Dave Roberts explains a problem with cost-benefit analysis:

As for modeling innovation, that’s always been the Achilles heel of economic forecasting. In this piece, Eban Goodstein and Hart Hodges trace a history of cost overestimations around environmental regulation. Again and again, models have underestimated the pace of business and technological innovation.

Today’s modelers surely do all they can to incorporate innovation. (As Brad notes, the CBO tries.) But there are constraints to how precisely this can be done. In 1980, McKinsey reported to AT&T that mobile subscriptions would rise to 0.9 million by 2000. The real number turned out to be ... 109 million. (This factoid is among many interesting tidbits in this presentation from Vinod Khosla.) What if a modeler had come along in 1980 and said, “There will be massive innovation and new infrastructure and new business models and costs will fall by orders of magnitude, so much so that the prediction of our friends at McKinsey is 121 times too low!”

They would have been roundly mocked. And rightly so. How could they presume to predict so many fortuitous twists, turns, and serendipities? They obviously couldn’t see the future.


We want to hear what you think about this article. Submit a letter to the editor or write to letters@theatlantic.com.