Liquidity and aggregate demand

Money is perfectly liquid. Other assets are not as liquid as money, but some are more liquid than others. One of the main features of the financial crisis is that some assets became less liquid than they had previously been. I want to look at the channels through which a fall in the liquidity of those assets could reduce aggregate demand.

We can define "liquidity" (or illiquidity) as the cost of a round-trip. Start with money, use it to buy the asset, then sell the asset again for money. How much have you lost in the process? Money by definition is the medium of exchange, and perfectly liquid. All other assets are less liquid. Short-term government bonds are the most liquid, next to money. And commercial paper (private bonds and stocks) are less liquid. Real assets are even less liquid.

Some days I don't want to buy anything. Other days I want to buy a small amount of stuff. Very rarely, I want to buy a large amount of stuff. We could say my preferences for consumption expenditure are stochastic. They look something like U = d.U(C), where d is zero some days, positive other days, and sometimes, rarely, gets very large. Perhaps I can predict how the parameter d will fluctuate over time; perhaps I can't.

We could probably think of firms in the same way. Some days they don't want to buy anything. Other days they want to buy a small amount of stuff. Very rarely, they want to buy a large amount of stuff.

Now think of an inventory-theoretic demand for various types of assets. Just like the Baumol-Tobin inventory-theoretic demand for money, but with more assets than just money and bonds. I will divide my portfolio between many assets: from the most liquid to the least liquid. On average, I will hold a small amount of perfectly liquid money for the frequent small expenditures; some less liquid government bonds for the less frequent medium-sized expenditures; and some even less liquid commercial paper for the rare large expenditures. [Update: even the consumer durables I hold could be sold to finance spending in an emergency.]

Firms may do the same.

In equilibrium, the yields on various assets across the liquidity spectrum will vary inversely with their liquidity, and the velocities of circulation (turnover rates) will vary inversely with liquidity. The most liquid assets will have a low yield, and high velocity. The least liquid will have a high yield, and low turnover.

Now suppose there is an exogenous shock: some assets (commercial paper) become less liquid than they were before. The shock might be the revelation that some of this class of assets were lemons. Knowing that some assets are lemons, people will want to take more effort to get information about an asset before buying it, or else suffer a greater risk that the asset is a lemon. Getting that information (or risking not getting that information) is a transactions cost, and it makes the asset less liquid.

The decline in the liquidity of commercial paper will reduce its value (increase its equilibrium yield) over and above any direct effect from some assets being lemons. In other words, even if it were revealed that some assets were worse than before and others better than before, so that the average fundamental value were the same, the fact that buyers cannot tell which is which, and it is costly to find out, will reduce the value of the whole class of assets. Because people value liquidity.

The initial decline in liquidity will be self-reinforcing. (It has a multiplier effect). With transactions costs higher, people will be less willing to buy and sell commercial paper. The velocity of circulation will fall. That makes the market in commercial paper shallower, with less volume of transactions, and a large quick sale will cause a bigger drop in price. That is because buyers expect, rationally, that a greater percentage of commercial paper offered for sale on the market will be lemons, compared to the percentage in the total stock.

So an initial small decline in liquidity will be magnified into a larger decline in liquidity. It is not impossible that the positive feedback parameter will be greater than one, and the market will freeze up totally, with all paper being held to maturity.

[Update: bank failures will also play a role in making assets less liquid. Banks buy illiquid assets as raw materials, convert them into more liquid liabilities, and make their revenues off the liquidity spread. When the reduction in asset prices reduces banks' capital, this reduces banks' ability to convert less liquid into more liquid assets.}

How will this decline in liquidity affect aggregate demand?

The first effect is a wealth effect. With higher yields on the now less liquid commercial paper, the price of commercial paper has fallen. If agents have finite lives (as in an overlapping generations model), they will reduce their demand for consumption due to a decline in the value of their assets.

The second effect is a substitution effect away from current consumption towards future consumption. If a household needs to sell commercial paper to finance a large purchase, like a new car, the higher yield on commercial paper will increase the incentive to postpone consumption, and save. Likewise, if a firm needs to sell commercial paper to finance new investment, the higher yield will increase the incentive to postpone investment.

Both the wealth effect on consumption, and the substitution effects on consumption and investment, mean that higher yields on the less liquid commercial paper cause aggregate demand to fall.

[Update: and if it were real assets, like houses, that became less liquid, that would directly reduce the prices of real assets, and reduce the demand for investment in newly-produced real assets.]

Now, the decline in aggregate demand will cause a response from the central bank. The central bank will want to lower interest rates to offset that decline, and prevent a recession. But the interest rate on which the central bank normally operates is the interest rate on assets which are nearly as liquid as money, like short-term treasury bills, or overnight reserves of the banking system.

Reducing interest rates on those assets will help, but it might not help enough, even if those interest rates are forced down to zero.

The wealth effect of lower interest rates on very short-term assets is very small, approaching zero in the limit as the term to maturity approaches zero. (The price of an infinitely-lived bond will double if the interest rate halves; the price of a one-year bond will rise by only 5% if the interest rate drops from 5% all the way to 0%).

The substitution effect of lower interest rates on very liquid assets will help increase consumption and investment, but only for those who would finance the marginal expenditure through selling those very liquid assets. That is a small subset of total expenditure.

We can think of "the" interest rate for the IS curve as a weighted average of all interest rates at which the marginal consumption or investment expenditure will be financed. Weighted first by shares in total expenditure; weighted second by the interest elasticity of that expenditure. That "average" interest rate is a very different interest rate from the interest rate on the most liquid assets. If the liquidity spread gets wider, the average interest rate may rise even if the interest rate on the most liquid assets (next to money) falls to zero. And then aggregate demand will fall. We enter a recession, and expected deflation, which raises the average real interest rate further still.

And we can think of "the" interest rate on the LM curve as a weighted average of all interest rate spreads between all assets and zero-interest money. Weighted first by the shares of assets in the portfolio; and second by the interest elasticity of demand for that asset in the portfolio. The fact that some of the total stock of all assets are now less liquid than before means that the average liquidity of all assets is less than it was before. The LM curve shifted left. It is not horizontal at zero. Only a small fraction of the weighted average LM curve is horizontal at zero.

Was it a fall in the Neo-Wicksellian "natural rate of interest" below zero which caused the "liquidity trap"? Depends how you define both terms. But it needn't have been a fall in time-preference, or in the return on investment, that caused the problem. It could have been a fall in liquidity.

And the solution is to increase liquidity. Either take the lemons off the market, or clearly brand them "lemons", or else increase the percentage of money, or money plus liquid bonds, in the portfolio.

Thanks to commenters, Adam and bob especially, and to a couple of posts by Robert Waldmann. They may not (probably won't) agree with what I say.

Comments

[bob]

Nick: au contraire. I think it is a great post

x-posted/paraphrased from another comment of mine:

One year ago, on another blog, under another pseudonym, I was trying to make the exact point that Patrick makes at 8:08 in yesterday's post and Nick makes in this post: if the credit system (all of the credit channels) is clogged with lemons ("AAA" securities that may or may not be junk) the credit transmission mechanism stops working. When you cut rates during this situation (before deflation starts) it ignites the threat of inflation causing people to rush into inflation hedges (the oil bubble) but it doesn't actually stimulate the rest of the economy, and hastens the onset of the liquidity trap. I was frothing-at-the-mouth, pulling my hair out trying to get this point across so that maybe someone would tell Bernanke "DON'T DO IT!!!!". It's too late now, but I am so so glad to see Nick and Patrick realizing this.

[Greg Ransom]

As liquidity changes, and the economy slams into the reality of the sustainable time structure of the economy, many assets change from juicy oranges to sour lemons.

This "oranges to lemons" story is both the ultimate and the proximate underlying cause of the boom-bust.

What you are discussing is the "secondary depression" -- the deflationary downward spiral which comes after the onset of the bust.

[bob]

Essentially what the TARP is, is a system whereby the government temporarily accepts counterfeit bills (AAAs that are junk) and gives out treasuries that are good. It's totally counter-productive because the secret is already out about the lemons, so the TARP can never really fool anyone, especially not sophisticated bond traders. Everyone knows the losses are there, and someone has to take them. If the government buys the junk, they take the loss. If the government is just accepting them as collateral, the banks still have to take the loss at some point. All of their counterparties know this, so their CDS blows out, they can't borrow at a decent rate except from the government, and they become zombified. as long as they are being supported by the government pretending that their counterfeits are good (temporarily) they continue on as zombies that can't extend credit properly.

[reason]

Greg Ransom
could you repost in English. I don't understand a word of what you are trying to say. I know you have an Austrian accent, and I never could understand that.

[bob]

on more point: the TED spread can't come down until the lemons are identified. The banks don't want to lend to each other if they know that everyone is holding a ton of lemons, but no one knows exactly where the bodies are because they are buried deep in the Level 3 assets, hidden from counter-parties.

The real nasty part of this problem is that there are so many lemons that the entire banking community is insolvent in the aggregate. This is why temporary nationalization is really the only semi-good option left. The banks can't absorb all the losses from identifiying and repricing lemons. I think the best way to handle this is to force all the initial losses on shareholders and bonholders (and the fixed income desk who bought all the lemons). The government will still have to take some losses, but at least they are minimized, and interference with the market punishment mechanisms (moral hazard) is also minimized.

[Andrew F]

Why not have the government buy illiquid/riskier assets and sell as many bonds as people will buy? Buy stock and bond indexes, and let the market allocate it through arbitrage. If the problem is credit markets drying up, flood them in a way that is transparent and avoids picking winners.

[Nick Rowe]

I have since read Patrick's yesterday's 8.08 comment, and have updated this post to bring banks in explicitly (they were sort of implicit before). Thanks Patrick!

Bob and Andrew F: there seem to be two policies that could work. Some sort of taking lemons off the market, or labeling them as lemons; or just (as Andrew suggests) a version of quantitative/qualitative easing where the central bank does an OMO for illiquid assets, mainly to increase the total quantity of more liquid (money+bonds) assets in the total stock, to satisfy the total demand for liquidity. And I agree: buying the index(es) would be the way to implement the latter.

[Nick Rowe]

Greg: so when I say "Now suppose there is an exogenous shock: some assets (commercial paper) become less liquid than they were before. The shock might be the revelation that some of this class of assets were lemons.", are you saying that the Austrian story about the unsustainability of the time-structure of production comes in before, and replaces my "suppose there is an exogenous shock"?

[Greg Ransom]

Nick writes:

"Greg: so when I say "Now suppose there is an exogenous shock: some assets (commercial paper) become less liquid than they were before. The shock might be the revelation that some of this class of assets were lemons.", are you saying that the Austrian story about the unsustainability of the time-structure of production comes in before, and replaces my "suppose there is an exogenous shock"?"

This can go two ways. 1) "the revelation that some of this class of assets were lemons" could come when a chain of investments in a longer-time consuming structure of production reveals itself to be non-completable -- e.g. the misallocation of resources is revealed when markets tighten, prices rise, profits are squeezed, and the constant additions of new money and credit run out. I.e. the housing / finance / real estate / mortage booms comes to a peak, then craters.

It could also come because the media begins investigating the unsustainable securitization and leveraging of tens of thousands of "liar loans".

In the current case, we've got both. Part of what Austrians want people to think about is the "how much" question. How much of the "liar loan" and housing bubble phenomena could have occurred in the absence of over-optimistic expansion of credit by the finance industry, in the absence of the Feds below natural rate policy, in the absence of a credit / money / finance bubble?

How much is the economic significance of a massive distortion in the time structure of relative prices across all production goods? Compare that to the number of "liar loans" written during the last 6-7 years. Which is economically "bigger"?

[reason]

Greg
I know the Austrians don't like math or empiricism, but if you are asking an empirical question, surely you should provide an empirical answer.

[reason]

And for heavens sake Greg, surely it is obvious that the housing problems didn't come from "markets tighten, prices rise, profits are squeezed" on the supply side but because without unaffordable credit the demand wasn't there. Housing is simple, building a house takes only about 6 months from start to finish. It is a simple gold rush problem. Somebody sees profits being made and everybody follows. Nobody realises that total supply exceeeds total demand, and there is no control via the markets because there is JUST ONE credit market for all industries. This was clarified by Kaldor in the thirties, I don't know why this story hasn't been abandoned.

[Scott Sumner]

Good post. I am not used to thinking about liquidity, so I have a couple of questions:
1. Is the problem you describe something likely to occur in most recessions, or just those associated with extreme financial turmoil?
2. When you say the interest rates on risky assets are relatively high, is that the rate assuming no default, or the expected rate of return?
3. Is there data on how liquidity has recently changed, such as bid/ask spreads in junk bond markets?
4. Regarding the policy of adding more liquidity, suppose you have 30-year T-bonds yielding 3%, but the market is highly liquid in terms of bid/ask spread. Does trading base money for these bonds help, or is the expected one year return on 30 year T-bonds effectively zero (as predicted by the expectations hypothesis?) There must be some futures market data to answer this question. If it is zero, then the proposed unconventional OMOs to buy long T-bonds may not help, rather they need to buy riskier assets.

[Nick Rowe]

Thanks Scott.

By the way, Scott Sumner has an excellent blog here: http://blogsandwikis.bentley.edu/themoneyillusion/ where he posts especially on monetary policy in depressions.

By the way^2, my computer lost about 2 months' worth of emails, and I have lost the email Stephen sent me on how to do links in comments. Can anyone remind me how it's done?

Now, to try to answer your questions:

Preamble: I wasn't used to thinking much about liquidity either. Money was perfectly liquid, other assets were less liquid, and that was all part of micro-foundations for money, and maybe interesting for finance-theorists, but we could ignore liquidity for macro. The central bank sets the rate of interest (sure, there are some liquidity and/or risk spreads somewhere in the background, but treat them as exogenous), and off we go into the Neo-Wicksellian model. Now (a couple of months ago) I realise(d) that was hopelessly inadequate. But I'm still trying to get my head around the concept, and its implications for macro/money.

1. I don't know. My guess (little better than that) is that liquidity is a problem in all recessions, but much more so in financial crises.

2. My guess is both (i.e. my guess is that risk-aversion has increased). But I'm not sure it really matters. Assume risk neutrality of lenders. If the demand for loans (for investment or consumption) depends on the rate of interest, and the supply of loans from lenders depends on the expected rate of return, then an increase in risk inserts a wedge between borrowing and lending rates, much like a tax.

But, there's the risk premium, and the (il)liquidity premium, and the two are different (see next answer).

3. There is one bit of very hard data available on the distinction between the risk premium and the liquidity premium. This is the yield-spread between "on-the-run" and "off-the-run" bonds. (I had never heard of the concepts until a couple of months back either). The two bonds are absolutely identical, in every (fundamental) way. Same issuer, same term, same face rate of interest. But the first is the new issue, and the second is the old issue (they might as well be painted different colours). By any fundamental logic, they ought to have the same yield in the market. And yet they don't. Off-the-run bonds have a higher yield, and that spread seems to increase in a financial crisis. Why? because trading seems to concentrate in the on-the-runs, so they have a deeper market, and are more liquid. Pure double equilibria. Pure application of Carl Menger on the origins of money. Here's my earlier post: http://worthwhile.typepad.com/worthwhile_canadian_initi/2008/12/trading-volume-and-financial-crises.html
But I don't have a source of continuous data. Nor data on bid/ask spreads, though I have personally observed widening bid-ask spreads on some stocks I watch. I am useless at applied stuff, like getting data.
And I'm not even sure that bid/ask spreads capture liquidity, despite what I wrote in the post above. There were no bid/ask spreads in the market for used cars I modelled (follow the links inside my previous linked Menger post), but it was clearly a liquidity problem.

4. I don't know. That's the sort of question I'm still trying to think through. Even if two goods have the same price, would changing the relative supplies of those two goods affect the price of some third good?

So, as you can see, this is all very much a work-in-progress. Mainly, I'm working backwards: suppose there were a reduction in the liquidity (not just an increase in the risk) of some assets. What would be the predictions of a model where people valued liquidity? Do these predictions seem to fit the facts? What are the (monetary) policy implications? But I think I'm making some progress by working that way. The predictions do seem to fit the facts, at least crudely.

[Nick Rowe]

Oooh! The links worked!

[jp]

Here are some papers that talk about bond spreads. They might give you an idea how to find such data:

http://ideas.repec.org/p/pur/prukra/1130.html
http://www.newyorkfed.org/research/staff_reports/sr73.html
http://www.cfapubs.org/doi/abs/10.2469/faj.v52.n1.1966
http://papers.ssrn.com/sol3/papers.cfm?abstract_id=936650

The last two seem to be the most relevant.

What you are talking about is the "means of exchange" function that assets have, since things like bonds and stocks aren't just stores of value. At some point one wants to trade them for something else.

[jp]

Brad de Long also talks about the effect here:

http://seekingalpha.com/article/72989-the-on-the-run-premium-on-treasury-securities

and provides a link to another paper.

[Nick Rowe]

Thanks jp! Good finds.

[bob]

whoops, in my second post there, I mean TALF, not the TARP. So many acronyms to remember these days....

[bob]

the Krug-man weighs in:

http://krugman.blogs.nytimes.com/2009/03/11/not-so-easing-wonkish/

[Nick Rowe]

bob: yes, I just noticed that. I haven't found a link to the original. The discussion in the Economist Free Exchange mentions various caveats http://www.economist.com/blogs/freeexchange/2009/03/red_shift.cfm

But my main reaction is "what does this mean?". If we compare the cost to the consolidated government/Fed's balance sheet of fiscal vs. monetary policy, we have to think about the assets the government gets in return, and whether those assets would go up or down in value if the policy is successful. If they go up in value, the policy has a negative cost. (From this perspective, the Bank of England policy of buying long gilts (bonds) seems unwise, because if the policy works, and expected future nominal interest rates rise, with higher expected inflation and growth, the BoE will make a loss.)

Dunno. I still need to get my head clear on the channels through which the policy might work, before getting any sense of reasonable estimates of the multipliers.

[Declan]

"Start with money, use it to buy the asset, then sell the asset again for money. How much have you lost in the process?"

If you'll forgive a dumb question, as I know little on the topic of liquidity, something that's always puzzled me is, where does the lost money go - is it what gets absorbed by middlemen (e.g. real estate agent commissions in real estate transactions)?

I always figured that a large bid/ask spread doesn't really have a net cost as one person's loss is another's gain.

[bob]

yes I'd like to see the actual paper as well.

I think there might be some sort of rational expectations thing going on. If bond traders know that the Fed has bought x amount of bonds, and is going to sell all of them at some point in the future (because they don't routinely hold that asset class), maybe traders discount the remaining bonds in the market to reflect this supply being out there? I'm not a big rational expectations fan, and I don't trade bonds, but this seems plausible to me. It would explain why Goldman's measurements found so little response in the bond market, and why the required balance-sheet expansion would be so high.

[Greg Ransom]

'reason' has Hayek -- and me -- wrong. Hayek and I LOVE math and we LOVE empiricism -- we HATE pseudo-science and we HATE the misuse of math.

The CORE causal / empirical element in economics is entrepreneurial learning in the context of changing relative prices and local heterogeneous resource conditions.

The core use of math in economics is to give us a picture of a pattern that doesn't in fact exist in the world, but with gives us insight into the structured order of the market. E.g. it give us insight WITHOUT providing us with any causal / empirical explanation.

There have been books written on the misuse of mathematics by "mainstream" economists. May I suggest you read one?

"Greg
I know the Austrians don't like math or empiricism, but if you are asking an empirical question, surely you should provide an empirical answer."

[Greg Ransom]

'reason' willfully misreads me and his remarks have NOTHING to do with what I said. Is 'reason' Brad DeLong?

"And for heavens sake Greg, surely it is obvious that the housing problems didn't come from "markets tighten, prices rise, profits are squeezed" on the supply side but because without unaffordable credit the demand wasn't there. Housing is simple, building a house takes only about 6 months from start to finish. It is a simple gold rush problem. Somebody sees profits being made and everybody follows. Nobody realises that total supply exceeeds total demand, and there is no control via the markets because there is JUST ONE credit market for all industries. This was clarified by Kaldor in the thirties, I don't know why this story hasn't been abandoned."

[Greg Ransom]

This is just another way to give a specifics empty explication of the Hayekian business cycle described in Hayek's _Monetary Theory and the Trade Cycle_: "It is a simple gold rush problem. Somebody sees profits being made and everybody follows."

[Nick Rowe]

Greg (@10.21, way back): Leaving aside for simplicity the "Liar loan etc. issues (which I agree were probably part of the problem, but not specifically Austrian, though not incompatible with the Austrian perspective).

A New Keynesian would say: "If the Fed sets the interest rate below the natural rate, demand for investment + consumption will exceed LRAS, and in the short run actual investment + consumption (both will increase) will exceed LRAS, but in the long-run this is unsustainable because it will cause ever-accelerating inflation, and the Fed will eventually have to raise interest rates above the natural rate, to bring down inflation. And it's the rise in the interest rate above the natural rate that causes the recession."

And the Austrian would say: "If the Fed sets the interest rate below the natural rate, this won't merely cause everything the New Keynesian (above) says it will cause (correct?) but will also cause a lengthening in the time-structure of investment, and when they try to complete the final stages of that investment, they find there's not enough savings(?), and this pushes up the natural rate temporarily higher (?), so there's a greater gap between natural and actual rate(?)...?"

Help me out please.

[kevin quinn]

Hi Nick. I followed the link to "lemons" and found what I think is a problem, but comments are closed, so sorry for being off-topic.

In the appendix the expected capital loss should be added, not subtracted, on the LHS of the price equations for the good and bad cars. In the equilibrium, the good car's price satisfies the no-arbitrage condition, since the foregone interest (7.458) plus the expected capital loss (2.54) = $10, the cost of the bus. But look at the lemon: The owner of the lemon incurs $5 in maintenance costs and $5.169 in foregone interest. From this we need to subtract the certain capital gain of $10.17, leaving the cost of owning the lemon at 0! Am I missing something? When I re-do it, I get prices which imply an unchanged fundamental value of 90.

[Nick Rowe]

Kevin: Curses! I think you may be right. (Oh well, at least I did say "Don't trust me" (on the arithmetic)!

I definitely have a typo in the appendix. It should read:

"Let G be the value of a known-good car

Let B be the value of a known-bad car

Let A be the value of an average car in the country

Let S be the market price of cars

A = 0.8G + 0.2B

S = (16G + 20B)/36

10%G + (not minus) 20%(G-S) = $10 (comparing cost of owning good car to annual cost of bus)

10%B + (S-B) + $5 = $10 (comparing cost of owning bad car, including $5 maintenance, to annual cost of bus)"

I have not re-solved for the solution, but I think you are probably right (I would trust your arithmetic over my own). So you get $90 for A? What do you get for G and B? Is S still $72.22?

If so, that says that the lemons problem would not change fundamental values, only "mark to market" values. I would need to make emigration endogenous to get lemons to reduce fundamental values (if you buy a good car, and can't sell it for its fundamental value, you forgo otherwise profitable opportunities for emigrating).

And kudos for actually reading through and trying to understand my model.

[kevin quinn]

Nick: I got G=91.54, B = 83.92, S = 87.32 and A = 90. I'm not sure what gives!

[Nick Rowe]

Kevin: I checked your arithmetic. Yours seems right. Mine was wrong.

This is what gives: my arithmetic was wrong (which doesn't bother me much, because I always knew my arithmetic was bad), and my theorists' intuition was wrong (which bothers me more). It just shows you need a math model to check the intuition, as well as intuition to check the math model.

If you buy a known-good car, you sell it and suffer a capital loss only if you emigrate, which means with 20% probability. I got that bit right. But if you buy a known-bad car, you sell it next year and make a capital gain whether or not you emigrate, which means with 100% probability. It's that last bit that I forgot when I was thinking it through in my head. And then my arithmetic slip ( a - instead of a +) unfortunately made my faulty arithmetic go in the same direction as my faulty intuition.

Oh well, stuff happens.

Here's the big lesson (apart from human fallibility and the need for someone else to always check your work): the cars in my model are illiquid (the market value is less than the average fundamental value), but that illiquidity does not cause any real misallocation of resources, only a redistribution of wealth. If I made emigration endogenous, then it would cause a misallocation of resources and real aggregate costs, and would reduce the real average fundamental value of the assets.

For example, people get random opportunities to emigrate, with an increase in income if they do. Owners of known good cars will be unwilling to emigrate, because they will suffer a capital loss if they do (it would be like making a gift of (G-S) to whomever bought your car -- an externality). So owning a good car means you forgo profitable opportunities. So you would be less willing to buy an car, and would take the bus instead. So cars are less valuable, on average, because owning one means you sometimes miss out on profitable opportunities.

Thanks kevin. This is helping me get my head clearer on the costs of illiquidity.

[Scott Sumner]

Nick, A belated thanks for answering my questions. I think you are right that the on the run and off the run bonds are a good way of getting at liquidity. I had forgotten about that anomoly. I seem to recall it as an issue in the LTCM fiasco of 1998. They went long and short (arbitraged) in a way that assumed the spread would narrow as soon as both bonds became off the run. But somehow the spread widened. (Black Swans again.) This is just based on memory, so I don't recall if it was a major factor in their financial crisis.

[Dioktos]

By the way, I think your statement that "velocities of circulation (turnover rates) will vary inversely with liquidity" is incorrect. Velocity varies directly with liquidity, as your next sentence states.

[Nick Rowe]

Dioktos: Oops. You are right.