I think what we need is better understanding of how to do risk analysis of a CDO, but that they still can perform a very valuable function because they can aggregate these risks and pass them around so that mortgages or other kinds of loans can be packaged and sold to investors all over the world, who in most times, would justify a small amount of each one.

I think that's something that investment banks have worried about for a long time and are continuing to worry about, but it's not an easy solution when you have lots of people betting the company's money, how do you really allocate those risks? How do you make sure that the people that take the risks are feeling the risks in an appropriate kind of fashion?

Allowing short selling is allowing people to sell - instead of having to buy the stock and then sell it, which doesn't do much; allow them to sell it, and then buy it. In which case they can express that information and the idea is that you would get more accurate valuation of companies by letting people express both their positive information and their negative information through either long or short selling.

I don't see what the alternative to risk management is. If it's just getting rid of the models and instead using the smart people who can figure it out? How do you train them? What do you teach them? Do you just put them in a cockpit and let them stumble for 10 years of their life and then after that they're good at it?

I'd like to see a little more action on the energy side of things. I've been pushing for some kind of a carbon tax for years, and it seems to me we've had lots of opportunities to do it.

I think we're fooling ourselves if we think that regulators are going to be able to outsmart the bankers. So, the task of designing regulatory reform is trying to make more or less foolproof regulation and that's one of the advantages of the systemic category.

There are a lot of ways that investment banking models work, but these risks are not internalized by the people that are taking them.