Are Analytics Responsible for the Financial Crisis?
February 9th, 2010 @ 6:00 am
Last week I spoke with Analytics at Work co-author and Babson College professor Thomas Davenport about the ways in which analytics can benefit organizations. This week, the conversation turns to why it’s imperative to use analytics responsibly.
BNET: The book briefly discusses the financial crisis, saying that a good portion of the financial services industry used analytics in the wrong way.
Davenport: It’s a very critical issue for business education. We basically have created two classes of people: quants and non-quants. They didn’t communicate very well in a number of companies during the financial crisis. The quants could develop these financial algorithms that made it look highly desirable to enter into a series of complicated derivatives and things like that. The non-quants — who tend to be senior managers — didn’t understand that, didn’t understand the assumptions behind them and didn’t know when the world was changing, so as to make those algorithms invalid. The quants haven’t been good at explaining what they’re doing in simple terms. Non-quants haven’t been diligent enough to delve into the assumptions behind the models and to know when they work and when they don’t. As a result, we have a really severe recession and a number of firms went out of business. So the stakes are high.
BNET: What are the keys, then, for using analytics more responsibly?
Davenport: One thing is for every manager to know something about quantitative analysis, so they can at least raise reasonable questions. The other thing is to realize that every model is a representation of reality. You need to know what are the assumptions behind the model. You need to know that this particular model only makes sense to base our operations on if housing prices are rising, which was the assumption behind a lot of the mortgage derivatives and so on.
Another common assumption that wasn’t widely known was that if you’re doing risk analytics, the most commonly used assumption is called value-at-risk. For different investments, their values are independent of each other. We found out in the financial crisis that if one class of investments fall, the others are likely to fall as well. So you need to state clearly what those assumptions are, and then managers can decide in straightforward terms, am I willing to take this risk?
BNET: What role does b-school play in analytics training?
Davenport: We have these quantitative courses in business school, but they haven’t really succeeded in creating a generation of well-equipped managers for this analytical world. There are some requirements now for business schools to do a better job of making the non-quantitative managers that they graduate more quantitative and making quantitative people better communicators and better at engaging in dialogue with decision makers about what the data and analyses really say. But I don’t think we’ve done a good job of preparing either group.






Harvard Business Review recently published its list of the 100 best performing CEOs in the world




