The purpose of a financial system is to solve a collective optimization problem whose solution we cannot guess a priori. If we are very sure that welfare is maximized by vastly expanding the housing stock and making homeowners of people who otherwise might not buy, then the government should just tax to build McMansions, and auction off the oversupply. More generally, one cannot judge a financial system by any particular outcome, because all financial systems make mistakes, and the mistakes always look good while they last. We judge financial systems by the performance of the economies they guide over time.
...Savers should not be investors, that is they should not be underwriting the execution of projects about which they have no opinion and whose risks they are unwilling to bear.
...Complexity is much more often a marker of snake-oil than of quality in a financial instrument. "Sophisticated" investors are almost always predators or fools. The real-world informational problems investors face -- what is it that should be done? how ought our resources be deployed? -- are challenging enough. Creating structures that cannot be understood except by applying complex models that may or may not adequately capture the behavior of the instrument is just idiocy, a mish-mash of quant hubris and pseudoscientific salesmanship.
He says much more, not all of which I agree with.
You cannot simply say, "We must have transparency." Requiring transparency means requiring people to disclose information that is costly to acquire. But that undermines their incentive to acquire information.
Take my example where the investment projects are fruit trees with susceptibility to disease. Suppose that it is costly to assess the soil in a particular area for its ability to nourish the trees so that they can avoid disease. If you force anyone who discovers information about the soil to be "transparent" and to disclose that information, people will not undertake investments to obtain the information.
Instead, what you need are incentive-compatible contracts. The entrepreneur who knows a lot about the soil offers a debt contract, which in effect tells the investor "I can promise you that the soil is at least good enough to enable trees planted here to pay off the debt." To make this incentive compatible, the entrepreneur puts a lot of his own wealth into the fruit tree project and retains a residual claim in the form of equity. Because of his equity stake, the entrepreneur loses if he overstates the value of the soil to the investor to whom he issues debt.
From this perspective, a good financial institution is one which encourages people to gather and use information in an incentive-compatible way. A bad institution promotes misinformation, typically because of problems with incentive compatibility. It could be that the vast majority of financial activity that we take for granted in the economy takes place within the context of what I would consider bad institutions.
Why do bad institutions survive? It could be that it is very difficult to learn what is a bad institution,, except through painful experience.
However, I suspect that the main reason we have bad financial institutions is that individuals are irrational. I can well imagine that it is easier to market a Ponzi scheme (suitably dressed up) than an incentive-compatible financial contract. People think they are smarter than they really are. As a result, bad financial institutions satisfy customer demand, just as politicians offering free lunches satisfy voter demand.
The issue of financial regulation is whether it tempers bad financial institutions or reinforces them. In my judgment, it does a lot of both. There are a lot of regulatory actions that serve to stop Ponzi schemes. However, there are other regulatory actions that encourage Ponzi schemes. I keep coming back to risk-based capital requirements, which were well intentioned but in fact had the perverse impact of promoting securitization of high-risk mortgages over the origination and holding of low-risk mortgages.