New Yorker writer John Cassidy has posted on his website a 1996 interview with the late Paul Samuelson. My favourite snippet concerns the use of mathematics in economics.
I asked Samuelson whether mathematics was now too important in economics.
Rather than answering the question directly, he talked about a lecture he attended in the nineteen-thirties by Lionel Robbins, a well-known professor at the London School of Economics. “Lionel Robbins gave an address saying this math stuff is just a passing fad. I was all of twenty-eight, but I thought, ‘Poor fellow, he just doesn’t realize that he’s missing the train.’ That was just a bad understanding of the dynamics of the profession. Math is a problem for everybody in the profession and it has been for years. We all say, math should be used just up to the point that I have used it, and no more…I always say to our graduate students when they are leaving: ‘As a graduate student at a top-notch university, you tend to lose touch with reality. You have been engaged in puzzle solving and learning a new language. When you emerge, you may tend to think you have been asleep for several years.’ The paradox is that the best people in practical terms are the Jim Tobins, the Bob Solows—the guys who are awfully good at the technical stuff as well.” Samuelson also brought up his colleague Modigliani, whose parking space he may have been occupying, noting “he has done more for Italy than pizza,” and the prevalence of technically adept M.I.T. graduates in the Clinton administration. (They included Lawrence Summers, Joseph Stiglitz, and Laura Tyson.)
“Like herpes, math is here to stay,” he said. “It takes strong math to defeat misleading math. For example, ordinary least squares”—a standard statistical method—“is misleading. It takes more mathematics than ordinary least squares to understand three-stage least squares, co-integration, or unit roots, all of which are improvements on ordinary least squares. But it does lead to a communication problem. The number of people who can communicate effectively, like Paul Krugman, is very small.