 Good afternoon, everyone. I would like to welcome you to the second session of this conference on monetary policy. My name is Cornelia Holthausen. I'm a Director General of Financial Stability and Microprudential Policy at the ECB. I'm very pleased to chair this session. So let me start by introducing the speakers, the two speakers and the two discussants of this session. So the first speaker will be Stain Verneuerberg from Columbia Business School. So Stain works on the intersection of housing, asset pricing and macroeconomics. So he has done research in many different aspects of real estate markets. But he also has a research agenda that focuses on government debt and fiscal policy, which will be important for the conference. Here he has been, he started his career at New York University Stern School of Business where he has been for 15 years and then joined Columbia Business School in 2018. And he's also a board member of the American Finance Association. Then we have Andrea Veduli from Boston University, an associate professor of finance there and also faculty research fellow at the MBR and research affiliate at the CPR and research scholar at Norris Bank. Her research fields are primarily asset pricing and macroeconomics and she's an associate editor of several distinguished finance journals. Then we would move to the second paper that will be presented by Guillermo Ordonez from the University of Pennsylvania, where he is professor of economics and finance and also Guillermo's research associate at the MBR, frequent visitors of reserve federal reserve banks and the central bank of Chile. And previously he was a faculty member at the University of Yale is co editor of the jet and an editor at the review of financial of economic studies. So a professor Ordonez is a macroeconomist specialized on the study of banking financial crisis and information imperfections. And he has also focused on how information processing affects the functioning of sovereign debt auctions. And then last but not least we have Federica from the University of Oxford who will be discussing that paper. She's associate professor at the University of Oxford research fellow at CPR and a member of the editorial board of the review of economic studies. So welcome all four of you to the session. And now we can give the floor to stain to present his paper. And I think you have around 28 minutes. Please, the floor is yours. I understand you will share the presentation. Thank you so much. Thanks so much for having me at this wonderful conference. Glad to present a new working paper in a series of research as Cornelia mentioned on the monetary and fiscal determinants of government debt. And this is joint work with Zeng Yang Yang from Northwestern, an elastic from Stanford and mini challenge from UT Austin. So in this paper we're going to focus on convenience yields and we're going to focus on the eurozone. So let me just spend 30 seconds trying to motivate why we think this this is interesting. Right. So just let's start at the start. What is the convenience yield. Well, there's certain bonds certain sovereign bonds where the safety and the liquidity of the government is such that investors are essentially willing to lend to those governments at, let's call it below market interest rates at interest rates are lower than what would be justified by the fundamentals. Let's call it the fiscal fundamentals of that country. Right. So, you know, that's what we think of as positive convenience yields. The study of convenience yields has for the most part focused on the US Treasury market, because the US government since World War Two has been the safe haven asset par excellence. And naturally, the focus has been has been on the unconvenience yields in the Treasury market. We think that it's interesting to study convenience yields in the eurozone as well. You know, for two reasons, first I'm going to try to argue and show you that there's actually a lot of variation in these convenience yields both across countries and in the time series. And second, because there's something special about a currency union, a monetary union, where essentially because exchange rates are fixed and because the countries share a risk free yield curve. We're going to see their convenience yields start to play an important role as absorbers of country specific fiscal shocks. Okay, so let me kind of give you a little bit more, you know, structure to think about these ideas. Right, so imagine that, you know, we can write down the bond yield, the sovereign bond yield on a government. YTI country I as the sum of three pieces a risk free interest rate, a default spread, which which captures the credit risk in the sovereign, as well as this convenience yield, which I'm going to call lamb that the I okay. And so that lamb that the I as you can see it shows up at a negative sign because in, you know, countries that have kind of this this safe haven status that earn this convenience yield, you know, they get to borrow at lower interest rates. And they otherwise would. Now, what's interesting about a currency union is that the risk free interest rate is common across the countries in that union. And so, you know, because of that, you could imagine writing down this equation for Germany, and you write this equation for another Eurozone country, let's say, Belgium. And then you can write the difference between the interest rates on Belgian debt and the interest rates on German debt as, you know, essentially, the difference between the default spreads between Belgium and Germany. And the difference between the convenience yields of Belgium and Germany. And note that the risk free interest rate drops out because Belgium and Germany share the same risk free interest rate. Okay. And so now we could kind of ask how much of the variation in interest rate differences between Eurozone countries Belgium and Germany is accounted for by, you know, these convenience yield differentials versus these default spread differentials. Right. And so what we're going to do in the data and I'll give you a lot more detail as we go through this, but just kind of big picture. You know, you could imagine measuring these default spread differences from the credit default swap market. And, and sort of then backing out what these convenience yield differences are by, by, by, you know, bringing the default part to the left hand side, and kind of calculating these convenience yield differentials as from this equation. And when you do that, what you see is that there's a lot of variation in convenience yields both across countries and over time. And, and, and so the one thing to sort of note here is that most of these numbers on this graph are negative. And that's basically telling us that Germany is sort of the safe haven country in the Eurozone and that most other countries have convenience yields that are lower than the German convenience yield as, hence the, the negative sign. You know, you also see large variations in the time series 2011 European debt crisis, for example, being an episode where these convenience yield differentials became particularly strong. So what we're going to want to do in this paper is sort of think about this variation over time and across countries and think about what you know some of the determinants of this of these intra Eurozone convenience yield convenience yields are. So that we're going to write down, you know, a very simple framework to help us think about, you know, what, what could be some of the fundamental drivers of convenience yields. And in particular, you know, we're going to take the, as our starting point the inter temporal government budget condition. And that budget condition is going to, you know, clearly point towards the fiscal determinants of the convenience yield. And what we're going to argue is that in a currency union, these convenience yields are going to play an important role as shock absorbers as absorbers of country specific fiscal shocks, right. And then I'm going to turn to the data, and I'm going to try to empirically quantify this shock absorption role of the convenience yields in the Eurozone. And what I'm going to argue is that these convenience yield differences account for the bulk of the variation in Eurozone sovereign bond yields. So I think that's sort of an interesting and a new stylized fact. And second, you know, may be more importantly, I'm going to show you that consistent with the model framework, the convenience yields respond to country specific fiscal news. And so one of the, one of the main messages of this paper is that fiscal news is an important determinant of convenience yields within the Eurozone and maybe more broadly. Okay. And so this, why, you know, why should you care about this? Well, because, you know, essentially these convenience yields are going to have large fiscal costs for peripheral countries in the Eurozone. I'm going to show you kind of a simple counterfactual. I'm going to ask a simple counterfactual question, which is, imagine that Italy and Portugal and Spain could all borrow at, you know, at the convenience yields of the German Bundt, right. How much more revenue would they have raised, you know, in the last 20 years, sort of since the adoption of the Euro, you know, relative to what they actually raised in the data if they had enjoyed these low convenience yields of, you know, of the German Bundt. Okay, and I'm going to argue that that's a substantial, a substantial amount of revenue that in some sense they missed, you know, because, you know, their fiscal status was not, it was not as strong as that of Germany. Okay, so let me begin by kind of giving you a sense of the theoretical results that we derive. And so, you know, the starting point of this analysis is a simple, you know, first order condition for bond pricing, which basically says that, you know, if all the bonds were purely risk-free, then the price of a government bond of maturity k plus one years and country I, you know, would be the expected discounted value of that same bond next year, which now has one year less to maturity, discounted by this, you know, stochastic discount factor. And you can think of this as this stochastic discount factor being common to the different countries in the currency union and the monetary union. Okay, now that's sort of not what bond pricing looks like because there's these two additional sources that we need to think about, the additional sources of price variation we need to think about. One is default risk, right, so there is potentially a default spread. You can think of this variable chi as, you know, the value of the bond in the case of a default. So this could be an indicator variable and then kind of there'll be a partial recovery in the case of default. And there's this additional term here, which is this Euler equation wedge and it's that's the, that's sort of our measure of the convenience he tells us how much extra investors are willing to pay for the convenience of holding this bond of high compared to other securities with the same pecuniary payoffs, right. So, and, and, you know, there's been a range of explanations for, for this convenience, you know, this could, you know, be reflecting a liquidity premium. Like in the work of amy hood and Mendelssohn or Krishna Morty, or long staff could reflect a safety premium. In the work of Krishna Morty and visiting Jurgensen and hey Christian Martin Millbrat. I could reflect the pledgeability of particular treasuries as collateral, for example, in the repo market, or it could be compensation for some non-pecuniary quality. But importantly, it's not necessarily an arbitrage opportunity because there's something special about this bond that people in the market value and that gets gets priced it. So, you know, taking that that starting point. That is our starting point where we can then do is we can write down the inter temporal government budget constraint, right. And so basically, you know, we can combine all the different bonds that the government has has outstanding of all the different maturities. And, you know, we know that the government has to pay off a certain amount of debt at the beginning of the period. It needs to finance some additional deficits that it runs this period, and all of that needs to be paid for by issuing new debt. So that's the one period budget constraint, and you can iterate forward on this budget constraint and essentially just, you know, make two assumptions. You know, first of all, you assume that there's no arbitrage that all the bonds in the economy are priced correctly. And second, you know, that there's a transfer salary condition that satisfied for the debt so that you know the value of that very far into the future converges to zero. And so under those sort of mild assumptions, what you get is that the value of all the outstanding government debt today. Think of this as the valuation of the aggregate government debt portfolio so it's basically all the different bonds, all the quantities of bonds of all the different maturities, kind of times their price, where I wrote their price here as kind of collecting risk free discounting, you know, credit risk discounting as well as these convenience yields, and then some of it across all the different, all the different maturities right so this is sort of the government has that of all a range of maturities outstanding this is the market value of the entire portfolio of government debt that market value has to equal by the government budget constraint has to equal the present value of all future surpluses. These are the primary balances of the country tax revenues minus non interest spending discounted bag by that same stochastic this confector blows an additional term which reflects the present value of all, we're going to call this senior rich revenue what we mean by that here in this particular context is all these convenience yields that the government is earning basically all the all the reductions and interest rate that it's receiving by nature of the fact that it's that is special. So that's sort of the present value of the government's earning this amount of convenience on this amount of outstanding debt, we can sum that across all the can all the maturities and then discount that revenue back. Right then so this is sort of like the, you know, the starting point of our analysis is basically just a government budget constraint there's not a lot of, you know, not a lot of surprising. This is not a surprising equation this just tells us the market value that has to equal the present value of surpluses plus future senior revenues. Now, if exchange rates are flexible, you know, think of a country like the UK, then, you know, and if there's now a shock about the present value of the surplus. So think of this as the UK government announcing a few weeks ago that it's going to permanently lower taxes. Right, so that's sort of like saying the present value of the surplus is not permanently lower, because we'll be running larger deficits for the foreseeable future. That's like a negative shock to the present value of the surpluses, then that needs to get reflected in the value of the market value of government debt. Right. And so in a country like the UK, this could mean that the interest rate has to go up. And that's in fact what we saw in the UK bond market a few weeks ago. Right. And so, you know, a country with flexible exchange and or there's going to be depreciation of the British right so this is sort of what we saw in the markets a few weeks ago. Right, so really real exchange rate movements could could adjust to make this equation hold. Now this becomes different in a currency union right in a currency union the exchange rate and the risk change rates are fixed, and the risk free interest rates are common. So now they cannot adjust in response to this country specific fiscal news right so this is sort of like, imagine, Georgia Maloney is, you know, a passing legislation that reduces future surpluses for Italy. There's a negative shock to the present value of Italian surpluses. But now the Italian government bond interest rates the risk free rate cannot adjust, because that's common across the eurozone the exchange rate cannot adjust. And so the only two things that can adjust now are either the default risk of Italy or this convenience yield. Okay, and so we're going to, you know, as an empirical matter we're going to want to ask how much of that adjustment to these specific shocks these country specific fiscal shocks, let's say to Italy, gets absorbed by default spreads on the Italian bond versus to convenience yields of Italian bonds. So that sort of naturally leads to this variance decomposition and the debt valuation. Right, so we can kind of as the question imagine that D is the market value of government debt. Is that present value of future surpluses and seniority revenue from the previous slide. So now we can say, you know, what, you know, how much of the variation in the market value of Italian debt that relative to some currency union average, or you think of this as Germany, instead, if you prefer, how much of that processional variation and time series variation between Italian and German or Italian and currency union wide debt valuation comes from fluctuations in the relative default credit spreads versus, you know, covariance of those of those debt valuations with convenience yields. Okay, with relative convenience yields between, let's say Italy and Germany. And so this is a first kind of variance decomposition that we're going to want to look at in the data. And then the main question we want to ask is, you know, what about the fiscal determinants of these convenience? So if there's a fiscal shock, to what extent, you know, or is that does that get reflected in these relative convenience yields. And for that we're going to make one important additional assumption, which we think is kind of well founded in the empirical literature and also in the theoretical literature, which is that there's downward sloping demand curves for save assets. Right. So, in particular, what that means is that when the government issues a lot more debt, then its convenience yields are going to start to go down because the debt is not a special anymore is not a scarce anymore than if there was very little debt outstanding. Right. And that's sort of the basic result in, in, you know, Krishna Murthy and Vincent Juergensen's paper, for example, which shows that if there's a lot of US Treasury debt outstanding, then Treasury has become less special relative to, let's say, corporate bonds of high credit quality. So that's sort of our main assumption here, which is saying that, you know, if there's more, if the value of the debt, if there's a positive shock to the amount of debt outstanding, and then reduces the convenience yields on that debt. And then we'll make one more assumption, which is, you know, just sort of for convenience, which is that if we think about convenience yields on bonds for different maturities age, we're going to assume that those are sort of all equal across the maturity structure. So think of this as an expectations hypothesis type of assumption for convenience yields across the term structure. And so under those two assumptions, what we can show is that if there's positive news to the surplus, right, so there's good news about future surpluses, then that's going to increase the convenience yields in that country. Okay. So in other words, you know, there will be an adjustment to the convenience yield in response to positive news about future surpluses. And so this is sort of the main theoretical prediction that comes out of the model. So, you know, just to kind of give that a little bit more content and make that make that practical, we're going to write down a simple numerical example where the government issues debt. There's an exponential debt maturity structure in terms of the quantities of debt of each maturity. And then we're going to assume that debt supply follows an AR1. And that this convenience yield, you know, this is our assumption that convenience yields are downward sloping in debt supply. So this parameter beta sort of this key parameter here, which measures how steeply downward sloping convenience yields are in debt. Right. So then the main thought experiment is imagine you now have a shock to surpluses. Right. So there's good news about about future surpluses. Then, you know, the convenience yield increases initially. So the government needs to issue less debt than it thought before. And then initially that increases convenience yields. And then these convenience yields sort of gradually come back down towards their long run average. Right. So, you know, in the long run, sort of a negative relationship between convenience yields and amount of debt, but in the short run, you can get this positive relationship to emerge, which is, which is sort of the one of the main. Things will be testing in the data. So now let me turn to the data and to the empirical results in the paper. But so the first thing we want to show is just, you know, to return to this convenience yields differential that I mentioned at the beginning of my talk and just show you how much variation there is in these convenience yields both across countries and across time. Right. So remember, we're going to calculate these relative convenience yields of each eurozone country I relative to Germany. And, you know, that is the sum of these credit spread differentials as well as the interest rate, the total yield to differentials. And so these, these, and so we can measure these delta tell us from the CDS market, you can measure these yields just directly from the bond market. We have data from 2002 until 2021. And then we're going to split our sample and the before 2008 and after 2008 sub samples. And you can see before 2008 these convenience yield differentials are fairly small. They're on the order of 10 to 20 basis points. And, you know, there's not a whole lot of variation across countries either. Now, after 2008 there's a lot more variation across countries and the magnitude of these convenience yield differentials also becomes a lot larger. You know, somewhere between 10 basis points for a country like France to something like 6070 4050 6070 basis points for the periphery countries and the eurozone. Okay. So, then the first, the first question we want to ask is, you know, back to our variance decomposition and the theoretical results, you know how much of the variation in the market value of that in the relative interest rates across countries. Can we account for by either the credit spread or the convenience yield, you know, differentials. And before 2007, the answer is that almost all of the variation in interest rates across eurozone countries comes from this convenience yield channel. Now, remember, you know, also showed you that there wasn't a whole lot of variation in interest rates in that period and in convenience yield in that period. So you may say, well, there's not a whole lot to explain. So maybe this is, you know, somewhat less interesting. The 2008 period. After 2008 period, there's a lot more variation in interest rates. There's also a lot more variation in default spreads. Remember, you know, the European, the great financial crisis is part of the sample, the European debt crisis is part of the sample. So even what we're showing here in the in the bottom panel is that even in this period in this period of, you know, the European debt crisis, even in this period, you know, more than half of the variation in interest rate differentials across eurozone sovereign bonds comes from this convenience yield differential and only 42% on average across countries comes from the credit spread differentials. Okay, so the bottom line is that there's a lot of, you know, there's a lot of variation in interest rates that's accounted for by these differences and convenience yields. These seem to be important, an important component of helping us understand the dynamics of interest rates in the eurozone. So then the, you know, the most important thing we want to do in the empirical section is to establish a link between these convenience yields on the one hand, and, you know, the default risk the fiscal, the fiscal news on the other hand. And so we're going to approach this a few different ways. The first way is we're just going to look in the cross section of countries at measures of fiscal conditions. So in the left panel here I'm plotting the average deficit, government deficit to GDP ratio, and the right panel and plotting the average debt to GDP ratio these are two different measures of fiscal conditions. And then on the vertical axis I'm always plotting the average convenience yields in that country. Okay, relative to Germany. Okay, so by definition Germany has a zero differential with Germany. You also see that Germany is fiscally the most sound country in the sense that it has had the lowest average deficit. It also has had a modest amount of debt. Compared to countries like Portugal, for example, or Italy, or Spain, which have had much larger average deficits and higher average debt. And you see that that relationship sort of implies that, you know, a one standard deviation increase in the average surplus, right, so reduce a reduction in the debt moving to the left, you know, increases that convenience yield differential with Germany by about 11 basis points just to give you order of magnitude. In the left panel, a one standard deviation reduction in debt, which is about a 22 percentage point reduction in the debt to GDP ratio that translates into about a seven basis point increase in the convenience yield relative to Germany. And so we think these are sort of reasonable magnitudes, but nevertheless, you know, substantial substantial magnitudes. Second, we can turn to the time series, you know, the results here are for the five year tenor but you know we can look at this across tenors as well. And so what we show in the time series that when there's changes in the surplus to GDP ratio, we see that, you know, those affect interest rates. Of course, they also affect credit spreads and but they also affect convenience right so basically improvements in fiscal conditions in the time series increases and surpluses end up increasing the convenience yield relative to Germany by, you know, about 11 basis points here. Okay. And so we can use these numbers to basically come up with a question of if you did a variance decomposition how much of the variance in interest rates. Can you in market values of that can you account for through this convenience yield channel. And so the answer that we get from this requestions is about, you know, about a third. Right so about one third of the overall movement in market valuations of that gets accounted for through this convenience yield channel the other two thirds gets accounted for through the through the default channel. Okay, so that's a substantial substantial fraction. Now that was using realized surplus data. You know, you could imagine that the bond market is forward looking and it wants to it looks at future conditions, future fiscal conditions. So we also collect data on from consensus economics about surplus forecasts, you know, as opposed to realize surpluses. You know, we have that for a smaller set of countries. But we find the same result basically is that, you know, when there's good news about future surpluses, either in the current year sort of like the, you know, we're standing in January we're forecasting the surpluses for the current calendar year, or for next year you know, we find that basically improvements in the fiscal situation lead to increases in convenience yields, relative to Germany. And again here the magnitudes are actually a little bit larger, they point to something like about 60% of the variation in interest rate of different differentials being accounted for by this convenience channel. So then the last thing we want to briefly talk about is, is, you know, why does this matter, and, and, you know, what do we, you know, what are the implications for the eurozone from that. For example, counterfactual exercise we want to ask, you know, imagine that every time a eurozone country went to the bond market in the last 20 years, imagine that every time they would have enjoyed the convenience yields of Germany, instead of the set of their own convenience yields that they had at that time, how much more revenue would they have been able to raise on these collectively on these bond issuances, right and so that's going to depend on the issuance amount on the duration of the bonds that they issued, and on that convenience would be differential at that time with Germany. Right. And what we find is that, you know, these numbers are substantial, they can be about one to one and a half percent for some countries per year, kind of at the worst, at the worst time second in the European death crisis. In other years are not as large but cumulatively, they add up to a substantial amount of lost revenue, and even for countries like France, right so cumulatively over this 20 year period, France sort of, you know, gives up about, you know, 1% of it's 20 GDP worth of revenue from not having enjoyed the same convenience yields of Germany for countries like Spain and Portugal that number is much larger it's something like five 6% of 2020 GDP. Right so on average for the entire eurozone. If you add if you add is all up this amounts to about 2.6 percentage points of 2020 GDP which you think is a substantial amount of revenue. So we think this is interesting when you think about, for example, the, you know, next generation EU bonds that started trading last year. You know, those are clearly common, common to the eurozone, they are enjoying the level of convenience yields of the German of the German bond. And so this is, you know, our results are telling us that this is, you know, because they're enjoying these convenience yields, they're generating, you know, substantial additional revenue, relative to what would have happened if each of the countries that issued their own bonds. And so this is kind of an important aspect of kind of a deeper fiscal union that we think is kind of worth, worth emphasizing. So, let me conclude what we tried to do in this paper is to build a theoretical framework that relates convenience yields to fiscal conditions. And to kind of make this point that in a currency union, the convenience yield plays this new role as a shock absorber of country specific fiscal shocks because exchange rates and the risk for interest rate cannot do this adjustment. And empirically, it turns out that, you know, even though default spreads could do all of the adjustment it turns out that convenience yields, in fact, do a good chunk of the adjustment. You know, they explain a large share of the variation in bond yield differentials, they rise when there's good fiscal news, and all of this matters because it affects revenues from bond issuance, especially in peripheral countries. Okay. Thank you very much. Thanks very much, Stain. So let's move right away to the discussant Andrea Widolien. And maybe just to say we have a Q&A just afterwards. So if you want to put in any questions into the Q&A chat, you're welcome to do so. Thank you. Okay, thanks a lot to the organizer for having me discuss this very nice paper. Let me start this discussion by saying I think it's a very important topic that the authors are working on, and I find it very interesting this extension to the Eurozone. Okay, so here's an outline I don't want to spend a lot of time summarizing the paper so I will have two comments the first one is is going to be on potential endogeneity issues when they run these regressions. And the second point is more like a philosophical point about like what's special about the EU and how should we think about the credibility of a government and how it affects convenience yields in their analysis. Okay, so let me put some context to the paper. So there is a very old term structure literature that tries to link government deficits are spending to interest rates so there are almost these like 30 year old papers by Evans and plus or. And usually these papers I found it very hard to establish a link between government spending and interest rates. Now what this paper does is different so they're not going to look at interest rates themselves they're going to look at convenience yield differentials between EU countries. And what they argue is that they reflect the relative fiscal conditions of member countries. Now, what is the convenience yield. So the convenience yield just represents how much risk adjusted return investors are willing to forgo to hold bonds. So this can best be illustrated by the simple no arbitrage condition. No arbitrage order equation just tells us that the price of any bond with maturity h plus one in any country I should just be the discounted value of that bond next period where obviously now the maturity has been reduced to age. Now because they allow countries to default think of this like psi I to capture the potential recovery in case of default. Now what's new in this paper is that no arbitrage condition has this extra term, which is going to be this wedge here, and this wedge here is a function of this convenience yield which is specific to each country, and for different maturity. So in particular this wedge is going to measure the extra safety and liquidity which is provided by these country I bonds compared to other bonds with identical payoffs. So what's the main results of the main result relates these convenience yields to fiscal conditions of a country. Stein did an excellent job explaining this very simple theoretical framework. So basically they start from the inter-temporal inter-temporal government budget constraint constraint which tells us that the market value of that so here you just sum up over all the bonds which got issued over different maturities has to equal the net present value of that so this is just the nominal tax revenues of country I minus the government spending and then because they study these convenience yield they also add these net present values of senior revenues which are captured by the CIs which are these convenience yields. Now they have two assumptions to derive the main results so the first assumption you need is that the expectation hypothesis holds for these convenience yields. The second one is that we have a downward sloping demand function for bonds and if you have these two assumptions you can show that the covariance between fiscal shocks. So that would be this here plus this copper times D which is just these senior revenues and shocks in the convenience yield has to be positive. So what they do now in the paper is that they test this result here in the data. So the way they measure convenience yields is very simple. So you just take the difference between say a five year CDS spread between country I in Germany and you subtract the five year yield spread between that country and Germany itself. They have three results so what they found is that on average convenient yield differentials are negative to Germany. They have a cross sectional results which tells us that countries with higher surpluses earn higher convenience yields. And the time series results is when a country improves its financial conditions, its convenient yield rises. Okay, so now let me get to my first comment which is about the endogeneity of some of potential endogeneity of some of the empirical results. So again what the authors do is that they want to show that fiscal conditions help explain the variation in convenient yield differentials across time. The empirical design looks as follows so changes in relative yields, changes in convenience yields or changes in default spreads are going to be regressed on the relative change in the surplus to GDP ratio of country I. And what we're interested in is in this coefficient beta which hopefully is going to be positive for the convenience yield according to theory. So what they do is that they link government surplus in year T with yield and CDS data at the end of June in year T plus one. So why is that? Because while we can measure the left hand side on a day to day basis, the right hand side obviously is only available at the animal frequency. What they write in the paper is that in doing so we allow six months time for the fiscal information to affect debt markets. Now what I want to argue in my discussion is that there are potential endogeneity issues, because imagine if there's something which affects both the left hand side and the right hand side of this equation. It will be very hard to establish a causal relationship between surplus GDP ratios and the convenience. Let me give you an example. So let's take spring 2020. At the time, as everyone in this audience knows the ECB implemented many unconventional monetary packet tools. Most notably this pandemic emergency purchase program PEPP, which was announced in March 2020. At the same time, or just a couple of weeks later, the EU announced a very large fiscal package, which was the EU next generation fund. So we can check what happened to yield changes during these particular days. So unfortunately I didn't have like the most recent CDS data. So what I'm going to show you here are just two day changes in five year yields for each of these countries relative to Germany. So as many, many papers and ECB papers have shown on the announcement of this PEPP, we saw a large reduction in particular in the peripheral spreads relative to Germany in the yields. There was another announcement, which was an extension of the PEPP, again a reduction in the yields. There was another German ruling which questioned the legality of the PEPP because of some proportionality rule, which led to an increase in these yields. Now if you look at the fiscal announcement, what we see here is that for each of these announcements for like this fiscal spending, which is basically an 800 billion Euro package, which goes into infrastructure, climate change and research. What you see here is that there's a large reduction and large cross-sectional difference in how these yields reacted on these announcement days. Now why do we care about that? Well, of course, we can all agree that changes in yields and hence changes in convenience yields probably happen for many reasons. What I'm showing you here is that unconventional monetary policy probably had the largest effect on these yield spreads during that period. And there's a very nice ECB working paper by Korodin Grimm and Schwab, which shows exactly that. So they also decompose the sovereign yield spread into convenience yield, liquidity premium and other things, and they show that this has had a very large effect at the time. What I also want to argue is that it's not just country-level fiscal announcements which matter, it's also EU-wide fiscal announcements which matter. And why is that? Because there's a clear risk-sharing motive which argues that fiscal risk is removed from weak countries such as Italy and Spain and moved on to shared budgets. And I will mention this again in my later slides. So perhaps it's hard to disentangle the effects and establish a causal relationship between fiscal shocks and convenience yields if we are at this very low frequency. Now what can we do? Now of course there's a very old literature that tries to identify fiscal shocks using a narrative approach or vector or to regressions, but fortunately we didn't have any wars in the eurozone in the last 20 years. So this is not something that we can do to study fiscal shocks in the eurozone. So what we can do is that we can just do what people do in the monetary policy literature and look for example at fiscal announcements, dates and see what happens. So as I showed you in the previous table, there was a large cross-sectional difference in how these yields reacted to EU-wide fiscal spending packages. Now there's an ECB working paper of many, many authors which argues that in particular Italy and Spain, their debt to GDP ratio will decrease by 10 percentage point by 2031. And then of course we can also, following the authors, we can also look at country-level fiscal spending announcements. However, I will probably distinguish between budget improving and budget worsening announcement just to give you one example. In September 2018, when the Italy's ruling coalition announced that the deficit is going to be 2.4 percent instead of a decline of 0.8 percent, what you saw is that there was a huge increase in the Italian spread relative to Germany, whereas there was no movement in the other spreads because of this basically budget worsening announcement. Okay, so this was my first comment. Now the second comment is... Sorry, we are running a bit out of time, so if you could try to be a bit faster on the second one. Okay, so the second comment is about the credibility of a government. So the EU is going to be special, I want to argue, and not just any cross-section because we have very specific rules on how debt and spending has to be in EU countries. Now of course we know that many countries actually are not compliant with these rules. So for example, only 50 percent on average are actually compliant with these rules. So what I want to argue is that it really matters what the credibility of a government is and what they announce what the budget is going to be. So here I just have very quickly two different examples. So Salvini mentioned that they're going to keep the budget, which led to a large decrease in the spread. However, when Mario Draghi was making a similar announcement and he basically called the EU fiscal rules obsolete, what you saw was that there was no change in the yield spread relative to Germany, even though basically it was bad news in terms of the deficit that people were expecting. So let me just summarize, so my second point is just going to be the EU is special because member countries spending and deficit cannot just be any number imaginable and I think the market takes this into account when they try and how bonds are priced in the market itself. So let me conclude. So my first comment is perhaps we can think about more carefully about what should be an exogenous shock when we study fiscal shocks on yields or convenience yields. And then the other thing is that obviously there's a large political component when we think about surplus to GDP ratios. Let me conclude by saying I think this is a brilliant paper. It's a very important agenda and had a lot of fun and I actually learned a lot reading that paper. Thank you very much. Thank you very much, Andrea. And before giving back the floor Tuesday, let me just read the two questions that we have in the chat and then you can reply to all of them together. One is by Philip Hartman and it makes a reference to a ECB working paper published in May 21 by Coradine and others. It looks at Euro area sovereign bond risk premier during that pandemic and it shows a more granular composition of sovereign yields including the convenience yield premium. This measure is conditional not only on default risk and the risk free rate but also on liquidity risk and re denomination risk. Hence could the authors measure be too broad or be biased. That's one comment and the other one comes from Wolfgang Lemke. With my product of your analysis, you would get a pure risk free rate, say the German yield plus its estimated convenience yield with what traded rate with that synthetic risk free rate correlate the most. I hand back to you. Thank you. Thanks so much. First of all to Andrea for, you know, some wonderful comments. Some of these things we have thought about, you know, very much like the idea of looking at fiscal announcements as sort of exogenous shocks to fiscal policy but, you know, as any I mentioned, we would want to do this for each country. And we sort of started working going down this path a little bit. So you might imagine it's sort of tricky to, you know, isolate an exogenous shock to each country's, you know, deficit over the over the years but I think it's definitely doable. And it sort of takes a page from that monetary policy announcements literature. So I think that's sort of promising route to make some further progress on, on this question of endogeneity. You know, I also think that, you know, the EU wide fiscal announcements are definitely important. You know, we do spend a little bit of time on that kind of in the paper, which I didn't go through and in the presentation. Thinking about, you know, there's this common component and convenience yields across countries, which is sort of the German convenience yield. And you can think about, you know, EU wide fiscal announcements moving that first principle component around. Right. And so, you know, that for that first principle component could sort of also capture flight to safety type of dynamics. And we show in the paper that it is indeed correlated with global stock market volatility or with us treasury convenience yields, for example. So that's sort of, you know, I think the level, the level of overall convenience yields in the Eurozone is presumably affected by a bunch of different things, like just flight to safety dynamics but potentially also EU wide fiscal announcements. So separating that out is a little bit more tricky, which is sort of why we focus on the, on the differential, the convenience differential with Germany. I like your point about credibility. I think that is true. I mean, I think we could think about some of these EU, some of these ECB monetary policy announcements, especially, you know, the quantitative easing programs that sort of affecting the relative default risk in the Eurozone countries, maybe absorbing some of that relative convenience, some of that relative default risk. And, you know, you could think of that as, you know, to, to, you know, just as, just as in your quote, Japan and sorry, Spain and Italy might have much lower that going forward, you know, in part because the ECB has sort of announced that, you know, they're not willing to tolerate large, a certain amount of spreads in the, in the bond yields. And so that is, I think, an interesting, an interesting thing to explore some more, like to what extent ECB announcements affect our decomposition. You know, to answer quickly to the questions that were asked. You know, I think the liquidity question and the Coradine at all paper is a good one. You know, we think of liquidity is sort of endogenously determined jointly with fiscal conditions. And now, as a practical matter in our empirical work, we do control for measures of liquidity in the bond market like bid-ask spreads and we show that our fiscal variables survive the inclusion. But, you know, I think more broadly we want to think about these things as jointly determined in equilibrium. And for that last question, I would say that the OIS rate is maybe the most closely related to the true risk-free interest rate. Thanks very much to Stain for a fascinating paper and Andrea for a very insightful discussion. So thanks very much.