The Trouble With Numbers



Let me begin by stating I have nothing against economists. I used to be one. In that life, we built some heavy-duty number-crunching programs that were supposed to predict the future. Type in figures for job growth and our software beast would blow fire out its nostrils as it computed what those figures meant for retail sales, or energy usage, or G.D.P. growth. Not a bad little machine.

I was a kid back then and I believed in that machine. I believed in it until I went to work on a Saturday, saw my boss poring over the keyboard and watched as he rejiggered the model.

When I asked him what he was doing, he told me he didn’t like the growth numbers that the model had spit out, so he decided to make some changes to the program so it would produce a better number. “But if you’re altering the model because you don’t like its results,” I asked naively, “why build a monster model in the first place?”

“Good question,” he said. “But I’m busy right now.”

I thought about that when the June unemployment figures were released in the United States by the Bureau of Labor Statistics. The figure looked lousy. Only 18,000 jobs were added to a work force of 131 million people.

There were reasons for such a lackluster performance, the analysts said. June’s weather was bad across the States, gasoline prices had been rising steadily for half a year, and the global supply chain continued to be disrupted by the terrible events at the nuclear power plant in Fukushima, Japan. As a result, some car makers had too few cars to sell, and some electronics and computer companies ran out of stock. Pretty basic stuff.

But what surprised me were the forecasts made by economists.

Each month, the top 85 economists in the United States are surveyed by Bloomberg News in advance of the jobs report to get their forecasts. This time, the consensus estimate was that the United States would create 105,000 jobs. The high estimate was 170,000 jobs; the low estimate was 40,000 jobs. In other words, not a single economist, out of 85, came close to getting it right.

The reaction of investors to the fact that the economy missed its number was to push the Dow Jones industrial average over the cliff. The markets tanked.

O.K., so nobody’s perfect, least of all a bunch of economists. And economics is hardly an exact science. What else explains the fact that only the tiniest handful of economists correctly forecast the global financial crisis of 2008, and its aftermath — the economic equivalent of sending a group of astronauts into space and watching in horror as they sail past the moon.

But analyzing the jobs numbers is no moonshot. The monthly statistics the economist failed to predict weren’t exotic. They have been compiled by the government and released once a month for a generation. As a result, the miss prompted me to wonder — could all 85 economists have run their models without taking into account the effects of gasoline prices, the weather and the problems in Japan?

That answer is difficult to surmise without wandering into the economists’ offices on a Saturday when their hands are dirty from working on their models.

We invest a lot of time and money trying to ascertain what’s next. And yet, our track record at doing this is spotty, at best. Had we simply focused on the extenuating circumstances of weather, gasoline and Japan, and not on the voodoo of a single number, we might have learned a whole lot more and reacted a lot less nervously. It’s not the forecasts that count. It’s what’s behind them that matters.

Download the PDF