Trading volume, liquidity crises, Carl Menger, and "off-the-run" bonds

An asset which becomes less liquid (the costs of trading it increase) will be traded less frequently. If an asset is traded less frequently, it will become even less liquid.....

An asset which becomes more liquid (the costs of trading it decrease) will be traded more frequently. If an asset is traded more frequently, it will become even more liquid.....

There is positive feedback between liquidity and trading volume. Carl Menger's 1892 theory of the origin of money is one example of this. Starting with barter, one good has slightly lower transactions costs than all the others, so is accepted more frequently in indirect barter, which makes it even more acceptable, which lowers the transactions costs even more....until every good is first exchanged for that good, which is in turn exchanged for others, at which point that good has become money. And because it is the most liquid asset, and people are willing to pay for liquidity in terms of lower yield (and most monies have a negative real yield), the value of money is far greater than any intrinsic value it might have, if any.

The same is true, though to a lesser extent, for all assets. Liquidity begets liquidity; illiquidity begets illiquidity. Money is just the most extreme case. Being the most liquid of all assets, it appears on one side of (almost) all exchanges. And we measure the liquidity of all other assets in terms of how easy or difficult it is to convert them into money.

This positive feedback, if powerful enough, means that liquidity crises can just happen, for no reason at all. There is one equilibrium with a lot of trading, high liquidity, and a high asset price. And there is a second equilibrium with little trading, low liquidity, and a low asset price. And any temporary shock, or even a change in expectations, can flip us from one equilibrium to the other, confirming those expectations.

Houses once sold quickly, so they were liquid, so flippers would buy and sell them, so they were liquid, and therefore worth more. Now houses sell slowly, so they are less liquid, so only people who don't plan to move will own them, so they are less liquid, and therefore worth less. Neither equilibrium is more fundamental than the other. We can't say the first was an irrational bubble and the second a return to reality. Both are self-fulfilling equilibria. But the first equilibrium was better, since houses were more liquid, so really were worth more then. (If a house were perfectly liquid, like money, the equilibrium price/rent ratio would be infinite, like money).

Empirical evidence shows that differences in trading volumes can make differences in liquidity and yields, which make bigger differences in trading volume, and so on. There is no fundamental difference between "on-the-run" and "off-the-run" bonds. They have the same duration, the same safety, the same everything else. "Off-the-run" bonds, by definition, are merely not the latest issue. That lack of fashion sense dooms them; everybody avoids trading them, because everybody else avoids trading them. Trading volumes typically drop by 90% when a bond goes "off-the-run". People who do more trades care more about liquidity, and so they hold "on-the-run" bonds, and so "on-the-run" bonds trade more frequently. People who trade less frequently care less about liquidity, and hold "off-the-run" bonds, and are compensated by a slightly higher yield.

There is one equilibrium, the one we observe, where on-the-run bonds are the more liquid and off-the-run bonds are the less liquid. But we can easily imagine a second equilibrium where it's the other way round. Probably, the reason why it's not the other way round is that when a new issue is introduced to the market, there is a very high initial trading volume by definition. And then it just stays that way.

The on-the-run/off-the-run phenomenon is the dual equilibrium in cross-section. The boom/bust phenomenon is the same dual equilibrium in time-series. Nothing fundamental changes over time, just as there is no fundamental difference between the two bonds. We just cross from one equilibrium to the second.

In normal times the yield differential between on-the-run and off-the-run bonds is just a couple of basis points. But in times of liquidity crises it gets bigger. Sometimes, like recently, it gets much bigger.

If there is a demand for liquidity, and if some assets become less liquid, the liquidity of the remaining assets becomes much more valuable (diminishing marginal utility of liquidity). US Treasury bills are the most liquid asset, after US dollars, and their yields have dropped to around zero. Small differences in liquidity, like between on-the-run and off-the-run bonds, cause bigger yield spreads, because it matters more. Illiquid assets, even if they become no more illiquid, become worth less.

Here's a model to illustrate my point. Just take the model in my previous post, and change the assumptions to make emigration (the need to trade cars) endogenous. Suppose 20 people per year must trade cars (their pay would be much higher if they emigrated). And suppose the remaining 80 people have a choice, and will emigrate (trade cars) each year provided they lose less than $12 on trading their cars, and otherwise won't trade.

There are now two equilibria. In the bad equilibrium, exactly as in my previous post, only 36 people trade cars per year (the 20 who must emigrate, plus the 16 lemon owners), and the market price of a car is $61.86. In the good equilibrium, everybody emigrates and trades cars once a year, so 100 cars are traded, and the price of a car is $90. The market price of cars can now flip between $61.86 and $90, with no change in the fundamentals. Everybody is happier in the good equilibrium.

It is easy to build similar models.  You need a demand curve for trade: a downward-sloping function of the transactions cost, with some people being more motivated and others being less motivated to trade. You need a supply curve for trade; which is also downward-sloping, because the transactions cost decreases with the volume of trade. Those supply and demand curves cross three times in my model (but we ignore the middle equilibrium, because it's unstable).

 

Comments

[Phillip Huggan]

I agree with your thesis/explanation, but housing isn't a good example. Housing experienced a genuine fundamental demand rise as a result of lower interest rates. It's boom bust has more to do with issues like regional unemployment in S.Ontario and USA rules that permit easy exit of out-of-the-money mortgages.

[alexcanuck]

If a house were perfectly liquid, like money, the equilibrium price/rent ratio would be infinite, like money
What? I can rent money all I want. At a price to rent ratio ranging from around 4% if it was a really good mortgage locked in at the best possible time, to 28% on a credit card, or even more from some guy named Luigi. Or a payday loan place.

[Nick Rowe]

Phillip: yes, perhaps I was pushing it a bit with the housing example. My guess is that the decrease in liquidity of houses as an asset will cause some decrease in price, but I don't know how much. I don't think the positive feedback on housing alone could be strong enough to flip houses from one equilibrium to another. But houses as part of the whole spectrum of assets? Who knows.

alexcanuck: 5,000 $20 banknotes pay no interest to the owner. A $100,000 bond pays (say) 4% interest to the owner. A $100,000 condo pays (say) $5,000 annual rent (or 5% interest) to the owner. If I own 5,000 $20 banknotes and want to earn interest, I have to sell them and buy a $100,000 bond, or $100,000 condo. When you lend money you are buying a bond (you get an IOU in return, and a bond is just an IOU). As you go from holding money, to holding bonds, to holding a condo, your interest rate goes up, but your liquidity goes down.

[jp]

I wonder about the difference in price between near futures and distant futures of the same commodity, and if the higher liquidity of near futures has some effect on price differentials between them. Arbitrage definitely keeps the prices of the two in line, though there are probably exceptions to this.

[Nick Rowe]

John: can you run that one by me more slowly? Do distant futures prices tend to be higher or lower than the actual price when the future arrives (on average)? And is this distant future "bias" is bigger than the near future bias? And do near futures contracts tend to have higher trading volume than distant futures contracts? Are these effects "big"?

[jp]

Hi Nick - if you follow this link, you'll see CBOT corn futures. The front contract is the March 09 contract. It has 10 times the volume that the May 09 contract does. Once March expires May will become the front contract and become the most liquid contract. This volume pattern is pretty standard in futures markets.

The difference in price between near and far contracts is governed by carrying charges - interest rates, storage costs etc. Traders engage in various forms of arbitrage to keep these prices in line.

I was just wondering if any sort of liquidity premium could ever exist for the front contract, given the fact that traders have the means to abitrage this way.

[Edited to make the click-through link - SG]

[Nick Rowe]

I never expected the difference in volume would be that big! I don't understand why volume would be bigger in the nearest than in the more distant contract: it might be for some fundamental reason; or it might be just another double equilibrium like in the on-the-run/off-the-run case.

Look, until a couple of weeks ago, I had very little sense of the macroeconomic importance of liquidity. I knew of course that people demanded money (the most liquid of all assets) as a medium of exchange, and that it's not a good idea to buy a house with a swimming pool or any paint other than "speculator beige" if you might plan to move soon, but I thought it was a reasonable working assumption to ignore differences in volume and liquidity in most other financial assets. What I still don't understand is: why liquidity matters so much; why transactions volume is so high (why don't people just buy and hold?); are volume and bid/ask spreads and market depth (the whole curve of bid/ask spreads as you trade larger and larger quantities) and brokers' commissions the whole story?

If liquidity is important, and valuable, then arbitrage will only partially eliminate yield spreads across near and distant futures contracts; just as arbitrage does not (normally) eliminate the yield spread between zero-interest currency and positive interest T bills.