Research Session 2: Monetary Policy, Inflation, and Crises: Evidence from History and Administrative Data
José-Luis Peydró, Imperial College Business School professor of finance, presented a paper examining the links between monetary policy and banking crises. Discussant Marc Giannoni, Barclays managing director and chief US economist, and Atlanta Fed economist emeritus Larry Wall joined Peydró for questions and an assessment of the work.
Transcript
Larry Wall: If we could go ahead and everybody get seated, we can get started. I'm Larry Wall from the Atlanta Fed, and in the last session we started discussing the links between monetary policy and financial stability to a limited degree. This session, we're going to take a more academic approach, but explore it from a much bigger-picture perspective. I wanted to start by taking a couple of minutes to explain how I think about some of these issues.
Financial instability has, as we all know, the potential for large adverse impacts on inflation and on employment. Thus, one could make a good case for adding a mandate for monetary policy to take account of financial instability. However, as the last session indicated, setting monetary policy is difficult enough when the central bank's mandate is limited to inflation and employment. Calibrating the effect of policy on existing inflation and the real economy is very difficult given the complexity and ongoing structural change in our economy.
Moreover, looking at it from the other side, our ability to estimate the effects of incremental changes of monetary policy on financial stability is orders of magnitude behind our ability to estimate its effect on inflation in the real economy. Thus, there is a case for delegating responsibility for financial stability to prudential regulation. Yet, I would argue prudential regulation suffers from some of the same conceptual problems as monetary policy, and arguably to an even greater degree, for a variety of reasons.
One that I'd like to highlight is that financial pressures work to cause the financial system to evolve in ways that reduce the effectiveness of prudential regulation. Regulation is costly to regulatees, so it creates an incentive to find ways to avoid or reduce costs. Moreover, if the entity is subject to the strictest regulation and can't avoid the cost, then the activities might very well shift to less-regulated entities.
Inevitably, the prudential regulators will lag behind the industry, and given political opposition to stricter regulation may not be able to catch up until after the next crisis. For this and a variety of other reasons, I would argue prudential regulation, micro and macro, is not and never will be good enough to assure financial stability.
That puts the financial stability ball back in the monetary policy court. Given our current and likely future understanding of monetary policy, we can't use it to fine-tune the stability of the financial system. Yet the cost of instability is so high, the ways of managing monetary policy to reduce fiscal and financial instability is something that we have to look for and consider.
Given this backdrop, I was delighted that José-Luis Peydró, a professor at the Imperial College of Business, agreed to present his paper on the relationship of monetary policy to financial crisis. He has up to 25 minutes to present the paper, "Monetary Policy, Inflation, and Crises: Evidence from History and Administrative Data."
José-Luis Peydró: Thanks a lot for inviting me here. I have the slides, if you can upload the slides. As Larry said, this paper is about monetary policy and banking crises, and we are taking a historical perspective. It is coauthored with Gabriel, Dmitry and Björn. Basically the motivation of this paper...we started writing this paper in 2022, the day that Russia invaded Ukraine. We thought that inflation was going to go up, interest rates were going to go up, and this potentially could create problems in the financial system after a period of low rates for long.
Currently, we have high inflation. A bit of high inflation we have had in 2022, 2023, and 2024, and raising monetary policy rates, and they are still high. There are many people that have talked about tradeoffs between GDP, recessions, the soft landing/hard landing, and monetary policy rates. We know a lot about those things. Milton Friedman was writing that in the 1960s, and a former vice president of the Fed, Alan Blinder, has a paper and a book on these issues.
What we know much less is the impact of these monetary policy rates on financial stability, especially after a period of low-for-long rates. Before I enter into my paper, let me just say: In the last two years, we have had some problems when raising interest rates. Here in this conference, we have talked about Silicon Valley Bank and the other banks after the Fed was raising monetary policy rates.
This is the United States, but in other parts of the world, there have also been problems. The ECB, the European Central Bank, raised interest rates in the summer of 2022. Immediately, some countries—for instance, Italy—suffered in sovereign debt because they were highly leveraged. The European Central Bank had to put an unconventional monetary policy instrument to stop those problems, the so-called "transmission protection instrument."
I work in London. In London, in September 2022, after a tightening of monetary policy by the Bank of England, with also fiscal policy from the government, we saw in some days no trading whatsoever in the sovereign debt of the UK, so-called "gilt," with big problems for non-pension funds. If you look at the data, also for hedge funds.We also saw that. We saw stable coin problems. We have seen commercial real estate problems.
Raghu Rajan, Steffen, and Acharya presented a paper in Jackson Hole about QT and QE, but we know very little about monetary policy and banking crises, and in particular about monetary policy rates. Before I tell you what we do in this paper, let me show you some very important case studies of important banking crises.
If you look on the right side of the panel, this is the Great Depression, the Japanese bubble, and Spain. What you see is that—and the number, for instance, Spain 2008, bottom right—the banking crisis in Spain started in 2008. What you see in all these important banking crises is that there was a "U" of monetary policy before the crisis. The issue of monetary policy is basically you lower monetary policy rates, potentially you create a credit boom and asset price boom, and this collapses and creates the banking crisis afterwards.
Since I am from Barcelona, let me talk first about the Spanish banking crisis. In 2002, the "sick man" of Europe was not Spain. It was Germany, based on the economies. The ECB lowered monetary policy rates, and those lower monetary policy rates were too low for countries like Ireland and Spain. Then, at the same time, there was a credit boom and asset price boom. Massive, even bigger than the United States. Then when the interest rates went up, the system collapsed and we had a banking crisis.
Still on the right of your panel, you can see Japan. In Japan, there was in 1985 the Plaza Accord in Paris for FX. This led to an appreciation of the yen. This was external to Japan, and the central bank of Japan lowered monetary policy rates. They stayed low for long, and then they raised monetary policy rates.
The bubble in Japan, what I explain to my students, is amazing. I don't want to give you the numbers here, but the numbers of the real estate prices in Tokyo were amazing. Then, when interest rates went up at the end of 1989, beginning of 1990, there was a crash in Japan and it lasted for many, many years. Spain's crash also lasted for many, many years.
On the top, and it is the last one that I want to comment on, the Great Depression. Since the previous ones I was commenting more on the lower rates, let me say a quote from Ben Bernanke, the former chairman of the Fed. He said, and he wrote, that in 1929 the Fed wanted to crash, or wanted to reduce, the stock market prices by massively increasing their rates. They did it, but they caused a banking crisis.
You have all these anecdotes, and all these important crises there. The thing is, let's have systematic evidence on this, and this paper is bringing systematic evidence on what is the impact of the monetary policy cycle on the probability of a banking crisis, and what is different from non-financial recessions, even the worst non-financial recessions.
Second, if there are these differences, what are the mechanisms? Are these mechanisms about credit booms and asset price booms and banks? Third, since banking crises are rare phenomena, what we want is to expand the time with covering many, many countries and many, many years. The data that I'm going to show you here—in the paper we have all the data, but given the time limit I'm going to show you only a panel of historical data. Lucrezia was putting some of these data on the balance sheet of the central banks before, but this data is basically 17 countries, including the United States, Japan, Canada, Australia, and some European countries, over the last 150 years.
We have more than 80 crises, and we have hundreds and hundreds of recessions. We're going to use the monetary policy interest rates, either the raw monetary policy interest rate, or over and above the systematic component of monetary policy. We will use, as I will explain later, also in all regulations, an ideal strategy following from the international trilemma.
Before I enter into the details of the paper, let me just give a preview of the results. I think it's super-interesting. Not only the preview, but to understand this preview of results. It's difficult to understand in this slide but let me start with the basic results. What we find is that these case studies, as I told you before, that you condition on the crisis and you look back at the monetary policy, it's not only those cherry-picked crises that I showed you before, like Spain, the Great Depression, Japan. It applies to all banking crises in these 150 years and 17 countries.
Moreover, if you are in a current situation like the American situation or European situation that you have lower interest rates, you have low-for-long, and then you raise interest rates, a crisis comes. The probabilities might go from just very low, 18 percent, to 50 percent. This is a big number for a banking crisis, and this is very, very different from recessions, even from the worst non-financial recessions, and I want to talk about this.
The big thing is going to be the mechanisms. Before I tell you that these effects that I was telling you before are larger, the stronger the view of monetary policy rate is if you lower monetary policy rates over and above GDP and inflation, effects are going to be bigger. Or, if you tie the monetary policy over and above GDP and inflation, effects are going to be bigger. They are bigger from the systematic part. So, what is the channel? The channel is going to be that when you soften monetary policy, you are going to create a credit boom and an asset price boom. It's not always going to be like that, but there is a high probability it's going to be like that.
When you tighten monetary policy after this big credit boom and asset price boom, you're going to have a stronger reversal and the crisis will come. Since these are credit booms and asset price booms, it's very complicated for me and for a researcher to put everything together in a paper, so I created this red zone. Look at the red zone bullet point. These red zones come from a Journal of Finance paper, which is basically combining high credit boom and high asset price boom.
When you soften monetary policy, you enter into this red zone of elevated credit and asset price boom. This is both businesses and households, both real estate and stock prices. This is consistent, and I will show you the results, with credit supply. In the data for this banking crisis, this is consistent that banks are expanding the credit supply rather than responding passively to the demand. I will show you this later.
Once you are in the red zone after this softening of monetary policy and you tighten monetary policy, the crisis comes. If you are in the red zone and you don't tighten the monetary policy, or you have a credit boom but you were not coming from softer monetary policy and credit supply from banks, you don't get the crisis. In a sense, the path of monetary policy is crucial for the crisis, this "U", and purely credit and asset prices without monetary policy do not get into the crisis. It's a combination of the two.
I will talk about bank performance over three years, and here I will want to highlight not only interest rate risk but losses for banks, credit risk with non-performing loans, reduction in banking prices, and a crash on stock prices, over a three-year period. Not immediately, because immediately banks benefit from higher rates, but over time as MP and interest rate risk jump in, losses come and stock prices go down. I don't touch anything on microdata, given the time that I have.
I just gave you the summary of the paper. Now let me go to the details; details are important to understand these things. The data, as I was telling you before, are these 17 countries, 150 years of history: Japan, Australia, Canada, United States, European countries like the UK, France, Germany, Spain. The crises that we take are bank crises. These bank crises can be a narrative. There are bank runs, forced mergers, defaults, and important stock price declines.
For the monetary policy, apart from having these 80 banking crises, we have hundreds of recessions because we want to make a difference between purely economic recessions, even the worst ones, and banking crises, just because of the mechanisms. For monetary policy, we are going to use the central bank monetary policy rate, but for instance, here in the Fed, you created the Fed in 1913. We saw yesterday that that was important, the 1907 crisis and previous crises. When there is no central bank interest rate, we will take the T-bill there.
We are going to have all the results before the First World War, and after the Second World War. It applies for the whole period, or only after the Second World War. Here, very briefly: the black line, which is in year 0, so, you have year -1. Year -1 will be one year before the banking crisis, year -2 will be 2 years before the banking crisis.
As you can see on the left-hand side of the black vertical line, these are years prior to the banking crisis. What do we see there? The "U", either for the whole 150 years or for the post-World War crisis, even for the worst banking crisis. What we saw in the case studies applied to all the crises.
Let me also say—I will not comment on this; this is potentially for a future paper—once the crisis comes, on the right-hand side of the black line, then what you find is that the central banks massively cut the interest rate to save the economy, similar to what we found in the Global Financial Crisis. Yesterday, Laura talked about fixed effects, and this is like that. Here is the same regression as before, with new local projections. We control for country fixed effects, for time period fixed effects. We don't impose any particular monetary policy cycle, but we find the same thing: left of the crisis, before the crisis, you have the "U" of monetary policy in all the periods, and here it's controlling for lots of things.
Now, panel A: this is different from long-term interest rate. We can talk in the Q&A about this, why it's different, long-term interest rates, than short-term interest rates. That's the panel A. The panel B is recessions, even the worst, non-financial recessions. What you see there is that interest rates matter that they go up, that they don't have this "U." They don't have this low-for-long, which I will tell you later why this is the case.
Before I tell you this, let me show you this in a different way. Before I was doing so-called "non-parametric analysis," I was data speak , so I didn't impose anything, and you get the "U" of monetary policy rates before the crisis. Now, let me take four potential monetary policy cycles: a "U," you lower interest rates, low-for-long, and then you raise rates, the inverted "U," you raise interest rates, then you lower interest rates, then raise versus cut.
If you look at the charts, you have on the left-hand side all the four different monetary policy cycles. On the right-hand side, column five, you will see that these four monetary policy cycles are around 25 percent of the time. It's not that one monetary policy cycle happens more than the others. All of them happen around 25 percent of the cases.
Amazingly, 55 percent of financial crises are preceded by this "U" of monetary policy rates. This goes to 71 percent after the Second World War, and if you look at panel A, column four, it happens 100 percent. After the Second World War, the worst financial crises, including Japan and many other ones, came from this "U" of monetary policy rates.
Despite that, only 27 percent of the cycles are this "U." If you go to panel B, column four, the worst recessions, but they are not banking crises, only 31 percent come from this "U." Non-financial recessions, even the worst ones, are coming 31 percent from this "U" while 100 percent of banking crises are coming from the issue of monetary policy rates. This is so far a condition on a crisis, and a look backwards.
In the United States and in Europe, it is the opposite. We have the monetary policy cycle. We see today that maybe the monetary policy cycle was not so tightening, and we will talk about that potentially in the Q&A. If you have the monetary policy cycle like a "U," then what is the probability that in the next three years you get a banking crisis? As you can see, this is 18 percent. This number will jump to 50 percent, but so far 15 percent. If you look at column four, this is four times bigger than the average conditional probability. You increased by four times the probability of a banking crisis.
I think these charts are very useful for here, but since I am an academic let me just put some IVs, some exogenous variation, some regressions...a bit what Laura did in the research presentation from yesterday. We take this trilemma. John Campbell today talked about these Scandinavian countries that they pick the currency, the other countries in the world. You get some variation of monetary policy. This basically is related to the Mundell trilemma, that if you have somewhat the capital account open, and you have a somewhat fixed rate, then your monetary policy is not independent.
If you do that, then we can get some more exogenous variation. This is very similar to what Campbell was saying in the previous presentation. Here, I don't want to put you a lot on the regressions. Camelia told me not to put many regressions but let me just say that here on the left-hand side this is the probability of a crisis in the next three years. Coming from a period of low-for-long, you lower the rates. They are still low, you potentially create a credit boom and an asset price boom, and then you're raising the rates.
The last line, which is in yellow, is if you cut for a long period of time and then you raise is when you get this probability of a banking crisis in the next three years. Not immediately, but in the next three years. These are crises, but for recessions it's very different. Look at non-crisis recessions, even the worst ones. Look at the last line: it's zero. It's not a question about recessions, at least in these types of things. It's a question of financial crises, of banking crises, which I will touch later on the mechanisms.
Before I do that, let me say Marc has worked a lot on the neutral interest rates. Here, we don't have the neutral interest rates, it's more complicated to do it in 150 years. We have the monetary policy rate over the systematic part of GDP and inflation. If you lower monetary policy rates and they are low given the GDP and inflation, and you raise them later on over and above the GDP and inflation, the numbers start being even bigger. Now, it's 28 percent of the probability of a crisis. It's not 18 percent, it's 28 percent, and this number will rise as we go forward.
Basically, so far, what I was telling you, the "U" here, it's not low rates, it's not high rates, it's a combination of the two. This leads to banking crises, with high probability and this is different from recessions. Why is this the case? What are the mechanisms?
There are papers, and I have papers on this, that when there are low rates, they create potentially credit booms, risk taking, and asset price booms. Other people have done that. If you have credit booms and asset price booms, it's complicated. I found this paper in the Journal of Finance by Greenwood, Hanson, Andrei Shleifer, and Sørensen: it's very nice, because they take one statistic, one sufficient statistic in a sense, that the red zone is highly elevated credit and highly elevated asset prices. This will make my life very easy. I can do it one by one, but I prefer to do it like this because it will be easier.
What we find in this regression, let me tell you the title on the top, is that when you soften monetary policy, when you lower interest rates, you are going to end up in this red zone of high credit and high asset prices. Moreover, on the bottom, and this is the regression showing that, what we are finding is that it's credit-supply driven, which means that you have more credit booms, but the spreads are low. You could have a credit boom and the spreads are low, because you think that the future is going to be good, but in the future you are going to have losses.
It's very difficult that this is demand driven because the spreads are low, credit booms are high, and then you have disappointing future results. There is another thing that when this is happening the bank stocks, the capital of the stocks, go up in real terms. Banks adjust to have a capital-to-asset ratio constant. If the capital is going up in real terms, then they expand their credit in order to have constant capital to equity.
Then you take this valuation, the bank valuation is going up over and above the profits, over and above the dividends. That is also consistent with overoptimism in the lending side. This expansion of credit and asset prices in this red zone is consistent with credit supply and overoptimism. Remember, I'm talking about banking crises. I'm not talking about all the financial crises.
You arrive to the red zone with this monetary policy. Once you're in the red zone, now we want to look on the top at the potential for a banking crisis over the next three years. What I highlight there is that if you are in the red zone and you raise the monetary policy rates, you are going to have a banking crisis.
One thing which I find fascinating: I like to put regressions, but these numbers are much better than a particular regression. These are, I tell you, all these 150 years, all the observations are there. Look at the third line. The third line says you have red zones, so you have a lot of credit and asset price booms, but you don't have this euro monetary policy. Look at column number four, row three, column number four: you have 71 red zones. Very high credit, very high asset prices, but only one of those ones end up in a banking crisis. Why? It's because there is not this monetary policy "U."
While row one has the monetary policy "U" and the red zones, and now it's one-third out of 33 red zones—13 end up in a financial crisis. It is a combination of the red zones and a combination of the monetary policy "U." Why is this the case? The first thing is what I was telling you before, and what I'll tell you in this slide. The first thing is that, remember when interest rates were going down, the expansion of credit and asset price boom was credit supply driven. The banks were taking more of this risk and high credit.
Once you are in the red zone and you raise interest rates, I want to touch on two things: on asset prices and credit and on bond performance. Let me start first on asset prices. If you have a red zone—high credit, high asset prices—automatically, there will be a mean reverted. You will tend to go later to lower prices and lower credit. That will be the coefficient so-called β2 in that regression. The β1 in that regression, analyzing the future—the future will tell you about what is the future—β1 is that if you have high rates, asset prices and credits will go down.
But look at β3. β3is a combination of monetary policy going up after you have this red zone. It's an "over and above" effect; it's an interaction effect. You can see, I plotted here the local projections both for household credit, house prices, business credit, and equity prices. The reversal, after you are in the red zone and you tighten monetary policy, it will be much stronger than just purely the upper side, which might lead potentially into a banking crisis.
Since these are banking crises, let me finalize before I do a wrap-up. Let me finalize in the last slide on the following about banks. Banks with higher rates immediately get more profits. But if you look at the profits of the banks, the nonperforming loans of the banks over three years, so the left-hand side variable here, if you see column one, column two, these are return on equity, nonperforming loans, stock prices for banks, and the last columns, seven and eight, crash of stock returns. All these things over the next three years. They go down if you can't raise for a long period of time, they were low...and then you raise rates.
You will get lower profits, more nonperforming loans, more losses over and above nonperforming loans, which is proxy for interest rate risk. You will get lower stock prices for banks, and you will get a crash. Remember, here is a free period. It doesn't go automatically because the interest rate losses, and the nonperforming loss, the credit risk, need time to do that. I just finalized my regressions. I only have three more slides. I think I am on time.
Let me just recap what I just said. The U-shaped monetary policy rate increases the probability of having a banking crisis, with strong numbers. In raising, a period like nowadays, with some caveats that we will discuss in the Q&A, this implies a crisis. This is unique to banking crises. Very, very different from the people that were talking about the soft landing of recession. This is stronger for the deeper "U," so the more you lower the rates, the worse this is going to be, and the higher the rates are over and above inflation and GDP, the worse this is going to be.
Briefly, what is the mechanism? Credit booms and asset price booms. This is very good when you soften monetary policy, but when you tighten monetary policy and you have these massive credit booms and asset price booms, they will collapse on prices, on credit, but this will imply losses for banks.
What is the contribution to the literature before I tell you some policy implications) There are people that have said that low rates are bad for risk taking. This paper, that here is called Jiménez et al., on the top, in 2014, is a paper published in Econometrica. Econometrica is a journal that takes forever to publish, but I presented that paper seven years before, in July 2007, to the executive committee of the European Central Bank, with the president, vice president and board members. That was July, mid-July 2007. This was three weeks before the banking problems in Europe that came on the 9th of August 2007.
These three weeks before, what we show, and I show in that presentation, is that in Spain those lower interest rates of the ECB were causing massive credit booms to very risky borrowers with a lot of risk-taking on real estate, and by the world's banks. People have done, like this paper that we did on low-level data, on bank-level data, on securities, on nonbanks. At the end, you need a banking crisis because you might take, on the margin, more risk. If there is not a crisis, okay. That's risk taking, but there is not a crisis.
When it comes to rate hikes, other people had said, "Okay, if you raise monetary policy rates, then you have a crisis." Other people, like my former colleague, Jordi Galí, has a paper with Boissay et al. in the studies that matches our empirical findings in a new condition model. There are other people doing that.
What is new in our paper is that it's not low rates, it's not high rates, it's a combination of the two and the two matter, because one creates a credit boom and asset price boom, and the other is going to create a crisis when you raise monetary policy rates.
The second contribution that I want to say is that there are many people that have worked on whether credit booms and asset price booms are the main determinants of banking crises. What we show in this paper: that's not the case. The monetary policy is crucial. That is, if you don't get a credit boom or an asset price boom, generating a banking crisis—remember banking crisis, not other crisis, a banking crisis—if you don't have the monetary policy "U." That is, you need the monetary policy, and the credit and asset prices, in order to create the banking crisis.
The last slide here is super-challenging, as you guys know much more than me about policy implications. Let me say, coming from our paper, we have the following policy implications, and many are in this paper I was telling you about, with Boissay, Jordi Galí, et al., and a paper by the Fed by Goldberg and López-Salido. What we find first is that the path of monetary policy rates matters.
Before entering into the red zone, before you have very high credit and asset prices, potentially one solution is to deviate from the Taylor rule on the margin, like in the envelope theory. You could deviate a bit on the Taylor rule, raising monetary policy rates, because you are not affecting much the GDP and inflation, but potentially you are reducing the probability of entering into a red zone of high credit and asset prices.
Another thing is macroprudential policy. Suppose that you don't want to use monetary policy rates. I think sometimes the Tinbergen rule is not perfect, but let's not talk about that. If you don't want to change monetary policy, you could still change macroprudential policy and use LTVs and countercyclical buffers. Luc Laeven from the European Central Bank presented yesterday here. We have a book at MIT Press on how to use macroprudential policy for these issues.
Now, suppose, just 30 seconds, suppose that you are in the red zone and you want to increase massively the monetary policy rates. Maybe you shouldn't, but if you have to do it, maybe it's important to have macroprudential policy and supervision. The supervision, given this conference and given what we heard yesterday, especially on interest rate risk, let me say that in history, credit risk is also very important. Potentially credit risk on commercial real estate and other markets could come, and that's basically my presentation. Thanks a lot.
Wall: Thanks very much. To discuss the paper, we have Marc Giannoni from Barclays Bank, and formerly from the Fed as well.
Marc Giannoni: That's right. Thank you very much, Larry. Thank you very much to the Atlanta Fed for having me here. President Bostic, thank you very much. It's really an honor to be here.
It was a pleasure to read this paper and really get to understand some of the stylized fact that this paper brought. Basically, what does the paper do? I'm going to be pretty quick on that because you described this pretty thoroughly. Essentially the question is what's the interest rate path that tends to precede banking crises? Here, as José-Luis explained, they use basically two approaches.
One is to focus on the so-called macro history database from Jordà, Schularick, and Taylor. They look at a lot of crises. At the end of the day, they look at 77 systemic banking crises across 17 countries, over a 150-year period, from 1870 until 2020. They look at banking crises as described in the literature here, as in the baseline case, at least, focusing on the narrative approach around banking crises.
In the second approach, and we didn't talk too much about that in the presentation, was a focus on the micro data set, a focus on Spain in particular, looking at both credit registry matched with firm and bank administrative data. This is for Spain from the mid-'90s until 2020. If there is one figure that you have to remember from that paper, it's this figure that I put here on the screen. That figure, I think, makes it very clear, I think it's very stark, that you have a so-called U-shaped interest rate path prior to systemic banking crises.
The U-shaped path really says that about seven years before a banking crisis what happens is interest rates start declining, then they stay low for a few years. In the three years prior to the banking crisis, interest rates move up again. This is really stark. I think it's a very important stylized fact, it's something we should remember, and policymakers should remember as they conduct policy, and as they assess the risk of entering in a banking crisis.
Now, another point that the authors talk about is that recessions are typically preceded also by interest rate hikes, but not necessarily a U-shaped path. In fact, they argue that they are not generally preceded by a U-shaped path. So, the U-shaped path is something that's more associated with banking crises.
The second finding is that this U-shaped path actually materially increases the risk of a banking crisis. It's one thing to look at all the banking crises and look at the interest rate path that preceded it. Now, the question is, if you have a U-shaped path, does it actually predict a banking crisis? The authors argue that yes, it does materially increase the risk of a banking crisis.
Again, they look at Spanish microdata to look more in detail as to what happened at the individual loans and individual firms when these interest rates increase. Another contribution of the paper, which I think is really important here, is to highlight the mechanism by which this U-shaped path could give rise to a banking crisis. In that sense, interest rate cuts tend to boost credit, tend to boost asset prices, and bring us in the so-called red zone that Greenwood and others have characterized in prior literature.
When you are in a red zone and policymakers start raising rates, and I like the way the paper describes that, all these vulnerabilities that have been developed at the time when interest rates were low start crystallizing. When they crystallize, it's basically what happens that returns on bank equity fall, bank stocks fall, and this raises the risk of a banking crisis.
Basically, rate cuts that do drive red zone booms, that are then followed by rate hikes, tend to increase risk of banking crises. What you need to remember, apart from the chart that I showed you before, is what's in red here. If you think about a banking crisis, it's generally coming with a combination of both a U-shaped rate path and being in that red zone, where you have asset prices that are moving up and credit that is expanding.
Just to give a sense of the magnitude, I picked one table from the paper that basically said that in any given period the probability of entering in a banking crisis or the frequency of entering in a banking crisis is about 9 percent. If you have this U-shaped path for monetary policy and you are in a red zone, that probability is basically multiplied by four. So, the frequency of entering in a banking crisis in the 2 to 3 years after that is about 38 percent.
A couple of comments here. Are the stylized facts robust? What does the U-shaped path have to do with banking crises or recessions? And what about the recent years? So, these facts, again, this is the chart. Is that a robust chart? What I tried to do is I tried to replicate the chart, and the great news here is that you can really replicate the chart. That's always a good thing for a research paper. That's what you get.
Now the question is, can you look around this 7-year window? Is it the same before, is it the same after? You can extend the window a little bit, and what you have is if you look at 15 years prior to the crisis. This U-shape, of course, is still there in the seven years, but you tend to have more U-shaped path than what looks to be the case when you look at just seven years prior to the crisis.
In fact, if you look 15 years before—on average, across all these 77 crises—you have another U-shaped path before. Rates go at 6 percent on average, they go down to 5.3 percent, go back to 6.1 percent, go down to 5.2 percent, go back to 6 percent. Bottom line, that first U-shaped path does not generate a banking crisis. This is to highlight the point that the authors mentioned as well in the paper, which is that if you have a U-shaped path in interest rate it does not necessarily trigger crisis, just that the risk of entering in a crisis is higher. That's what you have in here.
The other question is—and here we are in the US, and a very US-focused audience— how is it in the US, in particular? Now, we don't have many crises, fortunately, or haven't had so far many banking crises that are systemic in the US. We have about five or so in the sample that they are considering.
If you look at the interest rate path on average in the US preceding banking crises, you don't tend to see this this U-shaped path. The reason is you have to look at each of these crises. There are not that many of them, fortunately. In 1893, interest rates were kind of bouncing around a lot before the banking crisis. You didn't have really a clear, U-shape prior. In 1907, similarly, interest rates were bouncing around a lot, and then you had the banking crisis.
Now in 1984, the savings and loan crisis that lots of people here might remember, where it's not so clear you had a U-shape. You had one a few years before, you had another one after. In truth, it's really hard to pinpoint when does the banking crisis during the savings and loan crisis happen, because it's really extended very much through the '80s and early '90s. You have a bunch of U's that follow as well. Some of them might be related to the mechanism that they talk about.
There are two really big crises in the US where you have this U-shaped pattern, and one is the one that was mentioned in the presentation, in the 1920s, where you have this big period of declining interest rates, low interest rates in the mid-'20s, and then rising interest rates in '29 that then triggered the stock market crash and banking crisis. The other one is 2007, the Great Financial Crisis.
The two big ones that we think of as having something to do with monetary policy, like tightening, raising policy and bringing us to a banking crisis, are basically 1929 and 2007. In the US, if you look back for nominal interest rates over the whole 150-year period, you have a lot of back and forth in interest rates. Again, not that many financial or banking crises that are systemic. We have about six of them in this chart, but you have a lot of recessions. All these gray bars are like recessions.
If you notice, there's something peculiar about the recession here that's often documented, that we had a lot of recessions pre-World War II. We have fewer of them, or less frequent recessions. The time period between recessions has really extended a lot after World War II, and also after the 1960s, and 1980s again even more.
What that does, I think, is it confounds a little bit what the authors do when they look at fixed windows for looking for these U-shaped patterns, and they look at these fixed windows of seven years and three years and then again to seven years, when you had all these recession periods that were very much on top of each other before then. But at each of these recessions, in the early part of the sample, you do tend to see very condensed U's, if you want—they look more like V's, I guess—in interest rates as well.
I won't have much time to talk about real interest rates, but the point is, if you really think that monetary policy plays a big role in stimulating the economy, creating excesses, it is a little bit intriguing why you don't see this U-shaped path in real interest rates. You see it in nominal interest rates. You don't see it in real interest rates. Maybe there's something there about the nominal interest rate that triggers more risk taking and so on, that you wouldn't have necessarily taking place in the real interest rate.
The point is, there's no real evidence, or significant evidence, of a U-shaped path for real interest rates. What about the last couple of decades? Well, there is a very interesting U-shaped path, if you want, of interest rates from the Great Financial Crisis to 2019. That did not create a banking crisis but was followed by, of course, the COVID crisis. We wonder then, why didn't we really enter in this red zone at that point? Maybe this is because financial markets and households and others were recovering from the Great Financial Crisis, and maybe not engaging in these red zone types of activities.
Now, from 2020 until now, we do have another U-shaped path, narrower and shorter than what we've had in the historical context that preceded banking crises. I guess we haven't seen yet a systemic banking crisis, or have we? We had last year, with, of course, the SVB collapsing, Signature Bank collapse. Maybe that was the result of a U-shaped path, and maybe close to the mechanism that the authors talk about, in the sense that it drove banks to get concentrated. I mean, they did invest in long-term assets and got hurt when interest rates went back up again, by not protecting sufficiently against interest rate risk.
Also, we know that many other things went wrong as well in those particular banks that ended up in failure. Also, the policymakers addressed that crisis very promptly, putting in place a liquidity facility that likely helped prevent the contagion.
That's one way in which, maybe as policymakers recognize this pattern and recognize the fact that there is this U-shaped path of interest rates that actually does trigger these kinds of activities that brings us in a red zone, policymakers could take steps then to prevent the crisis from happening. As this crisis is prevented, that the pattern should disappear. We should no longer be seeing this U-shaped path causing banking crises.
All right. The final slide is like, so what? Let's say we have identified this U-shaped path of interest rate. The authors say it's important to address then the asset price and credit booms before the economy enters the financial red zone, for example, by leaning against the window by using macroeconomic prudential policy. Ideally, yes. If we have multiple objectives, we want to use multiple instruments. You would want to use, let us say, the interest rates for micro-stabilization even if it warrants following a U-shaped path for interest rates, but then use other tools, let's say micro-prudential tools, for financial stabilization.
A more interesting or relevant practical question is: What if policymakers don't have enough of these micro-prudential tools in order to mitigate the financial stability risks? Then maybe the question is, should they adjust the interest rate path, or have a slightly different interest rate path, that maybe has a less deep, less pronounced U-shaped path of interest rates, and does not lead us to this red zone?
There is a big debate here. A big academic debate has been held for many years now, from people like Lars Svensson, Ben Bernanke, Mike Woodford, and many others that have contributed to that, trying to assess really what the cost benefits are of these policies.
Conclusion: my time is up. It was a very interesting paper. I think it provides this stylized fact, we have to all go home and think about this U-shaped path of interest rates and keep it in mind. It's a very interesting fact to bring up and to make very clear here. It's a very interesting fact to remember, but it may not be sufficient to either cause banking crises, or avoid banking crises. There are lots of other factors that are at play as well that policymakers still have to keep in mind. In particular, bottom line: financial innovation, deregulation, mismanagement in some financial institutions, fraud, and all that, are things that we need to be on the lookout for, and it's going to be worth it for policymakers to continue to monitor the financial market much more broadly than just focusing on the U-shaped path of interest rate, even if this is a very useful tool.
I think the kind of work that we do see, whether it's the Financial Stability Report at the Fed or the Global Financial Stability Report at the IMF, I think is going to be still quite useful going forward.
Wall: Thank you, Marc. We're running a bit short on time, though the timer up here only gave us 55 rather than 60 minutes, so I'm probably going to run a little bit longer than what this timer is showing us. Would you like to respond, so we can get to some questions?
Peydró: Yes, briefly; very brief. First of all, Marc, many thanks for all your comments. They are super well-taken. Let me just say, on some of these things that Marc show the "V"s versus the "U"s. Our whole point is that it's not just lowering rates and then immediately increasing rates. You need to lower the rates, stay low for long, and then you create this credit boom and asset price boom, and then you raise the rates. If it is like a V, then you are not getting this credit boom and asset price boom, and then you don't get the banking crisis.
I don't want to enter into the history of the United States. It's clear that the Great Depression and 2007 fits the data. There are others, potentially, like you said, the Volker didn't fit sometimes. We have here the representative of Costa Rica, and they know very well that in 1982, and this was important in minutes of the Fed of Volker, that when you raise massively here in the United States, the monetary policy rates, this created a massive banking crisis in which your American banks were highly, heavily exposed. The big risk there was that—in fact, one of the charts that I presented, because I am in London, is from the UK at the end of the 19th century, with the Barings crisis, and the big risk taking of the British banks was in Argentina.
Sometimes the risk taking is locally, sometimes it's in other places, like the Germans, they put a lot of risk taking in Spain and Ireland, not so much in Germany. But very well-taken.
Then, just since I don't have much time, on the financial crisis, the last one on the financial crisis of 2010 and 2013: of course, what we have found, which is super-interesting and many people talk that when you lower interest rates after a big banking crisis, that might create the next one. But if you have a lot of regulation supervision because of the previous crisis, like the 2008 crisis, moreover, with people psychologically there is this: you remember about the previous crisis, then you don't get the crisis, but this was not happening in 2023.
In 2023, coming from Europe, I'm sorry to say this, or maybe I'm completely misguided, but you were starting to have a banking crisis but both the government, the FDIC, the Fed did so many things that you stopped the crisis. As you are saying, with this crisis over, I am not sure because then, maybe you are the United States. But if you pump the market with lots of liquidity, you might create an inflation problem. In countries in which the dollar is not the dominant currency, they might have a limited ability to rescue the system. That will be my comments. Thanks a lot. Very well-taken, your comments. Thanks a lot.
Wall: Yes, one could reasonably argue that the US would have had a banking crisis in the 1980s if there hadn't been so much confidence in "too big to fail" and the money centers had been allowed to get by with their Latin American losses.
I'm going to go with a few quick hits related to your study, questions about the study, so if you could be brief that would help. Shortening the question: Is basically what you're showing that it's always a CRE (commercial real estate) issue or more broadly, a real estate issue, and does that help explain the long lags?
Peydró: No. I might say, here, the credit booms and asset price booms, we have both credit to households, credit to firms, stock prices, and real estate firms. It happens to everything. I must say that on the margin, real estate that is a bit larger effects, but you get also this with other companies. I don't want to enter into the 1997–98 crisis in East Asia, but many companies in this were not on real estate and were borrowing dollars, et cetera, et cetera. So, applies to everybody, but bigger effects on real estate.
Wall: Does it matter in cutting the U, how far rates fall below neutral. I.e., did this level of stimulus make the crisis worse?
Peydró: Completely; completely. The effects double if you lower the monetary policy rates over and above the GDP and inflation. We don't have a neutral rate because it's more complicated in 150 years to have that and it is difficult to estimate. Marc has done a lot of work on that. But the lower the rates over and above the systematic part, the more credit boom and asset price boom.
Wall: And so, is there any insight into the duration: seven years prior and three years after? Why seven, why three?
Peydró: Marc showed it very well, that there were different U's, but in a sense, if you get a credit boom and asset price boom, then you stay there and you diffuse somewhat with macroprudential policies, supervision, by the time you raise the rates the risk taking is not so much there. It also features in the case studies that I showed you: the "U"s were 10 years. If you take 20 years, the effects are weaker.
Wall: I'll conclude with a question for both of you that brings us back to monetary policy more directly: Is the policy implication that it is extra-important not to fall behind the curve on inflation, necessitating a sharp rise in rates to catch up when policy rates have been especially low for a considerable period?
Giannoni: Sounds like a timely question. Look, I think the one implication of that paper in general, of course, the central bank's mandate is to maintain price stability. So yes, it's good to maintain monetary policy in order to be consistent with that price stability, at least that one part of the mandate here.
And have the appropriate tradeoff. If the tradeoff occurs with the other part of the mandate, the maximum employment side of the mandate, that's in general, here in the contribution of this paper, I think it highlights the fact that yes, if interest rates are too low, and then you face inflation problems and if you are forced to raise the inflation problem, then the question is, if in particular you enter in this red zone, you might see the risks of creating some trouble in the banking sector.
That's something that is worth keeping in mind as you set policy for too low for too long, that you have to keep in mind that you are potentially raising the risk to cause a crisis down the road. You are forced, then, to raise rates to, let's say, for instance, mitigate the inflation problem.
Peydró: Let me say that, compared to the previous speaker in the panel—David, who was critical of the QE on the balance sheet—we have shown in the European Central Bank, we have a paper with Luc Laeven, who spoke yesterday, and Carlo Altavilla, for the monetary policy, that you know you can tighten monetary policy, lower inflation a bit, and still have control of financial stability. Like the QE/QT, trying to, on the transmission protection instrument, trying to mitigate some of these effects if you care about financial instability.
Wall: With that, I think we'll eat lunch outside. The next session will start at 1:30. I'd appreciate it if you join me in thanking our panel, our presenter and discussant.