Tuesday, December 19, 2017

Money as a generally-accepted medium for short selling

Jim Chanos, famous short seller. We are all Jim Chanos.


Most people find the idea of short-selling to be incomprehensible. Buy a stock and hold it, that's what one does. To the majority of us it's just down-right odd to do the reverse, borrow stock in order to sell.

At the same time, pretty much everyone in the world is a short seller, even if we don't realize it. The credit card debt we wrack up, the lines of credit, the pay day loans, the mortgages—they're all examples of us going short. We borrow a certain type of security—dollars or yen or other types of money, either in paper or digital format—and immediately sell it. And then after a little time passes we cover that short, buying the dollars or yen back and repaying the loan. We are all Jim Chanos, the world's most famous short seller, the only difference being we tend to short different instruments than Chanos does.

The only time I ever sold a stock short was back in 1999. I was still in university and probably a little bit reckless. What I didn't know at the time was that there was still a year or so left in the crazy late 1990s bull market. The price of the stock that I had shorted immediately began to move higher, and I got worried. The problem with short selling is that because a stock can keep rising forever, losses are theoretically infinite. After a month or two I bought the stock back to cover the loan, then got back to my studies.

While I've only sold stock short once, I've sold money short umpteen times. Borrow some Canadian dollars, sell it for things like groceries or a plane ticket or tuition, wait a while, repurchase the money, pay the loan back.

So why don't I finance my consumption by shorting things like Netflix or bitcoin? Why do I short dollars instead?

The nice thing about shorting money rather than Netflix shares or bitcoins is that I know exactly how much it'll cost me to repurchase the necessary securities to cover my short. Our labour is almost always priced in term of money, say $30 per hour. So if I've shorted three one-hundred dollar bills, I know ahead of time that I only need to sell ten hours of my time to buy that $300 back.

What's more, labour tends to stay sticky for months. This means I don't have to worry about my per-hour rate plunging temporarily to $15 next Wednesday, forcing me to spend twenty hours of time instead of just ten to cover my short. And since the central bank sets an inflation target of 2% or so a year, I already know far ahead of time that if I'm making $30 per hour this year, I'll be making ~$30.60 next year. So whereas one hour of my labour allowed me to cover a $30 short position in 2017, that same hour will allow me to cover a $30.60 short position in 2018. That sort of long-term certainty is a great feature.

Not so with bitcoin or Netflix. The big problem with using these instruments to finance consumption is that labour is never paid in terms of bitcoin or Netflix, but in yen or dollars or pounds. So while the sticky nature of labour means I know ahead of time that I'll be able to sell an hour of my time for $30, I don't know how many bitcoins or Netflix shares this $30 will allow me to repurchase to cover my bitcoin/Netflix shorts. This would be less of a problem if Netflix and bitcoin were fairly stable in price, but both are terrifically volatile, as I described here.

Consider a scenario in which I short one share of Netflix in order to fund my weekly supply of groceries (which costs ~$200)—or alternatively I short 0.01 bitcoins—and the price of either of these two assets doubles over the next few days. I'll end up having to pay $400 to cover the short, or fourteen hours of labour. If I had just shorted $200 dollars instead, I'd only owe seven hours ($200/$30).

A worst case scenario is one in which Netflix of bitcoin start to rise to infinity. If so, I'd have to work every waking hour of my life just to repurchase Netflix/bitcoin and cover my short. No thanks, I'll take the predictability of shorting money to pay for $200 worth of groceries. Central banks can create and cancel whatever amount of money they need to keep the purchasing power of money on a narrow path. I'll never have to worry about the nightmare of working every day for the rest of my life just to cover a tiny short position.

-----

Without the institution of short selling, society is made worse off. The timing of people's incomes do not always coincide with their consumption plans, and a short sale is a great way to bridge the gap.

And if short selling is a vital tool for bridging the gap between our incomes and consumption plans, it is important that the various media we use for shorting are capable of facilitating this process. When the future costs of covering a given short are difficult to predict, people will shy away from shorting to fund their spending needs, and the benefits of trying to bridge the gap between income and consumption plans will go unexploited. Society is made worse off.

The best media for this purpose—those most capable of providing a predictable price—will become society's generally-accepted media for shorting. So maybe money isn't just a medium of exchange, store of value, and unit of account; it's also a popular short-selling medium. That's why we see stable instruments like Federal Reserve dollars or Bank of Tokyo-Mitsubishi yen deposits or Barclays pound deposits serving as the world's most prolifically-shorted media. Sure, bitcoin and Netflix short sellers certainly exist, but this is a very niche sort of transaction undertaken by highly-trained financial practitioners or fools. They aren't generally-accepted short media.

-----

Bitcoin is a particularly awful medium for short selling because its lack of fundamental value leads to jaw-dropping volatility. Anyone who borrows one bitcoin (which currently trades at $19,000) and then sells it to finance a $19,000 purchase, say a car, could easily end up owing $190,000 two or three months down the road. That'd be a mere 10x price increase in the price of bitcoin, which is just a regular month or two in bitcoinland. A price move of this degree could easily bankrupt the car buyer.

Which in turn could bankrupt the person who lent to them. A bitcoin lender must account for the potentially lethal effect bitcoin's price spikes will have on his or her customer base by incorporating a premium into the interest rate charged to their customers. This means that the interest costs of borrowing and shorting $19,000 worth of bitcoin in order to buy a car will always be more than the costs of borrowing $19,000 worth of Federal Reserve banknotes. 

Because this volatility is inherent to bitcoin, it will always be bad at helping people build vital bridges between incomes and consumption plans. Don't expect it to ever become a generally-accepted medium for short selling.    

-----

What other features make for a good generally-accepted medium for shorting? When it comes time to cover one's shorts by buying back Netflix or bitcoin, there may not be enough of these instruments available, which can lead to a huge spike in their price. These are called short squeezes.

Money is different. The supply of modern money is tiered, with the public at the top, commercial banks in between, and a central bank at the bottom layer. When a spike in the public's demand for money occurs (otherwise known as a bank run), private banks will try to accommodate that spike until they can't, at which point they can turn to the central bank for help. The central bank can in turn manufacture whatever quantity of new money is necessary to meet that demand. So short squeezes will always be prevented by the central bank. That's a feature that neither Netflix nor bitcoin can offer.

-----

If the central banker's job is to maximize people's ability to bridge long gaps between income and spending by ensuring the predictability of the generally-accepted medium for shorting, might there be a better rule for managing things than inflation targeting?

Say that a recession hits and a large part of the population loses their jobs. If the central bank has an inflation target, those who have jobs can continue to easily cover the same quantity of dollars shorted by selling their labour whereas those without jobs will not be able to cover their shorts at all. Aggregating these two groups together, there is a net reduced capacity for short covering. So we might say that the central bank is failing at its job of providing a predictable medium for shorting.

If the central bank were to temporarily boost inflation to 4% when a negative shock occurs, it would be easier for the unemployed to cover their shorts than under a permanent 2% inflation targeting regime. For instance, with inflation at 4% rather than 2% an unemployed person would be able to sell their car or couch at a higher price than otherwise in order to cover a short position. For those on the flip side of the coin—those who make their living lending the medium for shorting—a temporary increase in inflation to 4% means their loans will be less valuable. But at the same time, the increase in the odds that the unemployed will be able to cover their shorts means that creditors needn't fret so much about the hazards of bankrupt customers.

So everyone wins. The converse could work too. When the economy is booming and jobs plentiful, the central bank could reduce inflation to 1% or so, thus making it slightly harder for people to cover their short positions. By using a rule that improves predictability in both regular times, downturns, and booms, the central bank provides a superior shorting medium for bridging gaps between incomes and consumption plans than under a flat inflation targeting regime. I suppose this is an argument for NGDP targeting over inflation targeting?

17 comments:

  1. Great post! I’ve often wondered about these issues.
    One way to prevent short squeezes is for the stock exchange to insist that every short sale on their exchange have a cash equivalent clause. When a squeeze happens, short sellers have the option to deliver the cash equivalent of the share, rather than the share itself.

    As for bank runs, bankers traditionally deal with them by putting a suspension clause on their money. When a run starts, the banker reserves the right to delay payment by 60 days. The wise banker would hold assets that mature in 60 days or less, so assuming he is solvent, he can pay off all his customers in cash.


    The trouble with using the central bank to deal with bank runs is that if the bank in question is insolvent, then a loan from the central bank will not make that bank solvent, and the run will continue until the bank fails.

    ReplyDelete
    Replies
    1. Hi Mike, good points. I recall you mentioning the point about short squeezes in stock markets here:

      http://jpkoning.blogspot.ca/2014/04/short-squeezes-bank-runs-and-liquidity.html

      Delete
  2. Very good post JP. Yep, NGDP targets promote risk-sharing between borrower and lender in the face of aggregate supply-side shocks.

    Deflation in the Great Depression was a short squeeze.

    Some economists talk about money as the standard for deferred payment IIRC, which is sorta similar.

    ReplyDelete
    Replies
    1. Thanks Nick. Yes, when I was writing this post it made me remember money as a standard for deferred payment, the rarely mentioned fourth function of money.

      (Jevons referred to it as the standard of value http://www.econlib.org/library/YPDBooks/Jevons/jvnMME3.html)

      I never had a good intuitive understanding of this function until I wrote this post. Funny how that works. Sometimes you only get to understand something when you re-write it from scratch in terms of your own experiences and ideas.

      Delete
  3. Bernanke's ability to slow down the collapse was due to inside information, large firms were showing him their books, but not us. Thus, Ben to the rescue,but the accumulated risk just got stretched out, and mostly shifted back to the legislature in the form of inflation guarantees in government obligations.

    If lenders and borrowers had equal, operational access to the banks books, then Ben has no inside information, he is not the hero.

    ReplyDelete
    Replies
    1. The two posts are connected, anonymity and special central bank services. We don't have digital anonymous cash,just anonymous paper cash. It is easier to understand the information theoretical version is we assume digital cash, so let's assume it.

      In digital cash, the loan / deposit trade is anonymous, then, first, all parties can examine the distribution of loans to deposits equally without revelation. Second, a local algorithm can respond to observed change as fast as any other party.

      So, the central bank gets no advantage except that it keeps hidden information, it has discretion. We can model that as a three way trade, a subject implied in one of Nick's posts when he referenced Samuelson and the exact determination of rates. Samuelson investigated the three way trade. It is triple entry accounting, loan, deposit, banker fees, basically is the best semantic, or loans, deposit and seigniorage.

      Delete
  4. Shouldn't we count five functions of money then since you argued that 1 and 4 are not the same, JP?

    1 Store of value
    2 Unit of account
    3 Medium of exchange
    4 Medium of savings
    5 Medium of short selling

    ReplyDelete
    Replies
    1. Well I'm not sure about 4.

      If you receive $30,000 and you know that you won't need it for 1-year, would you hold it in the form of banknotes? You already know ahead of time that those banknotes will be worth 2% less due to inflation. Alternatively, you can invest the $30,000 in 1-year government bond and earn 1-2% interest.

      The bond will be more volatile than the banknotes over that 1-year time frame (ie it won't be as good a store of value), but it is a better savings instrument. You can of course save for the long-term by holding banknotes, but they aren't a great long-term savings vehicle.

      Delete
  5. medium of shorting ...

    or object of shorting?

    ('medium of exchange' is symmetric)

    ReplyDelete
  6. It is often asked whether it makes sense to short housing to go long equities, but it is far more common to short income to go long housing as wages are somewhat steady.

    ReplyDelete
  7. Hmm, this is interesting. It is similar to the answer I gave Alex Millar over in the Forum, I think.

    I'll rearrange it a bit to suit my brain.

    According to you, in a nutshell, money is the preferred medium for shorting because inflation is stable.

    Firstly, you have just implicitly acknowledged a causal relationship between the price of consumer goods and the price of labour. Good (you Marxist, you :-)).

    A further implication of your setup, one I'm less convinced of, is that the fact that money and labour tend to have a fairly stable relationship is somehow coincidental, or a function of modern central banking.

    You write:

    The best media for this purpose—those most capable of providing a predictable price—will become society's generally-accepted media for shorting.


    So first (hypothetically speaking) there are lots of media around, some of which are more stable in relation to hours worked and other less so. The best will tend to become society's generally accepted media for shorting (and all other functions associated with money, I suppose?).

    You then go on to qualify that coinincidentality by saying:

    When a spike in the public's demand for money occurs (otherwise known as a bank run), private banks will try to accommodate that spike until they can't, at which point they can turn to the central bank for help. The central bank can in turn manufacture whatever quantity of new money is necessary to meet that demand. So short squeezes will always be prevented by the central bank. That's a feature that neither Netflix nor bitcoin can offer.

    What about times before modern central banks came along? Was money then not a generally accepted media for shorting? Did occasional short squeezes, say in specie, prevent people from promising future labour output? And why would one medium be more stable in relation to labour output than another without the magical features of central banking? Why weren't they all like bitcoin?

    My answer to the last question would be that money actually IS the product of short selling labour (please correct my use of the term short selling if it's muddled, I'm not in finance). The medium called money only exists as such because it captures the promises by its users (debtors) to deliver labour output in future. It is not a feature of money, it is the cause of the value of money in the first place. And that cause, being forward oriented, is self-perpetuating and sticky by definition. The value of labour isn't sticky in ralation to money. Rather, money is today's labour output which tends to be rather sticky in terms of labour's output tomorrow.

    Does that makes sense?

    ReplyDelete
    Replies
    1. "What about times before modern central banks came along? Was money then not a generally accepted media for shorting? Did occasional short squeezes, say in specie, prevent people from promising future labour output? And why would one medium be more stable in relation to labour output than another without the magical features of central banking? Why weren't they all like bitcoin?"

      Great questions. You don't need the magical features of central banking to get a stable medium. The medium that is the most stable in relation to labour output will always be the one that is used to price labour. So if wages were set in gold, then gold would be the stablest medium for short selling. If you were a medieval peasant you'd want to short gold because you know you'll be earning gold in the future-- and because labour is sticky, you have a good idea how much.

      "My answer to the last question would be that money actually IS the product of short selling labour."

      Well you can't sell labour short because you can't borrow it and sell it. Unless slavery were legal. You can sell your own labour by entering into a forward contract with someone, I suppose. Where you agree to provide x hours y months from now. A wage contract might look a bit like that.

      Delete
  8. So if wages were set in gold, then gold would be the stablest medium for short selling.

    And if wages were set in bitcoin? Would that stabilise btc's value vs. labour? And vs. consumer goods? How does a growing labour force square with limited amount of bitcoins?

    You can sell your own labour by entering into a forward contract with someone, I suppose.

    Right, then it was the forward contract I was looking for, thanks.

    ReplyDelete
    Replies
    1. "And if wages were set in bitcoin? Would that stabilise btc's value vs. labour?"

      Yes, if wages were all set in bitcoin terms then it would be a fairly reliable medium for short selling. Because you'd know ahead of time how much you were going to earn, so you could safely short a chunk of it now. Because it is so volatile, though, I don't see it emerging as a unit of account.

      Delete
  9. What about counterparty risk where-in you break the forward contract / promise ? How do I trust you that you will go through with your promise?

    ReplyDelete
  10. Gold prices are more or less stable ... Bitcoins are not ( they are mostly increasing at a faster rate than that of gold ) ... Today's central banks still hold gold reserves as a hedge instead of Bitcoin ... That's you will always find willing buyers for gold (as it has "inherent" utility as an inert non metal corrosive.. jewelry etc ... But not Bitcoin)

    ReplyDelete
  11. In your analysis, you correctly explain that the purpose of "shorting" is advancing future income to the present. Which is completely different from betting on a decline in some asset's price, thus I think it would be better to use a different term, e.g. "borrowing".

    Indeed, borrowing is not a short, since it can be seen as the "real" creation and sale of a security, e.g. a bill of exchange. Unlike shorting, which could, in principle, be illegal, i.e. if securities' fungibility was broken, and the speculator would have to return the same (serial number) security as he has borrowed.

    Then you explain that the best medium for borrowing--to denominate the contract in--is the one that is the most "sticky" in relation to other assets. That is to say, the one that is already used to set prices, which is already a traditional criterion of money.

    ReplyDelete