I have written a number of pieces on fractional reserve banking and duration mismatch. I have argued that the former is perfectly fine, both morally and economically, but the latter is not fine. I have dissected the arguments made against fractional reserve banking, and pointed out that it is nothing more than a bank lending out some of the money it takes in deposits.
I have debunked the most common errors made by opponents of fractional reserve:
- Banks print money
- They lend more than they take in deposits
- They inflate the money supply
- Money is the same as credit
- Fractional reserves banking is the same thing as central banking
- It is the same thing as duration mismatch
Duration mismatch is when a bank (or anyone else) borrows short to lend long. Unlike fractional reserve, duration mismatch is bad. It is fraud, it is unfair to depositors (much less shareholders) and it is certain to collapse sooner or later. This is not a matter for statistics and probability, i.e. risk. It is a matter of causality, which is certain as I explain below.
This discussion is of paramount importance if we are to move to a monetary system that actually works. Few serious observers believe that the current worldwide regime of irredeemable paper money will endure much longer. Now is the time when various schools of thought are competing to define what should come next.
I have written previously on why a 100% reserve system (so-called) does not work. Banks are the market makers in loans, and loans are an exchange of wealth and income (http://keithweiner.posterous.com/the-loan-an-exchange-of-wealth-for-income). Without banks playing this vital role, the economy would collapse back to its level the previous time that the government made it almost impossible to lend (and certainly to make a market in lending). The medieval village had an economy based on subsistence agriculture, with a few tradesmen such as the blacksmith.
But I have not directly addressed the issue of why duration mismatch necessarily must fail, leading to the collapse of the banks that engage in it. The purpose of this paper is to present my case.
In our paper monetary system, the dollar is in a “closed loop”. Dollars circulate endlessly. Ownership of the money can change hands, but the money itself cannot leave the banking system. Contrast with gold, where money is an “open loop”. Not only can people sell a bond to get gold coins, they can take those gold coins out of the monetary system entirely, and stuff them under the mattress. This is a necessary and critical mechanism—it is how the floor under the rate of interest is set.
This bears directly on banks. In a paper system, they know that even if some depositors withdraw the money, they do not withdraw it to remove it altogether (except perhaps in dollar backwardation, at the end. See: http://keithweiner.posterous.com/dollar-backwardation). They withdraw it to spend it. When someone withdraws money in order to spend it, the seller of the goods who receives the money will deposit it again. From the bank’s perspective nothing has changed other than the name attached to the deposit.
The assumption that if some depositors withdraw their money, they will be replaced with others who deposit money may seem to make sense. But this is only in the current context of irredeemable paper money. It is most emphatically not true under gold!
There are so many ills in our present paper system, that a forensic exploration would require a very long book (at least) to dissect it. It is easier and simpler to look at how things work in a free market under gold and without a central bank.
Let’s say that Joe has 17 ounces of gold that he will need in probably around a month. He deposits the gold on demand at a bank, and the bank promptly buys a 30-year mortgage bond with the money. They assume that there are other depositors who will come in with new deposits when Joe withdraws his gold, such as Mary. Mary has 12 ounces of gold that she will need for her daughter’s wedding next week, but she deposits the gold today. And Bill has 5 ounces of gold that he must set aside to pay his doctor for life-saving surgery. He will need to withdraw it as soon as the doctor can schedule the operation.
In this instance, the bank finds that their scheme seems to have worked. The wedding hall and the doctor both deposit their new gold into the bank. “It’s not a problem until it’s a problem,” they tell themselves. And they pocket the difference between the rate they must pay demand depositors (near zero) and the yield on a 30-year bond (for example, 5%).
So the bank repeats this trick many times over. They come to think they can get away with it forever. Until one day, it blows up. There is a net flow of gold out of the bank; withdrawals exceed deposits. The bank goes to the market to sell the mortgage bond. But there is no bid in the mortgage market (recall that if you need to sell, you must take the bid). This is not because of the borrower’s declining credit quality, but because the other banks are in the same position. Blood is in the water. The other potential bond buyers smell it, and they see no rush to buy while bond prices are falling.
The banks, desperate to stay liquid (not to mention solvent!) sell bonds to raise cash (gold) to meet the obligations to their depositors. But the weakest banks fail. Shareholders are wiped out. Holders of that bank’s bonds are wiped out. With these cushions that protect depositors gone, depositors now begin to take losses. A bank run feeds on itself. Even if other banks have no exposure to the failing bank, there is panic in the markets (impacting the value of the other banks’ portfolios) and depositors are withdrawing gold now, and asking questions later.
And why shouldn’t they? The rule with runs on the bank is that there is no penalty for being very early, but one could suffer massive losses if one is a minute late (this is contagion http://dailycapitalist.com/2012/05/30/keithgram-contagion-defined/).
What happened to start the process of the bank run? In reality, the depositors all knew for how long they could do without their money. But the bank presumed that it could lend it for far longer, and get away with it. The bank did not know, and did not want to know, how long the depositors were willing to forego the use of their money before demanding it be returned. Reality (and the depositors) took a while, but they got their revenge. Today, it is fashionable to call this a “black swan event.” But if that term is to have any meaning, it can’t mean the inevitable effect caused by acting under delusions.
Without addressing the moral and the legal aspects of this, in a monetary system the bank has a job: to be the market maker in lending. Its job is not to presume to say when the individual depositors would need their money, and lend it out according to the bank’s judgment rather than the depositors’. Presumption of this sort will always result in losses, if not immediately. The bank is issuing counterfeit credit. In this case, the saver is not willing (or even knowing) to lend for the long duration that the bank offers to the borrower.
Do depositors need a reason to withdraw at any time gold they deposited “on demand”? From the bank’s perspective, the answer is “no” and the problem is simple.
From the perspective of the economist, what happened is more complex. People do not withdraw their gold from the banking system for no reason. The banking system offers compelling reasons to deposit gold, including safety, ease of making payments, and typically, interest.
Perhaps depositors fear that a bank has become dangerously illiquid, or they don’t like the low interest rate, or they see opportunities offshore or in the bill market. For whatever reason, depositors are exercising their right and what they expressly indicated to the bank: “this money is to be withdrawn on demand at any time.”
The problem is that the capital structure, once erected, is not flexible. The money went into durable consumer goods such as houses, or it went into partially building higher-order factors of production. Imagine if a company today began to build a giant plant to desalinate the Atlantic Ocean. It begins borrowing every penny it can get its hands on, and it spends each cash infusion on part of this enormous project. It would obviously run out of money long before the plant was complete. Then, when it could no longer continue, the partially-completed plant would either be disassembled and some of the materials liquidated at auction, or it would sit there and begin to rot. Either way, it would finally be revealed for the malinvestment that it was all along.
By taking demand deposits and buying long bonds, the banks distort the cost of money. They send a false signal to entrepreneurs that higher-order projects are viable, while in reality they are not. The capital is not really there to complete the project, though it is temporarily there to begin it.
Capital is not fungible; one cannot repurpose a partially completed desalination plant that isn’t needed into a car manufacturing plant that is. The bond on the plant cannot be repaid. The plant construction project was aborted prior to the plant producing anything of value. The bond will be defaulted. Real wealth was destroyed, and this is experienced by those who malinvested their gold as total losses.
Note that this is not a matter of probability. Non-viable ventures will default, as unsupported buildings will collapse.
People do not behave as particles of an “ideal” gas, as studied by undergraduate students in physics. They act with purpose, and they try to protect themselves from losses by selling securities as soon as they understand the truth. Men are unlike a container full of N2 molecules, wherein the motion of some to the left forces others to the right. With men, as some try to sell out of a failing bond, others try to sell out also. And they are driven by the same essential cause. The project is non-viable; it is malinvestment. They want to cut their losses.
Unfortunately, someone must take the losses as real capital is consumed and destroyed. A bust of credit contraction, business contraction, layoffs, and losses inevitably follows the false boom. People who are employed in wealth-destroying enterprises must be laid off and the enterprises shut down.
Busts inflict real pain on people, and this is tragic as there is no need for busts. They are not intrinsic to free markets. They are caused by government’s attempts at central planning, and also by duration mismatch.
You HIT bids and TAKE offers
I enjoyed reading your post. I have been a keen follower of Prof Fekete for many years and have tried to spread his message amongst Austrians and others. I would welcome your comments on my latest post "Why Austrian Economists are Wrong about 100% Reserve Banking" http://teconomist.blogspot.co.uk/2012/07/why-austrian-economists-are-wrong-about.html
I am trying to understand the controversy over real bills doctrine. It seems to me that your paper on duration mismatch is actually a departure from the RBD, at least as stated in its classic form, “so long as a bank issues its notes only in the discount of good bills, at not more than sixty days date, it cannot go wrong in issuing as many as the public will receive from it.”
The notes referred to above were promissory notes payable on demand. They were effectively the same as a demand deposit but accounted for as bearer instruments instead of an account on the books.
The RBD as stated above has banks issuing demand deposits backed by real bills maturing in 60 days – which is a duration mismatch according to your definition. The bank is borrowing short and lending slightly longer.
Following your duration mismatch theory, sound banking would be issuing bonds with a greater duration than the average duration of the assets on the books. Ie, 1 year or longer bonds, and investing the money by discounting 91 day real bills. Safe duration difference there…
If I correctly infer from your paper that bank deposits must be backed by assets with duration less than or equal to the deposits, then demand deposits must be backed by assets with duration less than or equal to zero.
It appears to me that in this paper you have proved Ebeling and Rothbard’s position that demand deposits, ie checking accounts, must be 100% reserved with cash in order to have a sound banking system.
Ebeling and Rothbard never condemned banks selling CD’s and investing the proceeds in shorter maturation loans or commercial paper.
Am I wrong or does your duration mismatch paper put you in Rothbard’s camp instead of Fekete’s?
No, I am not in that camp.
Real Bills aren’t lending. They are clearing, which is an entirely different thing.
There seems to be a misunderstanding here. Neither camp (Fekete vs. Rothbard) has an objection to the creation and circulation of bills of exchange in and of themselves.
The question is whether banks may issue bank notes or demand deposit accounts against bills of exchange instead of gold coin in the vault, and whether that practice is inflationary or not.
We both agree that the practice is not inflationary because the debts cancel each other out. The vendor draws a real bill against a merchant for 100 grams of gold in exchange for 1000 widgets, payable in 91 days. The mechant sells the bill to the bank at a discount of one percent. However, the bank rather than paying the merchant in coin, issues the merchant an IOU for 99 grams. So now the merchant has traded the vendor’s IOU for the Bank’s IOU.
But, the bill comes due and the bank collects coin in 91 days. So now the bank’s IOU, if still in circulation, is backed by actual gold coin. Therefore RBD in this form is not inflationary.
However, it does represent a small duration mismatch, because the bank’s IOU is a demand deposit, but the vendor’s real bill is not due for 91 days.
The following paper is very helpful in understanding the difference between Adam Smith’s Real Bills Doctrine, and the false doctrine that the Directors of the Bank of England turned it into.
Glasner, 1992, The Real Bills Doctrine in the Light of the Theory of Reflux
In your paper here you are proposing a principle that banks who borrow short to lend long will necessarily fail in the long run. Under Adam Smith’s version of the Real Bills Doctrine, an individual bank was issuing demand deposits against real bills with a 60 or 91 day maturation. Therefore your postulate here implies that Adam Smith’s version of the RBD is a recipe for bank failure – not because it is inflationary, but because it embodies duration mismatch.
Adam Smith himself did address the duration mismatch problem, but he did not state it as starkly as you have:
“To lend for longer terms [than 91 days] would mean that “the whole of the returns is too distant from the whole of the outgoings, and the sum of his repayments could not equal the sum of its advances within such moderate periods as suit the conveniency of a bank” (p201)
In Smith’s view the 91 days maturity of bills of exchange was close enough to the present as to present minimal risk of a bank run, but he agreed with your position on duration mismatch for longer durations.
To repeat, real bills are not debt and it’s not lending. This is no mere debate of definitions. I agree that 90 days minimizes the duration issue (as the average duration of the bank’s portfolio is of course half, or 45 days) but it’s more than that.
Real bills are clearing credit, where there is neither lending nor borrowing and the bill is not debt. The bill is nothing more than a mechanism that allows goods to flow from farm to factory to retailer to consumer while the gold coin flows in the opposite direction. It address the problem of clearing, that while the goods are moving from business to business, the receiving business does not have the gold coin yet.
Unlike borrowing, the bill is self-liquidating. That is the sale of the goods provides the gold coin to liquidate the bill. This is not true for borrowing, where the asset produces income that amortizes the debt.
There is no banking panic if there are bills and gold assets against demand deposits. One key reason is that long-term assets can fluctuate widely in value, and hence so can the bonds amortized by such assets. The bond market can go “no bid” if there is unusual selling, as banks are both illiquid and nervous and do not buy as they normally do. Selling becomes unmatched by buying.
The root cause of it is the extension of credit far beyond the intentions expressed by the savers.
Bills are both very short term but also do not represent an asset but goods in flow to the consumer. Within less than a season (90 days is no mere coincidence) the value of goods in flow do not vary in the gold standard (notwithstanding the incredible volatile we experience in the end days of the failing fiat paper regime).
The RB market does not go no bid. There is always a robust pool of buyers, especially if the discount rate ticks up. Fekete argues that there is no bid-ask spread in RBs. I am not sure I would go that far, but I think that the spread is de minimis and is hardly worth talking about as a phenomenon.
I don’t think that one should think of inflation as a rising money supply. For Fekete (and me) there is nothing to be gleaned by counting quantity of money, per se (I define inflation as the process of counterfeiting). Also, it is important to emphasize that bills and bonds and bank notes are not money. They are credit.
An expansion of legitimate credit is not inflation. Only an expansion of counterfeit credit. Which RBs most definitely are not.
You give an example of a merchant bringing a RB to a bank to get a bank IOU of the same size (actually discounted, one ounce less). Why would anyone trade a RB for bank paper? The RB is the highest quality earning asset, and the highest quality asset extant other than gold itself. The bank IOU is inferior and bank notes do not have a yield. In discounting the RB, the merchant will want actual gold.
Or, he may want bank notes because they provide small and convenient denominations to pay many small expenses, such as wages.
So the bank has RBs in its asset portfolio and notes are the matching liability. In other words, the bank issues notes to fund its portfolio of RBs. The bank earns the discount rate, minus its administrative costs and overhead to operate retail branches with tellers and other expenses.
While bank notes are lower quality than RBs, and they do not earn a yield unlike RBs that earn the discount rate, there is one reason for people to hold them. Sheer convenience. The convenience of having small, standardized denominations which are fungible outweighs the small amount of discount rate that could be earned. Just as today, no one keeps the bulk of their wealth in coins and bills, too in the gold standard no one would keep significant amounts of wealth in the form of paper bank notes.
It seems to me that the bottom line is that a merchant who sells his note at a discount should be paid in gold coin, or a deposit account backed with gold coin and payable in gold coin on demand.
If banks create ex nihilo bank notes or credit in their deposit account to buy the merchant’s bill of exchange then they are buying something with nothing, or rather with a claim on the asset which they are buying.
That would be like me offering to buy your car by handing you a slip of paper that says “payable with the title to your car.”
If it was morally acceptable for bills of exchange to serve as backing for a demand account, then the merchant would not be selling his bill of exchange, he would simply deposit it in the bank for credit to his account.
I tend to agree with Rothbard that deposit accounts should be 100% backed by gold coin, and that all other types of bank instruments can be backed by bills of exchange, bonds, and other investments.
This is basically what the Glass-Steagal Act achieved, with the exception of convertibility to gold coin. It forced banks to keep the demand account business completely separate from their commercial banking business.
Reforming the banking system would be encouraging banks to invest the majority of their portfolio in gold bills rather than in government bonds, and to require all demand accounts or bank notes to be convertible to gold coin.
“banks create ex nihilo bank notes or credit in their deposit account to buy the merchant’s bill of exchange”
It doesn’t sound like we are discussing duration here. It sounds like the topic is basic banking. There is a foundational concept of banking at issue.
1. Printing. Creating ex nihilo. A bank doesn’t have enough money, so it prints some notes and spends them on rent, payroll, or sending the CEO on vacation. This bank will soon be out of business. Aside from the discussion of fraud, it has liabilities exceeding its assets.*
2. Banking. A bank strictly matches assets and liabilities. It understands that it borrows in order to finance a portfolio of earning assets. The issuance of any liability–note, demand deposit, time deposit, etc.–is always and only to finance a matching** asset.
Those who say that banks “print” money (btw, it’s not money it’s credit) “ex nihilo” are often overtly hostile to the very concept of banking. Sometimes they reference the Biblical notion of “usury”, lending money at interest. Today, they often use the slur “bankster”. While of course there is crime and fraud in banking as there is anywhere else, the banking industry is not founded on fraud.
There is nothing wrong with borrowing, with or without interest. Banks are not special nor magical. They have a business model and they make money to the extent that they serve a need of their customers. They do not have (in a free market–I am not referring to today!) any special privileges, they are not immune to either market forces or the law.
*This, by the way, is an example of inflation as I define it. Borrowing without the means or intent to repay. The bank borrows when it issues a note, but if it does not use the proceeds to buy an asset then it demonstrates both lack of means and lack of intent to ever repay. It becomes a Ponzi scheme, issuing ever more liabilities to pay previous creditors in an accelerating game that can only end in one way.
** The duration discussion is a discussion about one criteria for the matching asset. It’s not sufficient that the asset match the liability in value. It must also match in duration.