He says,

in a variety of financial crises, I have arrived at the following guide to conduct: if you find yourself on the verge of imposing massive costs on an economy – that is on the people of a country or countries – by precipitating a crisis in order to prevent moral hazard, it is too late. You should not take the action that imposes those costs. Rather in thinking through how a system will operate in a crisis, you need to take into account the likelihood of facing such choices, and you need to do everything you can in designing the system to keep that likelihood very small.

This still does not settle what I see as the big issue, which is whether you think in terms of a system that is hard to break or easy to fix. I think that “hard to break” is an illusory goal. Wanting a system that is easy to fix leads me to prefer financial systems where debt finance is relatively more expensive and where banks are smaller. I would like to see more equity finance, and I would like to avoid having a situation where a bank is so big that its failure becomes unthinkable.

Thanks to Justin Lahart for the pointer. Fischer’s entire talk is interesting.