Fiscal Influences On Inflation In OECD Countries, 2020-2022

>> John Taylor: Let's get started. You can still eat a little bit of your lunch if you like, there's plenty to go around. And so pleased to have Robert Barrow and Francisco Bianchi here to speak to us both simultaneously.

>> Elena Pastorino: Acapella.

>> John Taylor: A little bit, but it's such a great treat to have you here.

There's lots of people on the video, a lot of people here. We've just finished a town hall meeting, kinda was thousands of there, but not quite as many as we like. Anyway, welcome, and it's great to have you. So however you wanna organize your talk, we have till 1:40 or so.

>> Robert Barro: Okay.

>> John Taylor: 1:30, I should say, 1:30, okay.

>> Robert Barro: Okay, I will start, I think I'm more comfortable standing up. So I wanna talk mostly about primarily empirical project related to the recent surge in inflation, and particularly trying to explore how that's related to the fiscal surge that seemed to accompany the extra inflation, at least in many countries.

So we're gonna use a framework which we think relates to the fiscal theory of the price level, at least as we understand that theory, but probably will get corrected. So this theory's been around since the early 1990s, or thereabouts, quite a while. And I would say it's not been taken very seriously, at least by mainstream macroeconomists, to think empirically about inflation and the determinants of inflation.

And I think partly that's because of the view that inflation has much more to do with monetary events and monetary policy. Also, there's the fact that inflation was low and stable, at least in most advanced countries, until quite recently, really from the 1980s up to 2020. But the situation changed with the surge in inflation over the last three, four years.

And I think that's made a lot of economists more interested in the possibility that this is especially related to what happened on the fiscal side, and to think about that within the framework of this fiscal theory of the price level. So, again, this is primarily an empirical study, thinking about the determinants of inflation measured by consumer price indexes.

And we look at headline and core CPI in particular, and we basically look at the full set of OECD countries. We look at 37 OECD countries, focusing on the period 2020 to 2022. Although we'll soon be able to extend this to 2023, but we don't really have the data to do that quite yet.

It's 37 OECD countries because out of 38, we're confused by Turkey, which had a 70% inflation rate in 2022, and we're not pretending to explain that. So that's the one country out of the OECD that's not in the sample that we're examining. So the empirical framework that we used turns out to be quite important in terms of the empirical findings at all.

In our analysis that comes out of a simple version of this fiscal theory of the price level, we think about a kind of frictionless version of this basic model. And the key element in that is the government's intertemporal budget constraint, which basically relates the initial real market value of the debt, which is on the left side of this equation, to the stream of primary surpluses in the future.

So tax is less spending, G is total spending, purchases plus transfer payments, the way we're defining it. And it's a present value of all those primary real surpluses, where we're thinking here about a constant real discount rate, little r in this expression. So B is gonna be the nominal market value of short and long term public debt that's outstanding at the beginning, which is the start of some period t, and the price level is measured at the same point.

And we have this stream of tax revenues and spending in real terms. Of course, we're assuming a basic no-Ponzi condition in the intertemporal budget constraint, so you can't just spend and borrow and never have to pay for it. There's no kind of free lunch in this framework. We think in this environment, of a surge in spending, which is intended to represent the reaction, particularly to the COVID pandemic and sharp recession that occurred early in 2020.

And most of this, empirically, is gonna be in the form of transfer payments. But here we're looking at total expenditure, being what's relevant. So we think about it at the beginning as that there's some surge. So Delta Gt represents what the spending is in the first period, period t, relative to what people were expecting.

And that's true for all these future periods. And this episode of the surge in spending is viewed as temporary, going out to some date in the future, t plus capital M. So we're thinking about a temporary surge in government expenditure. That's the experiment that we're looking at. So this was unexpected from before period t.

But once you get into it in this model, we're thinking about it as all being completely understood and fully anticipated, which might have some elements of not being completely realistic. Empirically in the period we're looking at, what matters is the surge in spending, particularly transfers. There's not a lot of action on the revenue side, although you can bring that into the picture, but it doesn't actually matter much.

So in the relevant period, 2020 to 22, if you're looking at the surge in spending relative to GDP across these 37 countries, that average is 12%. So the boost cumulative over those three years relative to GDP is about 12%, which is a big number. The US is more like 19%, so it's bigger than the OECD average, whereas the changes on the revenue side are actually positive but relatively small.

And we're not gonna focus on that. Okay, so I'm trying to get now some simple implications out of the model involving this government's intertemporal budget constraints. I'm gonna make a number of approximations to come up with a very empirically tractable formula that we're gonna bring to the data.

So this is a series of some approximations that we think are reasonable, and then you get an equation in the end that's critical for the empirical work. And I think it's very intuitive as to what that looks like. So we're gonna think about real GDP as growing at some constant rate, which over the relevant horizon, which is out to date, t+M, is the same as the real interest rate.

In which case you can write the surge in spending as the expression in 4, which is gonna be in terms of increments to the spending GDP ratio.

>> Robert Hall: Is that a question? Yes, the MS just occurred in the US, where we have up to date data.

>> Robert Barro: I'm sorry, I didn't understand that.

 

>> Robert Hall: The number of periods.

>> Robert Barro: Yes.

>> Robert Hall: So expenditure. I've been doing something that I have to give this away, but the expenditure is back to normal.

>> Robert Barro: Yeah, I think that's pretty. Well, I think that's pretty much true, yes. So I think bringing in 2023, for example, is not gonna make a big change in terms of what we're looking at, I believe.

 

>> Robert Hall: Yeah, sure. Well, by almost the beginning of 2023, expenditure was right back to normal in the US.

>> Robert Barro: I think that's about right and it's not just for the US. I think it's true more broadly in this sample.

>> Robert Barro: Okay, yeah Luigi.

>> Luigi Zingales: So at least for European countries in 2022, there is another shock, which is the Ukraine war.

And the fact that if you are in Europe, you understand you have to increase defense expenditure for a long future, okay? So the Europeans have underinvested in defense, and the wall clearly has made it evident to everybody they have to invest. So if I were to look at this, I said there is a permanent increase in government expenditure, and given the level at their taxes, there not a lot of margin to compensate with the tax element.

So there's a pretty significant increase in the permanent deficit, in my view. But that's mine.

>> Robert Barro: In our empirical work, we do have something related to the Ukraine Russia war, which kicks in in 2022. Not over the whole period, obviously. And I think it's empirically of some importance, although it doesn't much interact with what we're most interested in, which is the effect of the fiscal stuff on inflation.

So I'm gonna get to that later. And there are a number of interesting issues that we haven't sorted out related to that, particularly in context of the eurozone, which I'd like to talk about more, actually.

>> Elena Pastorino: Sir, maybe a related question, clarification about annotation. This is the exposter present values of streams of these.

So PI is realized. The realized PI.

>> Robert Barro: So we're looking at equation four is about the spending.

>> Elena Pastorino: Equation five, sorry.

>> Robert Barro: Equation five. So we're thinking about it as though at time t, when you know about the surge in spending. There's also a shift in inflation away from what it was previously and what people were anticipating for the future at that time.

So this is what happens just after that, in terms of the actual path of inflation. So it's assumed that there's some increase in inflation. And at that point people understand the whole ramifications for the future. So they're looking out at the future path of inflation, which is what's in this formula.

So this is trying to calculate the market value of the bonds outstanding based on the fact that people understand is this inflation. And with a given real interest rate, you have the nominal interest rates which appear in the denominator.

>> Elena Pastorino: Should I put a conditional expectation index at time t for set or, that was my question.

 

>> Robert Barro: So in this simple model that we're using, we're thinking about it as though it's fully understood at that point. So there's no issue about unexpected inflation in the future. It's all unexpected from the point of view of before the shock.

>> John Cochrane: Add to what Luigi said, and to some extent Bob Hall, it's important here that these are government spending, these are deficits that people don't expect to be repaid by surpluses in the long, far off future.

And that is an important part of the story, sort of Bob Hall's point. We're back to normal, but we're not running anybody's surpluses in order to pay back this debt. And I view part of the story, the expansions of the inflation Reduction act and the chip act and so forth are kinda telling us we're not going back to surpluses anytime soon.

Similarly, in Europe, if Europe might have been going back to surpluses to pay off the debt. Cuz we're gonna pay everybody's energy bills and we'll see if we start raising in military expenditures, that is important part of the story. And I think a reasonable one, but one that's important to recognize.

 

>> Robert Barro: I haven't gotten quite to that yet, but I think what I'll get to is consistent with what you were just saying. Actually, let me go a little further and then I can address the point more directly. So here I'm thinking about the nominal debt as having associated with it a time path of repayments of coupons and principals.

And that's what all the B's are in the numerator of each of these expressions. So these are assumed to all be denominated in nominal units in dollars. So you can think of all that stuff as being hostage. It's out there. And if you do something with the path of inflation, implicitly unexpected from the prior point at which people were holding the bonds.

Then you can affect the real value of all that stuff and you can do it at different points in time, and that's what's gonna show up in this equation.

>> Patrick Kehoe: It's a simple question. If you only had, say, coupon payments, like the maximum maturity or debt, say, was out to two, just for an example, and you had a one time unexpected surge, then you'd have to deflate.

If B3t, etc, were 0, B3t, B4t, B5t, then you'd have to have it all inflated away in the first three, two periods, right?

>> Robert Barro: Yeah, you only get credit for doing stuff out to the relevant maturity of the bonds.

>> John Taylor: Right.

>> Patrick Kehoe: Though in countries with relatively short maturity structures, the inflation must be kaboom right away.

And if it's long, then you could stretch it out.

>> Robert Barro: That comes, yeah, a little later.

>> Patrick Kehoe: Okay.

>> Robert Barro: That comes as a central element. And we spent a lot of effort measuring the duration of the public debt in all these countries cuz it's actually not a statistic that's readily distributed.

 

>> Patrick Kehoe: Mm-hm.

>> Robert Barro: But we have estimates of that for each of these OECD countries. Yeah, that's one of the things that matters. It's not the only thing.

>> John Cochrane: If you have instantaneous debt, then you have to have a price double jump, right? Fiscal thing, price level jump. But what we see is inflation.

So one device to get inflation rather than a price double jump is long term debt, cuz you can inflate away the long term debt. The other is to add sticky prices. If you have short term debt and just your standard new Keynesian model.

>> John Taylor: Obviously.

>> John Cochrane: Then two or three years of inflation will eat away the value of the debt, even though it's totally short term debt.

It's just a little more complicated model to do it. This way works, but it's not. I don't think we need. I don't think we need that complication.

>> Robert Barro: You don't cuz you're using long term debt.

>> John Cochrane: But we have this period of inflation. We don't have a price level jump.

One way to handle that is long term debt. Another way to handle that is sticky price. I just wanted to tell Pat, if you had short term debt.

>> Patrick Kehoe: Yeah.

>> John Cochrane: You don't have to believe in price level jumps cuz sticky prices also do it at a little more analytical.

 

>> Robert Barro: There's a lot of confusion in this area by looking at the annual change calling inflation the 12 month change.

>> John Cochrane: Well, we do that empirically.

>> Robert Barro: At this point periods are very short. You can take a lot of periods and they're all extremely short. That's really what the empirical framework is.

 

>> Robert Hall: But if you do that, then you get very close to instantaneous increase in the price level.

>> Robert Barro: But that won't typically be optimal. That typically won't be the best way to engineer this kind of surprising place.

>> Robert Hall: I thought we were talking about what is not what should be.

 

>> Robert Barro: Yeah, what is, is what we're doing empirically. So it's not true empirically that all the action is in a jump in the price level. That's certainly not true.

>> Robert Hall: Yeah. If you change from this silly thing of using the annual change, which builds in a lot of inertia, the amount of inertia is very small.

It's very close to just a hop in the-

>> Robert Barro: We're not assuming any inertia. It's just that it's not optimal to have all the price action come in the near term. That's not the best way to do this from a kind of public finance standpoint.

>> Robert Hall: I was commenting on John's intervention, not what you're doing.

 

>> Robert Barro: Criticizing John is fine actually.

>> John Taylor: Catch up to what you wanted to say.

>> Robert Barro: Okay, so to get some analytical formulas need some assumptions about what the structure of the debt maturity looks like in terms of how much debt is due at various points in time going out to the future.

And we make some simplifying assumptions that come up with a very nice formula that we can apply empirically. So we assume in particular that the structure of the indebtedness is that other things the same, the debt payments due is rising along with the nominal GDP, basically. So part of that is the inflation that was there in the system, and part of it is real growth.

So PI star is our measure of what the inflation rate was before we had the shock. And you can think about that for most of these economies as corresponding to an inflation target and was not too far from 2% per year, a little bit less in some countries.

So the Pi star is giving you the path of the nominal payments that are due, which based on bonds that were issued before the surge, before day till. And then there's a discounting involving the nominal interest rate, which is gonna bring in the inflation rate ex posts after you have the surge.

And that's why you have this sequence, PI t plus one, PI t two and so on. So then you get this formula which tells you how the nominal debt is related to the amount of payments in the initial period, which can be an extremely short length of time.

So it's not a lot of time there. And then you have the sum here, which involves the path of all the inflation rates. And not surprisingly, when each inflation rate is higher, given the target PI, the nominal value of the bonds is gonna go down. This is the nominal value exposed to shock is gonna be decreasing.

So effectively, you'll be able to reduce the real market value of the bonds by having the inflation, which is not too surprising. Some of that is in the near term and some of that is further out into the future, but it's reflected in the nominal interest rates, which is what's discounting stuff.

We have a baseline nominal value of bonds which was outstanding before the shock. That's what's PT star. When there's a surge, it's gonna affect the sequence of inflation rates. Gonna assume that the price level doesn't jump. But remember, this is over a very short period of time here, length of the period.

So we have a given price level. We could allow for a jump in the price level. It doesn't really matter much, but the response to the surge is gonna be this sequence of inflation rates, which is gonna depress the real value of the bonds outstanding. Equation eight tells you how much do you reduce the nominal value of the bonds by having this time sequence of inflation rates.

So this is comparing how much was there initially with how much is there ex post. And you're gonna get a sequence of expressions which involved a target inflation rate PI, which was expected inflation before the shock, and the sequence of inflation rates. Okay, so now the question is, what time path of inflation rates do you wanna choose, being the government?

So the government here is gonna be viewed as a composite of the fiscal and monetary authorities who are essentially cooperating to handle some kind of outcome that is deemed to be efficient. Then as John says, if you have a situation where the government does not want to either raise future taxes or cut future spending, there's another option here which you can effectively default on part of the initial real debt by having this sequence of inflation rates.

And that's essentially a form of contingent public finance that is contingent on this emergency represented by the COVID crisis and the associated sequence of spending. You can engineer this partial default on the debt through engineering inflation. That was surprising from an ex anti-perspective. Now, as John mentioned, for example in his classic 2001 paper, there's a whole set of paths that you can pick for inflation.

Given that you're gonna be choosing this inflation surge, one possibility is that you will smooth out the response of the high inflation so that you choose the same higher inflation rate in each period up till the time when you get to the debt maturity. Cuz there's no point in having higher inflation beyond that because you don't get any revenue from it, as was just mentioned.

So I'm gonna work out what is the formula you get if you assume that the high inflation is smooth out and is constant over this interval from t to t plus m, which is the relevant interval. And then you get really quite a remarkably simple formula in the end, equation nine is sort of an intermediate formula corresponding to the place where the future inflation is constant at the level PI.

And if you make some further approximations, which is that the surge in inflation rate is not that large in some sense, you get equation ten, which is close to what we wanna implement empirically. So let me tell you what's important about equation ten. First of all, you have this PI minus PI star, so that's the inflation rate that prevails exposed to fiscal shock, and now assumed to be constant over some interval.

And that enters relative to the inflation expectations that were present before, which you can think about as the inflation target, the pi star. So I'm thinking about Pi being bigger than Pi star. That's gonna depress the market value of the debt. That's why there's a minus sign here.

Two things are important in this result, which turn out to actually matter empirically in an important way. One is that the inflation rate interacts with the initial level of debt. So Bt star is related to the market value of debt before the shock. So not surprisingly, if you have more debt denominated in dollars, a given inflation rate is gonna generate more effective revenue by depressing the value of the debt if you have more debt to begin with.

So the level of the debt appears in that equation. The other thing that appears is something you can think about as duration of the public debt. So this is a simple model where debt is outstanding only up to some maturity cap, T, like ten years or something like that.

And T over 2 is kind of like the duration of the typical payments, coupons and principal on the debt, in this simple setting. So that's gonna enter interactively with the inflation rate. So if you turn it around, and we're gonna try to think about what determines the inflation rate, and you put the duration on the other side, it says that the inflation rate you need to get a given amount of effective revenue, the delta B, is gonna be smaller per year the longer the duration.

So as was said before, if it's a very short duration, you need to clobber the inflation rate right at the beginning, with the limit being where you have a jump in the price level. So the way the debt and the debt duration appear here is quite important for the empirical framework, and actually, for the empirical results themselves.

 

>> Elena Pastorino: And I can ask, I was curious about the interaction between Bt and Pi. I was drawing in my mind the second order Taylor expansion, with Pi as the denominator of the expression, or expanding at the second order terms, I would have Pi to the powers. These terms must drop out because you are approximating around Pi equal to 1.

So-

>> Robert Barro: If I need an approximation like Pi minus Pi star times t has to be much less than 1.

>> Elena Pastorino: And then I would have thought you would have wanted the higher order terms. And so just killing it. The second order is-

>> Robert Barro: I don't either want or don't want them.

But I think that that's a reasonable approximation. I mean, Pi minus Pi star, if you think of everything per year empirically, is something like 0.03 per year. And-

>> Elena Pastorino: It shows.

>> Robert Barro: T over 2 empirically is about six years or something like that. So that's the kind of approximation.

For the US, the duration is about five and a half years now, most of it.

>> Victor Davis: It seems like when you take this and take it to the data, you're using pretty strongly that there's no sort of anticipation of differential propensities to inflate versus rate taxes, which caused differences in the initial debt maturity, right?

So if you're Argentina or something, nobody's gonna lend you for 30 years, or no rational person is gonna lend you for 30 years. So you could imagine that in some sense, if you find this correlation, it may not be just this causality, but it's something about the difference on the government.

 

>> Robert Barro: In the end, we have three things that we say matter for inflation. The size of the government spending surge, the amount of the public debt, and the duration of the debt. And I think you can come up with good endogeneity stories for all three of those and maybe have an interpretation that's different from what we're giving to it.

But I think one of the strengths, and what we have is that we have a very strong constraint in the way these three things enter into the determination of inflation, which comes from the model, from, basically, the government intertemporal budget constraint. And we think that works. So some of the results are kind of counterintuitive, I think, right?

I think if I'd asked people what's the connection to inflation from having more public debt outstanding at the beginning relative to GDP, I think most people would have said that it would be positive. But it's actually negative, and it's actually quite strongly so, anyway.

>> Victor Davis: But it's like having a bigger base of tax revenues.

 

>> Robert Barro: Absolutely, it's the base. It's the base for inflation.

>> John Cochrane: To be fair, countries that have a lot of debt outstanding typically have more trouble promising to repay that debt, which is why they get into more trouble with inflation. Your statement is exactly-

>> Robert Barro: It's not so clear, because in order to have lots of debt, people had to be happy to hold it.

 

>> John Cochrane: But you're closer to that lateral limit or whatever. It's not so stupid to say higher debt engenders inflation. What you're holding constant is the ability to repay the debt, which if you're holding it constant, you're exactly right. If you're not, maybe they have a point.

>> Robert Barro: Okay, so let me put the results together now.

So as I said before, there's kind of another alternative here to either raising taxes or cutting spending, which is you can have this effective capital levy on the existing government bonds, which is like a form of contingent revenue. So we're gonna introduce a coefficient we call eta, which is the extent to which you're gonna finance the extra fiscal outlay through this inflation debt default mechanism rather than through traditional classic public finance.

We either raise future taxes or cut future spending. So we introduce a coefficient eta between zero and one to represent that, and then we end up with this equation that we wanna use in terms of a framework for the data. So on the left-hand side is the surge in inflation.

It's the inflation rate over some interval, which we look empirically from 2020 to 2022, relative to some kind of inflation target, which we measure empirically by the average inflation rate between 2010 and 2019 in that country. There's an alternative way to do it, I'll talk about later, but that's the surge in inflation measured per year.

So is eta gonna vary by country? Well, I think underlying it, I think you would tell a story where, because of differences in the political systems, eta is likely to vary by country. But of course, we're doing the estimation as though eta was a single coefficient that we're gonna estimate.

And I think there's a lot of interest in what that turns out to be.

>> Speaker 11: But you could model it if you thought there were a few variables that helped explain that propensity and then still be pretty easy to estimate.

>> Robert Barro: Maybe, particularly if we had some political structure type variable.

Yeah, that might also be a basis for having some instruments for the estimation, but we haven't done that successfully.

>> John Cochrane: I would also add, variation in data across countries is the error term.

>> Robert Barro: Yeah, it's the error term.

>> John Cochrane: So to the extent you see some graphs with some real nice r squared, I view the nice r squared of the graphs as kind of an empirical finding.

Hey, a lot of countries, pretty much the same ratio of how much we inflate a waiver Versus how much we spend, and that's an empirical finding.

>> Robert Barro: Okay, so just to summarize, I think I've said this, the stuff on the right hand side is the three terms I was talking about.

One is the size of the spending surge, which we're measuring here in terms of spending relative to GDP. And empirically, it's from 2020 to 2022, but then it's scaled by the initial debt GDP ratio, which empirically is in 2019, and by the initial duration of the public debt, which is also measured in 2019.

So that's what's on the right hand side here. Okay, so you can think about this as the kind of state contingent public finance which is in a well known Lucas and Stokey paper from 1983. And they were particularly focused on World War II and the idea that in some emergency situations like that.

It might be optimal from a public finance standpoint to effectively default on part of the public debt when there's a very clear emergency like that. So this is, I think, intended to be analogous to that kind of result that was in the Lucas and Stokey paper. And maybe in a much more ambitious project, you could have lots of different episodes that look like emergencies, but our empirical work here is focusing on the COVID period and the subsequent, particularly fiscal transfers.

Now, you might think that in most circumstances, Ada would be equal to zero. That's when you get the classic public finance, that whenever you have a deficit, it has to be matched by the present value of future net surpluses. So you might think that that's what holds in most times, in which case, in kind of normal times, there wouldn't be much predicted relation between fiscal stuff and inflation, and implicitly, monetary stuff.

So the idea is that this holds when there are clear emergencies and maybe commitment issues related to the government declaring an emergency are not such a big problem. And that's what we think fits with this COVID crisis, yeah.

>> Valerie Ramey: I was just gonna say, there's a recent paper by Eric Leeper and Bruce Preston testing fiscal theory of the price level with FDR leaving the gold standard in 1933.

So you might look at that, because, again, it's one of these event studies.

>> Robert Barro: I haven't looked at that.

>> John Cochrane: Great paper.

>> Robert Barro: Okay, that's the functional form we're gonna use in our empirical work, it's guided by that result. So I think I've already said enough about what that model implies.

Now, it is true empirically that the one thing that sort of stuck out as an additional effect on inflation, besides this fiscal influence, that I've been describing is something related to being close to the Ukraine/Russia war. We looked at a number of ways of measuring this, but it seemed like we really couldn't do much better than whether or not a country shared a border with Ukraine or Russia, of which there are eight countries among the 37 in the sample.

So I'll look at those results in a little while, and I think it's particularly interesting in the eurozone as to what that looks like, I'll get to that. Really, about the data, so, again, the sample is the sample of basically all the OECD countries. Headline and core CPI inflation rates are not too difficult to obtain, and the OECD is actually very good for providing that.

The data on government spending and quantities of public debt come from the IMF, the concept from the IMF that we're using is general government. So general government is all layers of government, plus social insurance funds, that's the definition from the IMF. Most notably, this does not include central banks, and there are a number of issues about whether you wanna consolidate central banks with the rest of the general government.

It's a little tricky because you need to have both the asset and the liability side of the central banks, and I think with respect to quantities of public debt, it's not gonna give you very much, it's mostly gonna cancel. It clearly affects the numbers on maturity of the debt, because the central banks tend to have liabilities that are much shorter term than their assets.

Anyway, we don't have those numbers at the moment, and IMF does not provide those data, they treat central banks as part of the private sector, and it's not part of this universe. Yeah.

>> Speaker 12: Marty Feldstein used to talk about some of these other nominal liabilities, like depreciation allowances.

I don't know if you have an estimate of order or magnitude, I guess in the US it'd be 5 to 10 trillion, something like that, compared to our public debt, which when this started was 20 something.

>> Robert Barro: I mean, something like depreciation allowances is gonna end up affecting effective tax rates on income, and of course, there are other ways you could do that if you really wanna do that.

 

>> Speaker 12: It's also a capital level.

>> Robert Barro: That's a question, it's a capital lever, you mean on the capital to which the depreciation is.

>> Speaker 12: All these companies are lost appreciation. It's kind of government debt, the government promised them that they can have these credits in the future, and now they're not worth anything.

 

>> Robert Barro: I have to think about that more but we certainly don't have that.

>> Speaker 12: Yeah, also emphasized the distortionary element of it, but for your purposes, he was talking about anticipated inflation, but you're of course about surprises. So is the levy part, that would be interesting for you?

 

>> Robert Barro: Well, the idea here, as in Lucas and Stokey, is that if you can really do these once and for all capital levies, it tends to be attractive from an efficiency standpoint, cuz it's not distorting. Obviously, if you keep repeating it, it's a different matter. But factoring into the public finance problem is the idea that this kind of levy here accomplished through inflation, has some efficiency aspects, and that's analogous to what Lucas and Stokey you're saying.

 

>> Elena Pastorino: Can I ask you? Sorry, reacting to with the lag, I was curious about your thoughts rather than the algebra, your comments about ETA, the elasticity turns, potentially being time or state dependent. I was thinking about the discounting structure, so there are geometric discounters, and whoever is out there being the planner is at least as long lived as the debt that has been issued.

Is it something that you think in practice? In theory, of course, is it all irrelevant in a way? But do you think these are considerations that matters at this stage of the implementation?

>> Robert Barro: So I was taking a standpoint of efficient public finance and not worrying about behavioral type elements related to public officials doing something different than what you would set up in an optimal tax problem.

So then the horizon of the politician is not a separate thing.

>> Elena Pastorino: Even like a limit of means type of planner, rather than a geometric discounting planner.

>> Robert Barro: Anyway, there are a few other issues here that I won't, there's a question what you should do with debt denominated in foreign currency, or in an inflation index form?

And I can say something about that, but I think I'll skip over that for a moment.

>> Patrick Kehoe: Just following up on Casey, so anything the government has promised that's in nominal terms. And it's not gonna adjust in the short term, say wage contracts, all the government people, etc.

In some of these countries, that's all another gain, so you get an extra book bang from that too. It's not just delta true b, it's delta all, nominal commitments that you surprised them on, at least in the short run.

>> Robert Barro: I mean, not all of them are gonna factor favorably into the government constraints.

 

>> Patrick Kehoe: That's right.

>> Robert Barro: Cuz that's what would matter here.

>> Patrick Kehoe: So if you have 10 million government employees and they have.

>> Robert Barro: That would, yes.

>> Patrick Kehoe: Do some nominal contracts that you're gonna chip, if I took a big salary cut cuz inflation went to 10%, I only got a 2% raise, that's

 

>> John Cochrane: But Pat, that would get measured here in the Delta G, I agree conceptually.

>> Patrick Kehoe: Would it?

>> John Cochrane: They measure government spending, so government spending.

>> Patrick Kehoe: Government, come consolidated G, some huge G.

>> Robert Barro: Yeah, we measured a total general government spending, including transfers.

>> Patrick Kehoe: So that would already be in there?

Do you get those for all those countries? I doubt it.

>> Robert Barro: I guess it would be in there.

>> John Cochrane: I think theoretically, G is also a function of, there's a good way to handle it, I think of the flows also being a function of the price level, which.

 

>> Patrick Kehoe: Yeah.

>> John Cochrane: So that the right hand side can move with the price level that way too, but I think they measure it straight out, so it's not a problem.

>> Patrick Kehoe: Okay.

>> Robert Barro: Okay, just one comment on the duration of the public debt, so, those data as I mentioned, were harder to find.

There's some information reported on what's called maturity of the debt, or average remaining maturity, which is only about principal payments. So we came up with some formula to approximate the duration based on information on what's called maturity. Given also some information about what the nominal interest rates are.

So we end up with estimates, particularly around 2019, of the duration of the public debt for all of these countries, so that's what we're using. It's funny the IMF had a table like this, which they abandoned for some reason in 2010. They used to report, maturity and duration of public debt across the OECD, not for all the countries.

And when I looked at it, the numbers look kind of nonsensical in some cases. So maybe that's why they abandoned this in 2010, I don't know, but they no longer report that information.

>> Speaker 11: Can I ask a question about this border with Ukraine thing? I mean I would have thought, God knows why they invaded Ukraine, but if you thought about Russian ethnicity, that seems to be what drives a lot of the fear of Russian invasion.

So that leads to more to the Baltic states, as the ones that might be most worried about this. They certainly talk about it the most, who also shares a border with Ukraine, they're probably not spending more money because of this. I mean, I don't think these matters, you have to be insensitive but

 

>> Robert Barro: Yeah, there's eight countries that have a border with one or the other.

>> Speaker 11: Right, I just think there may be, I mean, you could probably look at some data on what they've actually done to military expenditures. To figure out which ones seem to be responding more strongly or tie it more to.

 

>> Robert Barro: Well, we're gonna hold constant spending as far as we're able to estimate, at least cars.

>> Speaker 11: So the idea is that a fear of, so you're trying to capture some fear of likely conflict, that's gonna increase expenditures in the future, but we're not seeing it, that's the idea.

 

>> Robert Barro: We have some impermeable results related to this, which seem to be fairly strong. I wouldn't say we have a lot of detailed information in exactly how to interpret that, which something that we're trying to think about further, yes?

>> Valerie Ramey: I was just gonna say, you might also consider countries that have pipelines linking them for natural gas to the Soviet Union.

 

>> Robert Barro: Yeah, I think some of it works through that, yes I think so. Okay so, this is very old style econometrics, we're basically looking at cross sectional regressions. The only information here is, in the main sample is a single cross section of 37 countries. You can do the analysis, it turns out to be very similar.

Instead, you can use all the annual time series from 2010 to 2022, and then you can have a fixed effect. You estimate for each country, which is gonna give you an estimate of the inflation target. That's gonna produce results that are very similar to what we get from this simple cross sectional regression.

 

>> Speaker 12: Don't you kind of want no effect before COVID hits? I think you said before, right?

>> John Taylor: You might expect zero for Ada in quote unquote normal times.

>> Robert Barro: Ideally if Ada is just putting down the constant charm, which is related.

>> Speaker 11: But if you just estimated for that sub period, do you kind of get a much smaller ADA?

But you said before that in normal times, when the sort of typical public findings.

>> Robert Barro: Well I mean, you get zero explanatory now, you didn't from 2010 to 2019.

>> Speaker 11: Okay.

>> Robert Barro: There wouldn't be anything wrong, okay.

>> Elena Pastorino: It is surprising, isn't it?

>> John Taylor: Which?

>> Elena Pastorino: The longitudinal versus cross sectional.

 

>> Robert Barro: No, I don't think it's really surprising, cuz the 2010 to 2019, those data are really just pinning down the constant term, which corresponds to the inflation target, the PI star. Whereas in our other cross section, we just assumed that that corresponded to the average inflation rate.

>> Elena Pastorino: This way it validates your specification in a way.

 

>> Robert Barro: Yeah, I look at it that way.

>> Elena Pastorino: I'm not used to that.

>> Robert Barro: Basically, the fixed effect you get corresponds pretty much to the average inflation rate between 2010 and 2019.

>> Elena Pastorino: Amazing.

>> Robert Barro: Yes okay, let me talk about the sample a little more in detail. 37 OECD countries, so 20 are outside the eurozone, 17 are eurozone countries.

There are three OECD countries that are not on the euro and I'm leaving that as a trivia question, free to answer us what they are. Anyway, they're not in this sample.

>> Robert Hall: Okay, I was thinking about something else.

>> Robert Barro: Bob always knows every single question.

>> Elena Pastorino: Everything.

>> Robert Barro: Anyway, question is, do you wanna do something different with regard to the eurozone countries, which obviously have a common currency and a single central bank?

So we start here with a specification, where we're thinking of the eurozone as a single big economic entity. With some kind of inflation rate determined, which is gonna be a GDP weighted average of the ones we observe for all the individual countries. And then later we test whether this is a reasonable specification, or should we instead include all the euro area countries individually?

So, we have the results both ways and I'm kinda surprised at what the results are actually. Anyway, the initial specification has 21 economies where the euro appears only once, along with the 20 non euro countries. So, this is what the cross sectional regression looks like, the left side has headline inflation and the right has core inflation taking out energy and food.

Can I pin down?

>> John Taylor: Yes.

>> Robert Hall: What's the interval at the level of the price level, then inflation is the time derivative, are you measuring it over a whole year?

>> Robert Barro: It's measured per year, kind of as an average between 2020 and 2022, so it's over three years.

We have three annual inflation rates and it's the average of those-

>> Robert Hall: So what you're seeing is they're blurring, the timing, which seems like-

>> Robert Barro: Well, that's because the model speaks most compellingly to the cumulative price level effect. That's what's really wiping out real debt. And exactly the timing of when that occurs is much less central to this underlying theory.

And in order to get that result, we assume that the extra inflation was smoothed out, equalized over time, which is not really completely accurate. So I think it's the cumulative thing that's really a much more robust prediction from this model.

>> Robert Hall: It's actually the three-year change in the level of the price.

 

>> Robert Barro: Yeah, essentially, yes.

>> Robert Hall: It would be nice to check that out, cuz it seems superficially like people around the table are talking about very quick movement of the price level. But you're saying, well, it doesn't really matter whether that's spread over three years or whether it occurs in-

 

>> Robert Barro: But the duration of the debt is quite important in the model in terms of what it does to this inflation rate over the three years. And the duration of the debt is important in the model as to how that works out. And then we can check that out empirically.

 

>> Patrick Kehoe: The smoother the payments are, the more that would be a good approximation, right? If you had like a huge coupon payment coming in a year and teeny, teeny, and then another country had teeny, teeny huge, then you get some side action from the which way that inflation occurs.

 

>> Robert Barro: You want the price level change to conform to where the debt is.

>> Patrick Kehoe: Yeah, that's what I mean. You wanna hit it.

>> Robert Barro: Yes.

>> Patrick Kehoe: So that could be a secondary thing to look at.

>> Robert Barro: So we're thinking about more things along these lines in terms of the timing of the responses.

So there are a number of other considerations also. But this simple model just assumes that it's-

>> Patrick Kehoe: Yeah, it would be like a secondary check to see with simple variations, like decaying maturity structure, for example, we get very similar results. Now, Robert was actually suggesting, well, maybe you wanna kinda think even beyond the simple decaying in inflation path.

But for simple variations of this idea, like you were saying, that maybe there is a lot of debt, a short maturity, and then it declines over time in terms of cross sectional ability to explain inflation, you get very similar results.

>> Robert Barro: Yeah, I wanna discourage taking the timing of a model that has totally flexible prices, that seriously.

 

>> John Cochrane: You put a little bit of sticky prices in, and what happens is the inflation comes, the total price level adjustment is exactly the same, but it takes two or three years to happen. So a lot of our stimulus was in October of 2020. Inflation started January 2021.

No, there's something, of course, prices are a little sticky. So this is kind of a parable for what would happen if you threw a little bit of price stickiness in. And that encourages, look at the beginning and the end, look at the total price level increase. It always works as if the frictionless model over two or three years, but don't get really excited about frictionless models timing.

 

>> Robert Barro: I mean, what the timing looks like empirically is that the inflation rate is certainly building up, and it peaks in 2022. And I'm sure it'll be lower in 2023, but we haven't used that yet. Yeah, Casey.

>> Casey: Another advantage of your cumulative approach is, the data is not very good.

There are tons of goods and services not available, so as long as they're available in 2022, you're good.

>> Robert Barro: You are right, pretty much.

>> Casey: In my experience, living through it, there's a huge inflation in 2020 and then a deflation, as I was actually able to buy things again, so.

 

>> Robert Barro: The World War II is even more like that in terms of rationing and price controls. You'd have to take that kind of approach also, if you wanna apply this to that.

>> Valerie Ramey: I was thinking that this specification misses out on a key test in your theory, which is using the variation duration.

Your theory says there should be an interactive term, but it seems like to test that you should also have the level effect, that is, have the excess government spending, gross debt as a variable, and then the interaction. And then see if there is additional explanatory power from the interaction with duration.

 

>> Robert Barro: So we took the specification that the model said should apply, and then we look individually what happens if you knock one of these variables out one at a time. So you can tell whether there's sort of a separate contribution from these three pieces.

>> Valerie Ramey: So you do split them in your robustness check?

Because think of a Phillips-

>> Robert Barro: We try to test whether these are individually important for explaining inflation.

>> Valerie Ramey: Because an alternative would be some kind of Phillips curve aggregate demand thing, in which case the excess government spending would be in there, but not necessarily interacted with duration. So that's.

 

>> Robert Barro: What that looks like. But first, let me give you the main empirical findings in this table.

>> John Cochrane: Just that the duration business only matters if you're taking very literally this flexible price thing. If you do this with just the standard new Keynesian model stickiness, then you get exactly the same thing with zero duration debt, which is two years in which nominal interest rates are not moving, inflation is high, and that slowly inflates the way the value of short term debt.

So the duration business is really not central.

>> Robert Barro: Empirically, the duration as we measured it, is important for explaining. It's one of the things that's wonderful if it works. And if it didn't work, it wouldn't be a problem.

>> Robert Barro: Let me highlight the main results sorry?

>> Speaker 11: Hopper is turning in his grave, never mind.

 

>> Robert Barro: Okay, so let me focus on Columns 2 and 4 here. Headline core CPI inflation. These regressions include the spending variable that I've gone over several times, which includes these three pieces that are supposed to be important. It also includes this dummy variable for whether a country borders on Ukraine or Russia, which turns out to be important only for inflation in 2022, which I think makes sense.

Anyway, the results are kind of cleaner with that in there. But the basic findings with respect to the connection between fiscal stuff and inflation holds up, even if you just throw that out. So if you look at the coefficients we estimate on the spending variable, particularly in Columns 2 and 4, you can see it's somewhere between 0.4 and 0.5.

It's not that different for core CPI inflation versus headline CPI inflation, the fits are actually very similar. I think the r squared is surprisingly high, in fact. But the border dummy is also very statistically significant. So we're focusing on the results that include that. So let me show you what this looks like here.

So this is just a way of depicting those regression results, where the horizontal axis has this government spending variable with its three components. The vertical axis has the change in inflation rate, 2020 to 22, versus the previous ten years, measure of the surge in the inflation rate. And this is showing you the 21 economies That are included in this baseline regression.

So there's a very broad range actually in the extent of the fiscal stimulus and the extent of the inflation surge. But I think it's quite surprising as to how closely related these two things are. This is actually the first thing I did. This goes back so some time I looked here only at the spending part.

So the horizontal axis here is the excess government spending. It's not related to the public debt or to the duration. The vertical axis is the same, it's the change in the inflation rate. So here the two things are positively correlated, but it's very weak and at most marginally statistically significant.

And I remember when I first got this set of results and I was really kind of disappointed cuz I thought this was gonna work and this looked like a failure. And then I sent this result to four or five people and said, what do you think about this, Francesco, being one of these people?

And I didn't get back very useful comments-

>> Robert Barro: Except from Francesco, I guess. So now we started working on it together. So this result, which is pretty weak, kind of motivated us to go back to think about the fiscal theory, the price level in the context of the intertemporal budget constraint.

And then figure out what does that model actually say should matter for inflation. And that's when we got the result that I've already described, which included these scaling effects related to public debt and related to duration.

>> Elena Pastorino: I can ask you even when you are estimating maturity, and this is not a linear parameter.

Well, it's a linear parameter models, but there's, it's an error in variable models because he's estimated.

>> Robert Barro: So is everything. I mean, I talked to the IMF people about the public debt data. You talk to them too long, you get scared of any information in this-

>> Elena Pastorino: I was worried about an IV to clean that measurement error rather than the end of if it would made a difference.

 

>> Robert Barro: We haven't done very well with companies coming up with instruments. I thought that the extent of the COVID infection, measured in some ways would be a good instrument, but it doesn't have any, hardly any explanatory power, turns out.

>> Patrick Kehoe: Is there a correlation with things like QE or the central banks buying bonds as they're spending, like bringing a little monetarism back into this.

 

>> Robert Barro: So then you'd have to say what was exogenous about that. I mean, the spirit of this here is what you're thinking about is what you want for a path of the price level or inflation rate.

>> Patrick Kehoe: That could be the implementation of.

>> Robert Barro: And then implicitly, the central bank is cooperating with the fiscal authority to engineer the path of inflation, and we haven't gone into the details of the monetary instruments that are supporting that.

I don't know whether we would get more information by going into that more deeply.

>> Valerie Ramey: But YouTube can help you.

>> Robert Barro: What kind of help.

>> Elena Pastorino: Exactly, that's what I think.

>> Valerie Ramey: We have, one central bank and different. 17 fiscal policies. So I had heard that that's part of your way to distinguish it from a monetary point of view is to look within Europe where you have variation in fiscal, not monetary.

 

>> Robert Barro: So let me go.

>> Patrick Kehoe: So what is he gonna do when inflation starts different by country? His theory ain't gonna be able to handle that. If Greece has a lot different inflation than Italy, what's he gonna do in the model to deal with that? I wanna know.

 

>> Valerie Ramey: What does your model predict?

>> Robert Barro: Let me go into the results more on the eurozone, cuz we do have some other examples by region.

>> Valerie Ramey: Could I just say, that duration is a state dependent. It's the kind of stuff that I've been doing with government multipliers. And you just estimate it as a state dependent thing, where your state is the duration of your debt.

And you look, if that effect of the excess change in government spending, that effect matters according to the duration of your debt. It's the simplest thing to just implement in your system.

>> Robert Barro: I'm not sure about this. I mean, there's a lot of variation across countries in duration as we estimated it, but there's not a lot of action in the time series dimension.

 

>> Valerie Ramey: No, it's-

>> Robert Barro: Changing a lot.

>> Valerie Ramey: No, so you showed two beautiful graphs, the really clean correlation when you have the interaction term.

>> Robert Barro: Isn't that beautiful?

>> Valerie Ramey: And then you showed the one that was messier, that didn't have the interaction. That tells me that slope varies according to the duration.

And you can just easily estimate that in your regression.

>> John Cochrane: Wait, he also didn't have the level, the debt in the simple.

>> Valerie Ramey: Yeah, that's why you wanna do it in a regression.

>> Robert Barro: Before I talk about the Euro, let me talk about that. So I looked at the three components of the spending variable.

So we took some different approaches to how to do this. We took these variables one at a time and artificially set each of them equal to the sample average rather than having any cross sectional variation. So that's a way of constraining that this variable doesn't enter into explanation.

So we did that sequentially for the delta G variable, for the initial debt, and for the duration. And if you think about that as amounting to a restriction. It's not exactly right, but almost. Then you can test whether these are individually significant. And that's what you get here.

It turns out the delta G and the initial debt GDP ratio are about equally important. And they're both very statistically significant with p-values that are very close to zero. And the duration is also significant, but not quite as pronounced in terms of the p-values. So that's what we got when we did it individually.

 

>> Speaker 11: What if you just take logs of the expression, test the constraint.

>> Valerie Ramey: We can't take logs because-

>> Speaker 11: Out of the error process.

>> Robert Barro: First-off, there's a constant term in the baseline equation, and secondly,- Predicted as an additive constantly.

>> Speaker 11: Okay, that's right.

>> Robert Barro: An additive constant.

Yeah, that's what's in the model. But also some of the delta G and inflation variables are negative, which there's nothing wrong with. So we thought about doing something analogous to taking logs, but we haven't figured out a good way of implementing that. So you could sort of see three separate coefficients.

So this was our closest approximation to doing that. Let me talk about the Euro-zone results. The baseline model entered the Euro as a sort of GDP weighted average of all the stuff in that zone. And it was just one of the observations. So now here we're gonna enter each Euro-zone country separately, the 17 of them.

So the total sample is gonna be 37 countries. And we enter the fiscal variable in two ways, which appears in this. So, for example, if you look in column 4, which is core CPI inflation, there are two excess government spending variables that appear. So the first one is like what we had previously, where for non-Euro countries, you have the actual value of that.

For each Euro-zone country, you have the aggregate for the Euro-zone. That's what we had in our first specification. So that's the first variable. The second one enters the individual value for each Euro-zone country. So there's a lot of variation across the Euro-zone in terms of the fiscal surge and also in terms of inflation rates.

They're not at all the same. Anyway, here we estimate the two values of that. Now, if the only thing that matters for a Euro country is the euro aggregate, then the second Euro variable should have a coefficient of zero, and that's what we're checking out. So if you look at column four, you can see that's basically what the results is.

And column two is not quite as strong. There's some marginally significant positive coefficient on the individual Euro-zone fiscal spending. So let me say something which we're thinking about. An interpretation of this is that there's some fiscal aggregate for the Euro-zone in terms of the surge. And if you think there's a lot of mobility across the Euro-zone countries in terms of goods and factors, you might not expect that to affect the inflation rate differentially in one country versus another.

That is, you might not expect it to have relative price effects, basically. And that's what this result is consistent with. However, there's one other finding which bears on this, which would be great to have more reactions to. We had this border dummy variable, which, remember, is holding fixed, the size of fiscal spending that's being held constant.

And then we're looking at the effect on inflation from the border thing, and that was very positive. But we find if we think about these separate Euro-zone countries, the border thing is something that affects the Euro countries individually, not just through the aggregate. So I think that's because what the Euro-zone, excuse me, what the border dummy is picking up is stuff that does affect relative prices.

So if you thought about something as affecting cost of producing goods and services in one country versus another, let's say, within the Euro-zone, then you would expect that it would differentially affect the inflation rates in the different countries. And that wouldn't be true for just a kind of aggregate demand variable with the fiscal stuff.

So I think maybe in the end, it makes sense that the fiscal stuff, it's the Euro aggregate that matters for each Euro country. But for some other stuff, there might be individual effects. And that's what we're getting from this Ukraine Russia proxy for the effects of the war in 2022.

Anyway, we have those results.

>> Elena Pastorino: Then a question I was thinking about all these facilities that, including jointly issued debt, that the COVID crisis stimulated, finally or not finally, the EU to adopt. I'm thinking from the perspective of Italy, there were extraordinary lending facilities, some straight transfers and many non repayable grants that were rather substantial.

And in fact, I'm in countries like Italy, I know for a fact struggle very much whether or not meaning there was a lot of uncertainty as to whether there was really a huge amount of money with no string attached. Or it was part of the conditionality of the transfer that would have started a year later with the next generation geo plant, which for Europe is a time the Marshall plan.

So it's a gigantic amount of money that is being redistributed.

>> Robert Barro: How would that affect the results in terms of, let's say, inflation in Italy versus inflation in-

>> Elena Pastorino: I mean, to me it makes perfect sense that the individual country per se doesn't seem to be meaning or measuring much for the theory, but it's the joint aggregate, especially for the 2021 period, because all these joint coordinated facilities starting to measure even more.

 

>> Robert Barro: So you're saying we're not measuring the fiscal spending variable properly? Within the eurozone because of this facility?

>> Elena Pastorino: Yeah.

>> Robert Barro: It's huge.

>> Robert Hall: Can I get you just explain the second two rows. Which one is the own country and which one is Euro-zone?

>> Robert Barro: The second one is the own country.

The first one has the own country for non Euro-zone and it has the Euro aggregate for each individual Euro country.

>> John Cochrane: So column four, the result is that the individual country deficit does not matter, it's just the Euro.

>> Robert Barro: Yes.

>> John Cochrane: Okay.

>> Robert Barro: Which is wanted us to find some months back, as I recall.

Absolutely, and I mean, to the answer to the theory questions. How are countries in the Europe different? How are state's different? There are real relative price differences.

>> Patrick Kehoe: Sure. No, but I was thinking if you really wanted to do that, you could have like in the appendix, split out the price level into traded goods prices and non-traded goods prices.

And if you took the simple assumption all traded goods prices are the same, that would have almost by assumption be a common. And if you see anything, it'd be like restaurant prices, haircut prices, all that non trade. Maybe you get more action if there was some residual country specific effect in the subcomponent that's super non traded.

This is a check.

>> Robert Barro: We have core and headline, which doesn't exactly match up with what you were just saying. But-

>> Patrick Kehoe: Those guys are way more open than we are, like we have 10% exports of GDP. Some of those guys have 40, 45. So they're pulling out the non traded just for fun.

I don't know if we would change your theories. Just if I was looking for a differential effect.

>> Robert Barro: Well, one thing we're doing-

>> Patrick Kehoe: I look for, the non traded goods prices-

>> Robert Barro: Right now, we're applying this to the US states.

>> Patrick Kehoe: Yeah.

>> Robert Barro: And that's kind of analogous to looking at the Euro-zone.

It's just a delay in the data. It's gonna take another year to have the relevant data through 2022 for the US. But I mean, it's feasible, but I don't have the numbers yet. But the idea would be to test whether the US states are analogous to the individual Euro countries in terms of the response of individual state inflation to individual fiscal stuff versus aggregate stuff.

We're working on that.

>> John Cochrane: There is a potential effect, which you guys are finding out there, which I love. Potential effect is that if you run up big deficits in Italy, you're more likely to leave the Euro and therefore get your Euros. That would be a way in which you could see country deficits line up with inflation.

Even in Europe, you're not seeing it, you're gonna stick with it.

>> Robert Barro: Yeah, so that would be a kind of low probability disaster event, and in that event it would be different. These results would say if Italy left the Euro-zone, they'd be like one of the 20 non Euro rate, right?

 

>> John Taylor: So what's the theory that justifies the states? Looking at the states?

>> Robert Barro: Well, they have different fiscal variables, but there's a lot of, again, mobility across the states in terms of goods and factors, which, as we said, that might be different for traded and non traded goods.

But that might equalize the effects even though the fiscal influences are different.

>> John Taylor: But who owns the debt?

>> Valerie Ramey: Yes.

>> John Taylor: You are assuming each state owns its own debt.

>> Robert Barro: Yeah, I was starting as if I can think about separate government. I mean, there are some interactions between the central.

 

>> John Taylor: They assume the feds are gonna bail them out?

>> Robert Barro: I don't know what you mean by-

>> John Taylor: California they didn't.

>> Robert Barro: Do you mean literally in terms of bankruptcy or are you talking about something else?

>> John Taylor: Yeah, no, bankruptcy.

>> Robert Barro: I guess that would matter but the assumption all along was that there wasn't gonna be that kind of explicit default on the debt and bankruptcy.

We already assumed that that was true. And I guess in the eurozone we assumed that nobody was leaving it.

>> Valerie Ramey: They're debating what's happening?

>> Robert Barro: There's actually no provision for state banking bankruptcy in law.

>> Valerie Ramey: Sorry.

>> Robert Barro: There's no provision in law for a state bankruptcy.

>> Patrick Kehoe: They can default, they can't go bankrupt.

 

>> Robert Barro: I can default.

>> Patrick Kehoe: Absolutely.

>> John Taylor: Why don't you continue on your own paper?

>> Robert Barro: Okay, these are just figures when you separate the euro countries. So this is 37 countries and the blue line is for the euro countries. And you can see there's not much relation between the inflation and the individual fiscal stimulus.

So the red line is for the 20 non euro countries and the blue is for the 17 eurozone countries. And I'm graphing the relation between inflation surge and spending surge. This is just related to the regression that I already.

>> John Cochrane: Sorry, that's country by country inflation rates infinite when you split them out to 20.

 

>> Robert Barro: Yes, these are individual country, yeah.

>> John Cochrane: Country by country inflation rates and country by country, Fiscal. Fiscal.

>> Robert Barro: Yes, and as I say, it's not like you can see it here. It's not like the Euro-zone countries are the same with respect to their fiscal stimulus, quite contrary, yeah.

 

>> John Cochrane: Suggest that countries that peg to the dollar might be different than countries that pegging to the dollar is in some sense a fiscal commitment. Did you break that out at all or is there nobody left to peg to the dollar?

>> Robert Barro: So one thing about the Euro-zone is that they have a single central bank, only one short term nominal interest rate.

But there's also a lot of mobility across the countries within that zone, which is often thought to be a key ingredient of when you're gonna have a common currency. So if I was trying to look at that across these other countries in terms of pegs, I think I'd worry about that aspect of it also, I think.

 

>> Robert Hall: Denmark is pegged to the euro.

>> Robert Barro: Yeah.

>> Robert Hall: Yeah, that's at 0.13 throughout this whole period, so it's effective.

>> Robert Barro: So I don't know how much I could do a test just for Denmark. Should we treat Denmark as a Euro country?

>> Robert Hall: Yeah, I would.

>> Robert Barro: So I guess formally we could test for that, but I don't know if there's gonna be much power in this one observation.

So Sweden is almost pegged no?

>> Robert Hall: I don't think so, no. Sweden and Norway are much more independent than.

>> Robert Barro: Norway, for sure.

>> Robert Hall: Yeah So you haven't talked about indexation, which is not much of a factor in any of the countries, but there is one country that's completely indexed, and that's Chile.

So there would be no advantage to any of the strategies you're talking about. And therefore you're saying that there should be no change in inflation in Chile.

>> Robert Barro: But the numbers are not exactly what you're saying. So we've come up with numbers estimating the share of the public debt that's either inflation indexed or denominated in foreign currency, which is analogous for this purpose.

Chile is one of the countries which, as you say, has relatively a lot of inflation indexed. It's something like 30% to 40% of the total, but it's not the only one. There's about half a dozen that are very high, including Israel, Iceland, the UK is quite high. Anyway, we have estimates of those data.

I don't think these numbers are very accurate related to our previous discussion about whether the data are any good.

>> Robert Hall: Chile is indexation that operates through the which I think is universal. It's probably not being included.

>> John Cochrane: This is about price stickiness. It's not about, I mean, cash is not.

 

>> Robert Barro: You're talking about the bonds?

>> John Cochrane: Bonds are the prices in the economy.

>> Robert Barro: I think they have foreign currency bonds as well, so they don't have that much conventional bonds denominated in domestic currency. I think it's like 30% or something.

>> Robert Hall: They have two monetary units. And the is defined to be enough pesos to buy the cost of living bundle.

So it gives automatic indexation to all securities.

>> Robert Barro: You mean if the securities are paying off in that unit?

>> Valerie Ramey: They do, though, that's the custom.

>> Robert Barro: I'll have to check, I didn't think that was the case. I'll have to check that, anyway we-

>> Patrick Kehoe: That'd be an interesting thing to model ADA with, I think about ADA is depending on that, if you think the data reliable at all,

 

>> Patrick Kehoe: That ratio as a test.

>> Robert Barro: Well, what we did is that we remeasured the public debt, taking out the parts that were inflation indexed or denominated in foreign currency.

>> Speaker 11: Does this fit get better?

>> Robert Barro: It basically makes very little difference to the results. I thought it was gonna improve the fit.

 

>> Patrick Kehoe: That's a little weird. There's a huge literature in international macro about original sin, which is how much of your debt is in foreign currency and how much your government debt is in local currency. And the idea is, if it's in local currency, you can inflate it away, foreign, you can't.

And they make a big deal out of that. So you might touch on that literature to see if you want them to rethink how big a deal it is. Some of your colleagues in-

>> Robert Barro: I thought it was gonna matter. We spent a lot of effort generating the data, like debt maturity, that data is not so easy to find.

 

>> Patrick Kehoe: Yeah.

>> Robert Barro: I mean, it's not impossible, but it's not readily available. But we did something with that and we haven't really-

>> Patrick Kehoe: Like Ecuador, I guess, is right, that's dollarized. So, nothing Ecuador can do can change the US price level, I mean-

>> Robert Barro: But that's why we looked at the debt adjusted for these shares.

 

>> Patrick Kehoe: That's sort of weird that it doesn't. Trying to write down a theory where somebody's two-thirds issuing foreign currency, and it's the same as if you treated them in the model they're issuing all local currency.

>> Robert Barro: Right, it shouldn't be the case.

>> Patrick Kehoe: Yeah, it shouldn't be the case.

 

>> Robert Barro: I agree.

>> Patrick Kehoe: Okay.

>> Robert Barro: The results suggested we couldn't distinguish between the adjusted or unadjusted forms of that, but I'm not so happy with that either. Let me just finish up with some things we're currently working on. So the sample was basically OECD countries, which is really more related to the data than other things.

So these are nine other countries where you seem to be able to get the relevant data that corresponds to what we had. So I have the data almost complete for these nine other countries, which can basically be added to the sample, and we can also add the data for 2023, probably within six months or so.

As I mentioned, you're applying this to the US states. You needed the data from the census of governments, but there's quite a bit of delay in when they report these data. And in particular for 2022, they're doing one of these major five year surveys, which takes longer to produce.

So I think it's gonna be another year before we get the data for 2022, but then I think we can do it. The other part of that is we wanted to use the consumer price indices that Nakamura and Steinson have generated for the US, which I think was a big improvement on what was available.

The only problem is, in order to get the data, they had to physically go to the BLS to look at the numbers, and they were closed because of COVID and they only have the data updated to 2018 at the moment, but now they can go and get the rest of the data.

So Emmy promised me they were gonna do it, but don't have the data at the moment for the relevant time-frame. And I think that's an important part of that study. So I wanna say something about policy implications from this. Of course, there's a lot of inclination to say this inflation surge in the US and elsewhere was terrible, and to think about all kinds of mistakes from the Federal Reserve and other central banks and such.

But the approach we took in this paper is really not consistent with that. The approach in this paper is a kinda optimal tax approach in an intertemporal context. And at least if you take the government expenditure surge as given, especially the transfer payments, then the spirit of this is that the inflation response was part of an optimal tax response involving this contingent public finance, which means that you got to wipe out a big fraction of the debt.

So then in that spirit, it's not clear that this was a grave mistake. Now of course you can argue that the availability of this kind of public finance makes it easier to expand expenditure, especially transfer payments, but that's true for all kinds of tax instruments. And it's often been used as an argument against the value added tax, that it's too efficient, and therefore the government responds by spending more.

And this is analogous to that. So we were taking it as a given what the spending surge was, and then looking at the public finance response.

>> Patrick Kehoe: But you better have it, you mentioned it before, you better have it. Be mostly unanticipated if you start trying to do it every time that you just hit the inflationary, just take off.

 

>> Robert Barro: There's a commitment problem underneath this, which underlies lots of things. Let me just finish with this graph. I don't know, what do you think the path of debt GDP ratios looks like in the US or elsewhere, but this is what it looks like. The upper graph is for the US.

So this is general government, inclusive of social insurance funds, all layers of government, and then it's for the average for the 21 economies in our main regressions. So what you see here, there's initially a big upward surge in the public debt associated with fiscal deficits, and this is particularly in 2020, right?

So the big element there is the fiscal deficits, but then there's this effect on the price level and eventually nominal interest rates, which is the benefit you're getting from saying, I was only kidding about 20% of the debt. And that shows up in these numbers as a declining path of debt GDP ratios from 2021.

And I don't know what it looks like yet in 2023, but this is quantitatively a big deal. I mean, you're talking about 10 to 20% of GDP, depending on what you're looking at. So that's several trillion dollars. So this is a big amount of money in terms of revenue, effectively for the government.

And it does show up in the debt GDP ratios in the way you might think if you took this theory seriously, Bob.

>> Robert Hall: You had all these assumptions about no real effects. Would you care to conjecture on relaxing that assumption?

>> Robert Barro: I'm trying to think about what is the first order thing related to inflation.

And I think abstracting from all the complications related to real effects and new keynesian models and all this other stuff is, I think, a very effective strategy for looking at what's the main empirical prediction and then looking at how that stacks up empirically. I'm not saying there are no real effects from all this.

I'm saying that to study the main thing about the surge in inflation and how it's connected with the fiscal surge, maybe we can abstract from that. And that would be good, because we're not really gonna know about this other stuff when we end up trying to work it in, that's what I would guess.

 

>> Luigi Zingales: If I can add here, this is in part related to the questions also from Pat, with respect to why indexation of debt or the foreign denomination doesn't matter much if you have other benefits. You also touch on real wages and so on. If you have other benefits, that might explain why that doesn't seem to play as big as a role as you would think, just in this frictionless environment.

But I think the pretty interesting thing that I mean, at least it's actually remarkable how well it works, even starting from something playing vanilla in terms of what are the effects of inflation. So let's just think about this devaluation of the fiscal bargain effectively.

>> Speaker 14: Let's say you had a whole bunch of countries and they were symmetric and they all had some domestic debt and some foreign debt.

Wouldn't that all just kind of come out fine? I mean, I think they have less domestic debt. So at some level they have to inflate more, but they're getting benefit from the inflation that the other countries are doing. So maybe it all works out.

>> Robert Barro: I'm sorry, I didn't follow.

So we have a country-

>> Victor Davis: So let's say we had two countries and they each had 50% domestic debt and 50% foreign debt. So they've only got half as much domestic debt.

>> Robert Barro: None is inflation index.

>> Victor Davis: Pardon me, none of it's inflation index. At some level they've got, because they have half as much, they sort of have to do twice as much, but they don't because the other country is gonna do the same thing.

And so they both end up inflating just like they would if they each owned all their own debt.

>> Robert Barro: The first country being what you're owning the first country?

>> Victor Davis: Yeah.

>> Francesco Bianchi: It's like the US inflated half. If you're Brazil, you had half, US, half Brazil in debt.

And if the Americans who play in the same game, so they cut the half of the real value in dollars. So I think you were pegged, they'd cut the Brazilian piece, it was indexed to dollar go inflating at the same rate as you would if you all own all your own debt.

Yeah, especially.

>> Robert Barro: Not sure that's right.

>> Francesco Bianchi: I mean, I'm not sure either but I think there's decent probability I'm right. So we're figuring out. But it is true. Yes, if you were, well, suppose you have a lot of US dollar denominated debt and you can convince the US to, okay, let's inflate a bit then.

Then yes, you would benefit indirectly from it. And probably during COVID there was a global phenomenon, there was a bit of debt.

>> Robert Barro: I would say if everybody's symmetric, you're all gonna choose the same inflation rate. You're gonna think that, well, it doesn't do me so much good to increase my own inflation rate because only a part of my debt is domestic.

So in your model that means you need more inflation domestically.

>> Victor Davis: From the US and Brazil is holding part of my dollar debt.

>> Speaker 15: I mean, fiscal shock is the same. Let's say you all had this 12% shock and you all had a six year whatever. You have a capital gain on one half, a capital loss on the other.

 

>> Speaker 16: So everybody's gonna choose a symmetric inflation rate because they're all symmetric and in the end they're all gonna make up their 12% deficit and so they're all lost. But you're you lost by holding the American debt. They screwed you as the Brazilian holder of a deflated away.

 

>> Speaker 17: And you screwed them because they were holding some Brazilian debt.

>> Speaker 16: Then you would have had to inflate way more cuz that was a wash.

>> John Taylor: So we have to stop inflation