 Let me start first to thank the organizers for inviting us present our work. It's a pleasure to be here. This is joint work with Peerolivey Gonshaw, Verunika Penchilkova, and Nick Sender, and given the central bank affiliations, usually screen replacements. So let me of course start by stating the obvious by now that this shock is unprecedented in its complexity, unevenness, and severity. And I mean, at the start of the shock, and I think this is still the case to a certain extent, small firms, small businesses are especially at risk for failure for this shock. Why is that? Because this is not like financial crisis of 2008, 2009, 2010, but it is a shock to income, which is very even in terms of sectors and the firms. So you hit certain sectors and certain firms very hard in terms of basically lost revenue. So what happened since March 2020 is also tremendous government support, right? There was support implemented vaulting advanced economies and emerging markets to firms directly. So there was support to these firms, especially small businesses, many programs done for small businesses. And at the same time, of course, there was the colonialized stimulus, your standard macro policies, fiscal, and other students. So what we would like to do is basically, with this background, do several things actually, I'm going to present mostly the paper on the program, which is given this background, the COVID shock, and the support policies by governments, and the economy-wide fiscal monetary policy stimulus. What happened with the SME failures? Then I'm going to show a few results from our accompanying work with the same set of co-authors. One is looking at a future look 2021. Well, they're almost end of 2021, but of course, we wrote the paper beginning of the year. It's at AER papers and proceedings. The idea here is like we did so much support to save these firms in 2020. Do we have a time bomb in our hands in terms of zombie firms, wall of bankruptcies in 2021, 2022? So that's, I'm going to show a few results from that. And also we have a recent piece in the 2021, August 21 Jackson Hole Conference, taking a broader look at fiscal policy, not just policies to support some more firms, but broader fiscal policy and try to understand also its success, both in terms of firm failures and aggregate macro income. So some of the questions we ask in this research agenda, now we have been involved pretty much 18 months since the start of COVID is first and foremost, what is the impact of COVID on firm failures in a wide range of countries? What is the cost and effectiveness of government interventions aimed at saving firms? These policies, SME, small medium enterprise support policies, were they good, were they bad? Maybe they were good, they saved a lot of firms, but does this mean it's going to be a time bomb of failures in 2021 and maybe also 2022 once we start dialing back the support? And in general, is it the case that fiscal stimulus support added activity to demand? How big are fiscal policy flows globally? What happens in other countries? And also what are the implications for emerging markets of a two-speed recovery with global uneven vaccination? So let me try to get at most of these, if not all given the 30-minute presentation. Okay, the first thing I'm going to tell you is about the methodology, which is going to be extremely important if you want to understand the firm failures. First of all, this whole thing is about not having real-time data. Okay, so when we started in March, April, 2020, obviously there was no real-time data on firm bankruptcies and firm exits. And even though now we think there is, let me tell you there is still not real-time data. Why is that? Because first of all, when you think that the firm level, firms file these things with a two-year lag. So when you think in terms of firm exit in the sense of sense, it is going to be not before late 2022, we understand the full extent of the firm exit in 2020. That's number one. Number two, the bankruptcy filings, which is another way of looking out of this data, firm exit data, because you go and file bankruptcies, but bear in mind that these bankruptcies, filings and regulation differ a lot, a lot across countries. So it's not that straightforward that you will see every single firm exit in a bankruptcy filing. It's going to depend on the solvency, the good and all sorts of things. And here doing this shop, those were also stopped, basically, either because of congested courts or because of explicit regulation as happened in Germany that you are not allowed to file for bankruptcy. So the notion here stands from the fact that we just cannot measure in real time, but we do want to have an estimate of firm failures because that's of course going to help us to design the policy to save these firms. So we believe this is extremely important to provide a very simple model-based estimate for firm failures that is going to tell us something with a shock like COVID, what is going to happen in very short run in terms of firm failures. And the idea here is again to find that liquidity shortage, right? So if you think in a question like that, firms cash in hand plus the cash flow during COVID, then that is less than financial expenses. Based on the liquidity criteria, firm is going to be in trouble. So firm has to close this either through a bank by borrowing through credit markers or government has to close this gap, right? That's basically what we face in the spring of 2020. And the idea of our work is really how can we estimate this real-time cash flow during COVID, something that we don't know. So everything orange in the slide is basically what you need to be estimating using model and the data because you don't observe it. And our approach is exactly that, right? We are going to have a firm optimization model in the short run. And we are going to combine that with represent the firm-level financial data entering the poll, okay? So we observe firm-level financial data as of 2018, 2019 to a certain extent entering the COVID. So we combine those two letting firm minimize the cost of labor and materials given this COVID shock. And that's what is a COVID shock? COVID shock is going to be calibrated at very rich many of shocks, both at the four-digit sector level on supply and demand and also an aggregate demand shock. Because again, the whole thing with COVID is it is being a very uneven sector-level shock, different than our standard recession, different than our standard financial crisis. And that approach combining a firm minimization model firm-level financial data ex-empty COVID is going to give you an estimate equation like that. I'm going to go through details but you get the cash flow during COVID is by shocking the firm revenues entering the COVID with these COVID shocks and the firms cost of goods sold. And we are going to do everything in changes before and after COVID to eliminate some other costs like fixed costs and taxes that is not going to change because of COVID. Okay. So there's a rapidly growing literature here. I am missing a lot on this slide because I mean, unless you update this slide every week you will miss literature and the entire economics profession is working on COVID which is very good by the way. I mean, I think this is a very good development. Our contribution here is going to have basically model-based estimate, this is a structural model combining with firm-level data and other sector and aggregate data to get at the COVID shock because that's going to allow us to estimate heterogeneity in failure rates not only by firm, by sector and that will help us to quantify the effectiveness of the governments. Let me tell you about the methodology. So methodology is going to be, as I said, a simple model and then once we have this model we are going to introduce COVID shocks calibrated from the data. So basically it's a very simple setup. So firms are going to produce output. So firm is going to be denoted by I subscript and S is sector. So firms are going to produce output using, oops, I'm sorry, using idiocyclic productivity, capital, materials and effective labor. So A here orange, again, everything orange is going to be estimated so on the data that's going to be sector-level productivity shock due to most of us went home working and there's going to be some lower productivity with that. Demand is going to be standard as the CES demand function with sector firms within sector selling differentiated varieties, usual downward sloping demand. And here we are going to do everything in terms of hot culture. What does it mean? That basically we are going to do before or after COVID. So demand of firm product denominator is before COVID normal times and D prime is during COVID, COVID times. So D hat IS is going to be changing demand for firms products from normal to COVID times. And there's going to be the sectoral shifter and that's going to be actually can be positive or negative, right? We are going to capture the fact that you don't go to restaurants, but that demand will go down but you increase your online deliveries, right? Online grocery shopping, that's going to go up. So that's going to be your size sectoral demand shifter and PD hat is going to be this aggregate demand shifter which is going to come from other day thing. The firm minimization problem is going to be stayed forward. Firms are going to produce to meet demand. This is a short-term model. First, we are going to keep the prices fixed and output is going to be basic all demand determined. Later, I'm going to show you what happens when you relax this in the basic what we did in the Jackson Hole paper. The whole thing here is, of course, you are going to be subject to labor supply constraints. Certain sectors are going to be limited due to lockdown and also due to the health shock itself. You don't have the health shock here, but we are going to calibrate the lockdowns and then labor is not constrained. You have your usual first order condition and labor is constrained. You are going to have this supply constraint problem in those sectors. So going back to the original equation I showed you, cash flow is going to be firm revenue minus firm cost. And then by using our first order condition from this firm minimization problem, we are going to write this cash flow during COVID minus cash flow before COVID green. We observed that when labor is not constrained, that's the first order condition and you don't have the constraint the second first order condition when labor is constrained, that's going to be the change in the cash flow. And then your business failure is going to be based on this liquidity criteria based on this cash flow during COVID, which is the original equation I showed you. So how are these orange things determined from the data? How do we take this model from data? Okay, for labor utilization constraint, basically all non-essential workers are going to be a syndrome worker. So we are going to separate firms essential and non-essential based on several data set and basically the feasibility of remote work is going to come from on a data set or explored further by Dingler and Neiman. Then we are going to adjust the productivity based on ACS data. So that's going to be a more ad hoc way of looking at it. We are just wherever remote working which sector is having, you know, these remote workers, we are just going to adjust product down by 20%. This is not an assumption that makes a lot of difference. We do a lot of robustness on this. The critical thing is going to be how we calibrate the demand and supply shots at the sector level, demand shot especially. So we are going to here rely on face-to-face interaction to capture the idea that your demand for restaurants go down, your demand for online groceries goes up. That's how we calibrate. And under demand shot, we are just going to keep the GDP growth forecast and the actual GDP depending on when we do 2020 or 2021 numbers. And all sectoral shots are going to be defined at the four-digit sector level and firm failures are going to be, of course, at the firm level. I'm going to show you now figures aggregating a little bit, otherwise it's going to be hard to see. So those four-digit sectoral demand and sectoral supply shot, I'm going to aggregate them to two-digit so you can see clearly. So on the left, I'm denoting the supply shots at the two-digit sector level on the right of demand shots and the color coding is that orange sector are essential sectors and the dark blue is not essential. And you see the pattern you expect to see, right? On the sectoral supply shots, obviously the sector that affected a lot from the lockdowns are going to have very steep sectoral supply shots up to 15 percentage points, like accommodation on food. This is like closed restaurant story, mining, entertainment, recreation on that as opposed to electricity and waste observing smaller sectoral supply shots. On the right, you see this relative shift in the demand, that side parameter for some sectors, demand is going to improve, they are going to be mostly essential sectors, other sectors that rely a lot of customer-oriented interactions, again, accommodation for the entertainment and recreation, they are going to face very severe sector-specific dimensions, okay? All right, why do we assume certain things? Let me tell you about that because some of these are going to be relaxed later. First is liquidity criteria. So we are going to decide firm is failing when that equation I showed you fails, meaning cash attack and cash flow become less than your expenses and then use them. That's a liquidity criteria, not a solvency criteria. Why do we do this? Because again, our focus is on SMEs, right? We are not going to be working with Galaxos and Clines, Amazon, Google's off the road. We are not working with lists of companies. This is all SMEs defined as firms less than 250 employee in Europe, less than 500 employee in US. So basically, a semi-accessive credit market is needed even in normal times, let alone a shop like COVID that hits the record area, okay? That's our main reason. Our second reason, of course, is a data requirement. Reason, insolvency is going to be extremely hard to define for these firms. These are private firms. They are not listed on the stock market. They don't have a market capitalization. They don't have topist queue. They don't have an equity that you can value at the market price. When you look at an equity number on the book, they report basically book values in the balance sheet, you see a negative. You really don't know what that exactly is. It's very hard to pin down a negative equity for these firms. So we are going to go with the equity criteria. In the first round of facts, we are going to test imperfectly, just prices output is demanded over the short term because we really would like to do these weekly estimations and see what happens the first couple of months of COVID where obviously prices are not going to adjust. Later, now we are 18 months out in the crisis and of course inflation is becoming a big concern. We are going to lack this assumption of Jacksonville paper. Again, we are doing a partial equipment exercise to estimate the first round effects. We think this exercise is something policymakers can use when they don't have the real-time data but they have to put together a program like PPP in an image. So that's how we see our exercise being valuable. Input-off with network is going to be important, of course, because especially when price adjustment starts, price adjustment in one sector is going to affect the other sector. We don't do it in the first round of estimates in our first paper, but then we are going to relax this by introducing a full-fledged, not only domestic but a global input-off with network in our Jacksonville paper. And the calibration of shocks, again, this is hard. I mean, in the original paper, we are just going to do a eight-week lockdown on everyone and that's going to give you the supply shock and the demand shock is going to come from this face-to-face interaction. Later, we are going to do it more realistic because obviously the path of the economy, the way GDP evolves is going to feed back here. So we are trying to use a lockdown of stringent data for works for Google mobility data to really try to separate supply and demand shocks at the sector level back. So I will tell you what happens to the baseline SME failure rates when you do all these relaxations in the Jacksonville paper and you see that actually there are going to be some differences but overall, not a big difference. So I will mention that when I come to that slide. Let me show you the baseline failure rate. So this table is going to give you added a SME failure rate in three columns, non-COVID failure rate, COVID failure rate and the change. We are going to focus on this change because that is going to help us a lot to subtract things that are not going to be changing with COVID and not COVID like tax, right? So, and there are two roles here. What are those? In the original estimates, we are going to use 17 countries and some of those countries has very high coverage, meaning we almost cover universe of SMEs. That's what the high coverage countries are. And again, here the assumption is no government intervention. You want to show you the failure rates in the absence of government intervention so we can evaluate the effectiveness of government intervention. And the lockdown is an eight week lockdown on everyone, we don't get into heterogeneity with these lockdown stringencies and different timings of lockdown and all that. So same timing, eight week because these are the 2020 estimates. And what I'm showing you is the two letter failure rate at the end of 2020, okay? So basically it says 18% of the firms would have failed without the policy relative to a normal time failure rate of 9%. So you are experiencing a nine percentage point increase in SME failure rates, okay? So that's quite a big number. And later I'm going to show you that actually policy was very effective in reducing this counterfactual that would have happened in the absence of policy. This added failure rate is going to mask a lot of heterogeneity across sectors and across countries. So let me show you the sector of heterogeneity. Again, we do this for digit but I'm going to show you this two digit version so you can see it visually clearly. I'm plotting the delta failure rate here from COVID to non-COVID on the Y axis and you see that nine percentage point increase is actually can be as high as 25 percentage point increase in a sector such as entertainment and recreation but that's the sector that is sitting a lot both through a negative supply and negative demand shock perspective. And it can be as low as an essential sector such as electives, two percentage point increase. Similarly, counterwide heterogeneity is immense. I'm going to show you some countries here to make this point. A country like Italy, the change again so the SME failure rate from non-COVID to COVID can be as high as 13 percentage point whereas a country like Czech Republic is only experienced an increase from non-COVID to COVID and SME failure rate of around five percentage point. Why is that? There are going to be two factors here. First, of course, the balance sheet, the firms enter the crisis. So Italian firms are going to enter the crisis with weaker balance sheets so they are going to run out of cash pretty quickly once they got hit by this COVID shock a shock of their cost and revenue. And also the sector composition of different economies, right? Depending on the weight of the economy in the sector that are heavily hit by COVID you are going to be affected. But this heterogeneity is going to be extremely important later on both sector and country when we think about the effect of the policies and also international support of the policies. Okay, let me now show you what happens to baseline failure rate when I allow the prices to react. So basically, when you experience this shock you can increase your prices to make up for the increase in cost and also you can cut down your labor and material demands. As a firm, you can fire people, you can increase your price. So you can do the regular thing you do when you experience that shock like COVID through the prices before we already allowed you to fire people and cut down on materials demand but we didn't let you to adjust the price. Now we're also going to let that in the context of an IO icon, okay? We are also going to put a full-fledged IO network there because obviously price in one sector is going to be affecting price in the other sector and demand in the other sector. And here we are, this is also going to help us to allow a real location of firm demand. What is that? So the firms are going to be exited. Our liquidity criterion is going to stay the same. If it is extension, we will still tell that firm is exiting once that firm is having this liquidity gap but demand of that firm product can be reallocated to survive, okay? So that's what we call extensive marginal reallocation of firms there. So then we allow all these actually very interesting with that nine percentage point failure rate I showed you doesn't change much. So our original very simple setup is actually a very good approximation even so basically it says you don't really need this type of extension for that average number. Of course there are a lot of reasons for that and it might have gone differently for a different set of countries. First of all, we are increasing our 17 countries to 27 countries here. That is going to allow us to do real life lockdowns. We are not going to do this eight week lockdown on everyone anymore. We are going to try to capture real life experience of countries with different lockdowns stringes in 2020. We are going to use now 18 advanced economies and nine emerging markets that will allow us to do a serious decomposition between advanced economies and emerging markets. Our previous sample in the original paper we only have two emerging markets. And again, these numbers are no government intervention cumulative rate at the end of 2020. So that nine percentage point increase in SME failure rate over all average 27 countries is there but there's a big difference between advanced economies and emerging markets, okay? Advanced economy, a failure rate increasing only around six percentage point from non-COVID to COVID in the absence of government intervention again, whereas emerging markets is increasing 12 percentage point. And the reason for that is the additional input output. Maybe without the input output input output network, we wouldn't have this difference because when you look at that global trade and production network data you see that emerging markets input source is very concentrated. So a shop like COVID trickling down through this networks all the supply chain by the next stories that we read about every day. That is something that is going to affect emerging markets more than advanced economies which is giving you a higher SME failure rates in the absence of government intervention. There's also a mechanism here that works against increasing failure rates and that's the extent of much, right? I told you that then the firm exits the demand for that form products is going to be reallocated among surviving firms. So that is helping surviving firms to a certain extent. So we have two forces here working against each other in terms of SME failure rates, IO network and the extensive market. Okay, but overall we are going to have this number that SME failure rate is going to be increased overall nine percentage point in the absence of government support. What happens with government support? Now, there has been many, many different programs. So if you look at this firm support through Yale there's over 500 programs in our countries and most of these focus on small firms but there are programs for all the firms. So we are going to work on three, focus on three. Why these three the most common across our 27 countries, we can get data on them through ECB and ESRB. So basically these are pandemic loans, grants, direct grants and wages, okay, rent and tax waivers. And then we are not looking at like furlough schemes, other unemployment benefits of that just focus on these three policies. Then we do that, the first result we find is policy support was affected. What do I mean by that? In the first column here I'm showing you again that increase in SME failure rates, nine percentage point overall, around six for advance and around 13 for emerging markets. First I'm going to do a hypothetical cost of saving these firms. What do I mean with that? I have a model so I can do a counter section where I go and surgically like a doctor using tweezers, save the firms that are, well actress, right, because I know who they are, I know the firms with the liquidity gap and I can just save them. If I can do that, that will cost me almost nothing, like 0.8% of GDP. So it's a targeted, it's a fully targeted bailout, it's extremely cheap, 0.1 in advance, for me is 1.5% GDP in emerging markets, very, very cheap. Now actual policy here, I'm looking at the actual funds disperse and remember three policies, the pandemic loans, grants and waivers, they cost around 4% of GDP, 6% of GDP in advance, 1.9% of GDP in emerging markets, and they are effective why I call this slide policy support was effective because that's failure rate of nine percentage point increase SME failure rate now is 4.2. Okay, so basically the policy support, these three policies cut the failure rate in half, you are actually in negative territory in advanced economy. So basically you are doing better under COVID than non-COVID, right? This is a minus failure rate. Why? Because fiscal support was tremendous in advanced economies. I mean, the difference is huge. I'm going to show you a figure in couple slides on this, but the fiscal support is just so large that you fully offset SME failures in advanced economies. This doesn't happen in emerging markets, but still instead of around 13 percentage point increase in failure rates, you only have nine percentage point increase, even your fiscal support in emerging markets in terms of pandemic loans, grants, and waivers is much less, only 1.9% of people, okay? So in that sense, policy support was effective. However, all this doesn't mean policy was efficient, right? Because it wasn't targeted because this was like wartime, muscle field. You cannot go and do what I did in the month. Like you cannot go and do like a surgeon, take a tweezer and pick all these firms and save them. You cannot do that in real life. And nobody did that. I mean, most of these programs put together in matters of peak. So it clearly, it's a poorly targeted policy. That's no surprise. The good news is the silver lining is, there's no notification. A lot of the worry with these poor targeted policies is like we are throwing this money on everyone. Oh my God, we are creating all these zombies. This is going to be horrible in terms of medium and long-term productivity. It's all over again, back to the world of financial crisis 2008, 2000. This doesn't happen. I mean, the policy was not targeted, obviously. You only save 36% of the firms at risk. So before in my exercise, I was saving 100% by medical surgery. Here, you only save 36% of actress firms, which is around half of the jobs at risk jobs, you say. So that's why you still have, of course, unemployment, but only 2% of the funds and 2% of the funds go to zombie firms. And how many firms you are saving are zombies. Only 13% of the firms at risk are zombies. Okay, so in that sense, this is good news. But does it mean it's going to be good news in 2021? Maybe 2020 fine, okay, but then you are creating constantly creating zombies by keeping these policies up in place. We actually don't find that either. So we find that, although you don't save everyone at risk and you are not targeting well, most of the firms you save with policy are actually viable. So it is not all of the case that you didn't do any zombification in 2020. There is also no future zombification. Again, here's future. We only come until the end of 2021. I mean, 2022, 2023, who knows? I mean, this tends to change, but so far, no features of zombification. Why is that? First of all, by the end of 2020, failure rates are increasing only by 2.6% per each point, relative to normal. And this is even you make firms pay back their pandemic loans. Because pandemic loans were like super good at super good terms over a very long time, mostly five years, and most of the time government is taking some risks. So even when you make firms pay some of it, still you are going to have only a 2.6% point increase in the bankruptcy rate. Some of the firms survive till the end of 2020 because you make them survive with policy. They're also surviving until 2021, okay? In that sense, they are viable firm and only 22% of the firms are zombies that survive because of you, because of the policy, and only 13 are failing by the end of 2020. So this is not, this 13% of zombies, this is not a wall of bankruptcy that have been written a lot in the news that will happen by the end of 2020. So maybe we don't find that. Okay, let me now say one thing in terms of the broader fiscal policy and the global implications, okay? So so far again, although I use many countries, I focus on countries' own policies in terms of pandemic loans, grants and waivers. Of course, fiscal policy was much broader than that. Tremendous fiscal stimulus. And that's going to have effect globally through what we call global production, okay? I showed you a result with the Domestic Outlet Purp but of course this network is global. This is actually from my other paper on vaccinations making the economic case for global vaccinations from January 2021 that shows how much advanced economies are going to lose by not vaccinating the poor countries and actually recovery is never going to be full without everybody's vaccinated. That's basically what we said in January 2021. And here we are pretty much observing that situation. And that's very clear once you look at this picture, right? So on the left here, I'm showing you a readable version. When you put all the links, this is going to be like a big spider web. You cannot see anything but with the most important links of trade between countries and this is color coded, darker blue is more open countries, lighter blue less open countries. And so the circle, the boxes around countries are going to give you the vaccine inequality across countries. The important thing is this right-hand side figure on the sectoral, which is going to be embedded on this. So what is going to happen to firm failures and what is going to happen in one sector shot is going to affect all the other sectors and all the other countries, okay? That's exactly why it is not that easy to get out of this supply bottleneck problem. And this sectoral figure, you see, it's not about being a tradable sector or non-tradable sector. So this is color coded, dark sectors are tradable, light sectors are non-tradable, but it doesn't matter given the links, right? Construction sector, we all heard the numbers, sort of wholesale and retail, they have so many links with other sectors in the economy in terms of buying and selling that even they don't directly trade themselves, they are going to be affecting other sectors and themselves being affected to this. We incorporate that in our model and we combine that with this inequality in fiscal policies, okay? So this is now fiscal spending percent of GDP overall. So it is not anymore only pandemic loans, waivers and grants, but from the NF tracker, how much fiscal spending countries did. And of course, advanced economies on average have like a very large 16% of GDP number, whereas emerging markets has like a bare five percent, okay? There's huge difference between these kind of, this is the fiscal space we have been talking about and advanced economies went with whatever it takes approach, they have large fiscal space, they can afford that. Emerging markets just don't have that type of luxury. All right, this is going to have implications in terms of not only your own domestic economy, fiscal policy effectiveness and saving firms, but also globally, right? So those numbers I just showed you, 15% of GDP and advanced economies and 5% in emerging markets, that's around 11% on average of GDP. And our model is going to tell you that that's going to raise output by 0.67. This is going to imply a very low multiply, right? So when we focus on transfers, like we know that with an MPC of 29, the textbook transfer multiplier, now we are focusing on transfers in fiscal policy, that's going, that has to be 0.4, okay? We are getting 0.06 multipliers. It's almost like fiscal policy is not effective. But this is a misleading way of looking at it actually, because this, what you expect based on a transfer based multiplier of 0.4, improving only the demand in the demand constant sectors. We have a lot of supply constant sectors because of the IO network, so, right? All the 31% of global GDPs in demand constant sector, that means you should expect this multiplier of 0.13. We get 0.06 much lower than that. Why? Because of this trickling down effect of price. We let prices take chance, but once you increase price in demand constant sector, that is going to have an impact on decreasing demand in the downstream supply constant sector and that effect is going to reduce your multiply 0.06. So this doesn't mean to be a pessimistic message, right? This is not saying fiscal policy is not effective. Of course fiscal policy was effective, but in a different way than you said, it did stimulate output to the extent it can in the demand constant sectors. But it also shifted employment through sectors, right? So this Canadian employment story that has been now being studied in many papers, fiscal policy helped that a lot by elevating spending toward demand constant sectors from supply constant sectors. And that helped you actually improve unemployment. So there's a very big decline in Canadian unemployment that is thanks to the fiscal policy. So here we are showing you a kind of a even broader role for fiscal policy under a shock like COVID where we do need to be thinking about different demand constant and supply constant sectors. What does it mean globally? It means globally also we are not going to have our standard fiscal lower story, right? The standard story is U.S. spends a lot. That's good for everyone because U.S. consumer spends a lot, U.S. consumer demand increase. That is very good for all the other countries to a trade channel. They are going to sell more to the U.S. consumers. Okay, we find that that happens only for Mexico and Canada, although very small, all other countries it is negative. It's negative to bloovers because here there's going to be these other assets coming from the supply and demand constant sectors and the do I own it. Okay, so it's not that straightforward just because U.S. did a lot of stimulus and U.S. consumer spending a lot, this is going to lift all the time. It's going to lift some types correct with the ones that you have close trade relationship but not all the time. In terms of unemployment though, it is a very good thing. So these unemployment forwards there, U.S. spending, I'm showing you only the effects of U.S. fiscal policy here, that is going to reduce unemployment everywhere, right? Because it's going to help a lot to shift this unemployment problem between sectors. Although the numbers are small, so both output floors and employment floors small but you would like to highlight the fact that output ones are actually wrong direction, right? So they are bigger than it, which doesn't happen with a standard shock in a standard model. Okay, final thing I want to say that I want to conclude is these two speed recovery. So we do know that there's two speed recovery. I mean, that's largely because of the unequal global vaccinations in the world. So what our global model is going to tell you that during this phenomena, because the private savings is going to decrease because now U.S. advanced consumers are spending but not emerging markets consumers, that's going to affect on a global interest rate. It's fiscal policy, global interest rates can rise as high as 5% where the trade deficit is going to deteriorate in advanced countries. It's going to, of course, improve in emerging markets by construction. Now the interesting finding here is output is good. Recovery is good, it's positive on advanced economies but not in emerging markets. Output is actually going to decline in emerging markets. That comes directly out of our model, global model with IO networks and this higher global rates. And the big story there is, there's of course the terms of trade effect but there's also this differential risk premium. So what we did is we update this result from my 2019 Jackson home paper with the idea that okay, if the global rates are going to rise and there's this inflationary concern, advanced economy central bank is going to start raising interest rates, especially Fed. And that, this figure basically shows you the effect of a US monetary policy surprise tightening on the risk premium across advanced economies in emerging markets. And originally I showed this result in my 2019 Jackson home paper with data coming till 2018 here. We update the data and show you the same result. Risk premium is going to increase a lot in emerging markets here on the right with a US monetary policy tightening but it's going to decline in advanced economies. What does it mean? Emerging markets are going to be in even worse shape because their external financing costs is going to skyrocket much more than advanced economies in the middle of a two-speed recovery. This goes back to what Ken Rogoff said this morning at his keynote. So there is going to be a very dire situation for emerging markets. They are already in bad shape. And once US monetary policy and ECB monetary policy start normalizing, this is going to get worse through this differential risk premium channel in emerging markets and advanced economies. Okay, let me conclude. So the takeaways from this broad research agenda is first and foremost, policies prevented from failures. These are the policies that are targeted for the firms. So government support policies to the firms and they didn't create zombies. So this is good news. They reduce bankruptcies, they didn't create zombies but there are funds wasted. Actually, most of the funds, 88% of the funds went to firms, strong firms who didn't need it, okay? So that means moving forward, we have to find some sort of a mechanism to claw back some of this money through excess profit tax and schemes of that sort. The same is not strong and they don't need additional support. So this is telling us we should actually be dialing these policies back. The multiplier from fiscal transfers now when I move to the broader transfer of fiscal policy, they are going to be small but that's because of supply concern and IO linkages just the nature of this shock. It doesn't mean fiscal policy was ineffective. It's not effective actually. Definitely help reducing from failures and also allocated employment from the sectors that didn't need to sectors that need, okay? So in that sense, it is very fiscal policy with play a very important role in this crisis. The cross-border swallows are limited though not to the extent that even emerging markets have small fiscal place, space, they couldn't spend much. That's okay because US spend much. That's not the case unfortunately with this shock. So everybody on its own in terms of the fiscal policy which is going to mean in the future that emerging markets are going to face difficult times unless we close this vaccination gap that their economies recover because the air fiscal packages are always going to be small and that's going to create this headphones for emerging markets and that's going to get worse with the rising global rates and risk. So in terms of the key message, key risk to manage. So the key message in terms of what is the risk we need to manage moving forward is the financial market path, okay? So there's something very important here. So we don't know this yet because we don't have this type of detailed data for every country, real time. But if you look at US regulatory Y-14 data and now there are several papers documenting this what happened during COVID in terms of credit market versus government, okay? Who is tapping the credit market and who is getting help from the government? Large firms are the ones that tap the credit markets. If you're a large firm, you'll face a shortfall in your liquidity, in your revenue. You can go to a bank and close that. If you're a small firm, you cannot do that actually. And we shouldn't be surprised about that because small firms couldn't even do that. It wouldn't be the normal times. What happened with COVID and with this tremendous policy support is positive filled in for credit markets, for SNEs. But that's a very important result, right? In terms of moving forward, so we should still do fiscal tapering because fiscal policy did a job. More support is going to have this upward pressure on prices, rates, that is going to force monetary policy hands to increase rates. But we should taper fiscal, but monetary tapering has to be clear to communicate that slow because anything that is going to spook the financial markets, not only going to make life difficult for small firms who already didn't go there, but really basically survived because of the government, but that's going to make life for larger firms difficult. If there's some sort of financial market panic, then pretty much all firms are going to be in trouble and that's definitely a scenario you would like to avoid, not only in advanced economies, but also emerging markets because the implications for emerging markets are even going to be worse. Thank you very much.