This article by Mark Perry seems right to me -- the lightest touch (and probably the most effective) approach to bank regulation is to return to a regime that puts its major emphasis on capital requirements.
We can talk all day about causes of the recent financial crisis, but in my mind the root cause was taking real property with a volatile underlying value (e.g. homes) and leveraging the absolute crap out of it. In the initial transaction, home buyers were allowed to come to the table with less and less equity, until deals were being cut with more than 100% debt. This stupidity was a true public-private partnership, as the government kicked off the party and encouraged its growth via various community development policies as well as policies atFannie and Freddie, but private originators as well as home buyers eagerly jumped into the fray.
This debt backed by property that was already too highly leveraged was thrown into portfolios that were themselves highly leveraged, and then further leveraged again through CDS's and other derivatives. And then the CDS's were put into leveraged portfolios. I would love to figure out the effective leverage in the AIG portfolio. For ever $1 million in real property that secured the mortgages they insured, how much equity did they have? A thousand bucks? Less?
These investors felt protected by diversification that didn't really exist. The felt safe with AAA ratings from agencies who really didn't understand the risks any better than anyone else did. They relaxed assuming everything was watched by government regulators who were in way over their heads. But more than anything, they felt protected by history. The system of putting mortgage risks into tranches, such that the top tranches could only be affected by default rates consider then to be wildly improbable, had never to that point failed to deliver its promise. Default rates had always stayed withing expected norms.
And this is the most dangerous risk -- the risk that something will happen that has never happened before. Default rates that seemed impossible suddenly became reality. Tranches that were untouchable suddenly were losing large chunks of their value. Sure, there were warning signs, but at the end of the day what happened was that events occurred that were worse than people had thought was the worst case scenario (there is a whole body of interesting behavioral study on how humans tend to overestimate their understanding and underestimate the width of a probability distribution).
As new financial products are created and the economy evolves and the government pursues new forms of interventions in commerce, new failures can occur that have never happened before. And never has there been invented a micro-regulatory approach that guards against new-type failures (they don't even do a very good job against old-style failures). Capital requirements are the one approach that guard against catastrophic failures even for unanticipated risks.
It can be argued that this will raise the cost of capital, at it is true interest rates at any one point of time would have to go up. But one can argue that the low interest rates of the 2000's greatly understated the true cost of capital, and that those additional costs were paid in a sort of balloon payment at the end of the decade.
I am still thinking this through -- I don't think any regular reader would be mistake me for someone who favors regulation in general, but I am coming around to some extent on the notion that banks are different. I would ideally like to see a self-policing market where companies that choose to cut equity too fine just go bankrupt. But the reality of the political-financial complex today is that this never happens -- costs of large failures are socialized, and executives who made bad choices get fat gold parachutes and Treasury jobs.
Postscript: I have arguments all the time about whether the financial melt down was mainly caused by government or private action. Was it a public or private failure.? My answer is yes.
One thing that those of us who promote private action over public can never repeat enough is this: Our support for private action does not mean that private actors don't screw up, that there are not bad outcomes, that people don't make bad decisions, etc. They do. Lots of them. When these constitute outright fraud, there should be prosecution. For the rest of the cases, though, libertarians believe that in a free society there are automatic corrections and sources of accountability.
Make a bad product - people stop buying it. Sign a union contract with wages that are too high - you go bankrupt. Treat your workers shabbily - and the best of them go work for someone else. Take on too much risk - you will fail and lose all your capital.
The problem with our financial sector is not that it is not regulated -- it is the most regulated sector of the economy. The problem is that, as always happens, there has been substantial regulatory capture. There has been an implicit deal cut by large financial institutions - regulate me, but in return protect me. In a sense, as is typical in a corporate state, large corporations and government have become partners.
As a result, many of the typical checks and balances on private action in a free economy have been disrupted. In effect, certain institutions became too big to fail, and costs of failure and risk taking were socialized.
That is why the answer is not one or the other. Certainly the massive failures were driven by the actions of private actors. But they were driven in part by incentives put in place by the government, and their stupid behavior was not checked because traditional private avenues of accountability had been neutered by the government. This is why the recent financial crisis will always remain a sort of political Rorschach test, where folks of wildly different political philosophies can all find justification for their position.