Tuesday, April 8, 2014

Short Squeezes, Bank Runs, and Liquidity Premiums

This is a guest post by Mike Sproul. Many of you may know Mike from his comments on this blog and other economics blogs. I first encountered Mike at the Mises.com website back in 2007 where he would eagerly debate ten or twenty angry Austrians at the same time. Mike was the first to make me wonder why central banks had assets at all. Here is Mike's website. 

On October 26, 2008, Porsche announced that it had raised its ownership stake in Volkswagen to 43%, at the same time that it had acquired options that could increase its stake by a further 31%, to a total ownership stake of 74%. The state of Lower Saxony already owned another 20% stake in VW, so Porsche's announcement meant that only 6% of VW's shares were in “free float”, that is, held by investors who might be interested in selling.

Porsche's buying had inflated the price of VW stock, and investors had been selling VW short, expecting that once Porsche's buying spree ended, VW shares would fall back to realistic levels. Short sellers had borrowed and sold 12.8% of VW’s outstanding stock, but with free float now down to 6%, short sellers owed more shares than were publicly available. If the lenders of those shares all at once demanded repayment of their shares, then there would be 12.8 buy orders for every 6 shares available. In what was called “the mother of all short squeezes” share price rose until the short sellers went broke.

A short squeeze is bad news for financial markets, largely because the fear of short squeezes deters short selling, and thus inhibits the normal arbitrage processes that keep securities correctly priced. If I may make a suggestion to the owners of the world's stock exchanges, there is a simple way to prevent short squeezes from happening on your exchange: Allow cash settlement of all short positions, just like in futures trading. If the most recent selling price of VW was 250 euros, and if short sellers suddenly find no shares available, then allow those short sellers to pay 250 euros in cash (plus some small penalty) to the lenders of the shares, rather than having to return an actual share of VW. This would prevent the stampede to buy VW, and would assure that VW’s price would not skyrocket to crazy levels. (As a measure of short-squeeze mis-pricing, it is worth noting that VW briefly became the world's most valuable company at the height of the short squeeze.)

Short squeezes on stock exchanges are mercifully rare. Unfortunately they are not quite as rare in the banking world, where they go by the name of bank runs. Just as a short squeeze pushes short sellers to hand over more shares of VW than can be obtained on the market, a bank run pushes banks to hand over more currency than can be obtained on the market. And just as short squeezes can be mitigated by allowing cash settlement, so can bank runs be mitigated by allowing banks to settle their obligations in forms other than currency. Clearinghouses and other banking associations can issue loan certificates or scrip for use in clearing checks, or even for public use as currency. Some creativity might be required in the issuance of money substitutes, but in return banks are spared from having to sell their assets at distress prices, while the community is spared from the effects of a bank panic.

What I find most interesting about short squeezes and bank runs is that they are a clear case of market failure, where financial instruments are obviously trading above the value of the assets backing them. During a short squeeze, value is no longer determined by backing, but by the forces of supply and demand. I don't think that economists pay enough attention to this point. The price of financial securities is normally determined by the underlying assets, while the price of commodities is determined by supply and demand. When economics textbooks explain supply and demand, they speak of the supply and demand for apples and oranges or other commodities. They rarely if ever speak of the supply and demand for stocks and bonds, because stocks and bonds are not objects of consumption, and they are not produced using scarce resources. There is no production function and no consumption function, hence there are no supply or demand curves. When we examine a bond that promises to pay $105 in 1 year, we find the price of that bond by dividing 105/(1+R). If R=5% and we tried to sketch supply and demand curves for that bond, we would draw a pair of meaningless curves that were both horizontal at $100. This is what makes short squeezes so strange. The price of VW stock is supposed to be determined by backing, and not by the supply and demand for VW shares. But during a squeeze, supply and demand take over, and stocks trade at a premium relative to their backing. The same might be true of money during a bank run.

This is a problem that JP and I have batted around a bit. I usually argue that arbitrage prevents money from trading at a premium relative to its backing, while JP usually argues that money can trade at a small premium. I can never pin him down on the size of the premium, but he doesn't argue much when I throw around a figure of 5%. Well, here we have VW stock trading at a premium of 500%. Might such a premium be possible for money?

Apparently not. We never see comparably large premiums on currency during bank runs. Gerald Dwyer and Alton Gilbert (Bank Runs and Private Remedies, May/June, 1989) examined American banking panics that occurred between 1857 and 1933, and found that the largest paper currency premium (relative to certified checks) ever observed during bank panics was 5%. The average paper currency premium during bank panics was much lower, only about 1%. Other measures of a currency premium, such as a rise in the value of money relative to goods in general (i.e., deflation), are also in the modest range of 1-5%. Why the enormous gap, from a 1% premium on currency to a 500% premium on VW stock? My best explanation is that banks can get creative in devising alternate forms of payment, while the traders in VW stock simply did not have the time or the legal means to devise alternate forms of payment. Thus the market in VW stock failed catastrophically, while banks facing a run are able to muddle through.

The result of the banks' muddling with money substitutes is that even during stressful events like bank runs, the value of money is, at most, only 5% higher than its fundamental backing value. This makes sense, because any premium over backing value gives an arbitrage opportunity to investors. If the fundamental backing value of each dollar is 1 oz. of silver, and if the dollar somehow trades at 1.05 oz., then the issuer of that dollar earned a free lunch of .05 oz. This free lunch would attract issuers of rival moneys, and rival moneys would keep being created until each dollar traded at its fundamental value of 1 oz.

The idea that money is worth no more than the assets backing it is consistent with finance theory, and with the backing theory of money, but it contradicts the quantity theory of money. The quantity theory asserts that modern fiat money has no backing, that it is not the liability of its issuer, and that its entire value is therefore a monetary premium. Which of the two theories gives a better fit to real-life moneys? When we look around for moneys that fit the quantity theory, that have no backing and are not anyone's liability, we find very little. Just bitcoin and a few orphaned currencies like the Iraqi Swiss Dinar. When we look around for moneys that fit the backing theory, that are the recognized liability of their issuer, and are backed by their issuer's assets, we find every other kind of paper and credit money that has ever existed. I conclude that the backing theory beats the quantity theory.


  1. JP, I see you've moved on to another post... and I was a bit late and left a flurry of questions on the last one. But if you do still have time to look at just one of them, I'd chose this one:

  2. Great post, Mike.

    What makes this especially interesting is that the VW price spike was in the context of the Fall 2008 credit crisis, when everything else was getting whacked.

    By the way, have you ever read John Cochrane's Stock as Money?


    You mentioned introducing margin or allowing for alternative media for settlement to help control short squeezes. Another fix would be to allow short sellers to borrow stock for fixed terms rather the conventional callable stock loan. Just like term deposits help protect a bank from a run, a short seller who knows that they don't have to pay back their stock for, say, 1 year needn't be susceptible to a squeeze.

    1. Mike awesome article!

      Would fixed terms be any different than a European style option (where the option can only be exercised at maturity)? Because those European style options are available on VW stock, and investors chose to short through shares instead of buying optionality. They went Long Term Capital Management on VW and blew up.

      When you short a stock you're obviously jumping into a zero sum (well less than zero sum) bet whereas going long on a stock (well technically only the ipo) is providing essentially a loan to the company in exchange for ownership and a claim on future earnings.

      The majority of the people trading the stock (broadly painted as short sellers) and not providing any benefit for productive means transfer their wealth to the other participants (and brokerage firms via trading fees). Isn't that what should happen? Shouldn't capital transfer to the market participants who provide the most valuable service?

    2. Dave:

      Yes; fixed terms=European option. Along the same lines, a convertible currency is an American style option, while an inconvertible currency is a European style option.

      Short selling does provide a useful service in that it allows the opinions of short sellers to be reflected in the price of the underlying stock. Empirically, economists find that when short selling is banned, stock volatility rises.

      Your statement that stock buyers finance productive activity while short sellers don't is plausible, but if it were correct, then short sellers would systematically lose money, and that would violate the no free lunch principle. So here's an attempt at an answer: All investors (as opposed to producers) are merely placing bets on their favorite ponies. The guy who bets for a pony is no more or less productive than the guy who bets against the pony.

  3. Thanks JP!

    Short squeezes definitely seem to break a lot of rules.

    I wrote a little about Cochrane's paper in my 2011 paper entitled "The Fiscal Theory of the Price Level and the Backing Theory of Money". He did a good job on it. When I sent him a copy of my paper he responded that it was "a nice paper", but no details. I remember suggesting a while back that his paper should have been titled "Money as Bonds", to emphasize that money is valued more like bonds than like stock.

    I agree with your last paragraph. It might be more practical than cash settlement.

  4. I've never been a fan of cash settlement. To me it allows too much silliness in the market, where things like credit default swaps get issued at orders beyond the amount of underlying debt (only possible with cash settlement where the underlying debt doesn't have to be delivered to the insuring institution, just the market value in cash. More CDS than underlying could of course still be issued even in the event that delivery of the actual underlying was required, just as long as investors realize they're probably going to have to buy from the insurance company and sell right back to them. I don't think most people would find that proposition appetizing) that highly fragilizes the system. It seems like a cheap, semi-dishonest work-around ("tell me what you really mean" - I realize that's an overly harsh characterization but I'm having trouble finding the right words). That is why i prefer market structures like the one JP shortly outlined. Then again, I tend to be more paranoid than the average bear about market fragility and allowing structures that cause it.

    What I never understand: why would anybody short stock when its interest is in excess of the float? why not synthetically short with options or take only the short side of the bet by going long puts? To me the VW case is mostly an example of the arrogance and stupidity of some traders in the face of the market. There are maxims like "the only shorts you make money on are frauds" that are meant to get people scared of borrowing stock on margin for precisely this reason - I can only hope that most of them had protective calls or something loss mitigating for their positions.

  5. John:

    The problem I see with cash settlement is market manipulation, Some guy takes 1000 short positions in VW when the price is 250, then he sells 1 share for $1, and since the most recent sale of VW is now at $1, he delivers $1 each to settle each of his positions. The exchange would have to put controls in place to ensure that only bona fide trades count, but those controls might be expensive. JP's suggested alternative of fixed terms avoids this problem, but there might still be times when cash settlement would be helpful.

    Aside from this, cash settlement allows more flexibility and therefore makes markets less fragile, not more. With the right rules (e.g., cash settlement and fixed terms) in place, it's possible for people to take 2 million short positions when there are only 1 million genuine shares that exist. Just like it's possible for banks to issue $5 bil checking account dollars when only $1 bil of paper dollars exist. Banning such arrangements hampers short selling and makes markets less stable, not more.

    People might prefer short positions to puts because puts cost money and naturally drop in value over time.

    1. I agree, except that the increased flexibility allows for increased leverage, and this often leads to very serious problems.

      This is especially true with instances of speculation as opposed to hedging. That is why sophisticated brokers spend a lot of effort making sure they've identified and categorized their clients correctly in either the speculating or hedging camp.

    2. John:

      If people want to trade with a guy who has borrowed $99 to buy $100 worth of stock, then I'm OK with that. It's just Darwin doing his thing.

  6. Don't stock loan agreements effectively provide for a type of cash settlement anyway, in this sort of situation? Standard agreements like the GMSLA do not normally treat non-delivery as an event of default (thereby avoiding cross-defaulting the borrower) but simply require payment of a market value amount, calculated in favour of the lender as if it were an actual event of default.

    1. Nick:
      VW shares are traded on German stock exchanges, most importantly Frankfurt. I don't know Frankfurt's rules, but I'd bet that the exchange did not allow cash settlement when the Porsche/VW squeeze happened.

    2. That reminds me of Treasury markets prior to 2010 or 2011. If they couldn't find treasuries for settlement, short sellers always had the option to indefinitely fail to deliver. The only penalty was foregone interest. Garbade calls this ability to postpone the obligation to deliver without explicit penalty a "convention" in treasury markets.

      In a sense, the short sellers were issuing 'loan certificates or scrip' which would finally be canceled when they could deliver the treasuries.

  7. I don't understand how before the announcement, both the short sellers and porsche could have had the shares. Wouldn't the percentage have exceeded 100%? How is that permissible?

    Either the people writing the options which they sold to Porsche didn't have the shares or the people from whom the short sellers borrowed didn't have the shares. Something is fishy. Are we talking about a situation known as naked shorting, i.e. shorting without actually borrowing the shares? That should not be permitted.

    1. Prakash:
      Suppose you call a stock broker and tell him you want to buy 1 share of VW. He answers "OK, I don't have a share yet, but pay me $250 now and I promise I'll deliver you a share by 5PM today." The broker has just gone short and you have just gone long. You are both consenting adults. You could have just as well made a bet with the broker: "For every dollar VW rises, you pay me $1, and for every $1 VW falls, I'll pay you $1." That bet is equivalent to a futures trade in VW.

      Now, legalities aside, that kind of trade with a broker could happen 2 million times, so that there are 2 million short positions taken, even if there are only 1 million actual shares of VW in existence. Once again, you are consenting adults. This is also how fractional reserve banks operate.

    2. "This is also how fractional reserve banks operate."

      Good point.

      Naked shorting gets unfairly maligned. Before banking arrived on the scene, gold coin would probably have carried a large liquidity premium. But if bankers could short gold without actually borrowing any coin (just by printing gold redeemable notes), they could issue exchange media that was competitive with coin. This would reduce the liquidity premium on gold to some lower level, thus providing society with more liquidity -- at less cost.

      Speculators who engage in naked shorting in the context of the stock market are doing the same. If a stock has a large liquidity premium, a broker who doesn't have any shares but sells them short is providing a socially useful service. By providing the market with more stock and stock IOUs, they'll diminish the liquidity premium on that stock and provide traders with more liquidity -- at a cheaper price.

      It's good when entrepreneurs find new ways to reduce the prices we have to pay for things like televisions, health care, and transportation. The same goes with liquidity. The more of it, and the cheaper, the better.

  8. Might another way of framing the backing theory of money versus the quantity theory of money distinction be to say that money is an asset rather than a collectable. Bitcoin is a collectable whilst US dollars are an asset. Bitcoin gets its value because there is a fixed quantity and so it has rarity. US dollars get their value because "the system" ensures that the value of real goods and services are tightly pegged to the dollar. So the dollar is the supreme store of value over the short to medium term and that confers its asset value. As you say, during a short squeeze, stocks act as a collectable BUT in normal times they act as an asset in that their value depends entirely on what they deliver (dividends, chance of upside appreciation) not on their rarity.

    1. I think the asset value of money fits in nicely with Ole Peter's asset pricing model where every asset price trends towards the price at which nothing is gained by borrowing money to purchase more of the asset:
      The two attributes that contribute towards setting such an asset value are return and volatility. Volatility means that if a fixed level of leverage is used, the asset will need to be bought and sold as the asset price bobs around and that will lead to volatility decay trading losses. Money has the trump card when it comes to low price volatility. Money systems break down whenever it becomes profitable to borrow in them to hold another type of money and then exchange back to settle. I guess the backing theory of money would characterise that currency slide scenario as being symptomatic of a lack of "backing" ie contracts, wages, tax demands etc not being sufficiently fixed in terms of the money in question?

    2. Stone:

      If I own a paper dollar, or a bitcoin, it shows up on the asset side of my balance sheet. The paper dollar shows up on the liability side of the Fed's balance sheet, while the bitcoin does not show up on the liability side of anybody's balance sheet.

      A agree that bitcoins have value because of rarity--just like gold and baseball cards. An ounce of gold, or a bitcoin, or a baseball card, show up on the asset side of their owners' balance sheets, but they do not show up on the liability side of anyone's balance sheet.

    3. Mike Sproul, I wonder though whether money could have value without being rare and without being anyone's liability so long as price stability of real goods and services was assured in that money (by convention, regulation etc). I'm not convinced that historical examples of tax-token money (such as royal tally sticks used in England from the 1100s for a few hundred years, or the tax tokens issued by some US states or the tax tokens issued by colonial authorities in 1800s Africa) were tightly reliant on rarity. With all the QE we have had in the UK, USA and Japan, our money now seems less and less like credit money and more and more just like an asset that has value stability as its full attribute.

    4. Stone:
      I don't know of any historical token money that was not backed by its issuer's assets. But it's deceiving, since the assets aren't always obvious. For example, in 1690 Massachusetts was bankrupt and on the verge of collapse. It issued some paper IOU's, called them shillings, and paid their soldiers with them. They looked unbacked, since there was no money in the treasury, but people forgot that Mass still had the ability to collect taxes, and 100 shillings of "taxes receivable" can serve as backing just as well as 100 shillings worth of silver or bonds. Kurt Schuler recently made the same error over at his Free Banking blog, when he talked about the Argentinian currency board of the early 1900's.

    5. Mike Sproul, your inclusion of "taxes receivable" as a backing asset clears up everything for me as it brings tax token aspects of money fully into it,
      These are some links about tax token money:
      http://globaleconomicanalysis.blogspot.co.uk/2007/06/why-does-fiat-money-seemingly-work.html (see section "Tally sticks and Charles II")

      I also think that private debts denominated in the currency issued by the state monetary authority do a great deal to confer value to the state money. Basically the state just needs to ensure that the private sector uses state money as the unit of account and then private indebtedness does most of the heavy lifting of conferring value to the currency.

    6. Concerning the role of private debts, if central bank issued money is subject to legal tender laws, I'm not sure it requires any backing, even the taxing power of the government.

      Say I hold a bond issued by XYZ. Now if I issue a CLN (providing for physical delivery) where the underlying is the XYZ bond, the bond has effectively become a risk-free asset for me. I no longer care whether XYZ has any assets at all, because whatever the bond is worth, I can use it to settle my CLN.

      Legal tender laws effectively make all monetary debts CLNs for which the underlying is the currency. Whatever the true value of the currency, I can always use it to settle my debts. As long as there was debt (and in the real world, there is an awful lot of debt), currency would be valuable even if the central bank had no assets at all.

    7. Nick Edmonds, is the issue that there is more to ensuring people denominate their debts in the government's money than simply legal tender laws? In many countries, foreign currency denominated debts are fairly normal. Many companies not in the US nevertheless have USD denominated corporate debt. Households in Hungary had Swiss Franc denominated home mortgages. Don't some UK hedgefunds etc have JPY denominated debts? The trick is for the government to ensure that enough private debts are denominated in their currency. Perhaps taxation and other "backing" are all part of that. I guess a lot of what the US military and CIA does is ensuring debts and contracts outside of the US are USD denominated for much the same reason.

    8. Maybe. But I wasn't saying legal tender laws are all you need - just that I think they imply that the central bank doesn't need any backing assets.

    9. Stone and Nick:
      If money acquired value because of the denomination of private debts, then we'd have a violation of the no free lunch principle. If the government has 100 oz worth of silver, bonds, etc, as backing for 100 of its IOU's ("dollars"), then $1=1 oz., and everything's fine. But if the government has issued $101, and assets are only 100 oz., and the value of the dollar somehow stayed at $1=1 oz, then arbitragers would be all over it. The first $100 to be redeemed at the bank would get paid 1 oz worth of stuff, while the last $1 in line would get nothing. Just as arbitrage assures that stocks and bonds trade at their backing value, it does the same for money. This result is not affected by the fact that some third parties might have issued some IOU's denominated in dollars. The government's assets back the government's money, and the third party's assets back the third party's money.

      About legal tender: That can mean that the government declares dollars to be legal tender in payment of taxes. That's ok, as it provides actual backing. The government can also hold a gun to walmart's head and tell the public that walmart will accept dollars as equivalent to 1 oz. This also provides backing, although the backing was stolen from walmart, and if walmart goes broke, the backing is lost and the dollar loses value. The backing theory works whether the backing was stolen from walmart, or from taxpayers.

    10. Not sure I follow this (specifically the legal tender point). Say Company ABC has made some disastrous business decisions and has no assets, but still has $100 of IOUs outstanding. These IOUs are clearly worth nothing. Now let's say the government passes a law (puts a gun to everyone's head?) that these IOUs will be legal tender and may be used to settle any debt at face value. The IOUs will now be worth $100 (whatever the real value of a $ is).

      So the holder of the IOUs is suddenly $100 better off. Who has this been "stolen" from? Economically it has been stolen from everyone else that holds $ denominated assets, although those that are indebted in $ actually benefit. But I'm struggling to see how the continued solvency of those rich in $ has a bearing on its value.

    11. Nick:
      Start with the government declaring that ABC's IOU's must be accepted by walmart at face value. If walmart's net worth is only $90, then those 100 IOU's are worth $.90 each, since that's as much as the government could steal from walmart. Now suppose the government starts adding sears, kmart, etc. It's still the case that if the victims' assets total $90, then 1 IOU=$.90, but if those firms have net worth >$100, the IOU's will trade at par. This assumes that the government is physically able to steal from those firms. If it loses the ability to steal, then the IOU's will drop in value just like money issued by an insolvent bank.

    12. I think that in theory both Nick's legal tender argument and Mike's backing argument are sufficient to give bits of paper a positive value.

      In practice, I think the fact that central banks do choose to hold assets rather than sell them off is evidence in favor of Mike's point. If they didn't need them, they'd have sold them decades ago.

      Also, legal tender laws are far less draconian than one would assume if the positive value of central bank money depended on them. Given their leniency, I can only assume these laws' monetary role isn't very big.

    13. Mike Sproul, you say that if the dollar was at a value in excess of the value of the backing held by the central bank, then "arbitragers would be all over it". What precisely do the arbitrageurs do? Do you mean a shift in exchange rates versus other currencies or a shift in the yields of the bonds held as backing by the central bank? It all seems very much muddied by the fact that the assets held as backing by the central bank are fixed interest securities denominated in the very same currency that comprises the central bank liabilities.
      Perhaps stress testing the idea by setting out realistic scenarios where the USD lost all of its value would help to explain it all. I guess if people in America all started transacting in mpesa mobile phone money (as in Kenya http://en.wikipedia.org/wiki/M-Pesa ), M-Pesa had a floating exchange rate with the USD, all new private debts and wage contracts etc were denominated in M-Pesa, and the government only accepted taxes in M-Pesa, then the USD value would slide to zero. The fed would still be holding trillions of dollars worth of dollar denominated treasury bonds and MBS as backing for the currency but trillions of dollars would not buy you anything anymore. I guess it is a classic currency substitution hyperinflation scenario.
      I suppose that Hyman Minsky quote "everyone can create money; the problem is to get it accepted" is the key point. Successful currencies need to have all angles covered so as to ensure that they continue to be accepted. I also think a lot of it is probably down to trust in convention. Orphan currencies, shell money and I guess a large part of gold's value are driven by convention. There is a heck of a lot of gold stockpiled, it would swamp the market for wedding rings for a long time if people stopped viewing it as a store of value and sold off that stockpile BUT enough people trust that everyone else will continue to want it for its store of value role so it has successfully served that role for a very long time and probably will continue to do so.

    14. Mike - not convinced by that. Say Walmart's $90 net worth consists of a loan of $100 to me and loan of $10 from the bank. Then the 100 IOUs are worth $1 each. If I could buy them from someone else for $0.90, I would make a profit of $10, so I will pay anything up to $1 each for them. The same applies even if Walmart has negative net worth.

      Of course once all the debt the IOUs can be used for has been extinguished, they are worthless again, so an example with only Walmart and only a $100 loan is a bit misleading. In reality, the uses for legal tender are highly unlikely to ever be fully extinguished.

      JP - I should clarify that I'm not offering an alternative theory of money here - it just struck me that money subject to legal tender laws should retain its face value even if the issuing central bank had no assets.

      I think you mean that it would distribute out its assets if it didn't need them - selling them obviously wouldn't change its net worth. I don't know exactly why central banks don't do this, but in general I'd guess it's because they tend to be run as ordinary corporate entities that can normally only distribute out of post-tax profits. I'm not sure on this yet, but my feeling is that it has nothing to do with a need for backing.

    15. Nick, I guess in the days before QE, the central bank's holdings of assets allowed then to reel back in bank reserves by selling off assets to drain bank reserves as a way to increase interest rates whenever they wanted to. In the post-QE world where there is no scarcity of bank reserves perhaps it is just a historical legacy that that is the institutional structure?

    16. In the UK, the BoE's main asset by a long shot is its loan to the APF. In theory (but probably not legally), it could distribute this in specie to the Government, but then the excess reserves could never get extinguished. If the APF sold off its bond holdings and repaid the loans, the government would just end up sitting on the reserves. It would all be very messy and unnecessary. It makes much more sense to run the BoE as a solvent entity.

    17. "I should clarify that I'm not offering an alternative theory of money here..."

      Sure you are. It's a legitimate answer the question: "why do intrinsically useless/valueless pieces of paper circulate at a positive value?"... which is basically one of the biggest puzzles in monetary economics. Mike's backing theory is also a good answer.

    18. JP - maybe, but let me clarify. I don't think that pointing to legal tender laws is sufficient to answer that question, because the relevance of those laws presupposes the existence of monetary debts. All I think legal tender laws do is ensure the quality of the central bank's money as a risk-free asset, whereas the quality of commercial bank money does indeed depend on its backing. This is part of what enables the central bank to exert its special influence over money. If I was giving my answer to the question of why people accept worthless bits of paper (in general as opposed to specifically the central bank's), legal tender laws would only figure incidentally.

      Also, I don't disagree that the existence of appropriate backing would be sufficient to make money acceptable; I'm just not convinced it is essential (although I'm still open on this).

    19. Stone:
      If the money issuer holds 100 oz worth of assets against $101, and if it pegs $1=1 oz, then the arbitragers would line up to get their 1 oz for $1, knowing that the first $100 in line would get 1 oz, and the last dollar would get nothing. People would also short the dollar against silver and profit as the dollar fell.

      If assets are denominated in the same currency that the bank issues, then there is a problem of inflationary feedback, where, for example, a drop in asset values will cause the dollar to fall, which causes assets values to fall more, etc. I discuss this in my paper entitled "The Law of Reflux"). The Law of Reflux also answers your question about M-pesas. As people stop using dollars, they would reflux to the Fed, first for the fed's bonds, and ultimately for the Fed's gold. If the Fed failed to buy back the refluxing dollars, then that is tantamount to a loss of backing and the dollar would lose value. But if the Fed buys the dollars back at par as they reflux, the value of the dollar is unaffected.

    20. Nick:

      Those were ABC's IOU's, not walmart's. If walmart's net worth is just $90, and the government declares that walmart must accept all 100 of ABC's IOU's, then 100 of ABC's IOU's are backed by nothing but walmart's $90 net worth, and 1 ABC IOU=.9 oz.

    21. Mike. I think you misunderstand me. The existence of a debt owed to Walmart (representing all the debt to which legal tender applies), in addition to the ABC IOUs (representing currency), is essential to my point. I agree that if Walmart's net worth is simply $90 worth of silver and it has no financial claims, then the IOUs cannot be worth $100. But I also agree that, in a world where there were no debt, that legal tender laws might not be enough to give currency its value. My point depends on the assumption that there is a lot of pre-existing debt out there, because it's those debtors that generate the demand for the otherwise valueless currency.

    22. Nick:
      We need to distinguish two views:
      1) The State Theory or Chartalist view, which says that taxes (and legal tender laws) create a demand for money.
      2) The backing theory, which says that taxes (or assets seized through legal tender laws) are an asset to the money issuer, and thus back the money in the same way that a firm's assets back the stocks and bonds it has issued.

      Here are some reasons I like the backing theory better:
      1) it means money is valued for the same reason that that stocks, bonds, etc. are valued. There is no need for a "special" theory of money.
      2) As rival moneys (foreign currency, checking accounts, credit cards) are issued, the Chartalist (=quantity theory) view implies that the resulting fall in the demand for base money will cause inflation. Not so for the backing theory. If 100 base dollars are backed by 100 oz worth of the issuer's assets, then $1=1 oz regardless of what rival moneys come along.

    23. Thanks Mike. I think my own views are closer to the Chartalist version, but I'm still thinking it over.

  9. I guess what you are saying also goes along with this article about the Bank of England under Bagehot in the 1800s. Apparently Bagehot saw the role of the central bank as preserving price stability by undoing just the sort of squeezes on money that you describe:
    "Bagehot himself famously urged the Bank to accept as collateral “what in ordinary times is reckoned a good security” rather than attending to current market valuation. The point was to prevent troubled banks from being forced to liquidate fundamentally sound assets at fire sale prices."

  10. I'm not really convinced by 1800-era 5% premiums, because gold standard cash runs are very likely different than unbacked runs. There is the problem of survival sampling also -- cash runs for the USD are not the Weimar Mark, for instance. Unbacked runs are easy to handle, of course -- just turn on the presses; whereas gold standards likely had more defaults. But printing currency to satisfy bank runs is a huge event -- it swaps debt-deposits for base money. Base money is the ultimate deliverable for all debt contracts -- and the supply of base money affects the valuation of all deposit-debt-promise-to-deliver contracts. Bottom line, deposits are a very different beast than base money currency.

    Handing out base money to make debts whole is the essence of hyperinflation. In hyperinflation, there is never enough currency on hand to keep up with rising prices, and so more currency is demanded into existence. You have to pay debts back, and so demand base money -- the ultimate deliverable -- into existence, and so the central banks accomodate, rather than risk debt-deflation.

    The gold standard make bank run risk a real economy event, not a monetary event -- debts were purged fairly quickly alongside deflation, while money was stable. We're now looking at the converse -- a buildup of huge debts that inevitably will be monetized by base money production (the Fed is now doing it now as we write).

    The next step is for these reserves to be demanded into currency form, as happened once before in the 1930s-1940s. IOR can change this equation, by giving banks an incentive to hold reserves, but IOR does not affect the marginal demand for currency by the public.

  11. Anonymous:

    On backing theory principles, the monetization of debts is irrelevant, since it just replaces one government liability (bonds) with another (government-issued money). The transformation of reserves into paper currency is irrelevant for the same reason. That's not to say that an insolvent government won't see its currency inflate. It will, but it will see that inflation with or without monetization or reserves-to-currency conversion.

    Dwyer and Gilbert were pretty careful about that 1-5% premium of currency relative to certified checks, and the fact that measured deflation rates were in a comparable range reinforces the idea. Remember that I was asking whether money carries a premium above its backing value. Bank runs, which push things to extremes, provide a good way to answer that question.

  12. Mike, debt is not equivalent to base money. Debt is a promise to pay base money. (Also, both public- and private- debt demand matters, not just federal debts.) Debt extinguishes, as there is an asset versus a liability, but base money has no effective deliverable, or liability.

    Bottom line, debt is not a medium of account for all assets: base money is. I'd think about moving away from the issuing entity's perspective and consider the private economy perspective: we're not worried here about how the government funds itself.

    Printing bonds does not much change expected price levels, because cash is absorbed on issuance and issued as bonds mature: a time difference, but zero sum. Printing (zero-maturity) base money indefinitely dilutes the (universally deliverable) medium of account for all financial assets. (At present, there's not much price impact, as base money is in reserve non-public form, with very little borrowing demand (or else rates would not be near zero). If the Fed had done QE in currency -- buying bonds for physical cash -- instead of reserves, we would be seeing another price level entirely.)

  13. Anonymous:

    You are of course expressing the mainstream quantity theory view, but the backing theory rejects that view, and asserts that so-called fiat money is a true liability of its issuer, and is truly backed by the issuer's assets. Clicking on my name above will take you to my writings on the backing theory (aka the real bills doctrine).

    It's no secret that paper money issued by a government (or its central bank) always appears on the liability side of that government's (or central bank's) balance sheet. The Fed's balance sheet shows that, and even identifies its assets as "Collateral held against Federal Reserve Notes", in other words, backing. If the government issued paper money without a central bank, then any accountant will confirm that the paper money appears on the liability side of the government's balance sheet, and bonds issued by that government appear in the same place. The quantity theory asserts that the accountants are wrong, and the 'liability' is merely a fiction. The backing theory says the accountants are right.

    Since we have limited space here, just ask yourself why central banks hold assets at all? Its is because they truly back the money issued. If central bank assets don't matter, then why does every central bank we ever heard of hold assets that are equal in value to the money (+ other liabilities, if any) issued by that bank?

    1. Anonymous, you're saying that QE would be totally different if the bank reserves exchanged for the bonds purchased by the fed were held by the public rather than being held by banks. I think it is crucial to bear in mind that for QE although the fed buys the bonds from banks, most of those bonds were initially held by non-banks and then sold to the fed via a bank simply using the bank as a short term conduit for the sale. So the overall transaction is a non-bank getting bank deposits in exchange for bonds. eg see

      I also think the notion that it makes a big difference whether that money is bank deposits or paper folding money is mistaken. If an insurance company or family office or whatever has sold off millions of dollars worth of treasury bonds into the QE program (via a primary dealer bank) do you really think that they are going to have a different view of how to deal with that money depending of whether it is a bank deposit or a vault full of pallets of paper bills? Either way they are just going to see it as an account statement. I think the people selling bonds via QE largely hold cash for its "option quality".
      There is something about why Warren Buffett holds cash for just that reason:

      “”He thinks of cash as a call option with no expiration date, an option on every asset class, with no strike price.” It is a pretty fundamental insight. Because once an investor looks at cash as an option – in essence, the price of being able to scoop up a bargain when it becomes available – it is less tempting to be bothered by the fact that in the short term, it earns almost nothing. Suddenly, an investor’s asset allocation decisions are not simply between earning nothing in cash and earning something in bonds or stocks. The key question becomes: How much can the cash earn if I have it when I need it to buy other assets that are cheap, versus the upfront cost of holding it? “There’s a perception that Buffett just likes cash and lets cash build up, but that optionality is actually pretty mathematically based, even if he does the math in his head, which he almost always does,””

  14. Mike Sproul, did you see the stuff about a facility for fed "negative liabilities"? I thought it might be relevant for your theory.
    “There’s actually a Fed accounting facility in place, ready to go, that will allow for temporary losses by creating “negative liabilities” instead of negative capital on the Fed’s balance sheet.”

  15. Mike, long understood, but backing does not mean a thing unless there is a deliverable. There's no arbitrage available, and so the point is moot. Without deliverables, there is no difference between your "backing" and a state of being "unbacked", and so is unfalsifiable. Why bother?

    Also, to answer, the Fed holds Treasury assets versus base money liabilities because the Fed is a private institution -- take a look at their annual report. The Fed has equity capital, full stop.

    stone, yes they created demand deposits, but former-bondholder wealth is net zero changed -- they sold their bond and got cash. Just a portfolio mix issue. Again, doesn't signify -- moot.

    The real power of the Fed is to dilute the means of account. Bank reserves are undtradable in the real economy, and so do not affect the MOA for the broad economy.

    The only other form of base money is currency. Dilute currency, and you've succeeded in diluting the MOA for broad NGDP.

    Neat stuff about cash optionality -- cash is a long put on assets, paid in inflation!

  16. Anonymous:

    Suppose the Fed initially holds 20 oz of silver as backing for $20 FRN's it has issued. The silver is both the backing and the deliverable. Of course convertibility might be delayed by a day, a year, or a century. Delays don't change anything important, except that interest becomes a bigger factor.
    Next, let the fed issue another $80 FRN's in exchange for $80 worth of bonds (note: denominated in dollars, not ounces). The public now holds $100 of FRN's, which are still worth 1 oz. each. To see this, define E as the exchange value of the dollar (oz./$). Setting assets (20 oz + $80 worth of bonds worth E oz. each) equal to liabilities ($100 worth E oz. each) yields


    or E=1 oz./$.

    Now suppose there is a 30% reduction in the public's desire to hold FRN's. Thirty FRN's pile up, unwanted, in private hands. The holders probably intend to present them to the Fed and demand 30 oz, but if the fed is smart, it will instead sell $30 of its bonds, thus soaking up the unwanted $30 of FRN's. No need for the fed to pay out the silver. The bonds serve as the deliverable. This makes sense, since those last $80 were created in exchange for bonds, they can of course be retired in exchange for those same bonds.

    Not sure about your point as to the fed being a private corporation. Legally yes, it is. But of course the Board of governors are government appointees. If the fed were combined with the treasury, then the $80 worth of bonds would appear on the liability side of the treasury's balance sheet, but would effectively be cancelled as the $80 FRN's were issued.