The Demise of the Bond Trader

The big losses being racked up by the banks and brokers these days can be directly attributable to one thing: the demise of the bond trader. More specifically, it is the demise of the art of bond trading in favor of the science of bond trading. Since the advent of widespread use of computers on Wall Street for valuation purposes, the art of bond trading has been pushed into slow but steady decline. Twenty years ago, the most sophisticated piece of equipment on a bond trader's desk was a Lane or Monroe bond calculator. It was the explosive growth of repackaged MBS pass-thrus (CMOs) that created the need for more sophisticated methods of determining value. Back then, the real prestige item for Wall Street firms was a Cray Supercomputer (do they even exist anymore?), used in calculating all the cash flows and other variables that went into structuring a CMO. The rest is history. There have been hiccups along the way, most notably in 1993-4 when the mortgage prepayment models didn't adapt to the changing interest rate/product environment and in 1998, when genius failed, as the book was so aptly titled, to account for reduced liquidity. It turns out, or perhaps more correctly, will turn out that 1998 was just a warmup for 2007. When the art of trading finally pushed back against the science, liquidity ran for the hills.

I'm no Luddite. My writing this blog should prove that. Nor am I suggesting that the world go back to writing prices and spreads down on little pieces of paper taped all over the desk. The problem comes about when a mathematical model completely supplants common sense. Management, for the most part, loved this because a model provides a definitive answer. Management could then take the model-derived answer and show it to the CEO, Board of Directors, major investors, regulators, etc. and say "See, here is our exposure, here is our risk, and here is what we've done to mitigate it. You may rest assured, now." As is publicly known now, and privately talked about for awhile, the inputs that were used to generate these model-derived values, particularly with regard to liquidity, were faulty to say the least. The models, generally very complex mathematical formulas taking into account a whole host of variables, are developed by some of the finest technical minds in the world, but have little practical market experience. Many are PhDs in the "hard" sciences, Physics and the like. Most of the people that use the models are not finest technical minds in the world, but have some or much practical market experience. Therein lies the disconnect. The consumers of information, always looking for better and faster ways to make money (a good thing, by the way), have become so reliant on the neat answers that models provide them that they ignored common sense and forgotten how to trade and the risks of trading.

If there is one lesson learned by all of this, it should be that the decision makers and risk takers need to get in touch with their trading roots and apply good old-fashioned common sense to the model equation. Securities that are packaged and repackaged and repackaged and repackaged again deserve more scrutiny, on all levels including rating, then they have received in the past. Values need to reflect all the risks, including the risk that the value may not be able to be determined. Maybe then, the trader's "bid out" would have happened sooner, sparing many the problems of the past few months. Probably not, but it is a wishful thought. The good news is that some happy medium will develop, new methods will arise, and the bond market will go on its merry way.