Tuesday, January 29, 2013

Why is There Still a Shortage of Safe Assets?

JP  Koning wants to know why many of us continue to talk about a safe asset shortage five years after the financial crisis started. Shouldn't this problem corrected itself many years ago? He has asked this questions many times and most recently framed it this way:
Why do we *need* more safe assets? Why don't we just let the existing ones rise in value, thereby providing safety? If we wanted to express our desire for safety by buying fire extinguishers, then I'd agree that we need to produce more safe assets. After all, only some sort of increase in the supply of extinguishers will be able to meet that demand.

But things are different if we express our demand for safety by turning to financial markets. The great thing about t-bonds is that unlike fire extinguishers, we don't need to fabricate more of them to meet our demands for safety... we just need a higher real value on the stock of existing t-bonds. This can be entirely met by shifts in prices. Where is the problem that needs to be rectified?
This is a great question. Why haven't financial markets--the nimblest, most flexible markets of all--pushed treasury values to levels that would cause the market for safe assets to clear? Shouldn't arbitrage in these markets fixed this problem long ago?  

Let me begin my answer by recalling why the ongoing shortage of safe assets is such a big deal. Safe assets facilitate transactions for institutional investors and therefore effectively acts as their money. During the crisis, many of these transaction assets disappeared just as the demand for them was picking up. Since these institutional money assets often backstop retail financial intermediation, the sudden shortage of them also meant a shortage of retail money assets. In other words, the shortage of safe assets matters because it means there is an excess demand for both institutional and retail money assets. This excess money demand, in turn, is keeping aggregate nominal expenditure growth below where it should be. 

It is also important to note that although the crisis began in 2007 there has been a series of subsequent shocks that have kept the demand for safe assets elevated: the Euro crisis of 2010-2012, the debt-ceiling talks of 2011, the fiscal cliff of 2012, and concerns about a China slowdown. So there hasn't been for sometime a period of prolonged calm to ease the heightened demand for safe assets. Still, Koning's argument should still hold despite this spate of bad economic news. Safe asset prices should be able to adjust to reflect these developments.

The problem is that safe assets, treasury securities in particular, cannot make this adjustment when they are up against the zero lower bound (ZLB) on nominal interest rates. Given the large shortfall of safe assets, interest rates need to go below 0% for treasury prices to rise enough to satiate the excess demand for them. Investors, however, can earn 0% holding money at the ZLB. Consequently, investors will not purchase enough treasury securities to sufficiently raise treasury prices (i.e. lower interest rates) and clear the market.1

In addition, the heightened economic uncertainty and the ZLB means the demand for these transaction assets (i.e. money and treasuries) becomes almost insatiable. Investors, therefore, shy away from other higher yielding, riskier assets that normally would lure them. Portfolios get overly weighted toward liquid assets.

Note what is happening here: treasuries and money become increasingly close substitutes as they approach the ZLB, while the overall transaction asset market becomes increasingly segmented from other asset markets. In other words, as arbitrage becomes more powerful among transaction assets like money and treasuries, it becomes less powerful between the market for transaction assets and other asset markets. The short answer to Koning's question, then, is that the ZLB has segmented the transaction asset market and this is preventing the safe asset market from clearing. 

Below is my initial attempt to show this graphically. It shows that while treasury and money assets substitutability is always responsive to interest rate changes, market segmentation is not and only kicks in after some threshold close to the zero bound is reached.  In other words, at some point when treasuries and money become close enough substitutes, the market for transaction assets begins to segment from other markets. Where this actually occurs is unknown and the way I have drawn it is arbitrary. But hopefully you see the point.

Now market segmentation is a controversial idea. Many observers don't accept it. But it seems like a compelling story for the transaction asset market at the ZLB. Empirically, it provides an easy explanation for why BAA-AAA corporate yield spread, junk bond spread, and other non-transaction asset spreads are getting closer to historical norms, while the BAA yields-10 treasury yield spread and the S&P500 earnings yield-20 year treasury yield spread remain inordinately high.Welcome to the strange new world of transaction asset market segmentation.

P.S. Market Monetarists, including myself, typically downplay the importance of the ZLB for monetary policy. We argue the ZLB is really just an artifact of doing monetary policy with an interest rate; it should have no actual bearing on the efficacy of monetary policy.  Here, in the case of the safe asset shortage,  it does seem to be a non-trivial phenomenon that needs to be taking seriously.

1Given a sticky price level, the increased holdings of money at the ZLB will also continue since the price level does not adjust quickly either.


  1. Suppose the central bank just buys enough risky assets so that the quantity of resreve balances (and checkable deposits) is sufficient to meet the demand for safe assets?

    Clearly, those investment banks who underwrite new issues of secured commercial paper would find this undesirable. Dealers and brokers who buy and sell these financial assets might find this undesirable. Those doing cash management for all sorts of firms might find questions asked about why they deserve their high salaries if they aren't realy doing anything.

    From all of those folks perspective, it is important that there be lots of safe assets with someone high yields so that it makes sense to avoid the opportunity cost of just leaving money in checking accounts. Paying interest on reserve balances and checking tends to reduce their incomes.

    By the way, most money bears interest, and so the zero nominal bound does not apply. It is only hand-to-hand currency that creates the problem.

    If a shortage of safe assets requires an interest rate below the cost of storing currency, then a financial system based upon currency is going to have problems.

    Why is it illegal again from banks to pay interest on demand deposits?

    1. Bill, if a central bank signals that it intends to purchase enough risky assets to satiate the demand for safe assets, it should trigger a portfolio rebalancing that automatically satiates the excess safe asset demand. That is, if credible, the announced purchasing plan would raise expected nominal income and reduce the precautionary demand for safe assets. This is the whole point of a NGDP level target, no?

      Yes, most money bears interest but at different levels and each money asset has different degrees of liquidity. My point is that on both dimensions, these transaction assets will converge making it unlikely that nominal interest rates will fall far enough to clear the safe asset market. There will always be some transaction asset, say a saving account, that will earn at least 0%, that will prevent interest rates from falling on treasuries. So the ZLB is a problem here.

      Finally, I would note that needing a negative interest rates to clear the safe asset market is no different than saying the natural interest rate has fallen into negative territory. Nothing novel here.

  2. "Let me begin my answer by recalling why the ongoing shortage of safe assets is such a big deal. Safe assets facilitate transactions for institutional investors and therefore effectively acts as their money. During the crisis, many of these transaction assets disappeared just as the demand for them was picking up. Since these institutional money assets often backstop retail financial intermediation, the sudden shortage of them also meant a shortage of retail money assets. In other words, the shortage of safe assets matters because it means there is an excess demand for both institutional and retail money assets. This excess money demand, in turn, is keeping aggregate nominal expenditure growth below where it should be."

    I can't help but notice that your explanation seems to have deftly switched from non-money assets like short term securities and institutional assets at the beginning of the paragraph, to actual money and spending (NGDP) by the end of the paragraph. It looks to me like you equated excess nominal demand for financial securities, with an excess demand for money. I think they are different.

    Short term securities may behave like money in terms of liquidity, but they are not money qua money. Money is used to buy them. They are ultimately sold for money. NGDP is composed of the money spent on final output. Short term securities do not purchase final output.

    An excess demand for financial instruments just means there are too few financial assets available at the prices offered. This activity never was included in NGDP, so if $1 million is chasing 10 safe assets instead of 100 safe financial assets, such that we say there is a shortage of safe assets, then I would think that this doesn't necessarily imply there is less money being spent on that which makes up the goods that relate to NGDP.

    Now, if there is an increase in the nominal demand for financial securities as such (and a corresponding decrease in nominal demand for things other than financial securities, such as that which makes up GDP), then I would agree that NGDP would, all else equal, fall. But that can't be assumed to necessarily be the case solely on the basis of a shortage of safe financial assets.

    I think focusing on NGDP growth, which excludes financial securities, has unfortunately created a sort of bias against reductions in spending on that which makes up NGDP, and increases in spending on things that don't make up NGDP, such as financial securities. By "bias" I mean how the theory leads to calling for monetary easing when the demand for securities rises at the expense of that which makes up NGDP.

    If on the other hand market monetarists focused on a more encompassing NGDP, which includes spending on EVERYTHING, including financial securities, then this bias would disappear. As it should, IMO, since receiving money from the sale of financial assets pays the bills and provides profitable employment opportunities just like selling pizzas and TVs and other things that make up NGDP. NGDP targeting from this perspective is akin to targeting the spending of only one or two types of pasta in Italy. There's a whole other facet of the market.

    Focusing on more all-encompassing nominal spending aggregates also has the added bonus of helping to avoid financial bubbles, which could arise if the CB eases to the extent necessary to increase NGDP, at the cost of financial asset prices going through the roof. Nobody has proven that financial asset prices have to rise at the same rate as NGDP.

    1. I wasn't as thorough here as I have been in other post, so to be clear I do believe in a rather broad definition of money. Here is why. Gary Gorton and others have shown that repos, for examples, provide the same transaction services for institutional investors that checking accounts provide for retail investors. In both case, these transaction assets are highly liquid private financial liabilities. Both function as medium of exchange, which to me is the most important role for mony.

      Gorton is no alone is this view. See William Barnett's work at the Center for Financial Stabilty.

    2. I would be very cautious with this line of thinking, because when that logic is expounded, we invariably end up including a very number of heterogeneous commodities as mediums of exchange, as we look to other narrow spheres of activity to the particular confines in which a good is liquid.

      For example, in my house, we all use chocolate as a "bribing" tool. We all like chocolate, and whenever someone wants something from someone else, if the offer is for chocolate, then there is usually bargaining going on and whatnot. Here, in this artificially constrained context of "household", chocolate behaves very much like a medium of exchange. It is perhaps even more liquid than money. Yet chocolate isn't generally accepted outside of our house, and so it isn't money. Personally, I do not accept repos as final payment for my goods and services. Many millions more are in the same boat as me. But we all accept dollars.

      Similarly, while there are many highly liquid instruments such as repos that behave like money within the confines of certain financial activity, I don't think Gorton has shown that they are therefore a generally accepted medium of exchange. Nobody walks into a Wal-Mart or a Krogers with a repo, expecting to be able to trade with them. A medium of exchange has to be generally accepted for it to be money, or else virtually everything is money.

      A firm that issues a repo to another firm in exchange for another financial security, is not conducting an X for money trade, but an X for Y trade, such as one firm exchanging shares of their stock to another firm in exchange for that company's assets, or whatever. Within the confines of financial markets, stocks and repos may behave like money, but they're really more like barter goods when considered in the context of the whole economy.

      Another example: cigarettes in prison can be, within the confines of a prison, a medium of exchange, but they're not generally accepted as a medium of exchange inside AND outside of the prison. They are ultimately sold for money. You won't be able to walk into a Wal-Mart and buy apples with cigarettes. They want money.

      Sure, there are many people with all kinds of interesting theories about money. I would be surprised if everyone were on exactly the same page. Yet as far as I can tell, the people who say repos are a money, are engaging in the same special pleading, through arbitrary constraints, as those who claim cigarettes and chocolate are money.

      The reason I think we can't include repos as money in the argument you are making, is that you are relating your explanation to NGDP. NGDP is the sum total of dollar spending, not "repo as money" spending, on final goods. If you want to claim that repos are money, then I see no clear connection of this to NGDP such that you can conclude NGDP falls or rises when there is excess or inexcessive demand for safe assets that are repos, or traded against repos. At the very least, repos are contingent, not final payments. There is still a chance that the repo issuer reneges. With money, once it's paid, there is no credit promise.

      Perhaps you can make a post showing how supply changes to repos as financial market "money", affects generally accepted money spending and thus NGDP?

    3. Wow. I didn't think it was ever going to happen, but it just did. For the first time ever, I completely agree with a Major Freedom comment.

      One of us must be slipping.

    4. Sometimes, one can find oneself "slipping"...uphill.

  3. David, thanks for answering my question. I understand where you're coming from.

    Empirically, I agree that t-bills have hit the zero-lower bound. But don't longer-dated treasuries have a way to rise before they hit the zero lower bound? See my earlier comment here. Longer-dated t-notes or t-bonds are just as versatile as t-bills when it comes to serving as collateral, so if there is an excess demand for safe collateral, a further rise in longer-dated security prices should suffice to meet that demand, or at least part of it. The fact they these securities haven't risen to the point at which they yield 0 may indicate that we don't have a shortage.

    But let's assume you're right and the bond market can't meet safe asset demand because of the ZLB. Why won't the market for safety clear by the back door? All other liquid financial assets (stocks, MBS, corporate bonds) fall to such low prices relative to fundamental value that they now qualify for inclusion into the safe asset category.

    1. JP, here is my initial (probably incomplete) reply. Since the true weighted average maturity of treasuries is only 4 years and the median is 3 years, most treasuries are already close to the zero bound. Longer-date securities are just too small a share of the total for them to make much difference.

      On your second question, the answer as I laid in the post, is that the market for transaction assets is becoming segmented from other markets. Thus, the backdoor options are less viable. However, one sure way to clear the market for safe assets would be to do a NGDP level target and/or have temporarily higher inflation.

  4. If there is such a shortage of this safe "stuff" why has it all been going down in price the last 3 three months?

    The 10 Year is at 2.02. German, Swiss yields have gone positive. Even Japanese bands have been falling.

    Me? I think we are transitioning from a shortage to an excess. If I'm right, you'll hate it....

    1. Bruce, I see the increase in yields as a step in the right direction, a step away from the inordinately high risk premiums and the safe asset shortage. The day the bond vigilantes arrive, the day the economy will be liberated from its current morass. See here for more on this point.

      With that said, we are a long way from the normal 4-5% yields on 10 year treasuries. And on the short-end, where most of the public debt lies, yields are still very close to 0%.

  5. Question: Do the excess reserves deposited by banks with the FED represent the demand for safe assets? If so, shouldn't the FED pay negative interest rates on excess reserves?

    1. "shouldn't the FED pay negative interest rates"

      I don't not disagree with the opposite of the converse of that statement.

  6. Much of the "safe asset" demand is coming from entities that want to use those assets to do things that don't add much to the economy. Trading existing financial assets, for example, is a zero sum game.

    More to the point, the notional amount of financial instruments has grown enormously compared to the real economy. You have to wonder how it was that the economy of say, 2000, got along without all this paper.

    It seems to me that much of the business of Wall Street consists of concocting ever more complicated financial assets to screw the customer with. Have we already forgotten how the CDO squareds, those wondrous tranches of tranches, were going to eliminate risk?

    If all this paper does no one but Wall Street any good, why should the Fed endeavor to validate their business models?

  7. David,

    Generally I agree with you that the market appears bifurcated - "risk-on, risk-off" as Izabella Kaminska puts it.

    You may want to consider the impact of regulation on the availability of safe assets. Regulatory changes are forcing banks and institutional investors to hoard safe assets and shorten collateral chains, and the risk-averse regulatory stance ("we must protect savers at all costs") also discourages creation of private sector "safe" assets. The arguments of Gorton and others that the public sector should simply create more safe assets to meet demand is I think fundamentally flawed for reasons of fiscal responsibility. But we don't just have a safe assets shortage, we also have collapsing velocity (as Singh pointed out). The direction of regulation is distinctly unhelpful.

    1. Frances, that is a great point and one I should have mentioned. Steve Hanke has made the same point

  8. Hi David

    I am struggling to follow this discussion. I think that the term "safe asset" itself confuses me a lot. As such, I have some questions on and would be very grateful if you could comment on these:

    Literally, a "safe asset" does not exist. Even government issued money solely offers nominal safety, but not safety in real (i.e. purchasing power) terms. So is the term "safe asset" referring to a asset free of credit and/or interest rate risk? (put differently, is credit risk absence sufficient for an asset to qualify as safe?)

    You were talking about the importance of safe assets in transactions. Do you mean explicitly as a substitute for government issued money (i.e. money-like debt), as collateral in financing transactions, or as collateral for margin positions? (Or all three)?

    How is the concept of safe asset distinct from the concept of "money-like debt"? Is safe asset a broader concept than money-like debt (i.e. all money-like debt is a safe asset, but not all safe assets are money-like debt due to interest rate risk)?

    Can the shortage of assets be achieved by more government issued money? Or does it requires open market operations on non-treasury securities, so that money is not injected at the expense of treasuries which are "safe assets"?

    The difference between treasuries and money, is, in my opinion, only the interest rate risk. I don't understand why treasuries and money become closer substitutes. The interest rate delta only depends on the duration but not on the interest rate level? can you please elaborate on this closeness of substitution?

    I know, a bunch of questions. But It would be extremely helpful to get your opinion

    Many Thanks


  9. A good idea. Thinking safety first will lead you to a better deals and transaction.

  10. Thanks for sharing the information. That’s a awesome article you posted. I found the post very useful as well as interesting. I will come back to read some more. commodity tips