Saturday, December 1, 2012

Shades of a liquidity premium peaking through in stock market prices


In my other life, I analyze the stock market. I always find it interesting when the stock market reveals its often hidden monetary nature. The common assumption is that monetary analysis should be confined to a narrow range of coins, dollar bills, central bank reserves, and bank deposits. But this ignores the fact that all valuable things in an economy have a degree of liquidity, including stocks.

A stock's price can be decomposed into a "fundamental" component and a liquidity component. Fundamental value arises from a stockholder's right to receive any distribution of the assets of a corporation. The liquidity component is the premium on top of fundamental value that arises from the owner's ability to easily sell that stock. Knowing that a stock can be easily sold provides the owner with a degree of comfort that would otherwise be lacking if the stock was less saleable, or not saleable at all. This "comforting service" is built into the stock's price as a liquidity premium.

While in theory we can determine a stock's monetary nature by subtracting its fundamental value from its price, in practice this is almost impossible to do. Computing fundamental value is more art than science. Nevertheless, we can get indirect evidence of the existence of a liquidity premium by comparing different share classes issued by the same company. In Canada, it's common for large family-owned corporations to issue non-voting and voting shares. The family controls the company by holding 51% of the voting shares while the public holds the balance of the voting and all the non-voting shares. Both voting and non-voting shares rank pari passu, meaning they receive the same claim on the company's assets.

Given these properties, we'd expect the market price of voting shares to be equal-to or greater-than the price of non-voting shares. After all, since everything else about the two classes is equal, having a vote can only add to the value of a share.

The chart below shows the voting and non-voting share price of a major publicly-traded Canadian retailer. Below the price is the premium/discount on the non-voting shares. At the bottom are relative volumes traded.


Now it's evident from the chart that our hypothesis doesn't hold. For long periods of time, the non-voting shares of our retailer have traded above the voting shares. What might explain this? One reason is that non-voting shares are far more liquid than the voting shares. Just look at the much higher trading volumes in the non-voting shares. From 2008 to 2010, shareholders valued the liquidity bonus of non-voting shares more highly than the element of corporate control provided by voting shares. In other words, the perceived benefits provided by the liquidity of the non-voting shares caused their liquidity premium to balloon to the extent that non-voting share prices rose above voting share prices.

Why doesn't the non-voting share premium get arbitraged away? Well, the comfort that liquidity provides is a very real service, not a mistaken bit of irrationality that can be ironed away. A highly-liquid share is like a fire extinguisher - even if it's not being used, just having it there makes you feel better. So while buying cheap voting shares and short-selling expensive non-voting shares may seem like a risk-free arbitrage, it isn't. After all, if the market decides to put an even higher value on the liquidity services provided by the non-voting shares, then non-voting's liquidity premium will grow even more and the trade will lose money.

While risk-free arbitrage can't shrink a stock's liquidity premium, there are indirect forces that ensure the premium stays thin. The larger a firm's liquidity premium, the lower its cost of capital. After all, if you need to raise money, having a larger liquidity premium means that you needn't issue as many shares. Because a low cost of capital is a boon, firms will try to replicate the success of competitors who have attracted a large liquidity premium. They may do so by pursuing similar lines of business or marketing their shares to the same group of shareholders. This will have the effect of drawing trading activity away from shares with large liquidity premiums to those without, thereby destroying the underpinnings of that large premium.

All of this plays into the ongoing bitcoin discussion between Mike Sproul and I. While Bitcoin might have some negligible fundamental, or non-monetary, value due to its value as a curio, its liquidity premium is surely huge. There is no way to arbitrage this premium away directly, but over time competitors will peck away at it, causing bitcoin's price to deteriorate back to its fundamental value, which I'd guess is <$1.

13 comments:

  1. It looks to me like the premium (the yellow line) averages out to just about zero. I don't say this with much confidence, but I wouldn't rule out the possibility that all that graph shows is noise.

    Also, a bit of confusion: Wouldn't trading costs be exactly the same for voting/non-voting shares? So isn't the one exactly as liquid as the other?

    But on to the main point: Bitcoin could have curio value and it could have liquidity value. A short seller who borrows a bitcoin and sells it (say for $11) will issue an IOU promising the delivery of one bitcoin, or something of equal value. (The IOU is a hypothecated bitcoin.) This IOU could, in principle, be used to buy things in direct competition with actual bitcoins. This competition would drive down the price of bitcoins (say, from $11 to $7), and then the short seller could cover his short by handing $7 to the holder of his IOU, earning a $4 arbitrage profit. This profit exists as long as the liquidity premium is big enough to cover the transaction costs of the short sale.

    That short-selling argument leads me to think that most of bitcoin's value is curio value, like baseball cards and rare stamps. That kind of value would not be competed away by short-selling.

    So far, your argument has been that short-selling of bitcoins is hard, as evidenced by bitcoin interest rates in the range of 25-200%. Fair enough. Short selling bitcoins is hard, and the interest rates are crazy. That limits the effectiveness of short selling, but it is still there. BTW: Are there forward/spot markets in bitcoin? That would make interest parity arbitrage viable.

    But now, bitcoins aside, we get to things like gold, silver, and paper dollars. We can no longer make that claim that short selling is hard and markets are crazy. In these markets, a liquidity premium seems just about impossible. Suppose, for example, that the monetarists are right and the US paper dollar is a pure fiat money whose whole value is a liquidity premium. Everyone who borrows dollars and sells them is a short seller of dollars. These short sellers issue IOU's promising delivery of 1 paper dollar or something of equal value. Those IOU's take the form of checking account dollars, credit card dollars, etc. They compete with paper dollars and drive their value down, say from 1 oz. to .6 oz. Then the short seller covers his short by delivering .6 oz. to the holder of his IOU, earning a profit of .4 oz. That profit opportunity exists as long as the monetary premium is above zero. I conclude that there is no monetary premium.

    I also note that the Fed does in fact hold assets against the paper dollars it issues, does recognize those dollars as its liability, and issues new dollars only in exchange for bonds worth $1. Not only does the logic of that short selling argument tell us that the dollar is backed, the Fed's own balance sheet also tells us that it's backed.

    ReplyDelete
    Replies
    1. "It looks to me like the premium (the yellow line) averages out to just about zero. I don't say this with much confidence, but I wouldn't rule out the possibility that all that graph shows is noise."

      The stock spent almost 2 years trading at a $2 premium (at one point that equated to 20%!). So no, I don't think you can attribute it to noise.

      " Wouldn't trading costs be exactly the same for voting/non-voting shares? So isn't the one exactly as liquid as the other?"

      You'd pay a straight $35 commission (say) to get out of either. But if you put the non-voting stock up for sale at the offer price, odds are you'll be bought out in a few minutes. It could take you hours, days to get bought out of the non-voting.

      More later.

      Delete
    2. "A short seller who borrows a bitcoin and sells it (say for $11) will issue an IOU promising the delivery of one bitcoin, or something of equal value. (The IOU is a hypothecated bitcoin.) This IOU could, in principle, be used to buy things in direct competition with actual bitcoins. This competition would drive down the price of bitcoins (say, from $11 to $7), and then the short seller could cover his short by handing $7 to the holder of his IOU, earning a $4 arbitrage profit. This profit exists as long as the liquidity premium is big enough to cover the transaction costs of the short sale."

      I think your key word is "in principle".

      You gotta reread my old post that talks about the two things that bitcoin does. Bitcoins aren't just bits that have randomly lucked into becoming valuable. Bitcoin is also a highly evolved (and very impressive) decentralized record-keeping mechanism that allows for online "cash" payments.

      You can't just issue your IOUs and expect them to eat into bitcoin's liquidity premium without also replicating Bitcoin's record-keeping mechanism. Without it, your IOUs will not be able to do what Bitcoin does. My guess is that implementing a competing IOUcoiun system would be pretty tough to do. You seem to be assuming that it has already been done, and therefore the bitcoin liquidity premium has been arbitraged away, but this would be assuming too much.

      "But now, bitcoins aside, we get to things like gold, silver, and paper dollars.... I conclude that there is no monetary premium."

      I humbly defer to the example I used above. The non-voting shows a monetary premium.

      Delete
  2. Mike,

    you concentrate on "value" and ignore the transaction costs of trade. And the ultimate manifestation of an attempt to reduce transaction costs is money. Your model would only work if you ignored transaction costs, i.e. what makes money into money, in other words is from practical point of view useless. You ignore the whole payment processing industry, whose very existence proves your model wrong.

    An IOU denominated in Bitcoin cannot compete with Bitcoin on transaction costs. If it did, then Bitcoin would not exists, as Bitcoin would not be able to compete with fiat-denominated IOUs and thus would never gain sufficient liquidity for the network effect to manifest itself.

    There used to be a platform that allowed short selling of Bitcoin that had sufficient market share, but it is now in liquidation due to stupid management. As its market share grew, it appears to have increased liquidity of the bid/ask orders, and stabilised the price (others have claimed that during this time of several months, the price of BTC/USD was more stable than the price of silver/USD, but I did not verify this myself). Those platforms that exist now do not have a sufficient market share to have any real effect (in my opinion).

    ReplyDelete
    Replies
    1. I was just reading about this poor sap who borrowed 10,000 bitcoin for five years back in March 2012 at bitcoin=$5. He hedged on bitcoinica but it went under, so he's had full exposure on the runup to $12. Ouch.

      https://bitcointalk.org/index.php?topic=73599.0

      Delete
  3. JP:
    "But if you put the non-voting stock up for sale at the offer price, odds are you'll be bought out in a few minutes. It could take you hours, days to get bought out of the non-voting."

    But that doesn't make sense. Rational buyers won't be any less eager to buy the voting share, unless the voting share includes some obligation that the non-voting share doesn't, or unless there is some perversity of the stock that we don't know about.

    "I think your key word is "in principle". "

    Naturally, when short selling is hard, there will be less of a tendency for the money premium to be arbitraged away. But in liquid markets like for gold, silver, and paper dollars, short selling is easy and any premium will be competed away. The market for bitcoins looks like a case where short selling is hard.

    "My guess is that implementing a competing IOUcoiun system would be pretty tough to do."

    If the folks at bitcoin did it, surely other folks can take the even easier step of setting up a system where bitcoin IOU's can trade.

    "The non-voting shows a monetary premium."

    Too much about that premium doesn't make sense, or could be explained by perversities in the particular market.

    When I looked back at your last bitcoin post, you said bitcoin is doomed due to lack of assets. Did you change your mind about that?


    ReplyDelete
    Replies
    1. "But that doesn't make sense. Rational buyers won't be any less eager to buy the voting share, unless the voting share includes some obligation that the non-voting share doesn't, or unless there is some perversity of the stock that we don't know about."

      I used to "cover" this stock in a professional manner, so I know it fairly well. Apart from their relative liquidities, there are no perversities.

      Professional portfolio managers have good reason to avoid buying the illiquid voting shares because they fear that when the time comes to sell, there will be no buyers. On the other hand, when the liquid nonvoting stock needs to be sold, the buyers are likely to be more forthcoming. This service gets built into the price of the nonvoting as a liquidity premium. It makes a lot of sense.

      "When I looked back at your last bitcoin post, you said bitcoin is doomed due to lack of assets. Did you change your mind about that?"

      No, I don't think I've changed. In the long term, bitcoin I think bitcoin doomed because it has no non-monetary properties and therefore no lower bound to how far its price can fall.

      Delete
    2. "Professional portfolio managers have good reason to avoid buying the illiquid voting shares because they fear that when the time comes to sell, there will be no buyers."

      So buyers won't buy because they think that other buyers won't buy. The question that is still hanging fire is, why won't the other buyers buy?

      Delete
    3. It's a chicken and egg problem. Portfolio managers won't buy if there hasn't been enough liquidity in the past and if they find no evidence that others will increase their activity in the stock in the future. It's like Keynes's beauty contest.

      Delete
    4. It would make more sense if the voting shares had some obligation or inconvenience attached to them. For example, voting share owners might receive blizzards of paperwork from the corporation, and the mailing charge gets taken from their account.

      Delete
  4. Peter:

    When you refer to my model, I assume you mean my belief that the value of money is equal to the value of the assets backing it, just as the value of a bond is equal to the value of the assets backing it. Transaction costs certainly add complications, but we still find that the value of bonds is determined by the value of the assets backing them. I say the same is true of bonds that happen to be used as money.


    "An IOU denominated in Bitcoin cannot compete with Bitcoin on transaction costs."

    Each money will have its niche, just as we sometimes use coins, sometimes paper, checks, or credit cards. Any money that does have a money premium can have that premium arbitraged away, to the profit of the people who issued the IOU's that become used as rival moneys.

    ReplyDelete
    Replies
    1. Mike,

      Bitcoin as specie is form-invariant, a first historical example of such a monetary system. It can exist in almost any form (including coin and paper) without counterparty risk. This is why it is difficult for Bitcoin-denominated IOUs to compete on transaction costs with Bitcoin.

      If there were no transaction cost, there would be no money in the first place, and no liquidity. And they are also the reason why the liquidity premium cannot be arbitraged away, because different liquid instruments differ in transaction costs. The transaction costs add to the costs of doing arbitrage, and paradoxically, arbitrate increases liquidity, so that's why it won't work the way you propose. Because transaction costs as such exist, there is a steady demand for liquidity in addition to consumption / production demand for goods. You cannot arbitrage it away, only if you eliminated all transaction costs, and then money would cease to exist.

      Delete
  5. Today more accurate intraday tip
    BANK NIFTY TREND- BEARISH
    RES1:10800 RES2:10970
    SUPP1:10480 SUPP2:10290
    STRATEGY- SELL ON HIGH
    free stock tips on mobile

    ReplyDelete