Monday, May 6, 2013

Tell Me About that "Safe Asset Shortage" Again

From today's FT:
US Junk Debt Yield Hits Historic Low
The average yield on US junk-rated debt fell below 5 per cent for the first time on Monday, setting yet another milestone in a multiyear rally and illustrating the massive demand for fixed income returns as central banks suppress interest rates. 
Junk bonds and equities often move in tandem, and with the S&P 500 rising above 1,600 points for the first time and sitting on a double digit gain this year, speculative rated corporate bonds are also on a tear. ... With corporate default rates below historical averages, investors are willing to overlook the weaker credit quality in exchange for higher returns on the securities, analysts said. ... The demand for yield has become the major driver of investor behaviour this year, as they downplay high valuations for junk bonds and dividend paying stocks. 
“Maybe the hurdles of 6 per cent and now 5 per cent on high-yield bonds are less relevant given the insatiable demand for income,” said Jim Sarni, managing principal at Payden & Rygel. “People need income and they are less concerned about valuations.”
In other words, the demand for risky assets is exceptionally high, and this has driven up their price. Or to put it the other way, if you are a high-risk corporate borrower, now is a very good time to be looking for a loan.

Here's the same thing in a figure. Instead of their absolute yield, this shows the spread of junk bonds over AAA:

Difference between CCC- and AAA bond yields
See how high it is now, over on the far right? Yeah, me neither. But if the premium for safe assets is currently quite low, how can you say that it is a shortage of safe assets that is holding back the recovery? If financial markets are willing to lend money to risky borrowers on exceptionally generous terms, how can you say the problem is a shortage of risk-bearing capacity?

Now, the shortage-of-safe-assets story does fit the acute financial crisis period. There was a period in late 2008 and early 2009 when banks were very wary of holding risky assets, and this may have had real effects. (I say "may" because some large part of the apparent fall in the supply of credit in the crisis was limited to banks' unwillingness to lend to each other.) But by the end of 2009, the disruptions in the financial markets were over, and by all the obvious measures credit was as available as before the crisis. The difference is that now households and firms wished to borrow less. So, yes, there was excess demand for safe assets in the crisis itself; but since then, it seems to me, we are in a situation where all markets clear, just at a lower level of activity.

This is a perfect illustration of the importance of the difference between the intertemporal optimization vision of modern macro, and the income-expenditure vision of older Keynesian macro. In modern macro, it is necessarily true that a shortfall of demand for currently produced goods and services is equivalent to an excess demand for money or some other asset that the private sector can't produce. Or as DeLong puts it:
If you relieve the excess demand for financial assets, you also cure the excess supply of goods and services (the shortfall of aggregate demand) and the excess supply of labor (mass unemployment).
Because what else can people spend their (given, lifetime) income on, except currently produced stuff or assets? If they want less of one, that must mean they want more of the other.

In the Keynesian vision, by contrast, there's no fixed budget -- no endowment -- to be allocated. People don't know their lifetime income; indeed, there is no true expected value of future income out there for them to know. So households and businesses adjust current spending based on current income. Which means that all markets can clear at a various different levels of aggregate activity. Or in other words, people can rationally believe they are on their budget constraint at different levels of expenditure, so there is no reason that a decline in desired expenditure in one direction (currently produced stuff) has to be accompanied by an increase in some other direction (safe assets or money).

(This is one reason why fundamental uncertainty is important to the Keynesian system.)

For income-expenditure Keynesians, it may well be true that the financial meltdown triggered the recession. But that doesn't mean that an ongoing depression implies an ongoing credit crisis. Once underway, a low level of activity is self-reinforcing, even if financial markets return to a pristine state and asset markets all clear perfectly. Modern macro, on the other hand, does need a continued failure in financial markets to explain output continuing to fall short of potential. So they imagine a "safe asset shortage" even when the markets are shouting "safe asset glut."

EDIT: I made this same case, more convincingly I think, in this post from December about credit card debt. The key points are (1) current low demand is much more a matter of depressed consumption than depressed investment; (2) it's nonsense to say that households suffer from a shortage of safe assets since they face an infinitely-elastic supply of government-guaranteed bank deposits, the safest, most liquid asset there is; (3) while it's possible in principle for a safe-asset shortage to show up as an unwillingness of financial intermediaries to hold household debt, the evidence is overwhelming that the fall in household borrowing is driven by demand, not supply. The entire fall in credit card debt is accounted for by defaults and reduced applications; the proportion of credit card applications accepted, and the average balance per card, have not fallen at all.


  1. (It might appear there's a contradiction between this post about how "demand for yield" is driving up the price of risky assets, and the previous one about how "reaching for yield" by banks is not a problem. I would defend myself against this hypothetical charge by saying that the previous post was about how expansionary monetary policy is not dangerous or counterproductive. This one is about how it's not that helpful either.)

  2. Completely agree. The financial market disequilibrium theory looks dumber every day that the stock market makes a new high.

    Also DeLong is quite wrong about profits being depressed. Profits are good thanks to abnormally high margins.

  3. The whole thing seems quite weird and puzzling until you realize that DeLong is absolutely committed to the idea of rational expectations. The natural way to make sense of the fact that asset values -- especially equities -- fluctuate much more than variation in ex post returns can justify, is that people's beliefs about future returns, lacking any firm foundation, are subject to wild swings. But DeLong is committed to the view that there is a true present value of future corporate profits, and investors know it. So he has to fall back on big (unexplained) variations in people's willingness and/or ability to accept risk.

  4. i'm sympathetic to your multiple equilibrium interpretation of keynes, and your data on US household behavior does suggest a more complicated story of what drives aggregate economic activity . . . but i'm still not sure whether you can infer that there's no shortage of safe assets in a global context, from junk bond yields in the united states alone. not so long ago there was a time that spanish euro bonds and german euro bonds were treated as equivalent AAA, but that's no longer the case. perhaps dollar denominated debt of all credit qualities might be more attractive to global investors than euro-denominated sovereign debt that's subject to the non-neglible tail risk of an EMU breakup, or the renminbi denominated sovereign debt of a one-party state, autocracy whose need for political control throws into question both its capacity or willingness to absorb global capital flows at a sufficiently large scale . . .

  5. So, there's lots of stuff in Poszar's "Institutional Cash Pools and the Triffin Dilemma of the U.S. Banking System".

  6. Hi Rajiv.

    Glad to hear you're at least somewhat on board, since you've helped clarify my thinking on this stuff.

    So, fundamental point here, the position I'm challenging here hinges on the idea that safe assets and risky assets are complements. This is why DeLong is able to refer to "safe-asset shortage" and "risk-tolerance shortage" more or less interchangeably. The argument is that a lack of safe assets, limits lenders willingness to hold risky assets, which limits access to credit for households ad firms, holding down final expenditure. For this argument to work, you absolutely need either an absolutely high price for risky assets, or a high spread between risky and risk-free assets. Neither appears to be the case.

    I should have added in the post, not only is the price of high-risk bonds low, the quantity is very high -- $330 billion in 2012, more than twice the pre-crisis peak. I think most economists would agree, when we see price falling and quantity rising, the presumption has got to be an increase in supply. Of course no one datapoint is dispositive, but this definitely goes on the (rather large) stack of evidence that credit constraints are not what's holding back aggregate demand.

    Now, tho, there's an intermediate step. DeLong just takes the standard monetarist analysis, and does a find-and-replace of "safe asset" for "money." But you can't really do that. Safe-asset demand is specific to financial institutions. So there's two steps, first to establish that financial institutions have, in some sense, an excess demand for safe assets, and second, that the margin on which they respond to that is more restrictive credit to nonfinancial units. My position is that the second is certainly false now, and the first was almost certainly true in 2008. But there's a lot more to think through here.

  7. I think what I would say is that if there were no safe-asset shortage the current low level of economic activity would not be self-sustaining--that if there were no safe-asset shortage, at current interest rates people would be dumping safe assets and funding risky investments and buying currently-produced goods and services and we would be heading for a boom.

    And I would point to the enormous spread between real Treasury yields on the one hand and equity earnings yields on the other.

    Perhaps you would be happier if I said that, once you are in a depressed economy like this one, depression will continue until and unless you have a surplus of safe assets to kick the economy out of its low-level equilibrium?


    Brad DeLong

  8. Brad: "And I would point to the enormous spread between real Treasury yields on the one hand and equity earnings yields on the other."

    But the JUNK-equity spread is equally enormous! So if anything you're seeing a bond shortage, not a safe asset shortage. In other words, companies could increase their stock price if they levered up, but management is erring on the side of caution. But levering up has nothing to do with investing, it's just rearranging the financing.

  9. Hi Brad,

    Thanks for the response. I'm glad I haven't mischaracterized your views here too badly.

    Of course there's still a difference between saying (a) that when output is below potential, that means there is an excess supply of currently produced goods and services, which by definition implies excess demand for money, safe assets or some other nonproduced good; and (b) the balance of the evidence is that the only reason output is below potential at this particular moment is tight credit due to a shortage of safe assets in the banking system. I happen not to accept either of these claims, but the disagreement is logical in the first case and empirical in the second. It's not always clear to me when you are arguing (a) and when you are arguing (b).

    you would be happier if I said that, once you are in a depressed economy like this one, depression will continue until and unless you have a surplus of safe assets to kick the economy out of its low-level equilibrium?

    Yes, I'd be happier with that. Although you don't have to go as far as "will"; I'd settle for "may" or "can". The important point is that interest rate (or asset mix) required to maintain full employment, is not the same as the interest rate (or asset mix) required to restore full employment, once the economy has been well away from it for an extended period.

    1. Actually no, rereading, I think I would not be happy with that. Or at least, it depends exactly what you mean.

      If you mean that, *independent of current income*, there is some stock X of safe assets that is needed to produce full employment in 2013, then I would not be happier. The fundamental question is whether current or recent income affects desired expenditure, and therefore the interest rate (or money stock or etc.) required to reach full employment. The Walrasian vision -- which New Keynesians emphatically share -- is of an economy reconstituted de novo each day, so past levels of income don't matter. I think Brad shares this vision, but it may be I'm wrong. The comment here doesn't quite make it clear.

  10. As usual, I'm really just channeling Leijonhufvud. From the closing pages of On Keynsian Economics and the Economics of Keynes:

    The question of the effectiveness of monetary policy in depression remains to be considered. Here, asset demand prices are too low because entrepreneurial demand forecasts are attuned to a continuing slump. To restore a full employment rate of investment by "merely" monetary policy, market rate has then to be brought down much further than would be needed if entrepreneurial expectations had not already been adversely affected. ...

    Surely, the case for relying on monetary policy in such circumstances is not a strong one. ... The alternative would be a policy of government deficit spending designed to "correct" entrepreneurial demand forecasts. Direct expenditures on commodities will prevent the self-fulfillment of pessimistic prophecies. By falsifying the forecasts, a rise in asset values should be obtained without forcing down the market rate of interest.

  11. JW Mason, I see you tweeted... "The question is, is shortage of safe assets a sufficient explanation of why output is so far below trend?"

    The trend was falsely based on bubble values. I calculate with effective demand that real GDP is actually above potential real GDP.

    Unemployment is high because since before the crisis, capital income is up $500 billion, and labor income is down $400 billion as I calculate it. The national income is there for lower unemployment (paid workers), but the income is sitting with those with capital. Whether they are hoarding, investing in other countries or just simply not investing, is irrelevant. The liquidity for lower unemployment exists in national income.

    I posted a response to one of your posts from February...

    1. Edward,

      Thanks for this. Will try to reply soonish, or at least on the same 3-month schedule.

  12. I thought the Keynesian argument was that incomes, rather than yields, equilibrate savings and investment. In which case, you cannot point to yields and say that there is or is not an excess demand. The Keynesian would respond by saying, we have depressed incomes, so no there isn't any excess demands for bonds *now*, but there would be at the level of income that we should have.

    1. Right. But the people I am arguing with are not really Keynesians. Their argument is that the only reason we do not have a full-employment level of income, is a shortage of safe assets/risk-bearing capacity/trusted intermediaries. Evidence on the state of the bond market now is useful for evaluating this claim, no?

    2. I agree with you. But the other side has a kind of weird point to make, in the sense that if the economy was healthier, then the gap between the demand for safe bonds and the supply of safe bonds would be smaller. This does not mean that by increasing the quantity of safe bonds, the economy becomes healthy.

  13. JWM, you said above... "So households and businesses adjust current spending based on current income. Which means that all markets can clear at a various different levels of aggregate activity."

    This is true. I agree with you that there is no safe asset shortage.

    Can we say that there appears to be a safe-asset shortage because their low return rates lead to lower activity in their market?
    In which case, if the Fed funds rate was raised, returns on safe-assets would rise and the market for safe-assets would clear at a higher level of activity.

  14. If DeLong's "excess supply of labor" means a workforce that is too large to bring unemployment down, then is a possible lower labor participation rate a good thing? There seems to be a disconnect between the usefulness of the theory relative to economic reality. In other words, if the participation rate drops, then wouldn't there be a corresponding drop in output without higher productivity? Moreover, wouldn't net GDP lower as a result even if other factors such as government spending rise?

  15. See how high it is now, over on the far right? Yeah, me neither.

    Wait, so %-% on the vertical axis is [% returns on CCC] above and beyond [% returns on AAA], correct?

    If the demand for income by creditors is high, as supposed, then as you say, the price to the lender should increase. That would mean that CCC rated corporates get a sweetheart deal because any income is better than USTSY.

    Contrary to what you said, this seems totally consistent with Sarni's hypothesis. The price did rise, but it's the price paid by the lenders, not the price charged to the borrowers. Which shows up on the graph as "going down".

    I could be misunderstanding something, but I think you made a sign error.

  16. I'm not disagreeing with Sarni. I'm disagreeing with people like Brad DeLong, who think there is an excess demand for safety relative to income.

    No sign error-- DeLong's explicit thesis is that risky borrowers have an exceptionally difficult time borrowing right now. The opposite of getting a sweetheart deal.