Jim Hamilton on Econometrics, Energy Markets, and Low Interest Rates

Why are oil shocks such a good predictor of incoming recessions?

Jim Hamilton is a professor of economics at the University of California at San Diego. Jim’s research spans a number of areas including econometric, business cycles, monetary policy, and energy policy. Jim joins the Macro Musings podcast to discuss some of his work.

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: Our guest today is Jim Hamilton. Jim is a professor of economics at the University of California at San Diego. Jim's research spans a large number of areas including econometrics, business cycles, monetary policy and energy markets. Jim joins us today to discuss some of his work. Jim, welcome to the show.

Jim Hamilton: Glad to be here.

Beckworth: Well, we're glad to have you on. As I mentioned earlier, I have been following you since grad school, I work through your textbook on Time Series, your research papers, and I've been following your work as well as your blogging ever since. So it's a real treat for me to be able to sit down and chat with you. I want to begin like I do with most of my guests and ask how did you get into economics?

Hamilton: Well, when I was a freshman and sophomore, back in 1974/75, I got the impression that my economics professors really didn't understand what was going on in the economy. They really couldn't explain why we had high unemployment and high inflation at the same time. And I thought, well, those are important questions. If they can't figure it out, maybe I can. So I ended up going to graduate school and a couple years into graduate school, I figured out these questions are harder to answer than they might seem. But at that point, there it was. So today, I'm an economist.

Beckworth: Now, how did you get into Time Series because you have very famous textbook, Time Series Analysis that many of your listeners might have worked through themselves. So what was your journey into time series work?

Hamilton: Well, a lot of those discoveries were just taking place as I was leaving graduate School, and a lot of interesting stuff going on. Of course, there was disappointment with some of the traditional econometric approaches. And this seemed very promising to just describe the predictability of different Time Series. And so, as I was following the new developments, I realized, well, we really need a textbook now that's describing all these new tools that have become available. So I wrote the textbook, because that was something that was really needed at that time, I thought.

Beckworth: Yes, for those who haven't been exposed to Time Series Analysis by Jim Hamilton, it's the Bible of Time Series if you're a grad student. I know you've sold many, many copies around the world. How many pages is it? Remind us.

Hamilton: Over 800 I think.

Beckworth: Yeah, 800. So it's a very intimidating book when you first get it, but once you get into it, you're grateful for the detail in the working out of examples. Least I was, I needed all the help I could get, so. It's the Bible of Time Series Analysis. So it's great to talk to the author of the Bible Time Series Analysis. And you've done a lot of work with Time Series Analysis, I'm want to jump into some of the related work that touches on some of these tools you developed. Let's begin with your research on oil since you're well known for that, oil shocks, their effect on the economy. So let me begin with a loaded question. Sure has a long answer. But that is how important are oil shocks to recessions?

Hamilton: Well, I think there's no question that they are one contributing factor in at least some historical US recessions. When I first started talking about that, in 1980, I told people that six of the seven recessions we'd had in the United States since World War II up till 1980, were preceded by a big increase in oil prices. And I told people that they said, well, that can't be true. Well, today we've had 11 recessions and 10 of those were preceded by big oil price increase. I think the consensus today is that, as I said at least some of those recessions, the oil prices were one contributing factor.

Beckworth: What was the one recession that they did not go up prior?

Hamilton: 1960.

Beckworth: Okay. Wasn't aware of that. Interesting. Now, one of the difficulties in drawing a causal link between these surges in oil prices and recessions is identifying a true supply shock. So tell us about that. How did you in your work identify oil supply shocks versus some kind of response or endogeneity of oil to other economic conditions?

Hamilton: Well, prior to 1973, it was kind of straightforward. In part that was because the regulatory environment, the Texas Railroad Commission, would take responsibility for altering the production allowed from the state of Texas, a very important producer at the time, and would change production allowables in a way to basically keep the price from moving much from month to month. And when it did move, it was typically in response to a dramatic historical event. And that's true, to a lesser extent since 1973, either again a series of pretty dramatic events, the OPEC embargo, the Iranian Revolution, the Iran-Iraq war, the first Persian Gulf War, where I think you can make a pretty strong case that these were largely exogenous developments to the US economy. And some of the events we saw happen economically afterwards, I think should be attributed to those geopolitical events.

Beckworth: Okay, so a series of natural experiments makes it easy to measure the effect of these oil supply shocks. Has the effect of oil become less important over time and therefore are these oil shocks less important over time? Has our economy become less dependent upon oil?

Hamilton: Well, in most of the models you write down, a key parameter is the share of expenditures, total expenditures that goes to petroleum products. And that number has certainly changed over time. It goes up when the price of oil goes up because demand is relatively inelastic. And those changes over time and the expenditure share, account for some of the changes over time and in the apparent sensitivity to the economy.

Hamilton: And also there's been a chronic decrease in the importance of manufacturing for the US economy and that may insulate us somewhat. But I think we had a test of the hypothesis that it doesn't matter anymore in 2007 and 2008, when there was a big run up in oil prices and what we saw happen in that early phase of the 2007 and 2009 recession, looked a lot like some of the other historical recessions, namely the auto sales, particularly sales of the less fuel efficient vehicles were manufactured in the US really took a hit when gasoline got up to $4 a gallon. And you saw some of the similar dynamics there in the early part of the recession as another downturn. So I think people who said, it doesn't matter anymore for the US economy were clearly overstating the case.

Beckworth: Yeah, I remember those days very clearly. Pumping gas into the car $4, some cases higher than $4 a gallon. Since you're a student, and you follow closely the oil markets and energy markets, what has been your sense of fracking? How's fracking played into this discussion?

Hamilton: Well, it's really made it pretty phenomenal contribution to world oil production. I was surprised at how big a contribution it could make. It started out as a very expensive process and there was enough learning by industry that they were able to bring the effective cost of producing way down. Now, I don't think they got it down to 40 $50 a barrel, but now they were back above that, I think it's going to make a substantial contribution.

Hamilton: It's also very different from the historical production where for deep sea production, for example, you're talking about a very, very long process between the initial investment and actually getting the oil out. Whereas fracking is much more like a manufacturing operation really. And I think that will give some more stability to oil prices. Just because US production can ramp up when prices are high and will come down again, when prices are not sufficient to cover the costs. So it's a big factor. But also, we shouldn't understate the contribution that continues to be made in the Middle East. Iraq production and Iranian production increased substantially since 2014. And that was as big a factor almost as the fracking revolution in helping keep oil prices low.

Beckworth: I was thinking along the lines of that first point you just mentioned and that is will we ever see sharp spikes in oil prices again, because of fracking because it is so easy to ramp up production... I'll rephrase that, relatively easy to ramp up production relative to traditional oil drilling. So do you think the future holds in store smaller oil shocks?

Hamilton: Yes, I think it's certainly a mitigating factor.

Beckworth: Okay.

Hamilton: But we also have to pay attention to the total volumes. So we could get, I don't know, half million, a million barrels a day in relatively short timeframe. But if you had a much bigger disruption than that from, say, one of the main Middle East countries, I think you'd still see a significant move up in oil prices. So obviously not as big a move up, as would be the case if we didn't have that backstop.

Beckworth: Yeah. What about this idea of peak oil? And I know on your blog, you often would reference people who discussed this. And this kind of goes back to this discussion about economists will often have with environmentalist or others and as economists talk about the importance of the price signal incentives, prices go high enough it creates the incentive to seek out and find new ways of producing oil. Fracking being a good example of this. What is your take on the whole peak oil argument?

Hamilton: Well, obviously the people who were saying that the high price can produce incentives that will bring some additional production, turned out to be right. Of course took quite a while for that to happen. So prices were very high in 2008. And you really didn't see a supply response there.

Beckworth: Fair point.

Hamilton: So I think it's a combination of the incentives that economists reasonably focus on and what technology can allow. I think we were fortunate to discover, well, you can get oil out of the ground this way. It might be true, it might not have been true. And if it weren't true, I think we would have faced a pretty significant challenge, particularly with the growth in consumption from the emerging economies, China, and others. So I guess, I worried that the whole peak oil debate got over simplified. Incentives being emphasized by one side and the fundamental limits of geology by the other. I think both incentives and fundamental geology matter. And I think economists should be open to that. Obviously, we've had a long history now of always finding more oil. That doesn't mean that's how the future is always going to hold and how long fracking will continue, and how much additional production you could get with that technology from US and other countries? I don't know the answer to that.

Beckworth: Very interesting. So the truth lies somewhere between those two extremes.

Beckworth: Yeah. I mean, at least in the short run, elasticities may not be very high because limits of understanding technology, but over time, we see bigger responses. Let me go to one other area in this whole oil shock literature and research and that's the view that the link between sudden shocks, sudden oil price surges and recessions can in part be tied to the Fed's response to the rise in oil prices. So there's the Bernanke-Gertler-Watson paper in 1997. Others might point to, for example, like the European Central Bank in 2008, raise interest rates. The Fed didn't raise rates, but the Fed definitely was set on its hands between April I believe, in October that year I was worried about inflation coming from commodity prices.

Beckworth: And then 2011, the ECB raised interest rates twice, in part because commodity prices were again surging. So somebody looked at those and say, "Look, commodity prices were important, but they also confuse the Central Banks. Central Banks who target inflation, misread the inflation tea leaves and they hit the brakes too soon. I mean, where do you come down on that? How important is that story?

Hamilton: Well, on the specifics of the Bernanke-Gertler-Watson calculations, if you look at the details of how they develop that argument, they were calculating the counterfactual policy for the federal reserve that called for keeping interest rates at something like 4% through the 73/75 oil shock, and it's interesting question whether that's even feasible to keep it that low with all that was going on.

Beckworth: Okay.

Hamilton: But even if you thought they could, their specification was assuming that the main effects of the oil price increase all take place within six months. And the relation in the data suggests it can take significantly longer than that for the multiplier effects to work out. If you just redo the Bernanke-Gertler-Watson exercise with longer legs, even if you impose that strong counterfactual, you end up concluding that really the opposite of their conclusion that oil shocks would have still had an effect even if you have very, very aggressive monetary policy in response.

Hamilton: Now as far as the ECB, in the '08 and again in 2011, as far as '08, I think one key data point to recognize there and also in the other recessions is you can look at the composition of spending. Even within autos, it wasn't just that auto sales were going down, it was sales of a certain kind of auto were falling, namely, the less fuel efficient vehicles, which tended to be the ones manufactured in the United States-

Beckworth: Interesting.

Hamilton: ... were the ones whose demand was really falling. Now, I think it's straining things to say, well, that's all because of interest rates, particularly interest rates in Europe when I'm talking here about sales of autos in the US. So yeah there's something to that, the broad commodity price swings are clearly a factor. But I think there are details both in the economic response, and also details in the geopolitical events. So '07/'08 is not as clear an example as some of the others. But if you take the onset of the first Persian Gulf War, this was very dramatic night day thing, one day Iraq was there and Kuwait and two of the major world oil producers are now out of commission. And I think you can make a pretty strong case that, as I say in at least some of the recessions, there was a direct contribution of the oil prices themselves.

Beckworth: Okay. More recently, we had this sudden decline in oil price 2014 and 2016 quoted remarkable decline. Was it a 30% decline? It was a really large number. I forget the exact numbers, but I know you looked at that, and one of the interesting questions that come out of that is we didn't see and expected surgeon spendings. Some people were expecting to see the opposite that sudden oil price decline would lead to a surge. I know you've written on this, so what is the explanation for that experience?

Hamilton: Yeah, well, I've always argued it would be a nonlinear relation. What happens when gasoline goes to $4 a gallon is people say, "Well wait a minute, maybe I shouldn't be buying that new car I was planning on." And because the capital and labor for manufacturing those kind of cars is very specialized and hard to readjust. The result is a decrease in total output. Now, on the other side, if gas prices go down, it's not that people go out and buy a second car all of a sudden. And furthermore, there are dislocations in other industries, the oil producing industries when oil prices go down, and Baumeister and Kilian have an interesting paper where they looked at that episode 2014 to 2016. And noted that falling investment from the oil producing sector was a significant drag on GDP at that time.

Hamilton: And again, that kind of capital doesn't costlessly relocate, you can't take an oil drilling rig and start to make ice cream or whatever with it.

Beckworth: Right.

Hamilton: So I think the evidence has always been pretty clear that it's a nonlinear relation, but decrease is not going to stimulate the economy in the same way that an increase can bring spending down.

Beckworth: What about the time horizon dimension? So I could imagine oil prices drop dramatically. I know myself, I'm not going to quickly go out and buy an SUV overnight. It's not in the budget. But over time, maybe on the margin if I see year after year, maybe my elasticity changes over that horizon. Is there any evidence for that, that over the longer horizons, consumers tendencies to respond to oil prices get stronger if it's a persistent drop?

Hamilton: Well, it's certainly the case that eventually the economies kind of adapt and the fuel economy of the outstanding fleet, of course changes over a period of many years. I think there's a difference when you're looking at business cycle frequencies. You're asking, what's different this month, two months ago, and then these very abrupt changes in spending make a big difference. When you start to ask questions that are over a 10 year horizon, I think most of us would want to use a neoclassical kind of model to think about that, and take into account that the long run possible flexibility and substitutions you could make are incessantly greater than what you'd see over a horizon of a couple of months. But certainly the longer run elasticity of demand for oil is is much greater than the short run.

Beckworth: All right, very fascinating. So, key takeaway is we've got to keep our eyes on oil still. Still an important both for the business cycle horizon and longer horizons.

Beckworth: Well, let's move on to your research on interest rates. And you had a co-authored paper not too long ago, titled The Equilibrium Real Funds Rate: Past, Present, and Future. And you address the number of questions in that paper about the real interest rate. And let me begin with a question. Does the real federal funds rate converge to some long run trend over time?

Hamilton: I think certainly not. In a lot of our standard models it does. But that's really just an assumption of the model. And it's very hard to see corroboration of that. The longer you go back in time, the more different you see short term real interest rates become. They were much higher at the end of the 19th and early 20th century than they were later in that century. They're much lower today than they were 30 years ago. And you see a huge negative real rates during World War I and World War II. It's hard to look at that long run record and say there is this our star as it's sometimes called in theoreticals and it's a constant over time.

Beckworth: Yes, this has come up a lot in recent debates about monetary policy, which in our star or became the intercept in the Taylor rule that natural rates and John Williams, for example, has argued that it's fallen and John Taylor would say, "Uh, I'm not so sure." But recently John Taylor has been more flexible about it. But you do see a sudden drop change right around the time of the crisis. So there's been this trend that many people have observed real rates going down. But even within the crisis, there was a sudden drop in real rates. And of course, the question is, is that just a cyclical development? Is it more structural? Is it long term? And what your research suggests, if I understand correctly is that it could be a structural change, is that right?

Hamilton: Well, I think there are a variety of contributing factors. With the crisis, there was a flight to safety. And that continues with us today, which accounts for part of the low real return on things like US Treasury bills. I think there have been some demographic changes worldwide, which lead to higher saving rates, changes in income distribution, you can point to factors that have been underplay for several decades, and particularly became stronger after the financial crisis. But as I say, apart from the very recent period that the longer you go back in time, the more persuaded you become that this thing was never a constant.

Beckworth: Okay. So it's been falling since the 70s or the early 80s. But what you show in your paper is even that is a relatively recent phenomenon. That other periods, it's going up, it's going down. You mentioned World War II was negative. So there's no clear trend.

Hamilton: Not that I find, no.

Beckworth: Yeah. Okay. Now, what was the response to this paper? I mean, is that considered consensus? Is there any feedback on that?

Hamilton: Well, I think it did generate a lot of interest. Particularly for practical policy purposes, this is a really critical question for, we're going to follow Taylor, 'Well, what the heck you said number [inaudible]

Beckworth: Right, right.

Hamilton: My view has been that maybe we want to not have as much faith in something like the historical Taylor rules, economy guide policy. And I think we shouldn't be basing our interest rate decision in the current environment on an assumption that we know what that number is. I think we don't know it, and we need to look for other evidence. So basically, what's going on with inflation directly, or unemployment, the output gap? I think those are more important factors to be looking at right now.

Beckworth: Do you think it's worthwhile trying to estimate like a time varying natural interest rate like John Williams has done or others to help guide policy?

Hamilton: But Lombok and Williams have that and that's widely referred to. But as I say, no, my bottom line is I think we're looking at in the wrong direction. You can clearly see right down there is a theoretical proposition that if you don't know for sure what our star is, maybe you shouldn't be basing policy on that. I think that also makes a fair amount of intuitive sense.

Beckworth: Right. So stick to the basics, keep it simple, look at inflation, adjust rates accordingly. So and that actually ties in nicely to the paper we'll talk about next on Fed's QE programs.

Beckworth: What about the secular stagnation story? That also has received a lot of play. Larry Summers and others have talked about a persistent shortfall in aggregate demand. What was your conclusion on that story in your paper?

Hamilton: Well, we were looking at various observable factors that would influence the real interest rate over a period of a decade or whatever. And we found very little support for that conventional view that may be output growth is the single most important factor. So this secular stagnation story we have, slower output growth. And that's why we're going to see a lower average real interest rate. I don't think what anyway, we couldn't find much support for that in the data and looking at a long period of data for the US and looking at cross-section experience across different countries, they really is not that strong a correlation between the growth rate and the real interest rate.

Beckworth: Okay, which then suggests the secular stagnation story really isn't as important as some people make it out to be. You've also done work on term premiums and risk premiums, estimating those. There's been something on the New York Fed and the New York Fed has a website, you can download term premium estimates across the entire term structure of interest rates. Board of Governors has one, you've done your own estimates of it. Are those are useful exercise as well and helping inform policy? I know, for example, that the Federal Reserve will say, "We're trying to tweak that term premium." What do we gain from trying to estimate the term premium?

Hamilton: Well, in principle, it's a magnitude that we should be able to measure. And you can define it as the difference between what the yield on a certain security actually is, and what the yield would be if you just have a risk neutral forecast of the subsequent short term interest rates. And in principle, you can calculate a forecast. We know how to do that. We can can form forecasts of future interest rates and come up with a number. The New York Fed data that you mentioned is very popular because it's online. It's constantly updated, and lots of people want to have a number like that. The problem is, once you get into the details of it, you quickly persuade yourself that, "Gee, when I use a different forecasting model, that's also pretty plausible, I end up with a very different term premium." And that's a pretty big question. So it's easy to get lulled into thinking, well, it's the New York Fed website I could download it.

Beckworth: Guilty as charged. Yeah.

Hamilton: And if you go through the exercise, well, suppose I tried to use this way of forecasting rather than, for example, how you treat the unit roots, how persistent you think the interest rates are, whether you've got some small sample bias in that. Those those ends up making a big difference. And in addition, when you look at formal tests of how good these models are at predicting, say the excess return holding bonds, what you often find is that in sample, those models seem to look very good, but out of sample they do really poorly and turns out there's econometric reasons for that. I think the literature has this little bit of a false sense of confidence of their ability to predict excess bond returns. Yeah, you can find something in the given historical subsample that seems to be there, you can use that to construct something you call a term premium or risk premium. But I think any of those numbers coming from any of the sources you mentioned, including my own, should be taken with a grain of salt.

Beckworth: Now we had John Cochrane on and I've talked to him about this and he says, "Look at this they're huge on these estimates." So need to have a sense of humility when using them. What about market based derivatives of interest rates? So for example, the tips and the breakeven inflation forecasts that come out of those. Do you find those to be useful and informative?

Hamilton: Well, sure you have to look at tips when you're thinking about inflation expectations. So there are a series of issues with those having to do with liquidity of the asset. And whether there's some risk premium involved in inflation, valuation as well as others. And those are real issues. I think it gives us a good ballpark assessment, though. And I certainly look at those numbers all the time to get some inkling of what's going on. I think one other caveat that I've mentioned there is that in our standard macroeconomic models, we have this variable we call expectations of inflation. And it means something clear in the context of that model. It's a little less clear, when you try to relate that to reality. One of the things you have to come to grips with there is whose expectations are we talking about?

Hamilton: Even if we could get the expectations of bond traders. That's one number. Is that necessarily the same as the expectations of people who set prices? And are those the same as the expectations of consumers buying the products? Certainly when you look at direct survey data, the answer you give to that question is no. The people that are asking answering Michigan survey about what expectations they have for inflation, have a fundamentally different answer from the implicit Wall Street numbers. So my answer is yes, I certainly think tips are a very valuable source of information. But I worry about taking our models too literally, as to just what those numbers tell us.

Beckworth: Okay. Well, let's move on to another related area and that's the effect of the Fed's large scale asset purchase programs or QE, Quantitative Easing as it's more popularly known and you just recently had another co-authored paper made a big splash, widely discuss Twitter blogs, newspaper stories, lots of great publicity surrounding them. And one of the reasons is because it takes such a contrarian view about the ethicacy of the Fed's QE programs, the papers title is A Skeptical View of the Impact of the Fed's Balance Sheet. So right there in the name, you're very clear where this paper is going. So tell us about this paper and what you try to do in it.

Hamilton: Well, I'm not sure it was as controversial as you make it out. A lot of the people we talked to including a couple of discussions from the Federal Reserve were more or less acknowledging that, "Yeah, we do have some of these doubts." So a big part of what we were addressing was some of the evidence for why many people think that the Fed's program of large scale asset purchases had a significant effect on interest rates. And an example of the kind of evidence that had seemed pretty persuasive initially to some people, was when the Fed announced an expansion of quantitative easing or large scale asset purchases in March of 2009. Within 30 minutes of that announcement, you saw the yield on a 10 year US Treasury security fall by 50 basis points, really dramatic market response. And in such a short interval of time, you're almost crazy to say, "Well, gee, the Fed wasn't the factor moving that interest rate." And the indicated conclusion is maybe that the program, it just that one announcement, could give us 50 basis points, maybe the whole program did quite a bit more than that.

Hamilton: What a whole lot of studies have done is taken that date and some others and looked at the responses of longer term yields either on that day or up within a 30 minute window or whatever, and come up with a number that may be all tool, the program contribute at least 100 basis points, brought the 10 year yield one percentage point lower than it would have been in the absence of those programs.

Hamilton: Well, so what we did it our paper was say, "Well, how do we pick those particular dates?" So on this announcement in March, there was a Fed announcement, there was a dramatic market move. Suppose you look at all Fed announcements, not just a few select ones. Well, for example, at the next FOMC meeting, the Fed made another statement in response to that this time, the 10 year yield went up and they really didn't announce anything good. They didn't announce any new program, but something about what happened there seemed to make interest rates go up. So what was it? Well, one possibility is that people were really expecting more in March when they thought that March analysis would've meant something more that the Fed was going to do when they didn't seem to be doing it, next meeting, then they said, "Okay, well, this isn't that that big a deal."

Hamilton: Another possibility is that one of the reasons the market had that response it did in March was because the Fed and its staff have some information about the economy that's different from the information that private sector economists have. So now, that's not saying the same thing is that the Fed has better information, that the Fed knows, knows everything that private sector does, and then some. But if the Fed just has some information that the private sector doesn't, it's a totally natural reaction when the Fed makes a new statement to say, "Well, gee, what do they know that I didn't, and maybe the economy's in worse shape than I thought." And that could have been part of that March response. And then in the subsequent reversal, it may be that the Fed was communicating "Well, okay, it looks like the economy may not be doing quite as badly. And so that's part of why rates go up."

Hamilton: So anyway, when you look at the whole history of all the statements that the Fed makes, you end up coming away with a very different conclusion, which is that, on average, the Fed statements, were not bringing interest rates down at all. Another way we looked at the data was to look at news reports. So okay, let's take days when the interest rates moved a lot, what did a new service like Reuters end up saying was moving the market that day? Sometimes they attributed to something the Fed has done or people think the Fed is going to do. Sometimes is attributed to economic news or whatever. And there again, when we looked at the role of seemingly Fed-driven news, we did not find an indication that the quantitative easing programs were a major driver of interest rates. So as a result of that, we emerged with some skepticism that we really don't think we know these numbers as well as maybe some of the consensus had earlier claimed.

Beckworth: So just to be clear, the consensus view based on these events studies was roughly one percentage point cumulative decline from the QE programs. Is that right?

Hamilton: Yeah. And it is based not just on event studies, but I think event studies are a key and important to that.

Beckworth: Okay, so there you mentioned there's some vector autoregression, some panels data studies as well. But so the idea is that the Fed's QE programs cumulatively lowered, and this is the 10 year Treasury, I have to be clear, lowered those yields about 100 basis points or one percentage point. There's a couple of observations first, relative to the total decline of the 10 year Treasury yield, that's still relatively modest, even if that were the right number, the 10 year Treasury was like 5.25, five and a quarter or so prior to the crisis, and then it got as low as 1.4, 1.37, if I remember correctly, so there was a pretty steep decline.

Beckworth: Now, I guess, one percentage point is non trivial, but there's a lot of other things going on driving rates down during that time as well. And I know what they're trying to do with advanced studies tried to isolate that. So with the event studies, you mentioned the paper, they're looking at a very narrow window around that announcement. So, trying to isolate causality very carefully. One thing that they use to motivate this theoretically are different channels. And maybe you could speak to those. So I believe the most important channel they used to motivate these purchases was the portfolio channel, but there's also a signaling channel. So maybe you could explain to our listeners how those were supposed to operate.

Hamilton: Well, first in a plain vanilla asset pricing model, there shouldn't be necessarily any effect, even if the Fed buys huge volumes of some security. And the reason for that is that our standard asset pricing model says that the price of an asset is determined by the risk adjusted discounted present value of its return over time. And if there hasn't been something changing that fundamental pricing kernel or the fundamental's generating the flow of return from that asset, what the standard theory predicts is that if the Fed wants to buy a lot more of this stock. If that drove up the price save a 10 year bond, other people would say, "Well, I don't want it at that price," and there would be no effect on the interest rate, it would just change the ownership of the bond. So you actually have to work to tweak in some way the standard model to think these programs would do anything.

Hamilton: Now you've mentioned several of the possible ways that can happen. One is the portfolio balance idea, which is the claim that certain holders of bonds have a particular preference, for example, an insurance company may want to hold 10 year bonds because that's their business model and they keep holding them even if the Fed is buying a lot and then you do get an effect on the price from that or maybe some banks or financial institutions want to hold Treasury securities in particular, because of their liquidity services or because they are helpful and in as collateral for certain arrangement or regulatory factors they face. And again, if you have some inflexibility of the demand that when you alter the available supply that's available to those private agents by the Fed buying a lot more of it, you could affect the interest rate. So, those are the kind of models that we use to explain theoretically how it would have an effect. But those particular mechanisms are fundamentally limited. You think, "Well, okay, maybe I can change the interest rate a little bit." But it's not like you can make it any arbitrary number, just by having the Fed buy more and more of this stuff.

Hamilton: Now, you also mentioned the signaling channel. And there are a couple of ways that that goes. One, I mentioned earlier, which is that the Fed may be partly signaling its information about the economy. If the Fed has some information, the private sector doesn't, it can communicate those. For example, with this announcement, things look really bad. We're going to buy a trillion in securities. And as far as that's the channel, it's not a particularly helpful one, if everybody's just getting more glooming, because the Fed is gloomy themselves. Another channel that's people sometimes point to is that the Fed was really trying to communicate its future policy stance. He was trying to communicate what it was going to be doing with interest rates.

Hamilton: Now that we're away from the zero lower bound, and now that they have some choice, and he was trying to communicate that was going to keep those interest rates low, say in 2018, lower than they might have otherwise, and that somehow buying trillions of dollars of long term securities commits them to that, in a way they wouldn't be as committed if they had not acquired all those long term assets. So that's another way that the signaling channel works, where the claim is it's what really mattered was the Fed, convincing people that they were going to go very slow at raising rates once we did start to get the Fed funds rate back above zero, and that it wasn't the Fed physically holding the securities so much as the fact that by buying the securities they were able to more convincingly tell people that this was their future policy plan.

Beckworth: Yeah, there's a lot to unpack there. A lot of interesting things going on. I want to go back to the skepticism or the concerns theoretically with the portfolio balancing channel. And a Michael Woodford in his 2012, Jackson Hole paper is a huge paper. And one of the things he does in there is he speaks to this channel, and I remember one of the things he mentioned is and I think this is what you're getting at earlier, so correct me if I'm wrong. But he mentioned "Look, if the Fed buys a bunch of Treasury's long term securities off the market..." I mean, the argument is that they're taking duration risk, interest rate risk off the marketplace. And that's going to encourage these investors to rebalance portfolios and risk premiums will go down across the economy, a whole bunch of assets will be affected.

Beckworth: But what he observes is look, all the Fed has done is acquired those long term securities and put it on its own balance sheet, they haven't disappeared. And ultimately, who backs up the Fed's balance sheet? It's the public. It's the Treasury and in turn the public. So you really haven't gotten rid of that duration risk, you simply have moved it around. And collectively if you think of the US economy is one big balance sheet for the entire US economy is still there. Is that kind of the idea you're getting at earlier that you haven't fundamentally changed the discounted present value of those securities?

Hamilton: Well, yes, but that's one possible avenue whereby you would have a potential effect. So let's follow that through. So there's still some risk. And now that risk has been essentially transferred not just to the Treasury, but to the Federal Reserve and to make it concrete, it means that the Fed might have to tolerate more inflation in certain states of the world than they would have otherwise wanted. Because that's the only way to manage the outcome in that state of the world. And so to the extent that happens, then that could have real effects because you have a different forecast of inflation in that state of the world compared to the one where if the Fed was not absorbing that risk in that way. So that's one of the possible mechanisms involved in this signaling effect.

Hamilton: But no, what I was talking about was just take a standard asset pricing model, which says that a long term bond if you have no default risk, as we assume treasury bonds don't have, the price of the long term bond should be basically the discounted value of future short rates where the discount is with the pricing kernel. And if you haven't changed the pricing kernel, then you're not really changing the fundamentals of that, unless it's working through the signaling path that you're actually changing the expected future short rate that we talked about. And this portfolio balance is one way that you possibly could change the pricing kernel.

Hamilton: Now, Woodford also in other papers has spelled out models where you could actually have that effect. And it operates from this idea that there's some restrictions. There's some reason that a certain subset of the population just can't buy a certain kind of bond or won't buy a certain kind of bond. And that's the essential friction whereby the pricing kernel gets changed, potentially as a result of changing the supply available to the private sector. But the bigger point I was making was all these mechanisms are kind of subtle. They're not first order. They're not the thing that jumped out at you from a standard model. And so that's another reason why we could have a little skepticism. It's very different from the traditional control of short term interest rates. We all agree the Fed has tools to control the short term interest rate. It's another question whether the Fed has tools to control something like a term premium, and if so, exactly, what is the nature of that control or what's the nature that influence and how hard can you push it? That's the question in mind.

Beckworth: And so am I hearing from you then that even if it is there, the magnitudes aren't going to be very big. Is that right?

Hamilton: Well, that's certainly what suggested by theory. Now how big is big? That's something we want to look at the evidence for and that gets us back to, for example, thinking about these events studies and well, okay, exactly how do we interpret that evidence?

Beckworth: Right. And your empirical findings suggest that those effects aren't going to be that big. That was the takeaway I got from your paper.

Hamilton: Yeah, I think the evidence that they are big is not as persuasive as it was.

Beckworth: Okay.

Hamilton: ... or try to be.

Beckworth: Another issue that came up and I know we actually discussed this in the blogosphere some time ago, but you mentioned in the paper as well, is that during the period when the Fed was doing QE2, maybe QE3 as well, the Treasury was expanding the average maturity of its debt. So the Fed was in this race against the Treasury. So on one hand, the Fed is trying to shorten the average maturity of the debt that's outstanding, the marketable Treasury securities, on the other hand, that the Treasury is trying to extend it and clearly Treasury is a much bigger player than the Fed is and so the Fed in some ways may have been fighting a losing battle against the Treasury. Was that effect very important in this discussion?

Hamilton: Well, yeah, it is, if you're thinking this channel is one of the key ones. The portfolio balance channel that's supply of long term assets available to the private sector. And if the Fed is taking some of those assets away, and the Treasury is replacing them with the same number of new ones, you've had zero net effect on the private sector holdings of those and so it's another reason why you might be skeptical that LSAP has implemented in conjunction with what the Treasury did and what the Fed did, that LSAP maybe didn't have as profound an effect on long term rates as some people suppose.

Beckworth: Yeah, I always like to remind people making these conversations, people who think QE had this large effect, is that the Fed's holding the marketable Treasury securities never exceeded more than 19, 20%. As you've written about, it's actually declined when the Fed began to sell off some of its Treasuries in 2007. But it says some QE1, QE2, QE3 was just returning to the level of Treasury holdings that it had before the crisis, but it never became like a majority holder of Treasury securities. I think that fact is often lost.

Beckworth: Let me move to Japan though, because maybe it's easier to make a stronger case for Japan. But even there, you and your fellow authors express skepticism, Japan has been buying up a lot more of the government debt. It has pegged the 10 year Japanese Treasury, I think at close to zero percent. But you're skeptical of that result as well. Is that right?

Hamilton: Well, no, actually we offer that as one example. So there are two ways you can imagine implementing LSAP. One form is the way the Fed did it, which is to say we're going to buy so many billion every month, a quantity kind of target. What Japan has played with is a yield target, we're going to buy whatever it takes to achieve this number, which 1 trillion doesn't do it, then you go to 10 trillion, that's in principle what you're committing to. Now, that's a bigger bomb to throw at the problem, but it also brings some bigger risk. Now our assessment is it kind of seems at the moment that Japan has succeeded in keeping its rates from going up at the same time the US rates and the German rates are climbing, so maybe that is one option that you just say, "No, I'm really going to do this and stick with it, whatever comes." Of course, that takes more and more nerve as you-

Beckworth: Right.

Hamilton: ... billions you've committed to this game, it takes more and more confidence that these issues we were talking about whether this really makes any difference are important or not. And so, we give us another example, counterbalance to the Japanese example, the Swiss National Bank, which was again not targeting quantities but targeting a price in this case it was the exchange rate with the euro and the Swiss National Bank was saying look, buy and buy and buy euros, just to achieve this exchange rate target. And that turned out to be quite a few euros and they lost faith. They lost confidence. And when they did the franc appreciated tremendously and they had a huge capital loss.

Hamilton: So we take that as the combined impression from those two episodes as a cautionary tale, but getting back to Japan. For years, a lot of the Western economists were saying, "Well, you just didn't do this enough, just throw more at this," and scolding Japan for not getting the inflation rate higher. And I think when the US and Europe went through their own version of this, maybe the western economists developed a little more humility about just how successful these kind of programs can be. Yes, Japan's interest rates are very low. But where's the payoff in terms of growth in Japan, in terms of inflation? I would not point to Japan as a case study of here's how a Central Bank can fix these kind of problems if they're willing to buy enough securities.

Beckworth: Yeah, that's a great point. And I'm going to come back to that. Some people would argue inflation is a little bit higher than it would have otherwise been. Some point to real indicators is improving. But I too am like you, I look at Japan and I'm not sure that payoff has been that great. But I want to go back to Japan, whether this is working, because in the paper and I know you mentioned, Japan has kept its yields low even as other major economies have seen their yields go up slightly at the 10 year horizon. But you also mentioned in the paper, though, at least I thought you did, you mentioned that is it really clear why it's been at zero? In other words, they were already at really low rates, they've had this prolonged period of low inflation. Do we know for sure that it's the Bank of Japan's quantitative easing program that has the 10 year yield at zero or would they be there anyways?

Hamilton: Well, yeah, and that gets back to the point you were making earlier. Clearly, this was a long term episode, not just in Japan, but in the US and Europe, where the bad news persisted longer than people were anticipating, and with the economy remaining sluggish, that's something that's going to bring inflation down. Now, with inflation coming down and the economy so sluggish, all the Central Bank say, "Oh, we've got to do something. So let's ramp up our quantitative easing policy." So you see this correlation. Quantitative easing is being conducted around the world. At the same time, interest rates are all coming down. And so maybe the one cause is, but the question is, which causes which? Is the QE causing rates to go down? Or is it the weak economic performance causing low inflation, low rates and low policy response in terms of QE?

Hamilton: I think one very important episode on this if we go back to the US, QE1, I mentioned that in March 2009, when that started, you saw this 50 basis point drop in the 10 year yield right away. Well, within a couple months that 10 year yield was back up higher than it had been before the program started, it ended up actually 100 basis points, the end of QE1, the 10 year rate was one percentage point higher than it had been the day before the Fed made this initial dramatic announcement. Now, you can argue, well, if they hadn't had QE1, maybe would have been 2% higher instead of 1%, that these were other factors moving the rate up, and would have moved it up even more if it had not been for the Fed. At a minimum that I think makes the point that however important you may believe QE was, there are other factors that are even more important, and I think that basic news for the US economy and in particular, weakness of the European economy is as that later into 2011 and 2012, I think that was a factor in that long term decline in rates around the world.

Hamilton: After all, it is a world capital market is falling up on something you said about what fraction of the Treasury debt the Fed holds and pointing out it was never the major player. It's even less so when you think about the world capital market-

Beckworth: Fair point.

Hamilton: ... securities are just one part of that. And you just somehow got out of the whole supply of everything... interest rate. I think everybody would have to agree that other factors are surely making a key contribution to interest rates in every country over these periods when Central Banks were trying unconventional policies.

Beckworth: Yeah, absolutely. Absolutely. In the minutes we have left, I want to throw out one other explanation for why QE may not have been as effective as hoped by its proponents and its advocates. And this goes back to Paul Krugman's 1998 paper about Japan from the Brookings paper and then also Michael Woodford and Gauti Eggertsson. And then Michael Woodford resurrects his point in his 2012 paper. And this comes to the distinction between a temporary monetary injection versus a permanent one. So Michael Woodford notes in the 2012 paper, for example, he looks at the first round of QE that the Bank of Japan did in 2001 and 2006. And ultimately, it was reversed. So it increased and went down. There was some permanent increase in the monetary base, but it's more of just like a continuation of the trend that existed before.

Beckworth: And he looks at and says, "Look, this was ultimately temporary inflation. It didn't take off. The public was correct in believing the Bank of Japan would never do that." Paul Krugman says, "Just increasing the money supply isn't enough, you've got to believe that it's going to be permanently increased." If you look at the QE in the US, for example, the Fed is going to shrink its balance sheet. It will probably be higher than it was before. But a large part of the permanent increase in the Fed's balance sheet simply that trend growth, again comes from currency demand. And you highlight this in your paper. There's some questions about the final size. But the point is this some of this work by these folks have mentioned really stressed the importance that if you're going to inject money, it has to be perceived as permanent, or you're really not going to get much bang for your buck. And I'm wondering if you think that has any bearing on these very modest, at best, results for QE?

Hamilton: Well, in theoretical models, like Woodford has used, a key feature of monetary policy is the rule the Fed will plan to follow once they're able to follow rules and set the short term interest rate as something other than zero. And that's potentially a very powerful tool in that theoretical model and the question Central Banks struggle with was, "Well gee, can we somehow use that? Can we somehow tell people that we're going to follow rule number two instead of rule number one once we get out of this situation?"

Hamilton: And there was a notion forward guidance amounted to that. But it's philosophically in terms of the model, and as well as practically in terms of what a real world Central Bank can do, not clear exactly what that involves. If there was some reason for you to follow rule number one, why are you going to be able to tell people, "No, I'm going to follow Rule number two," even though rule number one would be the thing that makes sense? I'm just trying to persuade you now I'm going to do something else. So you so respond now. It's hard to do that, really explain what that means even in the theoretical model.

Hamilton: And when you get to the actual Federal Reserve, okay, what exactly are they saying? Are they a dossier in Forward Guidance that they're somehow finding themselves to the master of the ship saying, "No whatever comes we're not going to raise interest rates because we promised." And how do you do that? Janet Yellen isn't the chair of the Fed anymore. In what sense could she say such a thing? And what meaning could it have? So it's one of those things where in our theoretical models, it would be just so easy to do. You just write a different rule and there you go. But I think the practical problems with doing that and saying "No, this is really permanent. No, I'm really going to stick with it are very substantial."

Beckworth: So this comes down to a question of credibility. You may say this is a permanent injection, but can you credibly stick to that? And I think that's why some of the discussion over the monetary policy frameworks are moving to level targets, price level targets, nominal GDP level targets, and the emphasis being on levels because a level target would imply a permanent increase if needed in the situation like we were in 2008 and 2009. But I hear your point, though is credibility is really hard to generate, particularly when you're in the midst of a crisis like that.

Beckworth: All right, in closing, I just want to wrap up on this paper. You guys look at Forward Guidance, we've touched on that. We also look at negative interest rates, and your outlook for negative interest rates not very favorable as a practical policy tool. So you end up with kind of going back to the simple short run target interest rate as the main tool for monetary policy. You conclude if I understand correctly, that using the Fed's balance sheet is not as promising as a tool as we thought it was. And therefore, we end up back where we started before 2008. Is that the right takeaway?

Hamilton: Yeah, so one question was, should we think of this as part of the Fed's permanent toolkit? Should they be actively buying long term Treasury securities just in normal times? And our answer is no. It's not that effective. It's not that precise a tool and really the key to monetary policy is the short term interest rate.

Beckworth: All right. Well, on that note, our time has ended. Our guest today has been Jim Hamilton. Jim, thanks so much for coming on the show.

Hamilton: Thank you for inviting me.

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.