Sunday, January 20, 2013

I must be a dummy for not understanding the shortage of safe asset argument

I've never understood the global shortage of safe asset meme. I'm willing to be educated.

I know that Ricardo Caballero and Gary Gorton have written about the safe asset shortage problem. In the blogosphere it pops up in David Beckworth and David Andolfatto, and the folks at FT Alphaville can't talk about much else.

First, there seems to me to be definitional issues. What is a safe asset? Beckworth, for instance, describes them as "those assets that are highly liquid and expected to maintain their value." But liquidity and riskiness are separate concepts. There are many financial instruments that are very liquid yet risky—take the S&P mini futures contract, the most liquid futures contract in the world. There are many low-risk instruments that are illiquid—a 5 year non-cashable Canadian GIC being a good example. How are we to reconcile these oppositions into one definition?

Second, it seems to me that the concept of a safety is misspecified. How do we go about classifying safety? Is it a spectrum, or are there discrete categories of safety? If a spectrum, how are we supposed to measure degrees of safety or the lack thereof? Where do we get this safety data? Next, how do we get economy-wide gradiations of safety data?

If, on the other hand, assets are to be categorized into discrete buckets of safe and non-safe, how can anyone possibly know what asset goes in each bucket? Take a vote? Whatever Gorton says? Best 4 out of 7 rock paper scissors? At what point did Italian debt go from the safe into the non-safe bucket? Or the Zim dollar? The whole process seems ad hoc and impossible to empirically verify.

But ignore all these epistemological points for now.

Let's deal first with the word shortage. If society faces a shortage of safe financial assets, won't the prices of those safe assets immediately rise and thereby remedy the shortage?*  If so, where's the problem?

Imagine a basic economy in which gold is the only safe asset and everything else is risky. There's a sudden demand for safety which means that a shortage of gold simultaneously appears in individuals' portfolios. What happens? The shortage can't be filled by gold production, supply is more or less fixed. So the gold price immediately triples. There's now enough gold to satisfy everyone's demand for safe assets.

A big part of the safe asset shortage meme seems to apply specifically to collateral. In our simple economy, if for some reason the gold market is broken, just convert risky assets into safe collateral by requiring a slightly bigger haircut than before. This is RebelEconomist's point here. Next, recruit formerly uncollateralized risky assets like pianos and grandfather clocks into serving as collateral and slap a bigger hair cut on them. Do this until the demand for collateral is met. End of story.

What does a shortage in financial asset markets even look like? I've seen shortages at Toys R Us when some new item arrives and the store manager hasn't properly anticipated demand. Lineups and unhappy customers are the result. But has anyone ever seen a lineup at the bid-ask spread for AAPL? The idea is odd. If you want to buy AAPL now, just take whatever is on offer at 601. Then take whatever is on offer at 602. Then 603. Repeat until you've bought every share that has ever been issued. The same with t-bond markets. Your demand will never go unsatisfied. No shortage here.

But maybe I'm putting too much weight on the word shortage. Maybe the problem is not a shortage per se, but the underlying increase in preferences for safety. If  people's taste for risk changes, market prices adjust to a new equilibrium in order to satisfy those tastes. Why must this new distribution of prices be considered an aberration? Does the safe asset problem apply equally when people express their demand for more safety by purchasing fire extinguishers, guns, and home alarm systems? Would we then be talking about a shortage of safe consumables?

And lastly, back to the epistemological issue. I don't think we can get proof that we've got too few safe assets because we haven't properly specified the concept of safety, which means we don't have good data. Without good data, we can't be confident in any pronouncements concerning safe asset shortages.

So tell me why I'm a dummy.

[Update: David Beckworth responds]

*put aside the idea that some of those assets are "money", and therefore may be sticky.


  1. "The shortage can't be filled by gold production, supply is more or less fixed. So the gold price immediately triples. There's now enough gold to satisfy everyone's demand for safe assets."

    It seems like there's a logical gap there. How exactly do higher gold prices satisfy everyone's demand for safe assets? I can see how it might allocate the safe asset more efficiently (given some assumptions), but then lots of people's demand is still being unmet. My understanding is that the shortage is supposed to be remedied by a higher price because the higher price is supposed to incentivize production, but you have specified that the supply of gold is fixed There is no more gold than before -- how does this solve the problem?

    Anyway I don't know why (or if) you're a dummy, but it seems to me like there could conceivably be a shortage of safe assets, especially in the short run but also in the medium-long run if the supply is somehow fixed.

    1. The increased unit value offsets the lower unit count. No logical gap.

    2. When they use the word "liquid" it seems to me that they're really talking about the expectations of trading at face value, such that an asset can be used as collateral to borrow liquid assets. In a sense, safe assets are liquid by imputation. Following this, Gorton would define a safe asset as one that trades at face value and is expected to remain worth that value (i.e. "information insensitive"). I'd say that Gorton would agree that the riskiness of assets fluctuations, which is probably why he's so skeptical that the market can provide truly safe assets (instead of, say, U.S. sovereign bonds). I need to read Beckworth's (I think) post, because I'd think that U.S. treasuries would make up for whatever shortage of private assets there is. Also, I don't know if the only related problem is related to collateral for short-term credit, but if so it's possible that Gorton, et. al., are chasing something that maybe isn't necessarily necessary (i.e. repo and ABCP markets as large as they were in ~2005).

    3. Jacob, what Peter said. If the market wants $1T worth of safe gold, and there is only $0.5T, the price will immediately double and the market's demand satisfied.

      Jonathan, what is face value, then? Bonds have a face value, but what about stocks? What about gold, or land? This is the problem with setting up a category called safe... its arbitrary, and there's no data.

    4. Well, that's why I think gold would be the main form of backing asset, since ideally its value should be relatively stable.

    5. Sorry for the double response, but I'd equate "riskless" with "information insensitive." This latter concept is the most important.

  2. you're trying to figure out what it is from the outside, without comprehending what the effect is one the international banking system...

    on the purpose of govt debt, per BIS: “the purpose of government debt is not to fund government spending. It is to provide safe assets.” The BIS proposal for ensuring the supply of global safe assets in effect treats currency-issuing governments – especially the US – as the world’s savings banks.

    from what i wrote a year ago:
    the outstanding US debt & other safe assets are in such short supply that financial markets, & more specifically the shadow banking system, is unable to function properly...US debt has a special function in the financial markets; without it, the financial markets freeze up...what's important to understand is that short term US government debt has become, at least in part, the worlds money supply; a million dollar Treasury bill is used as money by the international banking system and by sovereign wealth funds in the same sense that you use a ten dollar bill in your wallet…thus, any contraction of the supply of the reserve currency (our treasury debt) has a negative impact on the world economy in the same manner that a contraction in the domestic money supply impacts our nation's economy (see IMF paper on this, PDF)

    a couple dozen links in the last 2 paragraphs here, in my first attempt to wrap my head around this...

    1. rjs, your response doesn't really grapple with any of my points. I'm still a dummy waiting to be educated.

      "thus, any contraction of the supply of the reserve currency (our treasury debt) has a negative impact on the world economy"

      There has been no contraction in treasury debt outstanding.

    2. the question of safe assets is not of just US Treasury debt, although that's the direction i approached it from in my's the universe of safe assets that has shrunk, as the credit suisse chart i posted with my explanation shows:

      what you are looking at there is, effectively, a contraction is the global money supply...

  3. In short, the safe asset problem is one of insufficient capital/margin, not inability to supply financial instruments.

    "Safe asset shortage" proponents don't like that view for one reason: raising capital ratios and margin requirements can temporarily depress velocity. Implementing the Chiliean model (forcing loss recognition, closing bad banks, recapitalizing good ones), for instance, might induce a brief recession. One way to think about it is that for real growth to accelerate, velocity must first fall temporarily. Instead, they prefer the idea that velocity must first be made to rise (with bail-outs, lower capital req., QE, etc) in order to restore growth.

  4. Jonathan's comment above is correct. If the price of your safe asset can suddenly double, then it's no longer low-risk. On the run treasuries trade near par and reflect more or less time value of money. If they were to suddenly trade at a dollar price of 200 or whatever due to supply and demand, then prima facie they could no longer be seen as being low risk.

    1. Going back to your examples in your post, if "gold" doubles in value, then it demonstrably is not something that can be expected to hold its value.

      As for haircutting unsafe assets, that is essentially how repo and stock loan work - and how tranched securities like CDOs worked as well. Unfortunately, when you do that, you end up taking on quite a bit of risk to correlation, which is not so good for you when correlations go to 1 during a crisis, which is exactly when you really want your safe asset to maintain its value...

    2. I think this gets at the heart of it: a "safe asset" is really one that is stable (or appreciates) during a liquidity crisis.

      The problem is that "shortage" proponents are not talking about liquidity. They ascribe the "shortage" to "excessive pessimism". Aside from being debatable, "excessive pessimism" implies "worries about solvency or earnings volatility". So we're not talking about liquidity risk, but instead credit risk. The problem is the Fed is creating duration risk in an effort to reduce credit risk. "Shortage" proponents assume duration risk is low; in reality, it can cause steep losses.

    3. Anon1, Gorton doesn't cease classifying off the run bonds as risky because they trade above face value. The high coupon compensates for potential capital losses, presumably.

      If a safe asset is defined as some instrument trading at face, what about all those instruments that have no face values: stocks, land, commodities, paper cash? Safety is a universal concept (or should be), not one confined to only that small sliver of assets that have face values.

      Diego, you say that advocates of the safe asset meme are not talking about liquidity. It seems to me that use both, somewhat blurring the line between credit risk vs liquidity risk.

    4. I don't fully understand the point you're making. Do you assert that e.g. "stocks, land, commodities" can be considered "safe assets"? I can't say I've ever heard of such instruments pledged as collateral for financial transactions. I think there's some shorthand going on here for "safe asset" as "asset suitable for a certain type of financial transaction".

      Do you agree with my point that, in the example your present in your post, "gold" does not satisfy Beckworth's criteria for safe asset, simply because an asset whose price "immediately triples" cannot also be "expected to maintain [its] value"? Fixed income assets with limited credit risk are special in this sense because (assuming rates are exogenous) their prices are not expected to incur that sort of volatility. I don't intend this as a defense of Beckworth's definition - I just mean to point out that your counterfactual scenario is incoherent, since your hypothetical safe asset is very clearly not maintaining its value.

      The reason that liquid, risky assets can't be substituted for treasuries is because of the associated risk involved with them. In the hypothetical scenario where you bought some stock and posted it as collateral with, say, a 20% haircut, then it is still the case that (1) you need to fund the haircut amount out of firm capital, and (2) you are exposed to the full economic risk associated with that stock. This is much less attractive than acquiring some amount of treasuries and posting those at close to face value, and not having to worry about taking undue risk.

      I'm not disagreeing with you that there are issues with the definitions of safe assets - clearly there are differences between what is demanded by (1) a bank's treasury group looking for collateral for a financial transaction, (2) a corporate treasurer looking for some way to store 100mm of cash, and (3) an asset manager required to maintain some weighted average rating of the bonds in his portfolio. But you are focusing on the generalities, when all the interesting business is with the details.

    5. "Do you agree with my point that, in the example your present in your post, "gold" does not satisfy Beckworth's criteria for safe asset, simply because an asset whose price "immediately triples" cannot also be "expected to maintain [its] value"?"

      If you don't like gold, I'm perfectly comfortable substituting bonds in my example. The price of bonds rises till they satisfy people's demand for security.

      "Do you assert that e.g. "stocks, land, commodities" can be considered "safe assets"? I can't say I've ever heard of such instruments pledged as collateral for financial transactions."

      It depends on the definition of safe. Beckworth says that safety is a spectrum, so why wouldn't blue chip stocks and land leased to good tenants be somewhere near the safer side of the spectrum?

      Your definition of safe asset uses discrete buckets. You seem to qualify something as safe if it can be pledged as collateral. Households pledge land as collateral all the time, so I don't see why land can't fall within the safe asset bucket.

      In any case, perhaps you can see my frustration with the definition of safe assets.

    6. "Do you assert that e.g. "stocks, land, commodities" can be considered "safe assets"? I can't say I've ever heard of such instruments pledged as collateral for financial transactions."

      Of course they are. Land, for example, is infamous for its role in the Japanese bust. It was used as collateral for many bank loans. It has been suggested that the use of copper as collateral is connected with copper hoarding in China.

  5. I assume that advocates of this theory would agree that the kinds of adjustment you talk about would work. They just believe that price-stickiness and other frictions prevent them from occurring fast enough so the economy gets stuck in a dis-equilibrium as peoples unmet demand for safe assets is translated into a demand for money.

    To use the gold example - if demand for gold doubles but frictions prevent the price of gold relative to other goods from adjusting quickly enough then people will hold money instead of gold and cause a monetary dis-equilibrium. if somehow alternative safe assets could be produced then this could be avoided.

    Its not immediately clear to me why safe assets should be subject to price stickiness, or why the market would not simply produce more to meet the demand but that is the only way I can make this theory make sense.

    1. Rob, well you and me are in the same boat. I've never seen a financial asset that was subject to price stickiness. Note my point on AAPL. But I haven't read the safe asset shortage advocates invoking price stickiness in order to generate a safe asset problem. Which is why I'm confused.

  6. this is all off on a wrong track...

    the difficulty here might be with trying to interpret Beckworth's understanding of "safe assets" in the context it's most often used are those financial instruments which are the equivalent of "money" for the banking system...

    ie, those "electronic" pieces of paper which are unquestionably accepted as a liquid medium of exchange in the international banking community in the same way that you accept a ten dollar bill as a medium of exchange in your neighborhood...

    1. You're demonstrating my whole point! The category "safe asset," or the degree of "safety" of assets, are both so poorly defined that we can't have good conversations about them.

    2. JP, i jumped into this thread because you expressed an interest in understanding a problem i grappled with understanding for while before i wrote about it, and all ive been expressing is what i've come to understand the problem is...i'm not advocating anything...

      again, quoting my old post, and the problem as i saw it then, for what its worth:

      economist david beckworth points outthese safe assets serve as transaction assets and thus either back or act as a medium of exchange…(they) have served as collateral for repurchase agreements, the equivalent of a deposit account for the shadow banking system…the disappearance of safe assets therefore means the disappearance of money for the shadow banking system”…this problem exacerbated the liquidity crunch that developed in the European banking system a few months back which prompted the Fed’s recent action to coordinate with 5 other central banks to lower the price of dollar denominated currency swaps in order to ease ongoing bank funding strains, and this week’s ECB lending of $641B to banks via a 3 year LTRO.…quality collateral has become so scarce that banks are hoarding it just as a miser would hide paper money under a mattress in times of uncertainty…but instead of creating “money” when its in short supply & needed, the world is stuck with this system of having to borrow it & pay it back; what steve roth at angry bear callsa stunningly byzantine and dysfunctional approach to managing the supply of money to the real economy that produces human-consumable goods and services...

  7. JP,

    Here is how I think about it. The "shortage" refers to a lack of commitment -- a lack of credibility -- a lack of credere -- a lack of credit -- a lack of belief on the part of creditors that they will be repaid.

    In a world in which everyone could be expected to make good on the promises they make -- there would be no shortage of commitment. But in the real world, commitment is in short supply. Do you disagree? (short supply relative to nirvana).

    How can commitment be enhanced when it is lacking? The use of some assets as collateral. Not all assets make good collateral. I think there is reason to believe that there is a limited supply of such assets (there are informational issues, legal issues, property right issues, etc.).

    If there is such a shortage, the equilibrium market price (a market clearing price) of good collateral assets will be higher than their "fundamental" value (the present value of their income flow). We call the difference between market price and fundamental value a "liquidity premium."

    Now, maybe you don't like the argument, or maybe you don't think it is quantitatively important, but I hope you can at least understand the argument. :)

    "...just convert risky assets into safe collateral by requiring a slightly bigger haircut than before."

    JP, let me offer my pledge to deliver $100 of gold to you next year. I'll let you discount my pledge as much as you want. How does this increase the collateral value of my promise?

    1. David,
      I'm not sure that is a good example of collateral.

      Maybe a more appropriate one is: "here's $100 of gold in exchange for $90 cash. In a year's time, I'll give you the cash and you give me the gold back. In case I don't, you can sell the gold for probably more than $90 and pocket the difference. What? You don't think you'll get $90? Okay, give me $80 in exchange for the gold."

      The problem in a liquidity crisis is that reasonable margin may not protect against a selling cascade. However, "safe asset shortage" proponents are not arguing that markets have abnormally high liquidity fears. The issue is not liquidity but pessimism over the risk to the value of the asset. This "pessimism" means the asset just needs to be repriced, not that it cannot be manufactured.

    2. "Now, maybe you don't like the argument, or maybe you don't think it is quantitatively important, but I hope you can at least understand the argument. :)"

      Yes, that is an argument I can grapple with. A few thoughts: one of my criticisms of the safe asset shortage meme is that it seems to jumble together liquidity and risk. The premium on top of fundamental value could be a liquidity premium, a risk premium, or both. Policy prescriptions would probably differ for either. Safe asset advocates offer policy prescriptions despite the fact that there is a lack of data for either risk premia or liquidity premia. Show me the economy-wide liquidity premium, for instance.

      As for the shortage language... liquidity is just a product, or a service. Same with credible commitment. If we put a higher value on both liquidity and commitment than before, the price of assets with these characteristics rise until the market has all the liquidity and credible commitment it desires. Why do we call this a shortage? Shortage implies an unmet demand for something, or a failure or aberration. High liquidity and risk premia are markets working, not markets failing.

      "JP, let me offer my pledge to deliver $100 of gold to you next year. I'll let you discount my pledge as much as you want. How does this increase the collateral value of my promise?"

      To me, the collateral comes in because you pledge to secure the gold loan with an economics textbook (for lack of a better idea) which I can hold over the year and take delivery of if you fail to deliver. If some sort of shock happens and I'm less confident in you, I'd need 2 textbooks to accept the deal. Am I messing up your example too much?

    3. JP...I'm just not thinking/writing clearly (I have a coauthor in my office right now working on a different problem!).

      Diego, yes, you have the better example. But note that if commitment is not lacking, then my pledge of gold is just as good as the gold, in terms of collateral.

      JP, as for your textook-as-collateral example, OK. But I still fail to see how discounting assets increases the amount of collateral in an economy (or did I misunderstand your original argument?).

    4. All I meant by that is that risky collateral can still serve as collateral. Just apply a larger haircut. Then find uncollateralized risky assets like your second book and use them as collateral. Supply of collateral should be forthcoming.

      But I'm more curious about the word shortage. I don't disagree that people probably want more liquid & safe assets than before. Perhaps if we get rid of the word shortage, that would solve all our differences?

    5. To painfully stretch the textbook example, the implications of your approach are:

      1) I would need to buy a second textbook. Ignoring rehypothecation, I'm going to need to go buy a second textbook. To do so, I'm going to need to borrow an amount of money matching the face value of the textbook. Assuming all of my other collateral is tied up, I'm going to end up doing so on an unsecured basis, which is going to be pretty expensive.

      2) I would now have risk to the value of an additional textbook, which I don't want to have in the first place. I could perhaps TRS it away (except that's expensive!), or try to hedge it some other way, but ultimately there's no free lunch there.

      If I'm misreading, please correct me, but for your scenario to work, the people who need to have collateral to pledge would need to own additional assets (risky or otherwise) that they can incrementally pledge as collateral. That those assets simply exist is insufficient, because they need to be in the hands of the people who need the collateral. To adequately transform your textbooks into something suitable for posting, I'd actually to need to e.g. buy the textbooks, tranche up the risk, sell (say) the 0-30 tranche that has the risk, and hold and pledge the 30-100 tranche that doesn't have the risk. If that's my job, fine, but it's not something that just happens for free, and apparently every once in a while it causes other problems in the economy.

    6. One extra note, in practice the way it works out is this -

      * If I own a bunch of corporate bonds, I will pledge them out for repo, but I'm doing this because it means I only end up needing unsecured funding on the haircut amount, rather than the full amount. I'm not going to go buy a bunch of corporate bonds solely to use as repo.

      * If I'm your repo counterparty, even _with_ the haircuts, I'd still rather have cash than your risky assets, just for operational reasons if nothing else. I may be roughly indifferent between having $100 in cash vs having $100 in treasuries, but I'd really rather not have $200 in textbooks, even if it's very unlikely that I end up taking a loss. This is roughly how the Lehman thing actually went down, right? It turns out that even with collateral (and especially with collateral that is not "safe", whatever that means), I'd still rather stop rolling your repo if it looks like you're going to go bust. That collateral may protect me if you suddenly go away, which is why I'm willing to effectively lend money to you cheaply, but in practice it's not everything, and that is reflected in what actually happens.

    7. Lol, well since we're going with the text book example...

      In my head I was imagining that David already had a whole bookcase full of unused textbooks, which I'm sure he does. So he could turn one of his dusty tomes into collateral. That's my point about recruiting new collateral.

      But let's say he's already pledged all his textbooks. Surely some of his books are less risky and more liquid than others. After a 2008 style crisis, some will rise in price as their risk + liquidity premia jump. Some will fall. He can use the unused bits of collateral from the books that rose in value to fund the undercollateralized loans secured by books that fell in value.

      Maybe David holds only risky books and is in trouble. But from an economy- wide perspective, in any flight to safety some books will rise in price and some will fall. On the whole, the books that rise will provide the collateral to fill the void left by the books that fall. Or so it seems to me. I like to think that my mind can be changed.

      Note that I am not denying your point that this happens at a cost, and that it doesn't cause various problems. But can these effects cause a so-called collateral shortage that lasts for years and weighs down the entire economy?

    8. I don't think that's a good approximation for what happened in 2008. My understanding is that

      1) As a whole, asset prices went down overall.
      2) The risk to assets previously seen as safe were not previously priced in, so haircuts increased.

      Assume I start with a treasury bond worth $100 and some mortgage-backed debt worth $100, and I can post both with no haircuts. In this scenario, I have $200 of collateral. Let's say 2008 happens, and my treasury goes up to $102, while my mortgage-backed bond falls to $98. However, because expectations of volatility have increased, there might now be a $20 haircut on the mortgage-backed bond, so while my portfolio is still worth $200, it will only collateralize $180, so now I have less collateral despite not having lost any money on my portfolio.

      This is actually something that Beckworth's definition captures. The collateral haircut is supposed to be set such that I make whole if you default, and I'm left with only the collateral. Holding liquidity constant, if the pledged asset is more volatile, I'm going to need a bigger haircut to reduce the chance that I lose money. Correspondingly, holding volatility constant, it's going to take me more time to liquidate the illiquid collateral, so I incur more variance. This is on top of any effect from change in asset prices. In the "gold" example in your post, there is no way I'm going to let you use the $300 worth of "gold" that was worth $100 to collateralize a $300 liability - I just saw it go up from $100 to $300, so there's some chance I'd get less than $300 for it if you defaulted.

      This is pretty closely analogous to writing a put, actually. So put it another way, even if asset prices don't change, if perceived volatility goes up, puts go up in price.

      Going back to the example of your AAPL scenario, if AAPL carries a 50% haircut, then to collateralize a $100 exposure, I'd need to get $200 worth of AAPL stock. This means that I need to fund an extra $100 (ouch!), and I'd end up sitting on $200 of AAPL delta (which I presumably don't want). This is a very, very significant cost - probably enough that I'm just not going to do the deal, unless it's extremely lucrative. Actually, now that I think about it, this is a non-sensical scenario in the first place - if I have $100 in cash (or can borrow $100), why am I not just posting it to you in the first place?

      So going back to the example of my portfolio above, maybe a better way to put it is this - before 2008, my $200 portfolio could secure a $200 exposure; after 2008, my $200 portfolio can only secure $180, so the amount of collateral available to me has gone down, which directly reduces my ability to intermediate. Now, maybe I have some amount of additional unencumbered assets that I could pledge - a liquidity reserve - and I could dig into that; but that's just a buffer - in the end, I can't secure as large an exposure with the assets that I own, so I'm just not going to intermediate as much.

    9. You didn't like the textbook illustration? I see a flight to so-called safety as causing a redistribution of asset prices, not a net fall in asset prices. That probably isolates one of the differences between us then.

      You also seem to be more interested in the point of view of an individual who happens to be facing a shortage of safe assets. I don't disagree with the scenario you've outlined. But I'm more interested in the economy wide perspective. Can the economy as a whole experience an unmet demand for safety?

      Lastly, if we use Andolfatto's language, then an asset can fall in price but enjoy a rise in its so-called liquidity premium. If so, its fundamental value has fallen more than its liquidity premium has risen. Nevertheless, that asset has enjoyed a flight to safety, that's why the premium increased. The point is, using Andolfatto's language, a change in prices isn't necessarily indicative of a shift in preferences for "safety". Which makes the empirical side of the safe-asset shortage argument tougher to verify.

  8. If I understand David Andolfatto correctly he is saying that higher perception of risk is leading to higher amount of collateral being needed to secure loans. This collateral is what is being defined as 'safe assets".

    I agree with JP that either
    1) the value of available and acceptable collateral will increase in value. If gold doubles in prices then each oz will greater collateral value. Perhaps though ceditors would be nervous about the higher prices being temporary and still value the gold as collateral at the old and not the new price.
    2) new collateral will be brought into play. The amount of potential collateral in the world would seem endless so hard to imagine this could be a limitation.

    I'm wondering if the real problem is more subtle. There is indeed a demand for collateral, there is not a shortage of available collateral, but borrowers are simply not prepared to put up the levels of collateral being required as they too see the borrowing as risky ?

    1. Correction: "There is indeed aN INCREASED demand for collateral"

    2. Rob, I like your rundown of things. I feel like there is something more subtle at work at the core of the safe assets theory but I just can't wrap my noggin around it.

      If you're theory right then the problem isn't a shortage of collateral, but a shortage of optimism.

  9. safe assets as defined by the Credit Suisse 2012 Global Outlook; basically, theyre talking triple AAA paper; if you want to include land, gold and common stock, then you are defining your own safe asset class that is not the same as is in common use by the Bank of International Settlements and the international banking community...feel free to do that if you want, but dont conflate the two definitions, or suggest you know what the often spoken of shortage implies...

    1. rjs, why do you think they don't include bank deposits? Seems to me that these are AAA and quite liquid.

    2. i dont know...can bank deposits be pledged as collateral in the same manner as Treasuries? in repo-agreements and the like?

    3. Dunno, but deposits are electronic pieces of paper which are unquestionably accepted as a liquid medium of exchange in the international banking community.

  10. But I'm more curious about the word shortage. I don't disagree that people probably want more liquid & safe assets than before. Perhaps if we get rid of the word shortage, that would solve all our differences?

    I think so. There are limits to the English language -- it is all so clear in the math!

    Btw, the argument has nothing to do with risk. Consider an economy with no risk. Following Kiyotaki and Moore, suppose that debtors can pledge to pay back a loan up to (but no more than) some fraction of an asset they own. Any money owed beyond that is not repaid (with 100% certainty).

    In the limit, send that fraction to zero (so that there is zero collateral). Credit markets break down.

    I'm betting that you understand this. The real issue (apart from semantics) is whether this story is quantitatively important.

    1. David, by Kiyotaki/Moore are you referring to Evil is the Root of all Money?

    2. Yes, that one...or any other one -- they're all the

    3. Ok, just read through it. Moore, by the way, writes:

      "Banks are not the only way of turning blue paper into red. In May 1970
      there was a banking strike in Ireland, which lasted over six months. It was
      feared that the economy might collapse -- because, without banks, there would be too little liquidity. In the event, even though over 80% of the money supply was frozen, the economy hardly blipped! Do you know why? The pub landlords took over, and started circulating IOUs through their bars. This is a nice illustration of the resilience of an economy, to create alternative supplies of liquidity."

      ...which is exactly what I've been trying to say. Shortages can only be ephemeral since the market quickly finds new "safe" assets.

      So θ1 is the borrowing constraint. You can only borrow a proportion θ1 of an asset's value. You get a "shortage" of "liquidity" if θ1 is set too low, so the price of paper will rise and gain a premium. Shouldn't this increase the value of uncollateralized assets and projects, since it now costs less to fund them? θ1 stays fixed, but since the price of assets is now higher, and once unprofitable projects are now profitable, more paper can be created, thereby quickly solving the shortage.

      By the way, I read

      which is unusually readable, and quite interesting.

  11. David,
    An interesting thought experiment, but for me it raises more questions than it answers. Are you saying an economy cannot function on purely unsecured credit? I could see how there would be more friction: taking freely traded collateral is more efficient than recouping assets in bankruptcy. From a systems perspective, however, increasing everyone's liquidity (by using liquid instruments as collateral) leads to less self-hedging against a liquidity crisis. The more efficiency, the more collective liquidity risk. Carolyn Sissoko makes the argument that interbank markets were more robust (less fragile) when they relied on unsecured loans:

    Anyway, further elaboration would make for an interesting blog post (as I'm sure you have loads of time on your hands!)

    1. Are you saying an economy cannot function on purely unsecured credit?

      It was a statement about a model economy, not the real economy. Unsecured credit (in reality and in theory) can be supported by credit histories and the threat of punishment for default. But when unsecured credit is lacking, additional instruments that support intertemporal trade can be useful (like good collateral objects).

      Thanks for the link, Diego...I'll go take a look.

  12. Hi JP,

    You're right on all counts. I worked on a trading desk of a major bank that dealt intimately with collateral and funding and the FT Alphaville concerns about "shortage of collateral" make no sense to me. In fact, all the explanations I've seen sounded very confused--if you have one that you found clear I'd be interested.

    In this post I define safe assets as "asset[s] whose price is stable in terms of the unit of account" and treat the very question of the "safe asset shortage", which is a horrible way to put it because, as you said, there isn't really a way to define the quantity of safe assets out there: you can always split a somewhat safe asset into a super safe asset and a riskier asset, so this is all very grey.

    That is true however is that if the real yield on safe assets is too high, then the price level will drop with all the unemployment problems that come with that. This is what's going on now with the "zero bound" non-sense and negative rates are the cure. At 0% nominal yield, safe assets are simply yielding way too much..

    You might also find this post interesting.


    1. I liked your post. You said "There always exists some rate of return for which the demand for safe asset equals the supply..." which is really all I'm saying in this post.

      Yes, if we've hit the 0 lower bound, then a shortage gets translated into a fall in prices. But the FT Alphaville crew don't seem to be saying this. They think we've seen a shortage for years, even prior to the crisis.

      But even if some short term t-bill rates are at 0, we've still got longer bonds yielding 1%, so an excess demand for safety can still be satisfied by a rise in existing longer term bond prices. As you point out, you can twin a longer term bond with an interest rate swap to build a truly "safe" asset.

  13. Thanks!

    "They think we've seen a shortage for years, even prior to the crisis."

    Yeah, that doesn't make any sense to me..

    "excess demand for safety can still be satisfied by a rise in existing longer term bond prices"

    Yes, using the future term structure to achieve something today is effectively equivalent to level targeting, i.e. commit to keep rates at 0% until back on path.

    It might work in a lot of situation but I like negative nominal rates better as they work just fine even if the entire real natural rate forward curve is lower than -1 * inflation target (or equivalent in NGDP space).