Arnold Kling  

George Selgin on Free Banking

The Problem with Democracy... Revenue Sharing...

He says in an interview,

Free banks compete, as it were, on an even playing field in issuing paper IOUs, which are basically what banknotes are. They have to redeem those IOUs on a regular basis: The competition among different issuers means that their notes will be treated the same way that checks are treated by banks today. They will be accumulated for a day or so and then sent through the clearing system for collection. It's this competition among issuers that assures that none of them has the power to lead the system into a general overexpansion.

Later, when asked about the prospects for free banking in the United States, Selgin answers,

Financial innovations tend to take us in the direction of free banking. Such innovations have already privatized the greater part of national money stocks, and will keep doing so in the absence of a wholesale nationalization of banks. It's only currency and coin that private firms have long been prevented from supplying.

So long as private currency remains illegal, and even if it doesn't, further financial innovation will tend to make us less and less dependent on any sort of paper currency or coins. Smart cards, debit cards, that sort of thing, have already made some inroads. And global pressures tend to favor the loosening of other kinds of bank regulations. There is, however, one kind of regulation that is growing instead of retreating and that market forces can't or won't resist, namely, government guarantees.

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.

Comments and Sharing

COMMENTS (22 to date)
Adam writes:

I highly recommend the Econtalk interview with Selgin, which I just listened to again recently.

On the issue of the business cycle, Selgin talks about how the Scottish free banks used to have an optional clause wherein they could choose not to give people their money in the event of a run, but instead to pay it to them within a specified time and with high interest. This clause was eventually made illegal; but Selgin's point is that in an actual free market banks would have every incentive to find ways to hedge against bank runs. The mere existence of the options clause also has (more effective) signalling effects along the lines of the FDIC; by knowing that other people will not be able to take all the money if they get to the bank before you, you have no reason to run on the bank yourself.

fundamentalist writes:

Hayek was pessimistic about free banking in his “Monetary Theory and Trade Cycles.” One of his main points in the books was that fractional banking can cause credit expansion without individual banks intending it or even knowing about it for quite a while. Mises supported free banking because he thought competition would force banks to be more prudent, but as Huerta de Soto points out in his book on banking, the larger the bank, the less prudent it needs to be. Modern banking is very lacking in prudence because the Fed creates the situation in which the entire banking system is essentially one large bank. Hayek later came around to the idea of free banking because of frustration with the Feds, but I think he saw it as the lesser of evils.

Mike Sproul writes:

What is so important about FDIC insurance? If I have a large amount of money then I can find my own safe places to put it. For example, I can lend it to a home-buyer and get a lien on the house. The lien is my insurance, and if it ever has to be paid off, the expense won't fall in the taxpayer's lap.

If there were no FDIC, then I'd keep a few thousand in my checking account, and the rest in stocks, bonds, and trust deeds. If the banks fail, then I'll say goodbye to about two-weeks' income and get on with my business. But thanks to the FDIC, I keep an excessive amount of cash in the bank--just waiting to become a huge burden to taxpayers.

Steve Horwitz writes:

As other parts of George's work and my own point out, before deposit insurance, banks could, and did, use a whole variety of ways of signaling their soundness, particularly by advertising their balance sheets and boards of directors. It's true that the FDIC stamp is such a signal, but to say it's "low cost" requires asking "in comparison to what?"

The costs of deposit insurance are large, as the interview with George notes, when we include the ways it encourages risk. Advertising and other methods for banks to provide similar, if not better, signals, are cheaper, all costs included. Plus, we know from history that they were largely effective when other regulatory interventions did not needlessly increase the risk of contagion.

Bob writes:

I'm glad to see Arnold staking out the middle ground. Too many people either accept deposit insurance without question or assume that there's a way to avoid implicit federal guarantees. I certainly accept that in an ideal world there is no public deposit insurance, only private insurance (which is largely a false distinction given that the public insurance can only pay large losses by tapping the private sector directly or indirectly). But the recent past should put to rest the idea that we can somehow prevent politicians from creating ex post insurance.

RE Adam's comment, all this does is force depositors to bear liquidity risk. It's not at all clear that's an improvement. If it weren't for the idiot regulations that seem to come out of each financial crisis, I'd be confident that a crisis every two decades is a reasonable cost for banks bearing most of the economy-wide liquidity risk on an ongoing basis.

David writes:

A bunch of points.

a) Economists are reputed to believe that when one subsidizes something, one will get more of it. FDIC (and other forms of moral hazard) subsidize risk.

b) FDIC subsidizes one source of bank financing, i.e., deposits. Consequently, one can expect that it will distort also bank capital structure.

c) Regarding the riskiness of free banks, Selgin noted in his superb Econtalk interview that, during the Scottish experience in the first half of the 1800s, confidence in the free banks was based on their very high levels of capital (30% of liabilities), not on their level of reserves. This allowed them to maintain very small levels of reserves (equal to 1 or 2% of deposits).

d) FDIC, like other forms of moral hazard, reduces the gains available from information search and analysis. Informational asymmetries are thus enabled and promoted by FDIC.

e) Isn't the term "insurance" (in FDIC) a misnomer? The system nature of banking (particularly under the incentives presented by FDIC for greater systemic risk and interconnectedness) suggests that the risks of a given bank's failure will not be independent of the risks of another bank's failure. The greater the moral hazard, the greater the dependency. It's a bit like insuring your house against fire in the knowledge that a) when your house burns down, most other peoples' houses will also burn down, and b) there is a high probability of fire. The only way to finance this sort of "insurance" is to have premiums which come close to fully recovering the average cost of replacing a house. Hence, FDIC tells people that they are "fully insured" when, in a very meaningful way, they are not. In this sense, FDIC provides a false signal. That's why we are seeing bank bailouts with public money and the recognition that the FDIC will probably need lots more money.

f) Purely from a system perspective, it seems that one must choose between a system in which there is a probability of a smaller and more isolated failure (and in which there are greater incentives for prudence) and a system in which there is no chance of a small isolated loss but a much greater chance of total system meltdown with horrific spillovers into the real economy. Personally, I prefer the former although I recognize that many other economists seem to prefer the latter. Not sure why.

Thus, I find it hard to agree that "putting the symbol "FDIC insured" on the door of banks is a very inexpensive signaling mechanism with a lot of social benefits."

Adam writes:


Selgin actually looks at how this was used in practice, not just from the perspective of theory. And it was optional, meaning that banks would only exercise the option if they felt the run was unwarranted and they would eventually be able to pay their depositors, given time. The added interest was to compensate depositors for their time.

Depositors always face a risk, regardless of whether the system is one of free banking or centralized banking. I offered the options clause as simply one example of how private banks would find ways to hedge against things, were the law to allow it.

David writes:

One more point (sorry).

It is a fundamental tenet of good regulation that it seek to emulate the result that an effectively competitive unregulated market would generate. Transferred to banking, that principle would presumably dictate that bank regulators study free banking experience and theory very closely to determine, at a minimum, whether regulations could be patterned after bank behaviour that arises naturally in a free banking environment.

Wouldn't it?

George Selgin writes:

A propos of Steve Horwitz' point, I think you will find that those FDIC stickers only serve to erode bank capital, which was long the private-market's principle way of signalling bank safety--and one that avoided the moral hazard problem.

Honestly, and at the risk of being "ultra," I don't see the merit in trying to find virtue in deposit insurance: there's simply too much evidence of banking systems having been able to function quite well without it, and of it's having done more harm than good. We must remember that no other country had it before the U.S. adopted it in the aftermath of the Great Contraction, as a poor substitute for structural banking reforms, and that no other country followed the U.S. lead before Canada did so in 1967. Since then, it has spread like...swine has the myth of its indispensability.

Floccina writes:

I like the idea of free banking but i am open to the following possibility:

Humans are by nature more risk averse than is optimal for maximum growth, the FDIC and Federal reserve subsidize risk and we get more risk and becuase of this we end up with more long term growth (although less even growth).

Philo writes:

“With free banking, I suspect that we would get cyclical movements in the perceived soundness of banks. . . . The same thing happens under regulated banking, of course. Which is where those government guarantees come in.”

Actually, the government guarantees come in earlier: they greatly decrease the pressure on banks to signal their soundness, by greatly weakening consumers’ incentives to pay attention to such signals.

The issue is not really “signaling”: everybody knows that the FDIC guarantee makes your (fractional reserve) bank account safer; there is no need to *signal* the fact (signaling is necessary only in the absence of a government guarantee, and even then it’s impossible to signal *that the government guarantees your account*, since it doesn’t). With the guarantee the bank is more likely to fail, due to moral hazard; but then the whole banking industry or, if necessary, the general taxpayer will make you whole. Of course, this is an added risk to you *as a taxpayer*.

Without FDIC guarantees, fractional reserve banking would be much less popular than it is. But that is not the whole of “financial intermediation.” Perhaps Arnold’s worry is that with free banking the financial system would cycle inefficiently through quite different mixes of fractional reserve banking and the other kinds of intermediation.

Adam writes:


You can make that argument for just about anything you want to subsidize. "Human by nature don't buy enough food to make it profitable for the optimal amount of agriculture to take place, so the government should subsidize it".

It's a nice "just so" story, but not much of an argument.

Kenny writes:

What good is a signal that everyone transmits? When was the last time you did business with a bank that wasn't FDIC-insured?

fundamentalist writes:

We should kill FDIC and let the free market supply insurance if there is a demand for it. Banks could by insurance against runs and depositors could by insurance against losses. The premia for the insurance would signal the soundness of the bank.

axiomata writes:

To me, modern free banking would require free market insurance providers. I can't see many modern consumers choosing to fractional reserve bank without insurance. As fundamentalist mentioned above, insurance providers would have to audit banks to calculate the insurance premium. As long as there is then no implicit guarantee to bailout insurance providers this should be a fair enough signaling mechanism.

Bob writes:


On what basis do you claim "that banks would only exercise the option if they felt the run was unwarranted and they would eventually be able to pay their depositors, given time."?

I apologize if this is an empirical fact, but it reads like a claim that is obviously absurd absent large penalties if the banks were not able to eventually pay their depositors (e.g., Arnold's periodic proposal to throw bank CEOs in jail). Bankrupt institutions are notoriously willing to gamble with other people's money.

Anyway, my basic point remains that there is significant liquidity risk in a modern economy and someone has to bear it. I accept that deposit insurance creates moral hazard, but don't worry about it as much as some because I don't see a way to prevent ex post bailouts, which create almost as much moral hazard. If we want banks to bear the liquidity risk without holding massive capital (which has its own cost), periodic bailouts seem inevitable.

Adam writes:

Well according to what Selgin said in the podcast, that's how it worked in practice in the Scottish banking system.

And of course it's absurd if there aren't penalties--the idea that contracts will create incentives of any sort is absurd in the absence of consequences for breaching them. The idea that property rights will provide incentives to increase the value of the property is absurd absence a mechanism for enforcing those rights.

And the bottom line is that if they aren't able to pay at all they just go bankrupt; something that is unavoidable in any system. And if they know they aren't able to pay then there isn't any reason for them to exercise that option and create more debt for themselves than they already had by adding interest to it. If a run occurs on a sound bank, however, they can exercise the option and not have to come up with the money right then and there.

I'm no expert on free banking, but it seems fairly intuitive to me.

There is no way to "prevent" ex post bailouts any more than there is to actually transition to a free banking system; these things are moved by individual actions of many policy makers over time, as well as the politics of the particular moment. But we can understand the theory of free banking, and understanding it helps us understand what happens under more heavily regulated circumstances.

johnleemk writes:

axiomata, that broaches an interesting point: instead of having a central bank and a central insurer, why not have free banking with private insurance regulated by the government? It has never occurred to me before, but it seems quite reasonable to expect that we could achieve better outcomes by getting rid of the government monopoly on bank insurance, while still requiring everyone to be insured.

Selgin writes:

Bob to Adam: "On what basis do you claim "that banks would only exercise the option if they felt the run was unwarranted and they would eventually be able to pay their depositors, given time."?

For a technical proof of the potential "incentive-compatibility" of option clause contracts (which depends on the interest obligation being sufficiently stiff), see Gary Gorton, "Bank Suspension of Convertibility," Journal of Monetary Economics, March 1985.

wintercow20 writes:

You should also note that free banks had every incentive to figure out ways to signal quality. The success of the Suffolk System between 1818-1858 was partially due to the fact that the country banks threatened with being taken off the list of banks in good standing was a serious stick regulating their behavior. Furthermore, when truly free banks have the ability to issue their own short-term non-assignable liabilities, whether for strategic or profit-seeking reasons, their interest in redeeming the liabilities of competitors substantially improves the soundness, acceptability and reliability of all banks/liabilities in the system. No FDIC needed.

Furthermore, if one wants to talk about whether having an FDIC is indeed efficient, you can't separate its existence from the presence of statutory reserve requirements - one would seem to be unnecessary, and both function as a tax on deposits.

"I suspect that we would get cyclical movements in the perceived soundness of banks."

There is another possibility: not only "the people" have to trust banks, but also _other_ banks have to trust them. And they will have a stron incentive to check each other out on how they back their IOUs.

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.

Comments for this entry have been closed
Return to top