Margin Call

The model worked. That is the film's first and most unsettling point, and it is easy to miss because every popular account of 2008 wants the models to have been wrong. The risk models at the bank were not badly built. They were built on historical data, calibrated within reasonable parameters, reviewed, and they performed exactly as designed. What they could not do was account for conditions that had never occurred in the data set they were fitted to. When Peter Sullivan takes his predecessor's unfinished file, extends the time horizon and watches the lines diverge, what he discovers is not an error but a boundary: the point at which the model stops describing reality and starts describing itself. Every model has this boundary — every forecast, every capacity plan, every recommendation engine, every estimate a team gives about how long the work will take. The question is never whether the boundary exists. It is whether anyone is looking for it, or whether a long run of accuracy has made it invisible. A model that has been right for years is not a model anyone is auditing; it is furniture.

The information then travels upward through the bank in a single night, and Chandor stages the ascent as a series of translations. The analyst explains to his desk head, who explains to the division chief, who explains to the head of the trading floor, who explains to the CEO — and at each level the listener understands less of the mathematics and more of the consequences. By the time it reaches John Tuld, the request is for an explanation pitched at a young child or a golden retriever, delivered by a man who has just arrived by helicopter and who will make the decision regardless. It would be comforting to read this as satire of executive ignorance, but the film is doing something harder. Tuld is not stupid; he is optimised for a different task, and he says so plainly when he claims that the only durable advantage in the business is being first. This is the management problem the film is actually about. In any sufficiently complex organisation, the people who understand the problem and the people with the authority to act on it are not the same people, and the understanding has to cross that gap intact — usually at speed, usually at night, usually through several intermediaries who each need it in a different language.

The fire sale is presented without melodrama, which is what makes it land. Dumping the bank's positions on its own clients before the market understands what is happening is not staged as a conspiracy or a moral crisis; it is staged as a long, tiring shift. Nobody wrings their hands. Sam Rogers objects — he tells Tuld that they are selling something they know has no value — and then he goes downstairs and tells his traders to sell, having been informed that there will be no future in the industry for any of them and that they will be paid extremely well for the morning's work. The survival imperative here is not a villain's plan. It is a gradient, and everyone standing on it slides the same way, each for reasons that are individually comprehensible: the mortgage, the divorce, the children's school, the belief Will Emerson articulates that the whole arrangement only holds up because people like him keep their fingers on the scales. Chandor's restraint is the most damning thing in the film. He shoots institutional self-preservation as ordinary work, joyless and competent, performed by people who are neither monsters nor martyrs and who will be at their desks again on Monday.

And then there is Eric Dale, who knew. He was head of risk management, and he was fired that morning in a routine round of layoffs, walked to the lift with a cardboard box, his network access revoked and his mobile phone cut off before he reached the pavement. The analysis that would have surfaced the problem days earlier was on his machine, unfinished, handed to a junior colleague on the way out with the observation that he had been working on something nobody was going to like. This is not an act of villainy either. It is a cost-cutting measure executed correctly by human resources, on schedule, according to policy — and it is the most expensive economy in the history of the firm. The firm spends the rest of the night sending people out into Brooklyn to find him, because the knowledge left the building with the person and there was no mechanism for separating the two. Every organisation that has made headcount savings by removing the people who held its institutional memory has taken a version of this bet. Most of them never find out what it cost, which is precisely why they take it again.