The Economist: Can forecasting provide anything worthwhile?
When I emerged from graduate school and went to work as the research economist for a big bank holding company, I determined to build the definitive computer model of the national and regional economies. It was going to be better than what anyone else had done and forecast with amazing accuracy.
For my first few years there, I worked on my model. R-squares and t-statistics improved, kurtosis declined and no breath of multiple correlation was allowed to persist. The models didn't forecast the future particularly well, but they were brilliant. And dialing up the giant mainframe in Lexington, Mass., and playing with its huge data base with my newly acquired skills was a lot of fun. (I also enjoy balancing my checkbook and working sudokus and crosswords.)
Soon after my experience with the bad employment data, I learned another important lesson that hadn't been taught in graduate school - pay attention to the people who have to meet a payroll every week, no matter what the official data are telling you. I was speaking to a group of homebuilders in a small community north of Denver at a time when Colorado and the Mountain West were in the grip of a serious housing recession. When my staff brought me the data for my speech, Longmont was a shining light in a black hole of negative statistics. Housing starts were up 225 percent over the previous year!
I introduced my remarks to the homebuilders by saying how pleased I was to be speaking to a group who had maintained their business at a high level when builders around them were struggling. I laid it on pretty thick, having taken the speech course that says start out by making your audience feel good about you. After a few sentences, a crusty old guy in the back of the room - faded jeans, boots and a tape measure on his belt - raised his hand.
"I don't know much about all these statistics," he said, "or about what it means to be up 225 percent. But I can tell you that last year we built four houses in Longmont, and this year we built 13 and one builder built 12 of those and homebuilding ain't a doin' well in Longmont!"
Today, I still look at the results of other economists' models of the economy. But I've long since ceased maintaining my own models. I take what they conclude, what I know about the current situation, what the fundamental laws of economics and business say occurs over time, what I hear from clients, the occasional newt eye or frog toe and stir them all together in a big pot. If what comes out makes sense, based on what I've learned over 30 years of experience, it becomes my forecast. If it doesn't, I try again.
There's a branch of forecasting called futurology (remember Herman Kahn and John Naisbitt?) that doesn't receive a lot of attention but is closer to what I do than model-driven predicting. Futurologists take a broad, careful look at what is going on around them. They point out that the future is more or less here, if only you can see it. Then they build scenarios, build a range of alternative outlooks, identify warning signals in each scenario and watch closely for these signals to occur. Finally, they encourage their clients to participate in visioning what they would like to have happen, rather than concentrating only on what might happen.
In my 33 years in the economic analysis and forecasting business, I've never seen more confusing or uncertain times than those we are experiencing today. None of us has lived through a financial-driven recession and its aftermath (the last one was in the 1930s). And the debt situation in Europe complicates an already impossible jumble.
Let me finish where I began - no one can foretell the future. But we can approach it thoughtfully with consistent assumptions that can be monitored as the year unfolds. That puts you way ahead of the competition!