StructuringHard
4 min read · 696 words
What the seat actually does
Structuring starts where the shelf ends. A client has a constraint — an accounting treatment, a capital rule, a currency they may not hold, a return they need to look a certain shape — and no listed instrument produces it. The seat assembles one out of things that do exist, prices it, and works out how the desk will hedge what it has just promised.
Every structured product is a bundle with a wrapper. Take a zero-coupon bond and an option and you have a capital-protected note. Take a short put and a coupon and you have a reverse convertible. The work is not inventing new mathematics; it is choosing the bundle that answers the constraint and being honest about what it costs.
- Reverse-engineering the ask — what the client says they want is a shape; what they need is usually one clause of it.
- Assembling — the components, from the desks that already quote them.
- Pricing the whole — including the parts that cannot be hedged perfectly, which is where the margin and the risk both sit.
- The document — which is the product. Everything above is a description of it.
A day, and where it goes
- Requests — from sales, usually as a shape rather than a specification: "something that pays if rates stay in a range".
- Modelling — the payoff, then the hedge, then what the hedge costs when the market is not obliging. The second and third take longer than the first.
- Termsheets — a page a client can read, and behind it a document a lawyer can defend. When those two say different things, the second one is what the product is.
- Post-trade — living with it. A structured book is a portfolio of promises whose hedges have to be adjusted for years.
What it is measured on
- Margin on issuance — the difference between what the components cost to assemble and what the note was sold at, which is the clearest fee on this site and the least visible to the buyer.
- Whether the hedge held. A product priced with an assumption about correlation is a product whose profit depends on that assumption being roughly right for its whole life.
- Reissue. The same structure sold again is worth more than a clever one sold once, because the second one costs almost nothing to build.
- Complaints, eventually. A product that behaved as described but not as understood is a cost that arrives years later.
What it touches on this site
- What gets built — autocallables, reverse convertibles, capital-protected notes and the rest of equity derivatives.
- What the wrapper changes — the same exposure in four different containers, which is this seat's whole argument.
- What it costs — costs and fees, and reading a term sheet.
- Where it is hedged — trading, which inherits every promise this seat makes.
How it goes wrong
- Complexity that is priced but not explained. A payoff with four conditions has four ways to disappoint, and a two-line summary describes one of them.
- Correlation assumed to be stable. A a note that pays on the worst of several shares is a bet on things not falling together, sold in the years when they did not.
- The hedge exists in a model and not in a market. The component that balances the book is quoted by three people and none of them on the day it is needed.
- Issuer credit forgotten. The note is an unsecured claim on the bank that wrote it, whatever the underlying does — see 2008.
Concepts to master
- Every structure decomposes. If you cannot write it as a sum of a bond and some options, you do not yet know what it is.
- The margin is in the assembly, not the idea. Components are quoted competitively; the bundle is not quoted at all.
- Path dependence is where the surprises live. A payoff that depends on what happened in between behaves nothing like one that depends on where it ended.
- A capital guarantee is a credit exposure. "Protected" says what the formula does, not who has to pay it.