Is a Tobin Tax feasible?

The Tobin tax--a tax on financial transactions--is very much in the news. Before one gets to the issue of the desirability of such a tax, one has to address the question of whether the policy is even feasible. I am skeptical. Financial transactions are easy to move: If two parties to a financial contract can just as easily sign and enforce the contract in the Cayman Islands as in New York or London, there is little point in US or UK policymakers imposing a Tobin tax. Unless, of course, moving the finance industry offshore is the policy goal.

In his column today, Paul Krugman raises and then dismisses this issue as follows:

On the claim that financial transactions can’t be taxed: modern trading is a highly centralized affair. Take, for example, Tobin’s original proposal to tax foreign exchange trades. How can you do this, when currency traders are located all over the world? The answer is, while traders are all over the place, a majority of their transactions are settled — i.e., payment is made — at a single London-based institution. This centralization keeps the cost of transactions low, which is what makes the huge volume of wheeling and dealing possible. It also, however, makes these transactions relatively easy to identify and tax.

This passage left me scratching my head. Even if most transactions are now settled in one place, that need not be the case in the future after a significant change in the policy environment. It is not as if London has some large, natural comparative advantage in financial settlement that would persist despite a tax on transactions there. The finance industry is set up to take advantage of very small price differences. If London became ever so slightly more expensive, wouldn't contracting and settlement quickly migrate elsewhere?

Update: NYU finance prof Aswath Damodaran reaches similar conclusions.