Thursday, February 2, 2012

The Cyclical Dimension of the Safe Asset Problem

An important problem facing the global economy is the shortage of safe assets, assets that facilitate transactions at both the retail and institutional level.  There is both a long-term, structural dimension to this problem as well as a short-term, cyclical one.  The structural dimension is that global economic growth over the past few decades has outpaced the capacity of the world economy to produce truly safe assets, a point first noted by Ricardo Cabellero.  The cyclical dimension is that the shortage of safe assets was intensified by the Great Recession, a point stressed by Gary Gorton.  I previously made the case that both the Fed and the ECB were an important part of the cyclical story by failing to restore nominal incomes to their expected, pre-crisis paths.  In other words, since 2008 the Fed and the ECB passively tightened monetary policy which caused some of the safe assets to disappear while at the same time increasing the demand for them. 

I still hold this view, but after reading some papers on safe assets and talking with Josh Hendrickson I have come up with a more general view to the cyclical dimension of the safe asset problem. It goes as follows.

Gorton, Lewellen, and Metrick (2012) show that safe assets have constituted a relative stable share of all assets since the 1950s. They also show, as does Bansal, Coleman, and Lundblad (2011), that public and private safe assets tend to act as substitutes in providing liquidity services.  Given these findings, it stands to reason that when the central bank is doing its job and nominal GDP (NGDP) is growing at its appropriate trend, then there will be enough safe assets being privately provided.  If, however, the central bank slips and NGDP falls below its expected path, then some of the privately produced safe assets disappear, creating a shortage of safe assets. The government steps in and creates safe assets that, if produced sufficiently, would restore NGDP back to its expected path. In other words, if monetary policy does not do its job then fiscal policy could substitute for it, but in a way not normally imagined. It would do so not by increasing aggregate demand via higher government spending, but by making up for the shortage of safe assets that facilitate transactions.  All this requires is running a budget deficit which may or may not imply higher government spending (i.e. it could also come from tax cuts).  Debates about Ricardian equivalence, crowding out, and other fiscal policy concerns become moot.  What matters is if there are enough safe assets, and if not, whether fiscal policy can provide them in the absence of a NGDP-stabilizing monetary policy. 

This understanding may serve as the basis for a paper, so I look forward to any feedback you can provide.


  1. Is the problem credit risk or liquidity risk? In other words, can the government help fix the problem by funding the national debt short?

    Of course, you know....

    Can a negative nominal interest rate clear the market for safe assets? Would that fix the problem?

  2. Interesting view. That reminds me of Gustav Cassel's criticism of the gold-standard in the 1930s in terms of the real growth as required to sustain full employment outpacing the increase of metallic supply, and so putting a monetary brake to nominal growth.

    I've always admired Cassel's explanation as one of the most sensible ever to give account of the Great Depression and how the world overcame it by replacing the gold standard with a part-gold, part-fiat money - the gold exchange standard. Yours might row in the same direction.

    What is not yet clear is the high-theory relationship of safe assets in the 2010s to gold in the 1930s. To be sure, is your concept of money all based on safe-asset support - whatever the assets though as liquid as possible - rather than money specifically being debt, as much as in the gold standard money was based on species? That would be really praiseworthy for it could pull the debate out of the boring landscape of debt deflation and the like.


  3. Thank you!

    I've been barking this same tune for a few months, but people just can't get their head around the idea that government could run a fiscal stimulus and reduce its spending simultaneously.

    Moreover, fiscal stimulus only works at all because it creates safe (and liquid) assets that are close substitutes for money. It really doesn't matter what is going on with interest rates or public spending. If, for some reason, government bonds were not considered safe and liquid assets, then fiscal stimulus would be wholly ineffective, i.e. it wouldn't dent average money balances (unless, I suppose, spiraling government debt induced fears of debasement, but that's another matter).

    It's (money) supply and (money) demand, stupid!

  4. Just as the paradox of thrift doesn't intrinsically have anything to do with higher rates of saving, fiscal stimulus doesn't intrinsically have anything to do with higher rates of government spending.

  5. Isn't money itself the ultimate "safe asset" that facilitates transactions? There can't be a shortage of transaction-facilitating "safe assets" if there is a sufficient supply of money.

  6. Hi David,

    We've been having a discussion on Gorton's work the last few days and a conclusion we've drawn but I'm not sure Gorton draws (at least from the paper) is there is never a lack of safe assets since the share of safe assets is a constant ~33%. Rather, the problem is the mix of safe assets in terms of insensitivity does not meet the supply/demand.

    For example, the problem in 2007 was not that there wasn't enough safe assets, but the demand for safe assets which were more insentive than those that were becoming sensitive was not enough.

    I don't think this necessarily takes away from your overall discussion (I largely agree with it), but a point of distinction I would make (assuming the above is true) is it doesn't have to be the government creating the safe assets; it can be anyone as long as the created safe assets are more insensitive than what is currently available.

    In the end, it's the mix not the supply.


  7. It appears that the nature of fast growing emerging markets and troublesome growth in maturing economies has created a shift. We are perhaps in the midst if this shift. The demand for low risk assets for both types of economies is on the rise. Mature economies are experiencing considerable stress pushing a natural level of the desire for safe assets to a higher level than would otherwise be the case. Emerging fast growing economies also have a need, a growing one, for safe assets (Everything can't be high risk. Some safety is naturally desired). At the same time, emerging fast growth economies are probably much larger, and creating an increasingly greater demand for safe assets, than in these types of shifts that have occurred in the past. This combination of events has thrown all of us for a loop, because we have not seen this type of convergence before, if not ever.

  8. The thesis is that the supply of private safe assets is primarily sensitive to NGDP. One test of this proposition is as follows: what would happen to that supply if banks were forced to recapitalize at much higher levels AND write down doubtful loans? I would argue that such a recapitalization would again render bank liabilities "information insensitive" for retail and institutional investors. The safe asset supply would rise independent of NGDP growth (causality would run from the recapitalization to more robust growth).

    The Nordic countries forced such a recapitalization and write-down after their real estate boom, as did Mexico following the '94 devaluation and Chile in 1981-84. I'm sure there are other examples. The point is that each of these countries experienced a strong recovery once the banking system "lemon" problem was fixed.

    We need to restore the ability of the banking system to produce safe assets. This is a problem of distributing losses: we can apportion them to wealthy bondholders, taxpayers, or middle class savers (through inflation). Some combination of all three would probably be ideal, with the heaviest burden falling on the first group.

  9. Sorry to post two questions. The other thing we're discussing around the office is how QE works into the equation. On one hand QE just subsitutes very insensitive assets with the most insensitive asset (cash which goes into deposits).

    At the same time, if we go back to the argument the mix of safe assets matters, it may be there is a relationship between return and insensitivity (that doesn't seem so far fetched) and the difference in insensitivity between cash and very insensitive assets (Treasuries) is that while cash is very insensitive the difference between the two (in terms of insensitivity) is less than the difference in return (even with such low nominal returns).

    If the second part is true, the unintended consequence of QE may be that the demand for a different level of mix is not being met and growth suffers.

    So the idea being kicked around is reversing QE would increase insensitive safe assets and possibly increase interest rates as demand is met. How high would rates go and would institutions/people start taking on more risk once their base of safety was met? Would the risk lead to growth that offset any negative aspects of the now higher rates?

  10. Hmmm. I want to say something, but I am not sure what to say.

    I think the globe is generating excess capital, for safe and unsafe assets. Globally, many people save regardless of interest rates or investment climate---they save for retirement, college funds, security, a future home, to start a business etc.

    And with the emergence of upper classes globally, some people save as they cannot consume enough. After a few cars and house and some baubles and a thick steak three times a day, you run out of things to spend money on.

    I knew a lady once who bought several pairs of $200+ jeans every week, and her hubbie never complained. Evidently, $1,000 a week for clothes was not worth raising an eyebrow over.

    Should the lack of safe assets be placed within the larger context of a globe that generates huge gobs of capital?

    Capital is no longer scarce. What does that mean? For returns? For growth?

  11. All, thanks for the comments and sorry the slow reply.

    Bill, no doubt there is both credit risk and liquidity risk and what I am suggesting focuses on the latter. One could argue, though, that some of the credit risk has arisen because expected nominal incomes were not realized and to the extent treasuries restore liquidity services, reduce money demand, and thus restore nominal incomes, then some of that credit risk disappears. For example, in Europe there is no question that Greece is a credit risk. But France?

    Enrique, I see the parallel you are making to the Cassel's criticsim and it is spot on. What makes it trickier today, though, is that safe assets that serve as a medium of exchange vary based across users. Retail money assets (e.g currency, checking accounts, retail money markets funds)are what we normally think of as money, but assets that facilitate exchange for institutional investors act like money too(e.g. repos, commercial paper, tbills). It is with this latter category where I think public securities can provide the missing liquidity services.

    Lee, you should have written this post! Want to do a guest post sometime? Email if you do.

    Anonymous, yes, but the safe assets being used by retail and institutional investors can all be thought of as money assets. And right now I think the biggest shortage of money assets is at the institutional level. With that said, should the Fed provide enough base money (or the expectation that it will contingent upon some target being met) such that expectations of the economy improve I suspect that the problems with money shortages at the institutional level would be vastly reduced.

  12. finn123said, You said

    "the problem in 2007 was not that there wasn't enough safe assets, but the demand for safe assets which were more insentive than those that were becoming sensitive was not enough."

    I think we are saying the same thing. I take safe assets to mean a high degree of information insensitiveness. How else could an asset be safe if lacks that quality? Yes, there are differing degrees of it, but the point is there enough assets that have information insensitivity below a certain threshold. To say the problem is the mix of safe assets, still implies a shortage problem.

    Yes, it doesn't have to be government creating the missing safe assets. In fact, in my ideal world where the Fed prevented or corrected the below trend growth in nominal incomes the private sector would have created new safe assets by now.

    Regarding QE, your point (I think) is that cash and tbill are close to perfect substitutes so all that is happening is an asset swap that has no effect. I agree in principle, but I think the point of QE2 and Operation Twist was to go beyond swapping cash for tbills,and instead swap cash for longer-dated treasuries where there are meaningful differences in liquidity and to do so up to the point that investors are satiated with safe assets and want to start taking on riskier assets. From this perspective these programs were not big enough.

    My own view is that expectations management is key here. The Fed should not announce an explicit dollar amount they will buy, but instead commit to buying as many assets as needed until the some nominal GDP target is hit. This would change expectations (improve outlook) and cause the portfolio adjustments to happen automatically. Here is my take on how this could work:

  13. Robert, I agree though, as I note above, what we are going through now seems to me to be more a cyclical development. Hopefully at some point the rest of the world develops the capacity to make their own safe assets in greater quantity so to put less pressure on the US do so.

    David Pearson, I agree that some kind of bank recapitalization is needed and would help, but to create more safe assets would depend on expectations of robust NGDP growth. Why would the safe assets stock grow over time if there were no growth in nominal incomes?

    I do, though, agree that some of the NGDP growth is endogenous--just like I think the good employment numbers last week improved expectations and endogenously created more liquidity--and the real economy itself can slowly over time restore some of what policy is failing to do.

  14. David Beckworth, why do you want to pander to the large financial TBTF institutions demand for safe assets? Dont they just use them as collateral for their OTC derivative transactions - the WMD as Buffet called them? Why would you want the taxpayer to subsidize this destructive behavior?

  15. Hi David,

    Thanks for the reply. Two from me:

    1) I agree that we are likely saying the same thing regarding mix and safe assets, but at the same time think there is a difference when using Gorton's work. Safe assets are a constant so there can't be a shortage of safe assets; only a shortage of varying degrees of insensitive assets relative to other assets.

    I know it may seem like I'm splitting hairs, but, at least to me, the distinction is important because the mix, and how it changes, becomes the next topic of discussion (figure 3 of Gorton's paper, I think). The issue isn't quantity of safe assets as a whole, but quality and quantity different types of safe assets.

    2) Full disclosure: repo markets are mysterious to me. That being said, going back to the QE discussion I agree 100% with you on the difference between cash and longer term yield curve assets.

    In terms of overnight repos, however, is the lack of very insensitive assets from the long term curve more of a problem? In other words, are 30 year Treasuries used in the overnight repo market?

    Again, thank you for the response.

  16. finn0123,

    Here is primer on the repo market. Hope it helps.

  17. Thanks David.

    In the mean time, did you catch this one:

  18. I think the whole shortage-of-safe-assets argument is a dead end.

    Economists shouldn't be using old intrinsic value methods but subjective methods. For the economist to qualify some assets as safe and some not is to fall into the old intrinsic mode of thinking. Only individual actors are capable of making the safe-or-not qualification, and a safe asset in one person's mind is often a risky asset in another's. You can't aggregate individual conceptions of safety so as to arrive at an aggregate category called "safe assets".

  19. Too many safe assets (or should that be claims?) chasing too many other safe assets or too few?

    Too many safe assets being turned into other safe assets?

    Too many claims on real safe assets?

  20. It seems to me Ricardian equivalence still matters. If you cut taxes, isn't that going to increase the demand for safe assets as well as the supply? If Ricardian equivalence holds, and I had optimized before the tax cut, then I can achieve the same optimum by holding the entire tax cut in safe assets. So wouldn't I be inclined to increase the proportion of my portfolio that I hold in safe assets, thereby increasing the demand for safe asests?