"Why does liquidity matter so much?
This question has been bugging me for the last few months. I see the financial crisis as largely a liquidity crisis. People only want to hold the most liquid assets, and shun the illiquid. So liquid assets have high prices and low yields; and illiquid assets have low prices and high yields.
But if we think of liquidity in terms of transactions costs, it is hard to see why people would pay so much to avoid holding illiquid assets. The transactions costs don't seem big enough to justify the yield spreads. A 1% extra transactions cost would only justify a 1% yield spread if you expected to hold the asset for one year. So you would need really big differences in transactions costs, or really short expected holding periods, to explain any significant yield spread between liquid and illiquid assets. Neither seems plausible.
Perhaps it's not transactions costs, or not just transactions costs, that determine differences in liquidity. Perhaps it's the fear that my need to sell the asset to raise cash may be correlated with other people's need to sell the asset to raise cash. If our needs to raise cash are correlated, the market price will fall at exactly the wrong time. A 10% fall in the market price if I ever needed to sell the asset would have the same effect on liquidity as a 10% transactions cost.
You can still think of illiquidity as high costs of making a "round trip". You buy the asset, then sell it again, and see what percentage of your money you have lost in making the round trip. Only this time, you consider not just the transactions costs but the probability that other people may be selling the asset at the same time you are, so the market price falls when you sell. An asset whose price is expected to fall 10% if you need to sell it has the same effect on the round trip as a 10% transactions cost.
Even if 1 year and 10 year bonds had exactly the same transactions costs, the former would be more liquid, even if you needed cash today, and could not wait 1 year till the bonds mature. That's because if the people holding them needed cash today, the price of the 10 year bond would need to fall 10 times as much as the 1 year bond to get the same increase in yield, and make them attractive to the people who didn't need cash today. So the difference in liquidity would cause a yield spread in the term structure, even if people expected interest rates to stay the same.
Even if simple bonds (understood by everyone) had exactly the same transactions costs as complicated bonds (understood by a few people), the former would be much more liquid. The complicated bonds would only be held by the people who understood them. If a substantial portion of those people needed to sell at the same time, the price would have to fall a lot before others would be willing to buy them. So complicated bonds are less liquid, and would have a higher expected yield.
Why should my need to sell an asset and raise cash be correlated with other people's need to sell the same asset and raise cash?
In normal times it isn't correlated (or will have a low correlation). Individuals' needs for cash will be idiosyncratic, and so uncorrelated. A big bill comes due. My car breaks down. I lose my job. I retire. If individuals' needs to sell an asset and raise cash are uncorrelated, transactions costs alone determine liquidity. Brokers' and agents' fees, bid-ask spreads, and things like that, are the only things that determine liquidity. And since transactions costs for most assets are fairly small, most assets are fairly liquid.
But in abnormal times, with a greater risk of aggregate fluctuations, individuals' needs to sell an asset and raise cash will be correlated. Many people lose their jobs. Many people face a margin call. Many banks run short of capital reserves. The risk of an asset falling in price just when you need to sell it, because other people (or firms or financial institutions) will also need to sell it, becomes much greater. So previously liquid assets become much less liquid, even if the transactions costs stayed the same.
If people believe the business cycle has been tamed, so aggregate risk falls, previously illiquid assets come more liquid, and the spreads narrow.
If aggregate risk reappears, liquidity spreads become much bigger, for two reasons. First, because any given asset, except those that are most like cash, will become less liquid than it was before. And second, because the average liquidity of assets has fallen, the supply of liquidity has fallen, people will place a greater marginal value on the remaining liquidity.
If my story is correct, the very low yields on short-term treasury bills are not due to a glut of savings and a dearth of investment opportunities (or not solely due to savings and investment). They are low because there is a shortage of liquidity, and people value liquidity more at the margin, and are willing to pay more for it with lower yields. The interest rates which matter for consumption and investment decisions, and for aggregate demand, are not zero. There is an "illiquidity tax" on aggregate demand.
And the policy implication is to reduce that illiquidity tax: use quantitative easing to increase the total supply of liquid assets (money is of course the most liquid), and to make illiquid assets more liquid by buying them when others need to sell.
It's the same story I have told before, of course, but I now think I understand liquidity better. It's not just transactions costs. It's much bigger than transactions costs, and big enough to matter. So I'm now more confident my story makes sense.
Me:
"In a pure liquidity story agents only need to believe the extra money supply will last as long as their demand for holding it,"
"But in abnormal times, with a greater risk of aggregate fluctuations, individuals' needs to sell an asset and raise cash will be correlated."
http://www.newscientist.com/article/mg20127001.200-why-money-messes-with-your-mind.html?full=true
"In his book Nudge, co-authored with legal scholar Cass Sunstein, also at the University of Chicago, Thaler identifies other irrational biases that lead to distortions in our mental accounting. Almost all of us, for example, are "loss averse" - it hurts more to lose £50 than it feels good to win £50. We also value money in relative rather than absolute terms - we consider £10 irrelevant when buying a house but not when paying for a meal. Similarly, finding £100 will give many people more pleasure than having a heating bill cut from £950 to £835, even though this gains them more in real terms.
We also have a well-known bias in favour of a little money now over more money later, which makes saving so difficult. Thaler has suggested - and many companies are now using - a scheme called "Save more later" that puts this bias to work. Employees can commit themselves to putting more money into their retirement savings in future years, rather than doing it now. It seems to work for the same reason that we are lured by offers of "no payments for the first year", but in a more beneficial way"
The Flight to Safety is just that. The escape velocity, if you will, to fear and aversion to risk, might well necessitate a compensating factor of permanence.
Don: I don't think we need loss-aversion to get the story to work (it would merely accentuate it). If you need cash when it rains, you don't want to hold an asset which falls in value when it rains.
Posts
No comments:
Post a Comment