In the March 26 New York Times, columnist David Brooks raises this question. The context is the financial market meltdown. Brooks writes, “Economists and financiers spent decades building ever more sophisticated models to anticipate market behavior, yet these models did not predict the financial crisis as it approached. In fact, cutting-edge financial models contributed to it by getting behavior so wrong — helping to wipe out $50 trillion in global wealth and causing untold human suffering.”
Unfortunately, Mr. Brooks has misinterpreted the role of models in the crisis. Way back in 1971 when I built my first econometric model one basic fact was drummed into my skull: don’t extrapolate outside the range of the historical data used to build the model.
That’s exactly where economists went wrong. There was no — zero — historical data on subprime mortgages, collateralized debt obligations, and the host of other derivatives built around the subprime market. Yet economists extrapolated existing data from reasonable quality mortgages into the subprime arena. Big mistake.
The fault, dear reader, is not the models. It’s their abuse by people who knew the results they wanted and used the models to justify those results.