McNamara and model risk

By Felix Salmon
July 7, 2009

Philip Delves Broughton notes that Robert McNamara, one of Harvard Business School’s most notorious graduates, basically did in the field of war what Wall Street quants did in the field of finance:

The journalist David Halberstam wrote that McNamara mistrusted people who did not speak his language of statistics and hard data. If it ever came down to one person saying something “just didn’t feel right” or that it “smelled wrong”, he would always go with his facts over their feeling. Fatally, in the case of Vietnam, the data he received was not accurate.

When Wall Street quants fail to account for model risk, they can end up losing hundreds of billions of dollars. But that’s an improvement over what happened when McNamara failed to account for model risk: those losses were much worse.


We welcome comments that advance the story through relevant opinion, anecdotes, links and data. If you see a comment that you believe is irrelevant or inappropriate, you can flag it to our editors by using the report abuse links. Views expressed in the comments do not represent those of Reuters. For more information on our comment policy, see

is crappy data “model risk,” or just crappy data?

Posted by bdbd | Report as abusive

JFK’s conduct in the Cuban Missile Crisis was a sharp rebuke of McNamara. Where JFK’s advisors said attack Cuba before their missile became active (they already were), JFK thought: no, we don’t know all the details of Cuba’s offensive capabilities (nor the climate consequence of Nuclear Winter). The same could’ve been applied to McNamara’s strategy of progressively increasing military involvement. There was a non-linear escalation of Vietnam War that was China aid, and that couldn’t merely be handled by escalating to war on China. And there was a total lack of matching war deaths (Vietnam could’ve been a tiger economy) with the objective (a flawed domino model of Soviet expansion). McNamara “learned” from his mistakes but never translated this knowledge into actions, keeping his qualms against Vietnam to himself. In the end he didn’t even bother to mention to destroy the records of USA allied South Vietnamese before advancing Vietcong siezed them. I’m surprised anyone would ally with such a rat after that.

Posted by Phillip Huggan | Report as abusive

Wall st quants failed to model the human element involved, that is a certainty…which is basically a lending industry that screamed “free lunch for any & all” to get that 2-car garage home, both the cars for the garage, and then the credit for all the junk inside the house.

The derision of modelling, and specifically, that of Wall St quants, is duly noted & well deserving in that respect. Please take to task the bewildering behavior of the huddled masses as well. I do not expect the average citizen in the US to equal the mathematical level of a 2nd year person at Stanford’s MBA program. But for heaven’s sake, can’t we just think a little bit harder about financial decisions going forward. Buying a house became as transactional as buying a darn toaster oven for $39.95.

Posted by Griff | Report as abusive

What about the ‘model risk’ in the global warming debate. We could be investing lots of resources on the results of computer models!

Posted by Eli Baker | Report as abusive

Many of the “quants” on the Street knew that the statistical models relied upon questionable assumptions and inadequate data. But when they expressed their concerns to higher-ups they were told to shut up because the money was rolling in.

Posted by Steve Numero Uno | Report as abusive

Based on my reading of Halberstam’s account of the War,
McNamara had reservations about America’s progress in Vietnam operations even in the Kennedy years (when the war was still at an incipient stage).

His shortcoming was not a tendency to implicitly believe modeled forecasts. Rather, it was the desire to please the boss (Johnson) with pleasant tidings.