Overseas QE probably does not explain low Treasury yields

Between the conclusion of the Fed taper in late 2013 and this past July, the 10-year Treasury yield got cut in half to a modern era record low of less than 1 1/2%.  The yield has since retraced slightly,  I think because the market is starting to sniff out a resumption of the Fed tightening program and the economic conditions that would motivate that.

The steep decline the Treasury yield — which has presumably been driven in part by a lower term premium —  has further raised suspicion that QE was not in fact the main, or even a particularly important, driver of the bond rally after 2008.  Closely related, QE is increasingly viewed as a weak policy tool, which is about time.


Please see note at bottom of this post for description of chart

For example, here is Larry Summers in a recent blog post making the point that the Fed’s toolkit is looking pretty empty,  and challenging the efficacy of QE in particular.

I don’t think the Fed has taken on board the lesson of the three year period since QE ended.  If longer term rates had risen after QE and forward guidance ended, this would surely have been taken as further evidence of their potency.  It follows that the fact that term spreads have fallen substantially since the end of unconventional policy…  …should lead to more skepticism about their efficacy. 

I cite Summers here mostly because I agree with 90% of what he has to say.  I would be less harsh on forward guidance and think it was probably a more important source of compression of the term premium than QE was.  But the broader point about the toolkit being largely empty seems right to me.

Still, there is another possible interpretation of the recent (presumed) decline of the term premium that might allow QE advocates to weasel out of the implication of Summers’ observation.

Maybe the US term premium and thus Treasury yields are being  held down by global QE.  In other words, the reason the end of Fed QE did not force the term premium up is that it was replaced by QE elsewhere, specifically in Japan and Europe. * QE works!  You just have to avoid being provincial about it.

Economists at Goldman Sachs, for example, have published research suggesting that there is a strong global component to the term premium in many advanced countries, including the US.   If Goldman is right about that, and if the global component is both dominant and heavily influenced by overseas QE, then the QE apologists might have an out here.

In the rest of this post, I will describe what sort of data would be helpful in assessing that issue, and give a sense of why for now I doubt that global QE explains the still-low US term premium.

Let me start with the latter, which I concede is not very empirical and relies on some priors that have built up while observing the comings and goings of the various dubious arguments for the efficacy of QE.  Please bear with me: I will get to some evidence, existing and required, quickly enough.

Advocates for the efficacy of QE seem often to proceed from “tight priors”, not good,  about the role of the program.  They start with the assumption that QE is very probably effective and then assign great significance to evidence that supports that view, while downplaying evidence to the contrary.  By far, my favorite example of this is Ben Bernanke’s description in The Courage to Act of how QE1 worked.

Longer-term interest rates fell on our announcement, with the yield on ten-year Treasury securities dropping from about 3 percent to about 2.5 percent in one day, a very large move.  Over the summer, longer-term yields would reverse and rise to above 4 percent. We would see that increase as a sign of success.  Higher yields suggested that investors were expecting both more growth and higher inflation, consistent with our goal of economic revival.

The idea that Ben Bernanke gets to decide for himself how he “would see” a steep rise of bond yields in the wake of a program designed to reduce them would seem absurd, unless you brought very strong priors to the question, which people clearly do.

Another example of the same phenomenon was the purely post hoc claim that QE worked because it drove up inflation expectations in the bond market. To the extent that the rise of inflation expectations in the bond market was not associated with a similar rise in the real economy, it actually implied higher real borrowing costs in the real economy.  Moreover, higher inflation expectations driving up bond yields was not part of the original pitch for QE. But QE advocates seem comfortable cherry picking the evidence after the fact  to all ow QE to seem effective.

Here is how I would relate that gullibility (if that is not too harsh) to the idea that global QE is why US yields made a low 2 ½ years after US QE ended. Let’s start with what makes sense and then get a little more critical about what people do with that bit.

It seems plausible that investors in, say, the core of Europe have an appetite for taking duration risk and will express that appetite in the US if conditions in Europe make it unattractive to do so there.  Alternatively, American investors might return home to do this same trade under those same conditions.

But why do we just assume that the only development overseas that matters in this context is euro QE.

After all, in Europe, the demand for duration risk has arguably been forced up by the fear of deflation and chronic economic weakness.  In that environment, duration risk diversifies other financial risks.  Meanwhile, the ECB’s promise to pin the short-term interest rate at zero or less for the next several years has surely contributed to a compression of the term premium there.

It does not make much sense to reduce all that complexity to the volume of euro QE. So I too have some pretty strong priors here.  I want to avoid getting snowed by global QE, just as I quite successfully avoided getting snowed by Fed QE. But, yeah, priors are priors. And it is ideological to rely just on them.

Screen Shot 2016-09-14 at 9.14.15 AM.png

It would be really helpful if we had a global analog to the analysis that Summers et al provided on the US situation, taken in isolation, back in September 2014 They showed, in the table copied above, that between the end of 2007 and the middle of 2014, the net supply of government rates duration actually accelerated, despite the effect of the QE purchases. Please pause to internalize that. It will get you into the 1%.

The reason is that the federal deficit was generally quite large over that period and that the Treasury was lengthening the average maturity of the federal debt. The latter was effectively inverse QE, once one pierces the various veils involved here. The point was not that the deficit was bad or (necessarily) that maturity extension was unhelpful.  Rather, again citing Summers:

…contrary to much of the discussion during the QE period the stock of longer term public debt that the market has to absorb went up not down.  The amount of longer term Federal debt that markets have to absorb is now as high as it has been in the last 50 years and long rates are extraordinarily low, as are term spreads.  This calls into question the idea that price pressures caused by changing relative supplies are likely to have large impacts at times like the present when markets are functioning.

I know from personal experience that if you trot this fact out in front of QE apologists you will provoke some pretty sophisticated epistemology. The problem with you, see, is that you are not considering the counterfactual. Were it not for QE, that huge surge of government rates duration would have driven up bond yields dramatically.

I guess that is possible, but it is telling that the case for QE was never made in those terms: we need to do QE to offset the Treasury; otherwise yields will spike! It is all so ad hoc, reflecting those very tight priors mentioned above.

The much simpler story here is that bond yields and the term premium collapsed despite a surge in default-free government rates duration. That probably means that the term premium is not determined by supply, just as Mike Woodford points out modern portfolio theory implies should be the case.

I have already been over this and probably sound tedious here.  But my point is that I would love to see that US accounting exercise applied at the global level or – more to the point – at a level aggregating all the bond markets whose securities are near substitutes for Treasuries, including the US.

That would not include Greek or Spanish bonds and might not even include JGBs, if you accept the premise that the Japanese bond market is not very integrated globally.  I would keep an open mind about how to aggregate the data, but to just assert that there are no bonds left overseas and that this is why people are effectively putting on flatteners in the US is to assume facts not yet in evidence (I know of).

Consider the case of Germany, which is the poster child for the thesis that overseas QE and related government rates duration scarcity is driving developments in the US.  Germany represents 18% of the ECB’s capital key and is therefore attracting about 14.4 billion euros of ECB purchases every month, which works out to about 170 billion euros a year of purchases.  Let’s use round numbers and call that $200 billion dollars of (mostly) bund purchases a year.

Clearly, this is contributing to a scarcity of bunds available to European investors, particularly given that Germany is now running a fiscal surplus.  There is a pretty lively discussion in Europe of how the ECB will act when it runs out of German bunds to buy, without breaking its own rules about investing in negative yielding securities.  Moreover, the private demand for bunds is quite high, presumably because bund holdings diversify other risks.

But it does not remotely follow from this that the ECB buying about $200 billions of arguably near substitutes for Treasury securities is why the US term premium is depressed.  Again, we really should get over the habit of attributing every economic development in a country experiencing  QE to QE itself.  That would be religion not economics, notwithstanding Paul Romer’s recent claim that these days there is not much difference between the two.

It is arguably unfair to limit the arithmetic to Germany simply on the grounds that Germany is the poster child. We should probably include the Netherlands and perhaps even France as well. And others.  But we probably should not include Spain and Italy, for example.  Recently, their bonds were clearly NOT substitutes for Treasury securities because they included a lot of perceived credit risk.  When Draghi announced whatever it takes and backed it up with euro QE, he radically reduced the credit risk in these securities.

That is the main reason I am, fwiw, less skeptical of euro QE’s fundamental importance than of the Fed’s.  In Europe, QE involves credit risk transfer into stronger hands, unlike in the US. So it is fundamentally more important and less a smoke screen, in my view.  But on the specific issue of the effect of euro QE on the US term premium, its application in the European periphery should have put UPWARD pressure on yields here.

Having said that, it would be great to see the arithmetic here.  I think there are a lot of guys just reasoning lazily from ideological priors.  I don’t think that includes me in terms of the laziness. I have at least looked into this in the US context. But yeah, I got my priors too.  It would be nice to see the relevant data on this stuff.


Note: Selecting the dates at which the various QE and Twist programs started and ended can be tricky because there is some dispute about whether you mark the dates when the programs were hinted at, formally announced or implemented. I follow the timeline provided by Bill McBride at Calculated Risk just to bracket that whole discussion.  This post is not really about how QE reacted to various announcements around QE, although that is an interesting discussion.I call the Fed’s announcement of mortgage and agency purchases in late 2008 “QE0” because the program was not widely called “QE” at the time.  In other words, the “first” installment of QE did not come until later, when the program was broadened out to include Treasuries,  with QE1, in March of 2009. But one could think of QE1 as really beginning at the vertical line indicating QE0, especially given that the Fed was hinting at broadening the program by the FOMC meeting in December 2008.

* Another explanation might be that it is the stock of QE or even the expected stock of QE over time that drives the term premium.  But there was little movement of the actual stock (relative to US GDP) between the middle of QE3, the  end of the taper tantrum, and the summer of 2016. Basically, it remained on its record high, which makes it hard to explain the sharp movements of the term premium or yields in terms of it.

Meanwhile, estimates of the future stock of QE, say five years forward, actually went UP during the taper tantrum.  The reason is that at the same time that the Fed was presumably drawing forward the date of the taper, it was also becoming clearer that it did not intend to sell down its portfolio. So if you think QE works through the portfolio balance channel, and that markets look forward, then the term premium should have fallen during the taper tantrum period. The reason it did not, I think, is that the Fed officials were describing the bond market as “overheated”, and strongly implying they meant to do something about it, including dropping rates guidance.  To say this point is overlooked would be a severe understatement, as is evident from the very term “taper tantrum.”