The stock markets are having a series of huge sell-offs lately, and investors are bracing for the bear market. Outside of the financial market, though the real economy is still doing reasonably well, people can't help but wonder when the next crisis will come.
More importantly, when the next crisis hit, do we have the tools to deal with it?
Central banks and many economists are confident that Quantitative Easing (QE) can be a reliable crisis-fighting tool. However, according to a recent research, the confidence in QE might have been a misguided one, as previous studies may have overstated the effectiveness of it.
In their working paper "A Skeptical View of the Impact of the Fed's Balance Sheet," economists David Greenlaw, James D. Hamilton, Ethan Harris, and Kenneth D. West challenge some earlier studies that concluded QEs have a significant economic impact. The major issue is that when those researches used simple event studies to quantify the impact of QE, the result might be biased.
Why might event study not be an excellent way to study the input of the QEs? Economists often use market reactions in the treasury yields shortly before the announcements of QEs, or other related events, to quantify the 'impact' of the asset purchases programs. But according to Greenlaw et al., this event study method might suffer selection bias, as economists only focus on the days with dramatic market movements.
Instead of focusing on some short time periods, they suggest a complementary observation technique: the authors are the Reuters News to double-check the result of previous research results based on event studies. They record all the days that the 10-year US treasury yield had movement larger than one standard deviation, and then they check the Reuters News report that day to identify the "source" of market movements.
If the Reuters identified the Fed's action or speech is the major reason for the market movement, it is marked as "Reuters Fed Day". Then, they add up all the market movement on "Reuters Fed Days" during the period of QE 1, QE 2, and QE 3, and use it to "measure" the impact of the Fed's unconventional monetary policies.
Here is the summary of all the 10-year treasury yield reactions at the "Reuters Fed Days" (dark blue line).
The result shows that 10-year treasury yield is cumulatively lower in all 'Reuters Fed Days' during the period, as one would expect from a monetary easing policy.
However, most of the fall in yield concentrated in the period before QE 1; During QE 2 and QE 3, the cumulative fall in l0-year yield actually got smaller, which means that the yield is trending higher even with all the Fed actions. This put in doubt the effectiveness of the event study method, as economists will miss out all the subsequent market movements that might be related to the QE.
In this particular case, the "effect" of QEs might have faded away very soon after it is implemented.
The research also distinguished the reason for yield changes on "Reuters Fed Days'" between "balance sheet news-related" and "interest rate news-related".
One the one hand, the yield change related to interest rate remains negative and stable during the period; one the other hand, the cumulative yield change on days dominated by news about the Fed's balance sheet fell before QE1, but the yield is trending upward since then and turned positive before QE3. Again, this is not a picture one should expect as QE should lower the 10-year yield instead.
This is why the authors make a bold claim that that interest rate policy is a better way to ease monetary policy, as compared to asset purchase programs as the effect is well known and stable.
However, the Greenlaw et al.'s claim is still hotly debated. As one of the authors of those original event studies, Joseph Gagnon of PIIE argued that Greenlaw et al. might have overstated their case. Gagnon argues that to use an event study method to measures the effect of a policy on an economic variable, including the "alternative" method Greenlaw et al. used, certain critical assumptions of event studies are needed: "(1) the policy in question was not expected prior to the news events, (2) the news events are the only times the market changed its expectations about the policy, (3) those expectation changes occurred entirely within the event windows, and (4) nothing else happened during the event windows to affect the economic variable of interest."
These assumptions are clearly violated in programs other than QE 1, hence those results might not be really indicative. As for the measured economic impact of QE 1, Gagnon argues that Greenlaw et al.'s result is actually in line with the one he gets in a 2011 paper he coauthored. So, the debate is still a lively one and the dust are far from settled.
If the advantage of interest rate policy is that the effect is better-known, how about we use the QE the same way as rate cuts. This is indeed the idea of an innovative proposal Gagnon co-authored with Brian Sack of D.E. Shaw Group.
What do you mean using QE the same way as a rate cut? You might ask.
Here is the fundamental point of their argument - according to Gagnon and Sack's summary of the academies studies, an asset purchase amounted to 1.5% of US GDP (about $300 billion) has estimated to have the equivalent economic effect as a 0.25 ppt rate cut by the Fed.
So, if the Fed wants to stimulate the economy with QE, they should purchase about $300 billion of long-term assets and treat it the same as a further 25 bp rate cut.
This suggested approach is in sharp contrast to the way QE programs had been implemented, which have either announced a more massive amount of purchases all at once in the beginning (as in QE 1 and 2 in the US) or declared a much smaller amount that continue indefinitely on a monthly flow basis (QE 3).
Gagnon and Sack argue that the primary purpose of asset purchase programs is to substitute standard interest rate policy, as the short-term interest rate was constrained by the zero lower bound (ZLB) and further cuts might not be desirable anymore.
If this is the case, a QE program that is designed to have a closely similar effect to a regular rate cut should be more natural for the public. It should be easier for them to understand the Fed's reaction function, compared to that of a surprising and significant in size program like the QE 1.
Also, many economists agree that the stimulative impact of QEs is related to the stock of long-term assets held by the central bank, the so-called stock theory of QE effect. Hence, as Gagnon and Sack argue, the Fed can use the level of its long-term assert holding as a proxy policy tool of the interest rate, which is stuck at the ZLB.
The two researchers suggested that the Fed should engage in QE exactly the way they cut interest rates - announce a $300 billion purchase after an FOMC meeting, then purchase the stated amount of long-term assets (preferably Treasury) during the interval before the next meeting.
One advantage of the proposed QE structure is that it has embedded "inertia," a characteristic that one-time large-scale purchase program like QE1 doesn't have.
"Inertia means that when you decide to move the rate, you tend to move it again in the same direction," Gagnon explained in an email reply to EconReporter's inquiry.
Having "inertia" in the monetary policy is important "in part because economic variables may have the same property of tending to move in the same direction for a while and in part because you don't want to move too fast."
A monthly flow-based program (like QE 3) also has inertia embedded, as "a monthly flow rate is like a promise to cut the rate by a tiny amount each month until some goal is reached," Gagnon said in the interview. But the difference is that their suggested new design allows the Fed to have the discretion to decide the appropriate action at the time of each meeting. Arguably this gives the Fed more leeway to adjust the policy direction in real time.
Also, a monthly flow-based program that the Fed implemented had a smaller purchase amount each period; for example, the monthly purchase amount of QE 3 is equivalent to a 0.07 ppt rate cut, what doesn't fit the norm of 0.25 ppt or 0.5 ppt rate changes the Fed usually practices.
So, maybe we can simple restructure the way QE programs are used and solved the potential unreliability of QEs, as Greenlaw et al. claimed.
Unfortunately, there is still one major problem to be solved. If the asset purchase program is used as equivalent to a rate cut when the time for rate hike comes, does it mean that there will be two interest rate level prevailing? One is the nominal interest rate that the Fed announced, another is the nominal rate minus the stock theory of QE effect.
"I agree it is more complicated to explain [to the public if there are two interest rates]," Gagnon agrees that this might be a potential issue that the Fed will face.
In fact, this "two interest rates" issue "is already causing some confusion" in the ongoing tightening cycle of the Fed, Gagnon said. As he observes, "some FOMC participants talk about remaining QE assets as an element of policy ease and others ignore that."
The Fed's System Open Market Account (SOMA) is holding near $4 trillion Treasuries. If comparing this to the less than $500 billion holdings the Fed had back in Sept 2008, that is a $3.5 trillion difference. As a $300 billion purchase is equal to a 25bp, the $3.5 trillion holding might have the impact of a 2.75 ppt "implicit rate cut", making the Fed fund rate still negative.
"[Fed Officials] are going to have to think more clearly about this [two interest rate problem]" Gagnon said.
Disclosure: I/we have no positions in any stocks mentioned, and no plans to initiate any positions within the next 72 hours.
I wrote this article myself, and it expresses my own opinions. I am not receiving compensation for it (other than from Seeking Alpha). I have no business relationship with any company whose stock is mentioned in this article.