Environmentalist Calculation [updated]

Suppose that you care about reducing carbon dioxide emissions. Should you:
a) implement a low carbon tax across all carbon dioxide emissions, or
b) subsidize solar power?

I've argued before that, absent comprehensive Pigovean taxation, you can get nonsense.

So, if you answered b, here's a nice little example of exactly that kind of nonsense.

Preliminary evidence shows some solar stations may have run diesel-burning generators and sold the output as solar power, which earns several times more than electricity from fossil fuels, El Mundo said, citing unidentified people from the energy industry. The power grid received 4,500 megawatt-hours of power from midnight to 7 a.m. in the months audited, El Mundo said.

Yes, this was fraud, and fraud could well also be a problem for a carbon tax. But it's hard to see how they could have pulled this one off if there were a carbon tax on diesel (and all other sources) rather than a subsidy to solar power. One wag suggested that the solar companies could have stayed within the letter of the law by using the diesel generators to power huge spotlights, which they'd then shine on the solar panels at night. This would never make sense under a carbon tax, but would be eminently rational if the solar subsidy were high enough.

Conditional on wanting to do something to reduce emissions, Club Pigou remains your best bet.

Update: Note, of course, all of the problems with Pigovean solutions relative to Coasean ones. Patri helpfully posts a reminder. In short, when the Pigovean solution is efficient, we're unlikely to get it because political processes will push us to cap & trade or other systems that benefit incumbents. And, if we got a Pigovean carbon tax, it's exceedingly unlikely that we'd have commensurate reductions in other tax rates; much of the collected taxes will be wasted.

I often use in class David Friedman's example about how the combination of a Pigovean tax and Coasean trading can lead to highly inefficient outcomes.