Red-Blooded Risk: The Secret History of Wall Street

Middle Office Again

The idea of the middle office is to create two information loops in the bank. The front office sends data to the middle-office risk manager, who processes it and sends it back to the front office. The key difference from my first idea is the middle office puts it in a form the front office can understand—and also that the front office can use to determine which source data to correct. Eventually, the middle office gets it right. At that point, it sends the data to the back office. The back office produces its reports and sends them to the middle office. Again, these reports make sense to the middle office, so it knows how to correct any input errors to get the back-office report correct. This idea actually worked.

The middle office did not replace the front-to-back-office data flow. There weren't enough middle-office people for that and it would have slowed things down too much. The middle office concerned itself only with information that was important for firmwide risk. Although this is a tiny subset of the amount of information flowing around a financial firm, it forces everyone to get things right. Everything affects risk. Any sorts of data errors can result in inconsistencies either in the risk data or between the risk data and what actually happens in the world. Either way, the errors are discovered and fixed.

That's what the middle-office did, but that's not why we built one. The middle office was built by risk people who wanted to tackle the problem ...