Lubos Pastor and Elisabeth Kempf on *Fifty Shades of QE* and the Implications of QE Research

Academics and central bank officials have published differing findings regarding the effectiveness of QE, and this may have implications for policy in the future.

Lubos Pastor is a professor of finance, and Elisabeth Kempf is an associate professor of finance, both at the University of Chicago’s Booth School of Business. They join David on Macro Musings to talk about their recent paper, *Fifty Shades of QE: Comparing Findings of Central Bankers and Academics*. Lubos, Elisabeth, and David specifically discuss the scope, design, and findings of their study, the theory behind QE’s impact on output and inflation, the career implications of QE research, and more.

Read the full episode transcript:

Note: While transcripts are lightly edited, they are not rigorously proofed for accuracy. If you notice an error, please reach out to [email protected].

David Beckworth: Elisabeth and Lubos, welcome to the show.

Elisabeth Kempf: Thanks so much for having us.

Lubos Pastor: Yeah, it's good to be on.

Beckworth: Well, it's great to have you on, and you have a very provocatively titled paper. And I've seen a lot of discussion about it, a lot of interest in it. And so, I had to get you on the show. And it's a very, I think, important topic, because, as you know, we've seen very aggressive use of QE, and large-scale asset purchases over the past few years. And we're now at a phase we're going to wind that down and potentially, go into QT or quantitative tightening, where the Fed will run off its balance sheet. So, how we understand that thing is very important, in your paper, you provide two different perspectives, which is very fascinating. And I think more generally, provide a discussion about incentives that researchers face. So, maybe just to start things off, give us the executive summary of what you did and what you found in the paper.

An Executive Summary of *Fifty Shades of QE*

Kempf: Yeah. I should mention that the paper is joined with our two great co-authors, Brian Fabo at the Slovakian Central Bank and Martina Jancokova at the ECB. What we do on the paper is we collected papers that study the effects of quantitative easing in the US, the UK, and the Euro area. We focus on papers that study the effects of QE on either output and inflation. And we ended up with a set of 54 papers. So, that's what inspired the title of the paper, *Fifty Shades of QE.* And we then went on to collect biographical information on the authors of these papers, and separated them into papers written by a central bank economists and papers written by academics. And what we wanted to know is, can we see any systematic differences in their research findings? Does one group find QE to be more effective than the other group?

Pastor: And indeed, that's what we found. We found that central bank papers report larger effects of quantitative easing on both output and inflation. We also find that papers written by central bankers are more likely to report statistically significant effects of QE on output. So, for example, while all the central bank papers report a statistically significant effect on output, only half of the academic papers do. And then, we also found that papers written by central bankers use more favorable language. So, for example, they use more positive adjectives, and to some extent, also fewer negative adjectives when describing their results in the abstract. So, overall, you might say that papers written by central bankers find QE to be more effective than papers written by academics.

We found that central bank papers report larger effects of quantitative easing on both output and inflation. We also find that papers written by central bankers are more likely to report statistically significant effects of QE on output...And then, we also found that papers written by central bankers use more favorable language.

Beckworth: So many questions this paper brings up. Let me begin by asking, were you surprised, or did you come into this with some kind of political economy theory that said you might find this result?

Kempf: Yeah. There's certainly a lot of different potential mechanisms that might expect you to see that central bank papers could be more optimistic in terms of their research findings. One of the mechanisms that we explore in more detail in the paper is career concerns. So, the fact that central banks, in some sense, have a skin in the game that academics don't when it comes to evaluating their own policies. And career concerns have been shown in many settings to be quite important. In fact, my previous research has been on financial analysts in the private sector. And there, you do see that career concerns can be quite important determinants of their research output. And so, one of the mechanisms we explored in more detail in the paper is precisely that. And we did find at least some suggestive evidence that perhaps central bank economists are worried about their future promotion prospects, if they find QE to be less effective, maybe they're less likely to be promoted. And we certainly have some suggestive evidence on that. But there are other mechanisms that also could explain the results.

Beckworth: Very interesting. And just looking back, not surprising, but just out the gate, it's a very again, provocative but interesting paper. And it reminded me of another paper that I read back in 2005 by Larry White, I just want to mention it briefly because I think it adds some perspective of what you guys are doing, and the title of his paper is, *The Federal Reserve System's Influence on Research in Monetary Economics.* And he goes through, and lists all the ways the Fed, in theory at least, could influence outcomes and research. Number one, they employ a lot of economists, probably the biggest single employer of all economists. He goes on to mention, they host lots of conferences, they have in house publications, working paper series.

Beckworth: Many of the economists at the Fed may sit on editorial boards, or maybe they co-author a paper with someone in a prominent journal, or there's also people who are affiliated, academics might be affiliated with the central bank. And through all these potential channels, maybe inadvertently, unconsciously, but there's this influence that seeps into the broader literature. And I want to just read a quote he gives from Milton Friedman about this, which I find fascinating. And again, very, I think, pertinent to what you guys are doing. He notes, “Milton Friedman, as reported in the Minneapolis Fed magazine article on the Shadow Open Market Committee,” this is back in 1993.

We did find at least some suggestive evidence that perhaps central bank economists are worried about their future promotion prospects, if they find QE to be less effective, maybe they're less likely to be promoted. And we certainly have some suggestive evidence on that.

Beckworth: "…maintains that since the Federal Reserve Board and its district banks hire a large number of economists in the field of money, the central bank has a sort of oligopoly on monetary opinion." So, I think your work bears that out. Let’s jump into that and maybe before we get into the details of your work, we could walk through some of the theories that are given. Your work is about the evidence, right? But I think let's just touch on, what are some of the theories for why QE should have an effect on output and inflation?

The Theory Behind QE’s Impact on Output and Inflation

Pastor: Well, there are theories why QE shouldn't have effects, and there are theories why QE should have effects. So, you can certainly find some simple theoretical benchmark models. Going back to Wallace, and Woodford, Eggertsson, and others, who show that under certain conditions, QE should not matter. So, if you make assumptions such as markets are complete, or there's only one agent that holds everything in the economy, including the essence of the central bank, there are no frictions. If you make certain theoretical assumptions, then QE will not matter. However, our evidence also suggests that QE does matter, and theories for that tend to assume something like segmented markets going back to the work of my friend, Dimitri Vayanos.

Pastor: If you have investors who have a preference for a certain part of the yield curve, let's say, then central bank buying, let's say seven-year bonds, will actually be able to move the price of the seven-year bond, because investors will not be willing to freely substitute seven-year bonds for other assets. So, I would start there. There are other channels through which QE can affect the economy. But one is since you're asking about theory, I would say preferred habitats/market segmentation type of theories, I think, are my favorites.

Beckworth: Okay. So, I've heard that theory. I've heard the signaling channel theory as well, that what the Fed is doing, what central banks do with their big balance sheets, kind of a signal, determination to keep rates low for a long time. But let me provide a little pushback, just so I want to get your take on this. So, one of the arguments, or reasoning provided is the central bank is effectively, with this portfolio balance channel, is taking duration risk off of the marketplace. So, the private sector is not holding any more. But I think that the Wallace critique would be well, all you've done is transfer risk away from taxpayers, under the federal government, but the federal government is backed also by taxpayers. So, you really haven't gotten rid of the risk, you've just maybe moved it around. But would you still contend there's still enough frictions, or incomplete markets such that QE does have an effect?

Pastor: We could debate this forever, as far as whose assumptions are more plausible. And Elisabeth, and I, and our co-authors, we just don't try to go down that path at all. We simply look at the evidence. We will look at other people's evidence on the extent to which QE matters. And we do find that it matters. So, certainly, on average, both academics and central bankers find QE to be effective, to different degrees, but they do.

Beckworth: And one last question on the theory, and this ties into your results, and we'll dive into your results. But another hot take is that QE works really well, and I completely buy this like in times of crisis. So, 2008, 2009, markets are crashing. The Fed literally is the backstop, it's a market maker, same thing in March 2020. Then, it has less effect in normal times. Did the results that you look at, did they look at both types of periods, or is it one over the other?

Pastor: Yeah, we look at both. And just like you said, we do find it early QEs, like QE one in the US, that seems to be more effective than the late QEs. And part of the reason could be that QE1 happened in a period of dysfunctional capital markets, unlike the others. But I'd say even in a modern period when markets are functioning, the theories you mentioned earlier, like signaling, could play a role. I personally think of signaling as a... it's like very credible forward guidance. If the central bank buys a lot of bonds, it is signaling that they won't be raising interest rates anytime soon. After all, it will lose money on its portfolio holdings if it were to do that. It's like forward guidance with extra credibility, even in normal times.

We simply look at the evidence. We will look at other people's evidence on the extent to which QE matters. And we do find that it matters. So, certainly, on average, both academics and central bankers find QE to be effective, to different degrees, but they do.

Beckworth: Let's move on to the scope and design of your study. You already mentioned 54 studies on QE. Tell us more about these studies. What areas, who was involved, and all the other important details?

The Scope, Design, and Findings of the Study

Kempf: Yeah. So, we started out with a search on Google Scholar and RePEc IDEAS looking for QE related keywords. And of course, we had to draw the line somewhere, right? There's an enormous amount of research on QE out there. And so, we required a couple of filters. So, one of them was, we required the studies to look at the effects of QE conducted in the US, UK or the Euro area. We also require the papers to study the effect on either output or inflation. Because these are two key macroeconomic variables that are of high interest to central banks. That means we're not going to be looking at papers that purely study the effect on asset prices. They have to look at either output or inflation. And then, in terms of the policies conducted, we focus on both large-scale asset purchases, as well as LTROs, Long Term Refinancing Operations.

Beckworth: And what are some of your key findings from this? In fact, would you call this a meta-analysis when you do this? Is that the proper name for this?

Kempf: Yeah. And I think that's certainly a byproduct of our paper is exactly a meta-analysis of the effectiveness of QE. And maybe before we get into the details of comparing the research findings, I think there are a few descriptive statistics that are also interesting. For example, we see that 60% of the authors in our sample are primarily affiliated with the central bank. So, that suggests yes, a lot of QE research does originate inside central banks. So, that gets back to the observation that Lawrence White made in the paper you mentioned earlier. And then, if you look, even just at the raw data, comparing differences and research findings, it's quite striking when you compare the group of central bank papers with the non-central bank papers.

Kempf: Just to give you one point of comparison, we see that 100% of the central bank papers, so these are all papers with at least one central bank author find that the effect of QE on output is statistically significant. And in comparison, only half of the academic studies do. So, already, and then we see similar striking differences, also, when you look at the estimated effect on output and inflation. And most of the main analysis is then about establishing that these differences are statistically significant. And indeed, we will find that most of them are.

Pastor: I can fill in some details on what Elisabeth said. So, she mentioned that we find different magnitudes of estimated effects for central bankers and academics. So, if you look at the overall average, like peak effect of QE on output, averaging across all studies. So, at the peak QE increases output by about one and a half percent. Okay. That's what our meta-analysis finds. But the average effect is actually 1.75%. So, 1 and 3/4 of a percent when you look at central bank studies, and it's only 1% precisely when you look at and academic study. So, it's a difference between 1 and 3/4 versus 1%. If you look at medians instead of means, if you're worried about outliers, it's a difference between 1.5% among central bankers, and 1% among academics. So, in both cases, the differences are not only statistically, but also, economically significant.

Beckworth: What about inflation, similar magnitude differences for inflation?

Pastor: Indeed. So, for inflation, the effects are actually comparable. The peak effect of QE on the price level is almost 1.5%, 1.4%. If you look at central bankers, you're looking at 1.8% peak effect. If you look at academics, it's more like half a percent. So, actually, the magnitude of the difference between central bankers and academics is even larger.

Beckworth: Yeah, very amazing, these numbers, the spread. So, it's both statistically significant and economically significant, the differences that you're finding here. Were you guys surprised by these differences in terms of magnitudes?

Pastor: Yeah, we didn't know what to expect. We had some ideas. But in the end, we let the data speak. I'd say the magnitude is larger than I expected.

Kempf: Yeah. I would agree with that, yeah.

Beckworth: Yeah. So, I was looking at, I believe this is from table one that you're reporting these numbers. And you guys also report a standardized effect, I guess, where you look at 1% of assets as a percent of GDP, is that correct? And see what… then we have these comparisons. And you see similar differences. But one thing that struck me, though, was on the inflation. I just want to run these by you. And correct me if I'm not reading this table the right way. But if you look at the standard peak effect on inflation, so from a large-scale asset purchase equivalent to 1% of GDP, it leads to a .24% increase in inflation based on central bank research. But a .05% for academics, is that correct?

Kempf: Yeah, that's right.

Pastor: And just to clarify, it's a .24% increase in the price level.

Beckworth: Price level. Okay.

Pastor: Yeah. And that's almost five times larger than the .05% number that you've mentioned for academics.

Beckworth: Yeah. The inflation number for academics is really small, given that size of purchases. And then, the other thing that's fascinating was the cumulative effect. So, we take that standardized approach from the cumulative effect. You have .18 for central bankers, and a -.01. So, if I'm interpreting this correctly, there is no cumulative effect on inflation from QE if based on the academics, or slightly negative, but it's close to zero effect on inflation. Is that fair?

Pastor: That's right.

Kempf: Maybe we should also clarify what we mean by cumulative effect would be indeed, the effect of the end of the time period studied. So, you're right, it would suggest that there is a peak effect. But then, it's temporary.

Beckworth: Yes. I ask this because right now, as you know, big debate going on in the US, and maybe to a certain degree in Europe, as well is inflation. Inflation is a hot topic right now. And the Fed is beginning to taper, beginning to do QE. And I guess the question if it's inflation that you're worried about, you're not as concerned about the employment side of the mandate, but inflation. If I look at the academics’ numbers, I'm not sure that tapering and balance sheet reduction is going to make all that much difference for inflation. But if I look at the central bank numbers, it does imply something meaningful can happen. Is that a fair interpretation?

Pastor: Yeah, I'd say so. I would also say that, if you look at the main drivers of inflation today, I wouldn't put QE anywhere near the top of the list of the primary determinants. There are so many things driving inflation nowadays, that what happens to QE… It wouldn't be at the top of my agenda, as far as my future inflation forecasts.

I would also say that, if you look at the main drivers of inflation today, I wouldn't put QE anywhere near the top of the list of the primary determinants. There are so many things driving inflation nowadays, that what happens to QE… It wouldn't be at the top of my agenda, as far as my future inflation forecasts.

Beckworth: I share that view entirely. I look at things more closely tied to fiscal policy and the pandemic. But a lot of the conversation seems to be we need to do this because inflation is running hot, and I understand the Fed wants to maintain credibility going forward, even if what we've seen behind us is more pandemic related. Okay. So, those are some of the headline interesting numbers. Also, discuss the different methods that were used. So, you looked at the differences in approaches to estimating these effects. Can you speak to that?

Kempf: Yeah. So, we looked at whether we can find differences in the types of methodologies used, and we see that central bank studies are more likely to use DSGE models versus VAR models, so these vector autoregressive models. That said, we can control for model choice. We still continue to find these differences. So, model choice doesn't seem to be explaining everything. But you could certainly hypothesize that the more modeling assumptions you have, the larger the range of possible outcomes. So the bottom line is, we do see differences in the methodology used, but it doesn't really seem to be explaining the differences in research findings.

Beckworth: Just to be clear, you're saying that the central bankers rely more on DSGE models, and academics rely more on the empirical VAR approach. Okay.

Kempf: Exactly. Yeah.

Beckworth: All right.

Pastor: On average.

Beckworth: On average, on average.

Kempf: On average, yeah.

Beckworth: And so, I guess, if you're being cynical, and I'll be cynical, you guys remain objective, because you're being academic. I'll be cynical, I'm the host here. They could build assumptions into these DSGE models that would give them an outcome that surprise, surprise, QE is very powerful. Now, again, I'm not saying they do that consciously. But we'll talk about this… incentive structure could be there. So, the model is designed in a way that it gives the outcome that they're after. Is that the cynical take?

Pastor: It is somewhat cynical, but in defense of central bankers, you might argue that somebody using a VAR model can also build in the relation they want by bringing in the 17th econometric variable on the right-hand side-

Beckworth: Fair, that's fair.

Pastor: ... to deliver the outcome of one.

Beckworth: Absolutely. And I use VARs. And there's a lot of art that goes into that as well. How do you impose restrictions on the data, and such. So, there's a little bit of hand waving on both sides. But the implication is DSGE is relied heavily or more heavily by central bankers. Okay. That's very fascinating. All right. So, we have these really interesting results. So, we've talked about the findings, the differences in methods. What about the implications for careers? Because I know you also went in, and you looked at the effect on career advancement based on what they found in their papers. So, first off, it sounds like a lot of work to put this data set together, a lot of a lot of intensive labor here, but what did you find in terms of professional advancement for these economists?

The Career Implications of the QE Research

Kempf: Yeah. So, we went ahead and looked at the career paths of the central bank authors in our sample. And so, converted their job titles into a numerical rank that allows us to capture how high up are they in the organization. And what we found was that, in particular reporting, larger effects of career on output was associated with more favorable career outcomes. To give you like a magnitude, finding one standard deviation, larger effect on output was associated with half a rank large improvement. Half a rank would be like moving from economist to senior economist, so a non-trivial degree of correlation there.

Pastor: And these are not causal results, it was a bit instead, these are associations. You could think of various reasons why such correlation might arise that would not be causal. For example, suppose somebody is a really hard worker, and that's why they end up being promoted, and that somebody cares so much about their central bank that they're willing to make adjustments in the results in ways that they believe might appeal to their bosses. In this case, it wouldn't necessarily be any genuine incentive. It would just be just a correlation, not causation.

Kempf: Yeah. One additional split that we did was by seniority. And we saw that this correlation was actually quite a bit stronger for more senior central bankers. So, that would suggest that perhaps their career concerns are even stronger than for junior central bank researchers.

Beckworth: So, these would be senior staff members at central banks. Okay.

Kempf: Yeah, exactly. Yeah. And we actually do, when we compare, so consistent with more senior central bankers maybe having even stronger career concerns. We see that actually, in the comparison with academic studies, the difference is larger when there's a more senior central bank author on the team. So, that will be consistent with what we see also in the promotion results.

Pastor: Yeah. And also, in this period of career concerns, we took a special look at the German Bundesbank, because unlike pretty much all other central banks, the leaders of the Bundesbank have expressed some skepticism about quantitative easing in the past. And if that's the case, you might conjecture that Bundesbank researchers might be more cautious in expressing support for the positive effects of QE. And that is what we find, we find that Bundesbank researchers actually tend to find weaker effects of QE on output, even compared to academics. So, not only compared to other central bankers, but even compared to academics. But I want to caution that these results while large economically, they're weak statistically, because we have only, I believe, four papers written by Bundesbank researchers in our sample. So, you have to take this with a grain of salt. Essentially, we have a bunch of examples of evidence that is all consistent with career concerns mattering. But we don't have a smoking gun that proves that this is the mechanism through which that generates our results.

Essentially, we have a bunch of examples of evidence that is all consistent with career concerns mattering. But we don't have a smoking gun that proves that this is the mechanism through which that generates our results.

Beckworth: The German example, the Bundesbank example was very fascinating, kind of a wow, that's consistent with what you're finding elsewhere, given what we know about the Bundesbank. And you can always rely on Germany to give you a good counter example. And I'm thinking back to an AEA meeting I went to. This was probably early 2000s. And they were revisiting the question of the 1970s inflation, what caused it? And so, John Taylor, and another co-author got up, and they said that was the Fed. It was easy monetary policy. Then, Alan Blinder got up in his paper, "Oh, it was supply shocks, it was the high oil prices." And then, the third paper was a German who got up and said, "Hey, we didn't have it in Germany."

Beckworth: And for me, that was a slam dunk case. Okay. That's two against one, the Germans came in to close the sale, I'm convinced. And this is what seems to be the case here. The Germans provide a nice counterpoint that's consistent with the findings you're finding, albeit from a different sign, different direction. Now, you also did a survey of research directors at the central banks, and looked into what they do, how they manage might play into these outcomes as well.

Pastor: Yeah. So, this is where the Central Bank of Slovakia, the National Bank of Slovakia was very helpful. At the Slovakia Central Bank, we've been interested in how to arrange research. And so, this survey was partly designed to help us understand how other central banks approve research papers for circulation, and things like that. And in part, it was very helpful for us. So, we reached out to a little over 50 heads of research at the leading central banks. So, all OECD countries, all European Union countries, the ECB, all the Feds. And we asked them a few questions about who proposes research topics to researchers? Is it, are these topics assigned by management? Or are they determined by mutual agreement by the researcher and the manager?

Pastor: We asked them questions about who approves these papers before public circulation, how these papers are reviewed. And if I were to summarize the results in a sentence, we do find substantial managerial involvement in research production. Happy to talk about more detail, but yeah. So, unlike in academia, where it's not that the dean comes to Elisabeth and tells her, "Elisabeth, you need to change your econometric specification on page seven." That just doesn't happen in academia. But it could very well happen in central banks. And frankly, that's probably a good thing. I actually think that as a researcher they are doing tremendous service to the profession, and they're helping out other researchers. But it makes you wonder, are there other ways in which this influence could manifest itself?

Beckworth: Yeah. You have to be careful if you are a central banker. What you say matters to markets. I remember Ben Bernanke, I believe, in the early 2000s, gave a speech about the yield curve. It was beginning to invert. Maybe this is late '90s. I forget where it was, but it was inverting. And he gave a speech where he said, "Don't read too much into it. It's just the term premium. Don't think the market is saying anything about a future recession."

If I were to summarize the results in a sentence, we do find substantial managerial involvement in research production.

Beckworth: And lo and behold, a recession did come. But I've always said to myself, "Well, what could he say?" Even if he believed it was signaling a recession, he can't be a responsible central banker and freak the world out by saying, "Run for the hills, a recession is coming, the yield curve is telling us this." And I think in the same light here, a good research director has to be careful of what's being put out so you don't rattle markets, you don't create confusion, you want a consistent message. But the drawback to that is that it can inadvertently lead to certain outcomes being avoided, even if it's not intentional.

Kempf: And I think it's certainly an interesting potential perspective on central bank research to view it as a tool of monetary policy, perhaps, which is then different from the goal of really producing independent research, where it's not ultimately taking a stance on “this is what central bank research should be about.” But I think, yeah, it speaks to this idea that maybe there could be reasons for them to produce research that tends to find monetary policy is effective.

Pastor: Yeah. And just to build on that, it's very important to distinguish research that is unbiased. The researcher's objective is to seek the truth, and research that is viewed as part of your policy, which then by its own nature, will have to be biased. And I very much hope that the overwhelming majority of research coming out of both central banks and universities is of the former nature. Otherwise, credibility issues arise.

Beckworth: Yeah, for sure. Now, Elisabeth, you've done work on this area, the conflict of interest in doing your research in other fields like finance, I believe you've touched on World Bank. Do you think you want to speak to that? How do these lessons or insights apply to other areas, other professions?

The Conflict of Interest Story

Kempf: Yeah. So, I think this idea that there's other interests at play, in addition to just producing accurate objective research is definitely there in a lot of different settings. In my prior work, I've looked at financial analysts, and saw that career concerns can matter for the assessments they make. Also, political partisanship can influence their views of economic conditions. But I think they're perhaps more similar to the central bank setting. There are examples that recently also received a lot of media coverage that can speak to these issues. One example would be a World Bank paper that found that following distribution of World Bank financial aid, there's an increase in deposits in financial havens, suggestive of part of the financial aid of the World Bank ending up in the pockets of the country's elite. As you can imagine, quite controversial finding for the World Bank. And according to reporting by The Economist, that paper initially passed an internal peer review, but then was held up at higher levels by higher World Bank officials. And so, I think this is just one example how institutions face these tensions between the ideal of producing independent research, and at the same time, protecting the interests of the institution.

Pastor: Yeah. And since you brought up the World Bank, we actually had a more recent example from the World Bank just three or four months ago. As you may remember, the World Bank recently discontinued its Doing Business report, which it had been doing for many years. And this was done in response to essentially, an accusation in a report commissioned by the World Bank, which suggested that the 2018 report was manipulated by the leadership of the World Bank. In particular, in favor of China. So, China was supposed to drop from, I think, it was number 78 to 80 something, 85, in the report, but ended up not dropping.

Pastor: And people have connected this to the possibility that the World Bank was in the middle of its capital raising campaign, China's cooperation was essential. And in fact, the current head of the IMF, Ms. Georgieva, was essentially accused by a report of instructing staff to help China achieve a higher ranking. Of course, she rejected the accusations. She basically said that she only asked the staff to double check and triple check the numbers. Now, you have to wonder if somebody is a chief economist, and they know what's going on, and they've been asked to double check and triple check the numbers, what are they going to do? This is a hard one. The IMF's executive board reached the decision that there was no conclusive evidence that anything bad has happened. And Ms. Georgieva continues running the IMF. So, this is more of an illustration of what may or may not be a bias in a setting that is also coming from... this as a public policy institution, not a central bank, but I think you see some similarities.

Beckworth: Okay. What about policy implications going forward? I alluded to this earlier, but we're now heading towards a stage where there's going to be less QE, maybe even central bank balance sheet run offs. They're going to get smaller. Are there any implications for that, or any other areas you think going forward based on what you found?

Implications Moving Forward

Pastor: We don't have any implications for whether QE should or should not be wound down. This is for central bankers to decide based on other criteria. We hopefully have some implications for how research should be conducted and evaluated, at least on some sensitive topics. Look, 90%, 95% of central bank research is not research that evaluates the central bank's own policy. So, our results are really only about that small subset of research where the central bank is conducting self-assessment effectively. And I think simply, were pointing out that there seems to be an issue there, and hoping that central bankers will take a look, and decide what they think is best.

Kempf: I think raising awareness of this potential issue is one of the things that we're hoping the paper will achieve at various levels of the organization.

I think raising awareness of this potential issue is one of the things that we're hoping the paper will achieve at various levels of the organization.

Beckworth: Yeah. So, going back to the comparison between academic and central banker, your deans don't come to you and say, "Hey, fix this paper, edit this here." But what they do is, they'll come to you and say, "Hey, I want you to make a plan for the next year." At least when I was in academia, I had to have a yearly plan, or a plan to work on this paper, maybe get this one published, teach these classes. And at the end of the year, I'd sit down with whoever my chair who was above me, and I'd have to answer, what did I accomplish? What did I not get done? But it was a way for me to be honest, to have a check on my productivity. And I have a colleague who's recommended something similar for central banks, have them set their plans up for the year.

Beckworth: If they don't hit their targets, have a report, why they didn't hit their targets? Why was inflation high this year? Why is the federal funds rate at a different level than they had expected? But given it like in a year-end annual report, and assessment of why things are the way they are relative to where they thought they were, even if these were conditional forecasts. So, I think this goes in the spirit of what you're pointing to as well. It's just more reflection on how things are done at the central bank, hopefully, leading to better central bank policy moving forward.

Kempf: I think, yeah, also perhaps involvement of more external reviewers could be beneficial if we had to speculate.

Pastor: Yeah. Just to build on what Elisabeth just said. If you look at that small subset of central bank research that has this self-assessment element to it, where they're evaluating their own policy there, I think engaging some independent external counsel could be useful. Here's an analogy with pharmaceutical companies. We generally don't want pharmaceutical companies evaluating their own drugs without any additional evaluation being taken. We want them to evaluate their own drugs because they have the best information about their own drugs. They have the best people to do it. And they care about the reputation very much like central banks do.

Pastor: But in addition to them evaluating their own drugs, we also want others to come in, and provide some external validation. And I think the same thing could be happening in the central banking world. We want central banks to continue evaluating the effectiveness of QE and other programs, because they are in an excellent position to do that, they have the right people, the best information, et cetera. But perhaps, if something truly important is at stake, then that some external counsel could be helpful.

Beckworth: But didn't we see some of that already? So, for example, I know the Fed did this Fed Listens tour before it adopted its new framework of average inflation targeting. And I think they stopped in Chicago. This conference may have been in Chicago even. And they had people get up and present papers on different approaches to monetary policy. And a lot of these were academics who came in and presented papers. There were also some labor groups that came in and stuff. But would you count that as a step in the right direction, or do we need more than that?

Pastor: It's a step. And remember, you're talking about the Fed, you're talking about, arguably, the most advanced central bank in the world. In our research, Elisabeth and I look at dozens of central banks. So, surely, there are many other central banks that could learn from what the Fed has been doing.

We want central banks to continue evaluating the effectiveness of QE and other programs, because they are in an excellent position to do that, they have the right people, the best information, et cetera. But perhaps, if something truly important is at stake, then that some external counsel could be helpful.

Beckworth: All right. Well, let me share with you a policy implication that I took from your paper. And again, this is not where you're going and I acknowledge it. So, listeners, this is David Beckworth speaking here. But given the huge differences in the numbers in table one, we talked about the effect of QE on output and inflation. Big, statistically significant, and economically important differences and magnitudes. To me, it doesn't give me much hope in trying to understand what's going to happen, again, with tapering, with balance sheet run off. We don't have Taylor rules for asset purchases, right?

Beckworth: And these studies, I think, underscore why there's not a good agreement on what a good number is. How does the Fed guide its decision making? I know they have their own internal models that will tell them how much, but for example, why did they buy 120 billion a month for a while, right? Why were they buying that with their large-scale asset purchases? And based on the meta-analysis in your paper, well, the answer depends on whether I look at an academic’s research, or if I look at a central banks research, which just gives me a little bit of concern about moving forward when we wind down that balance sheet. It's good to have a sense of the proper magnitudes of what will happen when we make these changes.

Beckworth: So, that's just my, David Beckworth, hot take off of your work. But it should give us pause, and I do worry about that. Well, let's move on from policy implications into the reception of this paper. Again, this is a very provocative paper. I know you've presented it in many places before. I was looking up all the different places, and I have seen pictures giving a talk, and I even saw one person's comments on your paper at a conference. But how has this paper been received? And first, before we get to the central bankers, let's talk about other academics and other people. What's your sense of the reception you're getting from this paper?

*Fifty Shades of QE* Reception

Kempf: Lubos, do you want to start on the academic or-

Pastor: Well, we have received helpful feedback from colleagues. We even have some colleagues who have had interactions with central banks. So, we've learned about the processes that inside various central banks, even from academics, from former central bankers, so that was helpful, at least to me. We also got a lot of helpful feedback in discussions and in peer review. So, we had referees, just like everybody trying to publish a paper. So, I think our feedback from central bankers was probably more interesting.

Kempf: Yeah. We certainly received, as you can imagine, quite a bit of interest from central banks in the paper, and we had several heads of research reaching out to us with thoughts and comments on the paper. And according to them, it had triggered a lively debate, I'm quoting here, inside the institution, which was really, as we said, what we were hoping to achieve with the paper. We had the opportunity to present the paper in front of central banks, the ECB was one of them. And I think, yeah, the paper has definitely gotten better also in the process, and we're grateful for having gotten the feedback also from the central bank side.

Pastor: Just to give you an example of the kind of feedback that we found helpful, we learned from one of the former heads of research who volunteered in email that there could very well be a channel, an alternative mechanism to the one that we investigated in the paper. Remember, we talked about career concerns quite a bit. It could just be that people have different priors. So, for example, suppose you have two graduate students, and one of them believes having grown up that policy interventions matter, and the other one doesn't believe that. Well, which of them is more likely to go and work in a central bank? Presumably, the one who believes that policy is valuable.

Pastor: And then, that person starts with a different prior compared to the other one who may end up in academia. And then, not only do they start with different priors, these priors can get reinforced by people around them. They present to colleagues, et cetera. If they have like-minded colleagues, then they not only start with certain priors, those priors evolve in the same direction. Whereas, if an academic starts out skeptical, and presents to skeptical colleagues, they may end up elsewhere. This is just one of several things that we actually incorporated in the paper, thanks to this helpful feedback from central bankers.

Kempf: And understanding these mechanisms, I think, is actually very important. We just spoke about policy implications. For that, it's really important to understand how much of this is driven by internal review process, career concerns versus these differences and priors that Lubos just mentioned. We're hoping that there's going to be more research on this in the future, trying to figure out the importance of these mechanisms. And actually, getting back to the request that we've gotten, there were actually quite a few researchers interested in our data set, either trying to replicate it, or doing additional analysis on it. And we are providing this, and so we're hoping that there's going to be more research on this in the future.

Beckworth: So, going back to your alternative explanation for your findings, basically, selection bias, people going to work in the central bank have a predisposition to finding QE to be very strong and effective. And that may be a part of the story. Another point before we end the show today, is your QE studies that you examined, they went through 2018, is that correct? You stopped at 2018. So, do you think anything different would emerge if we threw in the past two years? And I bring this up, because there were a lot of rule of thumbs for QE, even before the pandemic. Like Bernanke, I believe said 500 billion with lower .2% on a 10-year Treasury yield, these rules of thumbs. If I look at table one, and I plug in the doubling of the Fed's balance sheet, almost doubling of the Fed's balance sheet, and I get these really big magnitudes out that at some level seem a little big or large, did QE really make that much difference in output and inflation? Maybe it did. But what do you think would be different, if anything, if you had added in the last few years, where we've had really, really large-scale asset purchases?

Kempf: This is quite speculative, of course, but we have looked at it, is this disagreement between academics and central bank researchers is larger for the earlier versus later rounds of QE. And we actually do see the disagreement tends to be larger for the first few rounds. And so, if that trend would continue itself, that would suggest that as you add, maybe additional studies, maybe the differences would be smaller. Maybe that's convergence towards one way of looking at this, one particular type of model to use. But yeah, but this is quite speculative. I don't know, Lubos, if you want to make a prediction.

My prediction is that if we were to extend the data set sometime in the future, and if people were to look at this new wave of QEs in 2020, that they would find substantial effects.

Pastor: I'm happy to speculate any time. That's what I do. Well, look, one other thing I'd add, I agree with Elisabeth. I'd add that if we were to look at what the Fed did in March of 2020, and not just the Fed, but basically, all major central banks relaunched their asset purchases at that time in response to the COVID shock. I suspect we would find strong effects, or whoever analyze the QE would find strong effects. And the reason I believe that is that that period seemed at least somewhat similar to 2008. In that there was a temporary financial market disruption. It wasn't quite as big as in 2008, but there was some. And the Fed's announcement, for example, is they would also get into corporate bonds. For the first time, I think it was very important in calming markets down. So, my prediction is that if we were to extend the data set sometime in the future, and if people were to look at this new wave of QEs in 2020, that they would find substantial effects. Now, whether academics or central bankers would find larger effects is a different story altogether.

Beckworth: Yeah. I share that view. But again, it goes back to the crisis mode. QE has a large effect, because the central bank is actually doing something meaningful. It's the market maker of last resort that steps in. It would be interesting to maybe look at 2021, after we've gotten through the worst part of that, what has QE done? And I would be surprised if we didn't find these same results that you find in your paper, the academics, and central bankers maybe diverge. And I guess my point is, in normal times, the debate, in my mind, at least is in normal times, outside of market stress, what are the differences we find? Well, with that, our time is up. Our guests today have been Elisabeth Kempf and Lubos Pastor. Thank you so much for coming on the show.

Kempf: Thank you, David.

Pastor: Thanks, David, for having us.

Beckworth: Thank you.

Check out Conversations with Tyler, and subscribe to Conversations with Tyler on your favorite podcast app.

Photo by Olivier Douliery via Getty Images

About Macro Musings

Hosted by Senior Research Fellow David Beckworth, the Macro Musings podcast pulls back the curtain on the important macroeconomic issues of the past, present, and future.