When I was taking my MBA, I had the fortune to work for a very successful man.  I asked him “what’s the biggest mistake that MBA students make when they enter the workforce?”  His reply: “they believe too much in their own models.”

This advice has stuck with me and I’m very cognizant that when I’m doing something, whatever model I have (be it mental or an explicit spreadsheet) is only valid if two conditions holds.  First, I’ve got to have the right set of data/inputs.  Secondly, the logic of my model has to be sound: there’s an assumption in it somewhere and I have to be explicitly aware of what it is.

With this as my philosophy, I was fascinated to read The New York Times Magazine’s article, Triple-A Failure.  It talks about how the bond rating agencies, and Moody’s in particular, helped create the sub-prime lending disaster (current cost: an estimated $250 billion and counting).

The article is long (and well worth reading), but it comes down to this.  The bond rating agencies modeled how likely mortgages were to fail.  They had all the right data.  But they had one wrong assumption: that these mortgages were just like every other mortgage they’d ever modeled before.

Take a dash of “right data”, a pinch of “wrong assumption” and add “leverage” and you’ve got a recipe for financial disaster.

The bond rating agencies believed too strongly in their own models.  Don’t make the same mistake.