Model scams

Experts use mathematical models to give us predictions about the future. Rank the following events according to the ability of the model builders to give accurate forecasts:

(A) The winner of the 2008 presidential election.

(B) The size of the U.S. budget deficit for the year 2009.

(C) The official low temperature in Chicago on April 1, 2008.

(D) The number of hurricanes that will strike Florida in 2010.

(E) The rise in the sea level over the next 100 years.

(F) The average temperature for the year 2100.

People skilled in building mathematical models and with the aid of computers are capable of amazing computational feats. For instance, I am very impressed by the wizards that can deliver a robot to the surface of Mars at a precisely predetermined point. The reason they can do this is that they are dealing with relatively few variables — the speed of planetary bodies, their gravitational forces, etc. — and much is known about each of these variables.

In the above list of events where model builders use their skills to make forecasts, you might think that predicting the size of the budget deficit would be one of the easier tasks — turns out, not so. In the mid-1990s, only two years before the budget went into surplus, the Washington budget gurus were predicting “two hundred billion dollar deficits as far as the eye can see.”

By 1999, with the growing budget surpluses, there were many articles about how the surpluses would only get larger, which could cause huge problems for the U.S. economy. (“What would bond holders hold if there were no government debt?”)

By 2002, the budget surpluses had disappeared, and there were forecasts of endless deficits running to 4 or 5 percent of gross domestic product (GDP), yet last year the deficit was only 1.2 percent of GDP.

The budget deficit forecasters need to calculate future government spending, which they normally can do with reasonable accuracy, despite unexpected wars and Congress’ never-ending fiscal irresponsibility. Their real difficulty is in forecasting revenues, which are very dependent on the level of economic growth and people’s responses to tax changes. There are a great many variables that have significant effect on economic growth, not all of which are well understood or can be predicted with much accuracy, such as technological change — e.g., the Internet.

The government tax economists who are supposed to forecast tax revenues by projecting people’s responses to tax changes have done a dismal job, and in the case of capital-gains tax rate cuts, not even getting the direction of the plus or minus sign right at times in the past.

As an economist with some knowledge of how difficult it is to identify all relevant variables in making an economic forecast, let alone the magnitude of each variable, I am bemused when told that climate models predict this or that, given that they are usually of much greater complexity than economic models. Which climate models predicted the large increase in Northern Hemisphere snow cover this year? Or the record sea ice around Antarctica? Or that there would be fewer rather than more Atlantic hurricanes over the last three seasons? None did so, as far as I can find.

We know changes in solar radiation have major impacts on Earth temperatures. What causes variations in solar output and who has a model that can accurately predict these changes? We also know carbon dioxide traps heat, and we know burning fossil fuels, or volcanoes for that matter, produces clouds, which reflect heat. Where are the precise models of such interaction? And so on.

Story Continues →