The Coase Theorem explains - in terms that seem ridiculous until you really understand it, but painfully simple afterwards - why externalities don't inherently require state action to ensure an optimal supply. It explains why there is no inherent need to subsidise every public good, tax every negative externality. Too many nasty looking bikers? Pay them to leave. Trains setting your field on fire? Pay them to reroute somewhere else or stop running. Your child is going to have an abortion and you - as parent - don't like it? Bribe them, perhaps with offers of money for university. That final example is actually used in an academic discussion of the theorem, behind the academic firewall I'm afraid.
Of course, this only really works in the fictional land, populated by the imaginations of economists, where there are no transaction costs. In reality it would be a deeply unpleasant experience to negotiate over a pregnancy. All clubbing together and paying our 0.0000001 pence to a factory owner to attach a sulphur dioxide scrubber to his chimney just isn't practicable. We'd need to pay too much for it to be economical just to post him our tiny, fictional coins.
Coase wasn't ignoring this. However, if the problem is not an inherent market failure but, instead, a product of transaction costs making it impossible for society to come to a settlement through the market then negative externalities pose a fundamentally different challenge to the policy maker. After all, while transaction costs might make a market solution inefficient they are just as likely to make a mess of solutions relying upon state power. The case for Pigovian taxation is fundamentally weak.
Jim Manzi does a brilliant job of explaining how this relates to carbon taxes. Transaction costs make the obvious private sector solution to an alledged oversupply of carbon dioxide emissions, pay emitters to stop, impractical. However, there are also massive transaction costs to carbon taxes. He points to three important sources of transaction costs associated with a carbon tax. I'll paraphrase:
1. State intervention into major industries is extremely vulnerable to interest group lobbying. Oil and electricity generation companies have done quite well out of green politics so far. That isn't a freak accident. It's the result of the Logic of Collective Action: highly motivated minorities can exploit the use of state power even in a well-constructed democracy. This diversion of corporate talent to unproductive rent-seeking is wasteful in many ways.
2. We just don't have the necessary information to make an informed intervention in the market. We need to know what level to set the carbon tax. That should be the social cost but unfortunately estimates of social cost vary widely. If we get it wrong our policies will be too draconian or too lax.
3. Politicians don't use Pigovian taxes to correct for externalities but to raise revenue. They'll keep doing this and we'll wind up with higher taxes and the problems often associated with a high-tax economy.
Read Manzi's original post for a more scholarly explanation of each point.
I like to think that my report The Case Against Further Green Taxes is a part of the Coasean case against them. While Manzi can illustrate in theory the transaction costs associated with green policies you can see those problems in practice in Britain today. Corporate subsidy and green non-jobs wasting taxpayers' money, carbon taxes set too high and imposing an unpleasant, not to mention regressive, burden on ordinary taxpayers wanting to do socially useful things like get to work or move goods. The proper comparison to a free-market economy which emits too much carbon thanks to missing markets isn't an angelic, perfect government intervention but one with its own, often even more pernicious, flaws.