The standard economic reaction to an externality is to tax it so as to include it into the price system. We'll then get the appropriate amount of said externality, a balanacing of the costs and benefits of it. In economic terms the argument is well nigh unassailable but I am beginning to reconsider my support for the idea. For taxes, by definition, have to be imposed by politicians and there's almost no good idea that a politician cannot screw up.

The straw for the camel is Osborne's increase in the bank levy this budget.

The move to increase the bank levy – from a charge of 0.088 per cent on banks’ global balance sheets to 0.105 per cent from 2013…….The Treasury retained its goal of raising £2.5bn a year from the levy.

The thing is, the bank levy is not supposed to be a tax, a method of filling the Treasury. It's supposed to be, in theory it is, an insurance premium on those bank liabilities which are not already covered by some other deposit insurance scheme. If, for example, bank borrowings from the wholesale markets (these are not covered by other schemes and so pay the levy) fall then the revenue from the levy should fall: we citizenry are providing less insurance to the banks and thus should get lower premiums. Similarly, if such wholesale funding moves from overnight borrowings to 5 year bond issues, this is less risk and thus lower premiums should be charged (the system does differentiate between these two scenarios).

But here we have the Treasury targetting an amount to be received: and a Chancellor seeing the levy as a convenient place to raise revenues, not to set the appropriate insurance premium.

And I'm afraid that the more we see of entirely righteous Pigou Taxes the more we see of this behaviour. I pointed out in these very pages some years back that if we applied the Stern Review to petrol taxation then fuel duty should fall by 12 p a litre: since then it has risen another 5 or 6 p still using Stern as the justification. Air Passenger Duty was set (amazingly, by Gordon Brown) at the Stern level of some $80 per tonne Co2-e: it has been doubled at least since then purely for revenue purposes.

I've not really made up my mind as yet, just rethinking my position. But it could be that Pigou Taxation just won't work because of politics. In the same way that Keynesian economics doesn't because of politicians: even if the entire economic analysis is correct there is never the political will in the booms to put aside money for the busts. And so it is with Pigou Taxes. They are correct in theory, of that there's no doubt.

But giving a politician a justification for a new tax is like giving a child a loaded machine gun: noisy, dangerous and very definitely a very bad idea.