It is not very often that I find the products of conservative think thanks worth discussing, but today seems different. I swear I am not entirely sure how I found my way on Examining the Social Cost of Carbon, a post by Paul Knappenberger, who apparently is an associate director of the Center for the Study of Science at the Cato Institute. In the post, Mr. Knappenberger argues that it is disingenuous of the Obama administration to impose a social cost on carbon due to uncertainty about the effects of carbon dioxide on the society (the post also has some more specific arguments about climate science, which I am not qualified to comment on, and normative claims suggesting that America’s carbon price should not consider the damage of climate change outside the United States, which are not worth commenting on):
“The social cost of carbon is a poor concept from the start. It is an ill-conceived, one-sided supposed measure of the damages associated with climate change resulting from human emissions of carbon-containing greenhouse gases (such as carbon dioxide and methane). Or, rather, it is a measure of the damages predicted to occur by a collection of computer models — computer models which themselves largely fail at capturing the climate evolution during recent decades.”
I have serious reservations about the Obama administration, and I would love to join the bashing just for fun, but this is a funny argument. Suppose we had no carbon price. That would be the same as if we had a carbon price of zero. This non-policy is also a policy. It can only be justified with respect to some model saying that carbon dioxide emissions are harmless. So, the social cost of carbon is not a “poor” but “fundamental” concept. Without such a concept, systematically approaching the complex challenge of climate change would be next to impossible.
The greater point is that uncertainty is itself not an argument for inaction. If there is uncertainty, damages could be either lower or higher than expected. Policy should consider both the expectation and variance of an estimate, without an implicit or explicit status quo bias.