I would like to see this whole issue analyzed in terms of signaling. With free banking, I suspect that we would get cyclical movements in the perceived soundness of banks. People would grow to trust the signals of at least some banks more and more, until those banks abuse that trust. Then, when people lose money, the trust will drop way off.
The same thing happens under regulated banking, of course. Which is where those government guarantees come in.
The question I have is whether a system of government guarantees produces an overall higher level of economic growth. It could do so by giving people more confidence in financial intermediaries on average through time.
It could be that putting the symbol “FDIC insured” on the door of banks is a very inexpensive signaling mechanism with a lot of social benefits. It could be argued that without that signaling mechanism, financial intermediation would be much more costly than it is today.
I am not saying that this is definitely true. However, I do think that it is the key issue in banking theory. It is all about the costs and benefits, including moral hazard and regulatory costs, of different signaling mechanisms.
Signaling is an important issue in the evolution of cooperation. Signals that work by providing accurate information are hard to fake, usually by being expensive: inexpensive signals do not work well. The best signals are simple and sincere ones: if a bank wants to signal financial soundness it can do so by being financially sound and proving it with good and clear accounting. Trust is hard to gain, and entities like corporations with long lives and stable, continuous and repetitive relationships cannot risk jeopardizing their reputation: abusing trust is not a long term successful strategy; signaling is an investment in trademark.
The possibility of cheating always exists, but government regulation is not only not a good solution, but probably the worst possible one. The best way to minimize cheating is to make people responsible for their own dealings, so that they stay alert and can at least learn from their mistakes. Having many decentralized motivated cheater detectors is far more powerful that having a few technocrats in charge of a rating bureaucracy. Confidence is not necessarily a good thing: trust is useful when the trusted entities work efficiently; simply promoting confidence creates an environment of naive trust ideal for cheaters to prosper. Those people not interested in investigating the true situation of the entities they deal with can use the services of professional raters, but there is no guarantee that the raters will be completely honest or impartial. Guarantees are never perfect or final, but government guarantees create an illusion of safety. Free competition is necessary in the provision and supervision of all goods and services: government insurance not only socializes the cost; what is more problematic is that is establishes unique official standards of conduct that are not known with certainty to be adequate (and most certainly are not), but are generally and wrongly perceived to be so (there are no alternatives to compare with).
Regarding cyclical movements, they are probably a feature of systems with a strong dependence on a single important entity, and this is the case with government regulation: since it cannot be perfect maladjustments will tend to accumulate until they are no longer sustainable and the system breaks down. A monopolistic regulator trying to stabilize the system it controls will probably destabilize it even more by failing in the intensity and the timing of its interventions.
A system with many competitive elements and interactions can compensate the fluctuations of some elements with the opposite fluctuations in other elements. Complex adaptive systems are like homeostatic organisms: too complex to be purposefully designed or controlled, they adapt by means of multiple decentralized interactions. Banks are not only controlled by their depositors and shareholders: they are also limited by the competition of other banks which can inform potential customers of the dangers associated with the risky behaviors of the least trustworthy firms (and of course companies can lie about their competition). The distributed information of market watchers can be gathered in the secondary markets of private bank notes where all people are free to prove their knowledge by means of their financial bets.