New BSDE tools price options after multiple defaults strike

When several assets can default at once, pricing an option stops being a tidy spreadsheet problem. This paper tackles that mess with backward stochastic differential equations, or BSDEs: equations solved from the option’s final payoff back to today’s price. The model mixes ordinary Brownian motion with p default martingales, one for each default time, and even allows a generalized driver that includes an optional finite-variation process. The authors prove existence and uniqueness of the solution, and they establish comparison and strict comparison results under a suitable condition on the driver. In the linear case, they go further and write down an explicit formula for the first BSDE component using an adjoint exponential semimartingale, with the formula changing depending on whether the finite-variation process is predictable or only optional. They then use the theory to price and hedge a European option in a linear complete market with two defaultable assets, and in a non-linear complete market with p defaultable assets. They also give two examples where a large seller can influence default probabilities: one for a single asset, and one for all p assets.

If you have ever watched one bad news story hit several stocks, this is that problem in math form. One default can shake a market. This setup handles p of them at once. It does that for a European option, a contract you can use only on one final date. The surprise is not just that prices move. The surprise is that the model still keeps prices in order. It can even give an exact formula in one linear case. A small timing choice matters too. A process can be predictable, which means seen one step ahead, or optional, which means known when it happens. That is enough to price and hedge the contract without losing track of default risk.

When defaults pile up

The main result is a well-behaved solution. The BSDE has one solution, and only one, under the stated rules. The model also compares prices. Lower inputs cannot jump above higher inputs when the driver, the rule that pushes the equation, meets the right condition. A stricter version also holds. If the inputs differ, the outputs differ too. The linear case goes further. It gives an exact formula for the first part of the BSDE. That formula uses an exponential companion process. Its shape changes if the extra finite-variation term is predictable or only optional. The same theory then prices and hedges a European option in two market settings. One is a linear complete market with two defaultable assets. The other is a non-linear complete market with p defaultable assets. Two examples show how a large seller can shift default odds.

How the backward equation works

A BSDE is a backward stochastic differential equation. It is solved from the end back to today. The random engine has two parts. One is Brownian motion, a smooth random wiggle. The other is p default martingales, bookkeeping terms for each default time. A martingale is a fair-game process. The model also adds a finite-variation process, a term that changes in a controlled way. That term can be predictable or only optional. Predictable means the change can be seen one step ahead. Optional means the model only learns it when the event arrives. The analysis builds bounds first. Those bounds show the solution exists. They also show it is unique.

Why the timing detail matters

This matters because finance often lives at the jump points. Hedging means building a trade that offsets risk. The model gives that trade a place to live, even when defaults arrive one by one. It also lets the seller be large enough to move default odds. That is a better fit for some markets than a tiny-trader story. The optional-versus-predictable split is not a footnote here. It changes the formula itself. So the timing of the extra term becomes part of the price, not just decoration.

What the exact formula opens up

The big surprise stays in view. Several defaults do not break the order of prices. In the linear two-asset case, the exponential formula gives the first part directly. That makes the price path easier to read when default risk jumps. The most concrete next test is the p-asset non-linear market with the same optional-versus-predictable split watched closely. If that split keeps changing the answer in the same way, then the model will stay useful for markets where one seller can move many default chances at once.