquidity in a thirty-six-minute window. Enron and Madoff did not exploit exotic instruments; they exploited the gap between what auditors, rating agencies, and regulators were checking and what they said they were checking. The Long Depression of 1873–1896 and the Silk Road's trade network extend the same question backwards: how the structure of commerce itself determines what can and cannot happen inside it.

What this topic covers

• The specific mechanism that makes the structure work on quiet days
• What it costs to maintain, and who pays
• What happens when a single node in the graph fails
• Which parts of today's plumbing are load-bearing in ways most users do not realise

A timeline of canonical structures and their failures

Why structures fail in characteristic ways

Each structure in the hub has a vulnerability that traces back to the assumption it was built on. Lloyd's was built on the premise that unlimited-liability Names would price marine risk; when asbestosis claims started appearing in volumes nobody had imagined, the Names lost their shirts and the Equitas reinsurance solution was retrofitted. The Buttonwood self-regulatory model worked because brokers shared a small physical space; it took a century for the same mutual-monitoring structure to start failing as electronic trading detached transactions from the trading floor.

LIBOR's vulnerability was structural: any benchmark that depends on judgement-based submissions from competing institutions is rigging-prone. The Flash Crash's vulnerability was emergent: the modern microstructure of electronic markets was nobody's responsibility to test as a system. Settlement risk had a name only because Herstatt's failure revealed that a multi-trillion-dollar daily process — banks wiring foreign currency at one time of day, expecting return wires at another — had no centralised counterparty if either leg failed. Each failure is a story about an assumption that everyone made and nobody audited.

Structures that mostly work

The hub spends a lot of time on failures because failures are visible. The structures that work tend to be invisible — the FedWire payment system handling trillions a day for half a century without a public outage, the SWIFT messaging network passing roughly forty million messages a day across more than 11,000 institutions, CLS Bank settling about $6 trillion of FX trades a day under a payment-versus-payment model that was Herstatt's direct response. None of those have stand-alone articles in the hub yet, partly because successful infrastructure is not eventful and partly because the documentary record is sparser. They are nonetheless what most of finance runs on.

Modern parallels

The 2023 collapse of Silicon Valley Bank revealed a structural feature few people had foregrounded: deposit concentration risk amplified by mobile-banking-speed runs. The plumbing assumption underlying FDIC's $250,000-per-account insurance was that retail depositors run slowly. When a venture-capital text-message group can move tens of billions in a morning, the assumption stops holding. The structural fix — the Bank Term Funding Program announced 12 March 2023 — was a patch on the same wholesale-funding plumbing that has needed patches since Continental Illinois in 1984.

Start here

• The Buttonwood Agreement — twenty-four brokers, one tree, the architecture of the NYSE
• Lloyd's of London — a three-century coffee house that insures almost everything
• The LIBOR scandal — what happens when a benchmark turns out to have been a group chat

For the failure cases, follow with Herstatt and the Flash Crash. Together they bracket fifty years of payment-system and microstructure failure.