Cato's Arnold Kling has a thought-provoking article in National Review arguing for breaking up the large banks:

It is the political economy that most concerns me. Freddie Mac and Fannie Mae represent everything that is wrong with the politics of big banks. They acquired lobbying prowess, their decisions were distorted by political concerns, and they were bailed out at taxpayer expense. All of these developments seem to be inevitable with large financial institutions, and all are deeply troubling to those who value economic freedom. Unless there are tremendous advantages of efficiency or systemic stability from having large banks, their adverse effect on the political economy justifies breaking them up.

If we had a free market in banking, very large banks would constitute evidence that there are commensurate economies of scale in the industry. But the reality is that our present large financial institutions probably owe their scale more to government policy than to economic advantages associated with their vast size. Freddie Mac and Fannie Mae were created by the government, and they always benefited from the perception that Washington would not permit them to fail — a perception that proved accurate. Similarly, large banks were viewed as “too big to fail,” which gave them important advantages in credit markets and allowed them to grow bigger than they otherwise would have.

The case for market failure here is pretty simple. If very large financial institutions fail, they can bring the whole economy down with them. Therefore, they have an incentive to undertake risky investments, the profits of which they keep entirely to themselves, and the losses of which can be bailed out by taxpayers.

I agree with Kling that the ideal solution would be to simply limit the size of the banks. The second-best solution, which is currently being pursued by Democrats in Congress, is to regulate the banks to prevent them from engaging in risky behavior, and/or tax the large ones to reduce the advantage they gain over small institutions that aren't too big too fail.

Kling, I'm sure, would reply that regulation won't work, because the large banks gain so much political power that they can defeat any regulation. I'd respond in turn that that, while this may turn out to be true, they already have so much political power that breaking them up has zero political feasibility. So we're in a second-best world where it's regulate, and hope regulation works, or do nothing. My skepticism of Kling's argument is that, like some principled right-wing arguments that acknowledge climate change, it argues for an ideal solution that lacks any chance of happening, while favoring the status quo over a second-best solution.

Still, Kling's piece is a very interesting read, and I'd urge everybody to check it out.