Skip to content

A Tale of Two Models

There’s been a lot of media coverage this year about an apparent dispute between meteorologists and climatologists regarding the evidence of climate change. (Example.) Precipitating the most recent storm of opinion has been the release of a survey by the Center for Climate Change Communication at George Mason University. Of the nearly 600 members of the American Meteorological Society who completed at least part of the survey, nearly half responded that they did not believe global warming was happening (25%) or that they did not know whether it was happening (21%). Given the visible role that meteorologists play in informing the public (they are largely TV weather forecasters), their perspective plays an important role in how the public forms its opinions on climate change.

On several occasions, I have heard a meteorologist defend denial of climate change through some variation of the argument “I know all too well that beyond four or five days out, our forecast models are meaningless . . . how on Earth am I supposed to believe a model that goes 40 or more years into the future?” For example, see CNN’s Chad Myers. (In this particular clip, Mr. Myers goes so far as to accuse climatologists of inventing climate change for their own personal financial gain). And in a sense they are right—it would be foolish to use a short-term local forecast model to evaluate what might happen 40 or more years into the future. But that doesn’t mean that NO model can be used to forecast 40 years out.

The factors that go into modeling long-term trends and dynamics are different from those of modeling short-term dynamics. Different relationships are important. Different trade-offs matter. Different models are needed.

Consider business models. The model needed to manage a business over the next two weeks is different from the one needed to manage the next two months, which in turn is different from the one needed to manage the next two years. Over a two-week period, cash flow may be the most important consideration, and a good model is going to track expenses and income in meticulous detail. If my horizon is two months, finances still matter, but operational concerns are likely a more important element of the model. And while finances and operations will make an appearance in a strategic model looking at the next two years, competition, market changes, and technological evolution are bound to play a much more central role.

The two-week cash flow model will be largely worthless for forecasting cash flow two years out. Too much will have changed. But that does not negate the value and validity of a two-year model that focuses on long-term concerns.

We already know this, though. Even the climate change deniers know this. And here’s how. Suppose someone is a motor sports fan. Ask her to forecast the winner of the next big race. See if she thinks her forecasted winner is guaranteed to win. “Well, no, there are no guarantees” will be the likely answer. Then ask her to forecast the season champion. She will have no trouble identifying a small number of likely champions, maybe even just one or two. But how can she predict who will win the championship, if she can’t predict the winner of the next race?

You can do the same thing with any sports. Can you predict with certainty who the winners will be for this week’s baseball (or football or basketball) games? No. But does that stop you from predicting who will make the playoffs?

Card games, board games . . . you can use most any situation that plays out over time to highlight the fact that short-term forecasting is fundamentally different from long-term forecasting, and that limitations related to a short-term forecast in no way negate the ability to forecast over the longer term. Other factors might, but mere shortcomings of short-term models are not among them.

I don’t know about you, but I’m not going to rely on climatologists to tell me about the weather over the next few days, and I’m not going to rely on meteorologists tell me about climate change.

This article originally appeared on the Pegasus Communications blog. Pegasus Communications is now Leverage Networks.

Contact Us