Related Links

• Will Interest Rates Remain Elevated Even as Monetary Policy Normalizes?
• Charting the Path of Interest Rates
• Lubik-Matthes Natural Rate of Interest

Transcript

Tim Sablik: My guest today is Felipe Schwartzman, a senior economist in the Research department at the Richmond Fed. Felipe, thank you for joining me.

Felipe Schwartzman: Hi, Tim, it's a pleasure to be here.

Sablik: Your research focuses on how the economy adjusts through business cycles. Today we're going to be talking about your most recent Economic Brief which asks the question, "Will interest rates remain elevated even as monetary policy normalizes?" It's a question that's on the minds of many Fed watchers.

In response to the high inflation and the post pandemic recovery, the Fed raised its policy rates significantly. Now that inflation has fallen closer to the Fed's 2 percent target, the FOMC has begun reducing rates. In September, the FOMC cut the Fed funds rate by half percentage point and again by a quarter percentage point in November. But even before the Fed cut rates, FOMC members signaled their expectation that interest rates would eventually come down after inflation had abated.

However, there's been a big debate among policymakers and economists about where interest rates will settle in the long run. This long-run equilibrium rate is sometimes referred to as "r-star" by economists. We actually kicked off the year with a podcast episode on this topic, and so it seems fitting to return to this question again now at the end of the year.

To start, could you share how you define r-star, and how should listeners think about this concept?

Schwartzman: R-star is a concept that I think has become confusing to a lot of people, including very knowledgeable people. I think the reason it has become confusing is because different people tend to define it differently.

[Economists] from the New York Fed define it as the interest rate that, if you put it on and you stay with it, it's going to make inflation be where we want it to be. What helped me trying to think through this is to instead of thinking about r-star is to think about the natural rate of interest, which is a very closely related concept but that has a very tight theoretical definition.

The natural rate of interest is defined as what interest rate would we get if we lived in a world where monetary policy doesn't matter. In the kind of models that central banks use 99 percent of the time, monetary policy matters because prices are sticky. So, just make prices not sticky and you get what the natural interest rate is.

You can think about a price that is defined by supply and demand, just like the price of bananas. It's not something we set. It's just something that the market sets. The difference here is that it's the price of some goods today versus some goods tomorrow. In other words, it's a return that you get for saving or the cost that you pay if you want to borrow.

On the demand side, you can see this natural rate of interest move around and go up because of demographics. Middle-aged people save more than young or old people. Households accumulate a lot of debt. They will try to pay off these debts, so that will push this interest rate down. Or, even more of a squishier attitudinal shift, some people will talk about how after the pandemic, there was this "revenge spending" — people just wanted to seize the day with travel and experiences.

On the supply side, the natural rate will also reflect opportunities for investment that are available for firms. If those opportunities look really good, firms might be willing to borrow at high rates.

All in all, if the economy is growing strongly and if firms and households feel confident in spending, this is going to be an economy with a higher natural rate. In practice, that's not how interest rates are determined, especially in the very short run.

Sablik: Why is r-star so important for monetary policymakers?

Schwartzman: The Swedish economist Knut Wicksell, more than a century ago, hypothesized that the difference between the actual interest rate and the natural one would create imbalances in the economy that would result through increased prices or inflation. This view has proven to have remained a very useful one in academic research and informs monetary policy today.

There are caveats. First, inflation may happen even if the interest rate picked by the central bank is always equal to the natural rate. The reason is that inflation can fluctuate, for example, because consumers become more willing to accept higher prices — something that some people say may have happened after the pandemic and then firms will reach for higher markups — or because cost pressures coming through commodity prices or supply chain disruptions will raise inflation.

Second, the natural rate of interest is not necessarily the best for everyone or even for the economy as a whole. For example, the choice of interest rates will have distributive implications. If you're a saver, you like high interest rates. If you are a borrower, you don't. Also, sometimes you don't want to have your interest rate equal to the natural one, like if you have a supply chain disruption. If you really try to fight it by having a very high interest rate, that's going to have other effects in the economy. But you do want to increase your interest rate a little bit as a counter pressure, even if that's not the natural rate.

Most importantly, though, the natural rate is very hard to model and estimate reliably. It depends on the specifics of what model you have in mind. If you want to have a grip of how it varies from meeting to meeting, this is not super actionable at this kind of frequency.

Sablik: This raises a question that I think a lot of Fed watchers have asked, which is, "If it's so hard to model r-star reliably, why does the Fed pay attention to it at all?"

Schwartzman: One thing that the Fed can try to do is to track where it thinks the natural rate will converge to over many years. Even if we want to deviate from the natural rate for short periods of time — for example, to keep the economy a little bit tight in order to countervail a supply shock or something like this — most models will agree that it's not a good idea to try to deviate from the natural rate forever. If you do that, eventually you're going to get inflation pressures that are going to accumulate.

A second main reason to track [the natural rate of interest] is that policy acts with lags, so we want to target a long-range destination point. That's a saner choice than trying to get it right month by month.

Of course, the very last reason is that the very idea that interest rates will eventually converge on a natural value provides a basis to estimate the expected natural rate.

Sablik: That's a good segue to speaking of forecasts and trying to estimate r-star. As I mentioned, we had Thomas Lubik and Christian Matthes on the show at the beginning of the year to discuss their model for estimating r-star, which listeners can find on the Richmond Fed's website. That's an exercise that you try to do in your Economic Brief to some extent. How does your approach to estimating r-star differ from the one used by Thomas and Christian?

Schwartzman: The first main distinction I would make is I think they do a much more rigorous job than I do. I would barely call what I do really an estimate because I don't have a sense of uncertainty around it, and I don't follow a super-clear, replicable formula. It's more of a back of the envelope calculation to get some approximate numbers. It is, in fact, a rough version of a more refined method advanced by [Jens] Christensen and [Glenn] Rudebusch in a couple of papers.

I think what the more interesting question is, what is the conceptual difference? And I think there's some very interesting differences here, but also similarities.

Let me start with the similarities. Both approaches start from the idea that one way to estimate long-run r-star is to obtain a forecast for the interest rate several years in the future. So that's what I do, that's what they do. The main distinction is how we obtain that forecast. Thomas and Christian develop an econometric model to do theirs. Their model is fairly flexible. It has very little structure imposed. So, their forecasts are pretty theoretical. Largely, their model provides the best forecast for economic variables given their current values, based on historical relationships and some estimate about how these relationships may be changing over time.

Rather than build my own model, I decided to use that of bond traders. When bond traders buy long-term bonds, they need to have a view about where they think interest rates are going to be over the life of that bond. This is true not just if they plan to hold these bonds to maturity, but also if they plan on selling it at some point to another bond trader who is also going to be doing those calculations. So, these forecasts are reflected in the bond yields.

In a world where bond traders are not concerned about risk, liquidity or other conveniences brought by holding government bonds, this forecast is all that's going to matter. If their returns are higher by holding the bond versus rolling their money over — into short term funding markets or vice versa — there's an arbitrage opportunity that they ought to be taking advantage of. If you want to look at the yields of 10-year bonds and compare them to, say, five-year bonds, you can have a rough sense of what bond traders think interest rates will be between five and 10 years in the future. So that's kind of the general idea.

It gets better because the U.S. government also issues inflation-adjusted bonds. These are the TIPS. This allows us to disentangle changes in bond yields that are stemming from shifts in inflation expectations from the kind of real factors behind the determination of natural interest rates. In the end, real inflation-adjusted bonds are the ones that I use in the EB [Economic Brief].

Sablik: Based on your analysis, once you account for all these factors and do your calculations, what has happened to r-star since the pandemic?

Schwartzman: In my calculation, it went up by more than 1 percent. Like I said, it's a coarse calculation. You can have errors on both sides.

It is hard, though, to square the fairly large increase in 10-year yields that we've seen with just things like term premia and convenience yields and other factors. A big chunk of that, in my view, has to be the market forming a view about long-term interest rates, which is different.

Sablik: In your paper, you look at some trends in the economy that could further support the idea that r-star might be higher in the future. Let's take a look at some of those, starting with consumption. What's happened to the personal savings rate since the pre-pandemic period?

Schwartzman: That has gone down. Perhaps there was some kind of pent-up demand coming from the pandemic. Some people at first were flush in cash, both from perhaps lower consumption that they had during the pandemic and also transfers that they received. Perhaps there was just a change in attitudes. So, we're going to see how long that's going to stick.

Sablik: How about investment? How would you expect changes there to affect r-star?

Schwartzman: I think the big question there is how new technologies could affect the long-run growth in the economy. This is a very speculative debate that we're in the middle of, in terms of new technologies and new processes that firms developed during the pandemic, but also the innovations in AI that are starting to come out. There's a question about perhaps we're just going to see higher, longer-term growth. If we do, firms will want to invest and households will feel more confident about the future and perhaps be more willing to spend now and earn later. Both of these things would point to higher r-star.

Sablik: Government spending is another factor influencing long-run rates. What do you expect to happen in that case?

Schwartzman: That one is very hard to tell. One aspect I mentioned in the Economic Brief is that geopolitical risk seems to be increasing worldwide. This could lead many countries to increase their military spending. More government spending abroad could have an impact on asset prices here in the U.S. That is, of course, highly speculative.

Sablik: Is the sustainability of government, then, another potential factor, too?

Schwartzman: It's different for different economies, right? If you're talking about the U.S. economy, which is really the topic here, this tends to show up — if it does at all — through inflation risk. The worry is if the U.S. government has a lot of debt and there's not a clear plan for how it's going to pay for it through higher taxes or lower spending, then you pay off the debt through higher inflation.

To the extent that this is a risk that gets anticipated, this may affect bond prices. This would not show up in the TIPS because that's inflation adjusted. That could also affect the term premia, because if inflation is volatile, that may also make monetary policy more volatile and interest rates more volatile.

Sablik: Your Economic Brief came out at the end of August, which was before the Fed began cutting its policy rates. Has your outlook for r-star changed since then?

Schwartzman: Well, I'm just going to say, based on those calculations, I don't think there's many reasons for them to change. Bond prices have remained reasonably stable.

Again, we're all here talking about rough calculations, right? Is it 0.1 above, 0.1 below, who knows? But I think the general trend seems like we're going to have to live with higher interest rates. This is what bond markets seem to be telling us.

Sablik: When it comes to this question that everyone's asking about where interest rates will settle, what sort of things will you be paying attention to as we move into the next year?

Schwartzman: I'll be looking at the bond market, of course, but not just that. In the meantime, of course, we need to keep paying attention to inflation, unemployment and all the rest.

Sablik: Felipe, thanks very much for coming on the show to discuss your research.