Updated at bottom to include richly-deserved additional and new derision of the officially-sanctioned wrong view.
- QE does not operate as indicated on the label and is at best only marginally effective.
- It is worth understanding QE, even though the program has been wound up for now.
This second is actually the more important, and so I will skip the suspense. The reason it is important to understand that QE does not do much is that the inefficacy of QE fits into a broader theme: the cupboard of remedies to deal with the next recession is pretty bare.
Accordingly, the Fed will want to err on the side of stimulus now, in order to risk manage away from an “early” return to recession. The Fed leadership wants to allow traditional business cycle pressures to force a rise of core inflation and inflation expectations before the next business cycle downturn.
A renormalized inflation backdrop would obviously not be a cure-all for the US economy. But it would lower the risk of – and damage associated with – a return to the zero lower bound on interest rates. This is important because other remedies for the zero bound, including QE and h money, are unreliable or even implausible (in the case of h money).
Fed funds and Eurodollar futures prices fully discount this point — or something that rhymes. And market observers (analysts, journalists, etc.) seem to accept it as well.
But, leaving aside what is priced in fixed income and accepted by observers, the absence of obvious remedies for the next recession means that the odds of an imminent monetary squeeze are quite low. This in turn would seem to reduce the risk of the US economy being forced into recession in the next year or so. Recession risks more than a year forward are probably above average, but that is a separate discussion.
The QE story
What QE is even supposed to be has had a tendency to evolve over time, as circumstances have required. For example, it started off under Bernanke as a manifestation of Friedman’s quantity theory of money and ended up under Yellen as having been motivated by James Tobin’s research into portfolio balance effects.
That the Fed leadership cannot even get straight why it might work is to the discredit of the policy, although Yellen seems closer to the mark than Bernanke. In the video linked above, she discussed the “intellectual foundations” at around 40:00.
Thankfully for the Fed leadership, many on Wall Street and in academic circles remain quite deferential to the Fed’s authority. So it is possible that QE might have had some transitory, greater-fool effects on risk tolerance, which might have been stimulative in the short run. You may recall those charts showing the monetary base compared with the S&P500.
But suspending disbelief for just the moment, QE is now meant to operate through one or more of four channels.
By taking default-free rates duration out of the market and thereby reducing the term premium, which is the market price of rates duration
By indirectly signaling that the Fed would leave short-term interest rates on hold
By showing that the Fed was willing to “break some eggs” in order to meet its objectives
By raising inflation expectations (although this one was a belated rationalization)
Let’s consider these four channels in turn. The first one is the so-called portfolio balance channel, and is probably the most important argument currently advanced by QE advocates.
The first thing to point out about the portfolio balance channel is that it is extremely weak tea, compared to the strong claims that Bernanke made for QE at the turn of the century. Advocates claim that QE might have reduced the term premium in the 10-year Treasury yield by perhaps 75 bps, which is hardly a sea change. They also concede that the effects were probably ephemeral (i.e. would unwind on their own) and that subsequent rounds of QE would be less effective, because it is harder to reduce the term premium from -50 bps to -125 bps than from 25 bps to -50 bps.
So what QE advocates actually claim for the portfolio balance channel is not even that exciting. Beyond that, they are probably wrong even in their limited claims. Bond yields and term premium fell during the QE era, but more than all of this decline was achieved while QE purchases were turned off. While QE was actually being applied, bond yields tended to rise.
Event studies showed short-term impacts on bonds, but event studies are unreliable and it is telling that this method of assessing QE was invented after the need to rationalize QE was in place. More technically, event studies rely on market efficiency, which the logic of QE — such as it is — assumes away. And event studies clearly do not work in other contexts. For example, the FTSE’s short-term reaction to news about the Brexit vote ahead of the event, implied that a Leave would deliver a 15-40% decline in the market there. In the event, the FTSE went up.
Separately, as Larry Summers and colleagues have pointed out, QE did not actually succeed at making default-free rates duration scarce, because the effect of Fed purchases was totally dominated by the large fiscal deficit and by the Treasury’s policy of lengthening the average maturity of the debt. Bond yields fell, despite a surge of government duration supply, not because of the dearth of it.
That QE advocates choose never to address this point is also telling. They could argue that bond yields would have surged in response to all this supply were it not for the partially- mitigating effects of QE. But they don’t. Instead, they imply wrongly that QE made bonds scarce.
The simpler story is the older story and the mostly true story that Mike Woodford still tells: marginal efforts at duration supply manipulation – either way – are just not that important to the US Treasury market. Bond yields and the term premium are determined by other fundamentals.
The second channel is much more important in reality, but tended to get much less emphasis among QE advocates, at least until near the end of the program. Applications of QE probably did have a major effect on the term structure of interest rates by signaling that the Fed intended to remain on rates hold for longer than markets might have previously judged.
But this presents a case for rates guidance, not QE. A key historical fact here is that the bond market was perfectly complacent when the Fed began to wind up QE while simultaneously choosing to offset the signaling effect of that by strengthening its formal rates guidance. On the evidence, it was the rates guidance, and not the QE purchases per se, that mattered.
The third, egg-breaking argument may or may not have merit. I do not have a strong view on that. But the strength of it would depend on whether the Fed’s willingness to “sacrifice” was linked to a direct means of stimulating the economy. I think we can agree that it would not have been very important if Ben Bernanke has promised to hold his breath or pray the Our Father until deflationary pressures ebbed. There had to be some reason to believe that the sacrifice was in service of something that actually mattered.
The fourth argument, regarding inflation expectations is actually so post hoc as to be irritating. At the time of its implementation, QE was meant to depress bond yields via the portfolio balance channel. So the post hoc assertion that we see QE’s success in a rise of bond yields driven by higher inflation breakevens in the Treasury market is pretty rich. The rise of inflation breakevens was probably just part of transitory risk-on in the capital markets, and did not translate into a rise of inflation expectations in the real economy, as measured by, say, survey data. Accordingly, higher inflation break-evens actually RAISED the real cost of borrowing the REAL economy. If this is an argument for QE, the program certainly requires no critics.
In any event, the Fed leadership was clear when it initiated QE that it had no intention of allowing the program to generate unusual inflation. And just three years into the program, they formalized their commitment to 2% as the long-run inflation objective. Since then, actual inflation has undershot and inflation expectations in both the bond market and real economy have declined.
I am tempted to add that the current enthusiasm for h money is a concession by the activists that QE did not work. After all, if it worked, why do we need another tool? I think there is something to that. But, I guess you could argue that QE worked up to a certain point and cannot beyond that. That would not be my take, obviously. But I guess it would not be logically incoherent.
Does this skepticism even matter?
So much for the story, then. But isn’t that just dwelling in the past and score settling? Do these points have any relevance looking forward?
I think there are two forward-looking implications, one of which is generally applicable and important, and the second of which is more a personal pet peeve.
The important point is discussed at the top of this post. The Fed leadership, like everyone not in denial, realizes that QE did not act as originally indicated on the label and was marginally effective at best. Moreover, whatever the effect of the initial applications of QE, later applications would – by all accounts – probably be less effective still. This fits into the practically important point that the Fed must be careful not to derail the economic expansion now, as discussed above.
Second, and much less importantly, it is striking how unwilling the consensus among Fed watchers is to call out the Fed for their recurrent fibs around QE. As mentioned above, the instinct to defend the program by pointing out that bond yields went up is particularly egregious and borders on cultish. For me, this means that there might be a business in some independent thinking, rather than, say, the application of what we might call access economics.
Update on May 10, 2017
This comment is heavily delayed because I have just learned a bit about the ACM term premium, in large part because former Fed Chairman Bernanke has highlighted it on his blog, as the relevant measure of the excess returns investors demand for extending duration in Treasury securities.
If you put two and two together here, you get four units of hilarity, at least if you are interested in logical consistency.
As I have mentioned on a couple earlier occasions, Bernanke tried in his book (The Courage to Act) to have it both ways regarding QE announcements. To the extent that yields went down on the days of QE announcements, that was evidence that QE worked as indicated on the label. But to the extent that those yield declines were predictably transitory (because event studies are dumb), Bernanke interpreted the subsequent rise in yields as evidence that policy was working at lifting growth prospects and reducing deflation risks. Here is the quote from The Courage to Act that I have used in the past. Bernanke is celebrating the success of QE1.
A new era of monetary policy activism had arrived, and our announcement had powerful effects. Between the day before the meeting and the end of the year, the Dow would rise more than 3,000 points – more than 40 percent – to 10,428. 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.
Leaving aside that lower yields and higher yields both prove Bernanke is right, there is an additional source of great humor here that links back to the ACM term premium. Subsequent work by those same authors allows them to decompose the inflation component of the bond yield into the inflation expectation and the inflation risk premium. And it was the latter, mostly, that moved in Bernanke’s phase two.
You see, one of the ways QE was so helpful was by making people uncertain about the inflation outlook and driving bond yields up that way. This according to its main advocate, Ben Bernanke, leaving aside — again — whether this approach to disentangling bond yields is even valid.
I shit you not, although you have to put two and two together for yourself. He does not actually do that for you.