Conversations with the Trading Desk

Most crises are crystal clear after the fact, and the subprime crisis is no different. At least as far as it has affected the various banks and investment banks, the problem is that they had too much inventory in something that was risky and illiquid. How did they end up in that situation? Is it something that they should have recognized before they were hit with multi-billion dollars write downs?

I often replay these sorts of problems in my mind. Putting myself in the situation of the risk manager at one of these firms, here is one way I could have seen things playing out over the course of meetings with the trading desks.

Session 1
Me: “Hey, guys, have you noticed that our CDO-related inventory has been growing. It wasn’t too long ago that we had just a few billion, and then it went up to $20 billion, and now nearly $40 billion. That seems worrisome to me.”

Them: “Maybe that’s because you aren’t sitting here on the desk watching these things all day long like we are. Look, you can’t make money in this business without ending up holding some inventory from time to time. At least if you can, be sure to let us know how. And in this case a lot of this inventory is rated AAA. You wouldn’t have a problem if we had a bunch of AAA corporate bonds, would you? I mean, that stuff is a better bet than our company is.”

Session 2
Me: “You know, even if this stuff is AAA and you aren’t worried about defaults, spreads are really low right now. If they go up for some reason – and even up to somewhere in the range of their historically “normal” levels – we could see a multi-billion dollar write down on this inventory.”

Them: “Oh, right. Good catch. We’ll put something in place for a hedge against credit spreads widening.”

Session 3
Me: “This stuff that you are saying is AAA. What if those ratings are off? I mean, it is not like the rating agencies have any experience with these sorts of structured products. They are basing the ratings on historical default probabilities and historical correlations. Don’t you think with all the changes we are seeing in the market – the very explosion in the issuance of CDOs, for one – that these correlations could be different than they have been in the past? And, for that matter, why should we be betting on what rating agencies have to say?”

Them: “Well, if you have a better way of assessing the risk of these things, let us know. Otherwise, do you mind if we get back to work. We’re trying to make money here.”

Session 4
Me: “I’m sorry to bother you again. But, maybe we have been focusing on the wrong thing here. The issue for us is not so much if these are really AAA or not; our concern is not just with defaults being higher than we expected. The issue is whether these could trade substantially lower for any reason. And I think I have one scenario that could lead to much lower prices: The instruments in our inventory are not very liquid. That is, after all, the reason we have ended up holding these things in our inventory rather than pushing them out to a client. So what if someone suddenly is forced to liquidate, a hedge fund, say, that gets into trouble. The price in the market could plummet. And it wouldn’t take much of a price drop to hit us hard. When you have $40 billion of this stuff, a ten percent drop will lead to a mark to market loss that will wipe out all the profits you guys have made over the last few years.”

Them: “Good to see you are cranking out the scenarios. But if we worried about every little “what if” that you can cook up, we wouldn’t be doing anything. We are risk takers. That’s how we make money. So if you want us to stop making money – or if you have a better idea on how to do it without taking risk – then let us know. Otherwise, do you mind if we get back to work here?”

The point in this set of conversations is that the risk manager is always at a disadvantage when dealing with the trading desk.

First, the traders obviously know their market better than the risk manager can ever hope to. So if a concern gets elevated to the point of an us-versus-them debate with the traders on one side and the risk manager on the other, the traders will be able to run circles around most risk managers. So a firm either has to get the traders to share the same mindset as the risk manager, or the management of the firm has to handicap the risk manager if things come to blows.

Second, for each blowup that occurs, there will be ten cases where there was a legitimate concern but nothing happened. That is, if measured based on the realized outcome, the odds are the risk manager will be wrong more often than he is right. So there is the risk of looking like the boy who cried wolf.

Another point which is not expressed in the set of conversations above is that in most large firms the risk manager makes himself too busy to really focus on risk management. If he lets himself get sucked into making the role look weighty, he will end up spending his time running an organization, worrying about having adequate face time with senior management, and elbowing his way into all the right meetings. What he should be doing instead of all that is spend time trying to think outside of the box. There are possibly hundreds -- in some cases thousands -- of people in his organization, reams of risk reports to run through and meetings all day long. The risk manager can end up looking really, really busy while not actually doing his job.

On this score, I have suggested to a number of people at Citigroup (and the same points would hold true for Morgan Stanley and probably any number of other firms) that the real risk manager should not have people management and report generation responsibilities. He should be able to have the time and space to question and think. He should be able to use all the risk data as an input and demand other types of analysis he deems necessary, but then have the time to sit back and think. In this respect, his role would not look much different than any number of very successful portfolio managers. (I treat Citigroup in a chapter of my book entitled "Colossus" as an example of an overly complex, and thus crisis-prone organization).