Predicting the future is hard. It’s so difficult that even teams of analysts using fancy models get results like this:

eia 2007














This isn’t back-of-the-envelope stuff. This is the US Energy Information Administration’s official prediction for oil prices, circa 2007. According to the “high price” scenario, oil may reach $100 per barrel some time around 2030. But wait: oil was at $127 yesterday. So, not only was the EIA projection wrong—it was wildly and completely wrong.

Okay, everyone makes mistakes, even energy analysts. In 2008, the EIA cleaned up its act and produced this forecast: eia-2













As you can see, in the “high price” scenario, oil will almost reach $120 per barrel around 2030. At the risk of repeating myself, oil was at $127 yesterday! So, the EIA projection is still wildly and completely wrong.

Ordinarily, this would be jolly good fun. But it’s not so funny when these kind of projections are used to make decisions with big consequences. These two charts, in fact, come from the Western Climate Initiative’s economic analysis, which is being conducted by consultants. (The forecasts shown in the charts are the data inputs for a model that is supposed to inform policymakers and thereby guide the design of the cap and trade program.) But the data inputs are laughably wrong.

Of course the consultants are well aware of the problem. And they’re already trying to develop new forecasts to use in their modeling exercise. Plus, in their defense, the reason they planned to use EIA data was that it’s probably the best available and it’s considered highly credible.

In any event, the WCI consultants will probably use new and improved EIA data to update their models. That would be fine, maybe, but the EIA seems to have a hilariously persistent tic. Check out this chart via Kevin Drum:

  • eia - 3











    The black line is actual prices and the colored lines are monthly forecast revisions. The black line (the price) keeps going up, and every month the EIA forecasts a decline, starting immediately. It’s as if the forecasts are impervious to reality.

    On the other hand, who knows? The EIA could eventually turn out to be right and the current high prices may evaporate. But even if that happens, it’s a bit worrisome that our climate policy might be based on forecasts that are so wildly and consistently inaccurate. And this stuff matters. A lot.

    All else being equal, forecasts of lower oil prices will tend to raise the predicted costs of climate policy. In part, this is because cheap oil means lots of consumption and intractable behavior. But if oil prices are high in the future, as they are now, then a cap and trade program might add only small costs—or even result in savings. So the EIA’s perpetual sunshine-optimism about future oil prices might create a big distortion of the cost of sensible carbon reductions.

    All this to say that it’s tough for even the best minds to predict the future. Not that we shouldn’t do our best to calculate costs and minimize impacts—we definitely should! But our projections for the future should be leavened with a heaping spoonful of skepticism.