 This paper is called On the Effectiveness of Climate Policies and it's joined with John Hussler, Parke Sel, Michelle Reiter, and there's a usual disclaimer there. Okay, so I will give a quite lengthy introduction here. We have a like a motivating question you could say is what can macroeconomists do to contribute to the aim of controlling climate change? So we have, you know, since sometimes we have Nord houses integrated assessment models, the DICE model and the RICE model, and these models are good because they have allowed us to understand the interconnection between the climate and the economy. Okay, but the question is if these models are useful. I think so, but there are people who have criticized them, you know, quite brutally. So Pindike, for instance, in Journal of Economic Literature says that these models have, they have many problems, so many problems that they might actually be, you know, unusable. And similar criticism is laid out by Stern and Stiglitz recently. There has been some progress made though, like by my co-authors, for instance, and others. So we now have more accessible and decentralized versions of, for instance, the DICE model, and we can get clean formulas for optimal policies, such as optimal carbon tax. But still, you can also say that, as Oswald and Stern point out, economists are very largely absent from this research area. And why are they absent? Well, one thing could be that the problem of climate change is conceptually trivial. Humans cause warming as a byproduct of burning fossil fuels. And this is a classic case of a pure externality. Already 100 years ago, Pigot figured out what we can do in such case. We should just apply a tax equal to the total marginal externality damage that the polluter is otherwise not paying for. And this is rock solid logic. So no research has basically questioned this basic insight. And in addition, thousands of economists have signed a petition urging for the adoption of a carbon tax. So you can conclude that maybe there is nothing conceptually interesting to work on here. But we say that we want to make a case that there are plenty of important and non-trivial things that macroeconomists can and should work on. And we want to illustrate this by example, and show how to address important questions and hopefully yield interesting answers. But of course, we will not do this in a perfect way. So that just indicates that there are much more work that can be done. So a key message we have is that these integrated assessment models, they are actually super helpful. Maybe perhaps not for what they have been used for so far. But for the analysis of suboptimal policy. In particular, we can do cost-benefit analysis of policy packages for what we should do around the world. And these packages might be very far from Pigou's recipe. And we can compare them and tell the policy makers what we find. Which of these policies would have an effect, a quantitatively important effect, which would be to expensive and so forth. Some policies might actually be meaningless. They may not have any effect. So these are the questions we can answer. And this is an opportunity for economics. Because natural scientists, they actually have no idea how to do this. But we do. We are the experts on understanding how markets operate. I mean, the decisions of households and firms and how different regulations and taxes act. So we're not really here pushing for a classic case of a second best policy. But we are pushing for quantitative analysis. For instance, we might want to compute how much more certain policy packages would hurt our welfare than, say, a global carbon tax. If we have some specific goal that we want to achieve, say that we would not like to increase the temperature more by more than by X degrees by a certain year. So we have several contributions. One contribution is that is like you could call it methods. We offer a framework and integrated assessment model for quantitative analysis of policy designed so that it can be built upon further. And the key features are that it's decentralized, which is key here when we want to look at sub-optimal policies. It's rich enough to ask important questions. It's bad enough so that you can see what more we could work on, although that's not on purpose. And it builds entirely on modern macro. This is in principle a complicated model. You will see it will have many regions. And there will be many like energy inputs and so forth. We have a representation of the climate. There's a carbon cycle. Still, the equilibrium is very easy to solve for. You can almost solve it with pen and paper. And it solves in a computer in under a second. I mean, this is not key, but it's, of course, a great thing. And we're going to use this framework to ask three quantitative questions. And these questions are the following. What if the right kind of policy is used, but at a level that turns out to be wrong exposed? So this is highly relevant because there is so much uncertainty when it comes to global warming. We don't know exactly how much the temperature will increase when we emit carbon dioxide into the atmosphere. So there's huge uncertainty here. And there's also large uncertainty about the costs of these damages. So what we're going to look at is we're going to compare a global carbon tax that turns out to be too high exposed. And compare that to one that is too low exposed. And we're going to show that the former error is not very costly, but the latter is, meaning that to underestimate global warming can be very costly. But to be slightly more ambitious is not particularly costly. We will also look at if you use a particular form of bad policy design. So this is an argument you hear that maybe we want to have different carbon taxes around the world. Maybe we want to let some countries off the hook. Some poorer countries, maybe they should grow first before they implement policies. And we show that this is actually not a good policy at all. It can be super costly if we let big regions off the hook. The last thing we look at is if we can promote green energy instead of taxing carbon. And the answer that we find is one that many people don't like. But we find that it's very likely a bad idea. It's clearly not a substitute for taxes or anything. So it would be very risky to rely on such a policy. I say here that we look at endogenous technical change and we do that in the paper, but I'm not going to cover that in this talk. Okay, now I'm going to go into the model in a transparent way, hopefully. So there are our regions and each region I features a representative consumer with standard preferences. One region, region one, is the oil supplying region. They sell conventional oil and they have a finite oil reserve that and it's extracted at zero costs. Regions two to R, the remaining regions, they are oil consuming regions. They have no conventional oil, but they import it from region one. So in the oil exporting countries, they have a stock of oil and the low motion for that is given on the bottom page here. So the stock of oil that you have in period T plus one equals what you have in period T minus what you sell in period T. Energy services in in the oil consuming regions are produced by local competitive farms that combine different energy sources as inputs. And we're going to allow for the inclusion of unconventional fossil fuels that are highly substitutable with conventional oil. So here we have an oil composite consisting of conventional oil. That's the E1 argument there. And then we have L sources of unconventional oil. We will only look at one, so we think of that as fracking. And then energy services are produced by a representative firm in each region that combine the oil composite and the additional energy inputs. Could be coal, green energy sources, etc. And now there is the raw cap just the elasticity of substitution between these objects here. In each region, there is also a representative firm that produces a final good according to a Cobb-Douglas production function that you can see in the middle of the page. And as we have shown in other work, Cobb-Douglas production when it comes, you know, to incorporate energy is not good if the time horizon is short, say up to a year. But here a period will be like 10 years. And then it actually is not a bad approximation. There is also a resource constraint at the bottom of the page. So what we consume in each region plus what we save has to equal income minus the costs that go to energy payments. This is the natural science part. So each region emits carbon dioxide emissions because they use fossil fuel. And these emissions are denoted by M. And then the stock, the atmospheric stock of excess carbon is denoted by S. And it's given by the equation on the middle of the page there. So it's a function of emissions and also historical emissions. And there is some depreciation. So carbon dioxide leaves the atmosphere with some according to some process that we capture with this parameter D. The temperatures we steal from North House. So this is North House's temperature models. The energy body has in its dyes and rise models. There are two temperatures, one for the oceans and one for the surface. And at the bottom of the page, we have a productivity component A. So it's a function of set, which you can think of as TFP growth and gamma times S, where S again is the stock of carbon dioxide. So here's where the externality comes in. The gamma is positive. So that means that if you increase the stock, it will actually low, you know, it will take away some of the production. Governments, they in each region sets carbon taxes. They run balanced budgets. They are local. So no intergenerational, sorry, no interregional transfers and they rebate tax revenues in proportion to income. Some limitations. There is no trade between oil consumers. And there is no borrowing or lending in them. I think that trade is a limitation. And that's something we're working on, but it complicates things quite a lot. Now, the properties of the equilibrium, all firms maximize profits, all agents maximize utility, all markets clear. Now, our combined assumptions that we make here that I have shown you, implies that the equilibrium is determined sequentially without any forward-looking components. So that means that given a world market price for oil, all equilibrium conditions have closed form solutions. So in each period, finding the equilibrium is only a matter of finding the equilibrium oil price. And this is very simple because the supply is predetermined and given by 1 minus beta, where beta is the discount factor, times r, where r is the stock of oil. So it solves in a millisecond, even in a program like Excel. But again, we want to be quantitative. That's our first aim. But the fact that it solves quickly is of course good. Here is the equilibrium. I will not go into that, but here you can see the closed forms. Okay. Now, I will move to the calibration of this model. We set up a specification with eight regions and four sources of energy. But it's straightforward to just, if you want to have 25 regions and many more sources of energy, that's straightforward. It will not be a problem really. So we look at US, Europe, China, India, Africa, Oceania, South America, and an oil producer. We think of the oil producers as it's basically OPEC in Russia. We have four sources of energy, conventional oil, fracking, as I said, the unconventional oil, coal, and renewables. We only allow for fracking in the US. The key aspects here, the elasticity of substitution between oil, coal, and green energy is set to 0.95. And that's taken from a study by David Stern. And the elasticity of substitution between the different components of the oil composite is much higher. So it's around 10, we set it to 10. And we calibrate the model so that China subsidized coal production and that the modern product of capital is equal at time zero. We also include convergence in the model. So regions like Africa and India, they will grow faster and eventually catch up. This is to just show you that we can calibrate this model to match the data. Here we have a share of carbon dioxide emissions in the data and in the model. And at the bottom we have share of GDP, world GDP in the data and in the model. And you can see that it matches the data well. Now, this is just to show you some stylized experiments that we can run against model. I will get to the three questions I have, but just show you this. This is just some simple output from the model. You have degrees Celsius, that is temperature on the y-axis, and you have a time scale on the x-axis. Now, there are three lines basically on top of each other at the top there, one blue, one red, and one black. The blue line is the LACIFER. That is no policy whatsoever. And we set the climate sensitivity here to three. We'll talk more about that in a second. Anyway, this doesn't look good, as you can see already by the year 2100, we break the three degree line, three degrees Celsius line. And after that, the temperature keeps increasing up to eight degrees at the end of the period we consider here. So not a good scenario to be in. Now, the red line shows carbon taxes in the European Union. So we set kind of like an optimal tax, but only in EU. And as you can see, that doesn't really take us far. Europe is too small to have a big impact. It's like 10% of the global emissions. And you can also see the black line, which is cold taxes in Europe only. So then we don't tax oil. And those two lines, the black and the red, they are actually exactly on top of each other. So it doesn't matter if you tax oil or not in this model. Because oil is so profitable. So if Europe don't want to buy the oil, the suppliers of oil can just sell to other regions easily. They can lower the price slightly. They will still make a profit. The purple line is that if the whole world would implement this moderate tax on carbon dioxide, and then we see we can get the temperature down to three degrees. That's why economists love this tax. It's really effective. If we can implement it and get broad support for it, it can really do the job. The blue line or whatever it's called at the bottom, that's the Swedish tax rate. So that would take us even further. Anyway, now I will move on to the three questions I had. So suboptimal policy number one, the right design, but the wrong magnitude. So there's a large uncertainty about parameters. For instance, the IPCC, they compute this measure called the climate sensitivity. So the climate sensitivity just tells us how much the temperature we increase if we double the amount of carbon dioxide in the atmosphere. And the IPCC says that this sensitivity is likely between 1.5 degrees Celsius and 4.5 degrees Celsius. This is a huge range. It's a wide range because it matters a lot if it's like 1.5 or it's 4.5. And we don't know. So in the last IPCC report, they even don't give a me. They just say that it's in this range. So we don't know. And in addition, there are many papers that look at the costs associated with global warming. And they also report very large uncertainties. So very large uncertainties about the climate response and about how costly it will be. So that means that we're going to have to implement some policy in the face of this uncertainty. And probably we'll make something wrong. So here we just say that assume that we go with the climate deniers and go for 1.5 degrees. Say that the climate sensitivity we assume it's 1.5 and we assume that damages are low. Later, it turns out it was the opposite. The climate sensitivity was 4.5 degrees Celsius and damages were costly. Okay, then we have made a mistake and we can compute the cost of that mistake. And we have the opposite mistake. Assume we go with the climate sensitivity of 4.5 and that costs are high. Later, it turns out that it was 1.5. We have made a mistake and can compute again the costs on making these mistakes. So here you can see the result from this exercise. So if you want to focus here now on the top left graph, which is aggregate losses, the black line is the cost of underestimating global warming. So again, going with the climate deniers when the climate sensitivity actually is high. Now, these numbers are huge. Like, you know, there is a consumption loss of 20 percent. There's a consumption loss of 20 percent at the end of the interval. I think in the great in the great depression in the 30s GDP fell by like 15 percent or something like that. Here we have like a permanent, you know, I have a great depression every year if you underestimate global warming. And then the red line is the cost of overestimating climate change to impose taxes on carbon when they are not needed. And this is something we hear from the politicians that that's very harmful, they say. Here we say, well, it costs something that the line is about zero, but it's nothing basically, at least compared to the other mistake. So better to are on the ambitious side. The next thing we do is to look at non-uniform carbon taxes. We will let some regions off the hook, but we will have to increase the carbon tax in all other regions because we want to aim for, you know, that the temperature should increase by a certain amount. We set it to 2.6 degrees in the year 2165. I mean, this is just an experiment, but it's 150 years after the starting year in the model. And in the first case, we say that Africa and India do not have to implement any policy at all. That requires the carbon tax to increase by 5.3 times in all other regions. And in the second experiment, we let China off the hook and say that China only participates marginally. That actually requires the carbon tax to increase 20 times in all other regions. And in fact, if China does not contribute at all, it's impossible to even reach that 2.6 degrees Celsius. So here are the welfare costs from this experiment. So in the top you have where Africa and India are let off the hook. And you see that they gain something, but all other regions lose a lot, in particular the U.S. And that's because they hurt the fracking industry. And yeah. Connie, you have five minutes. Okay, perfect. Thank you. At the bottom, we see when we let China off the hook. So China actually gained quite a lot, but you can see, you know, all other regions lose substantially more than that. So it's not a good policy to just let big regions off the hook. They should participate ideally, but maybe there has to be some other, yeah, the thing that no one likes side payments or something like that. The last experiment we do is to look at green technological progress. So here we're a bit like reduced form. We just say, assume we can implement a policy so that there is faster technological change in green energy. So that the price of green energy falls by 2% per year. And assume we can stop technological progress in cold production so that the cold price increased by 2% per year instead. Then we get this. So here are a number of lines. So let me guide you through them. The first line you want to look at in the top graph is the pink one at the bottom. That's actually the global cold tax that we looked at before. That's just for reference. That's the optimal tax. That's what it gives us. Now the blue line shows you if you can have faster green growth and stagnant cold. So that's a bit surprising because the blue line is very close to the pink line. You can get very far by that if you could do that. But the problem was that I actually changed two things. I had faster green growth and slower stagnant cold. So if I only allow faster green but have neutral progress in cold, we get the green line. And then you see that doesn't really work. And unfortunately I stopped the graph here at 2140. But if you if you go out a bit, you will see that it goes exactly basically as in the last year. It goes up to eight degrees around the year 2200. So it's not enough. The black line is the opposite. It's neutral in green but stagnant in cold. So then you again get back close to the first best. So what this shows you is that it's not enough to make green energy cheaper. You actually have to hurt the coal industry. You have to make fossil fuel more expensive. And the bottom graph just shows you that it that so in the top graph we have actually kind of green and fossil are you know they are complements actually because the elasticity of substitution is below one. Meaning that if green becomes cheaper, you will actually consume more also fossil fuels because they're complements. But if you say that well I don't believe they are complements at least not in the medium run, you can look at what happens if we have a higher elasticity at the bottom graph and it doesn't make things better at all. And it's because fossil fuel at this point is still so much cheaper. So it takes too long for green to actually catch up. So I am done here. So I will just conclude and say that we construct this integrated assessment model and use it to obtain quantitative answers to important questions. Some of these answers were surprising to us and we think that they should be of relevance to policy makers. There are many things we can do and there are many weaknesses of the model. There's only one sector. There is no trade and there is no intertemporal markets. All right. Thank you. Thank you very much for being so disciplined, Koni. Let me just remind all the attendees. Please submit your questions in the chat. And now I would like to give the floor to our discussant, Emmanuel Muenich from Deutsche Bundesbank. So Emmanuel is shortly hindered. So I'm standing as a discussant. So I would like to share the slide if I could have that. Thank you, Sophie. Apologies. I was looking at the program that's on the website. Please go ahead. Sorry. I'm not able to share the slide yet. Okay. That should work now. So yeah. So hello, everyone. And it's been an honor to be able to discuss this paper. It's a very important paper. And also thanks to the organizers for being flexible. And let me discuss this paper where Emmanuel is hindered. So yeah. So let's dive into the discussion of the paper. So since we have all seen this paper, the Koni's presentation, let me be very brief when summarizing. So let me be very brief summarizing the paper. So in this paper, the authors develop a state-of-the-art dynamic multi-region, multi-energy source, general equilibria integrated assessment model with climate and carbon cycle modules. Using this model, the authors evaluate different suboptimal policies. And this is an important contribution to the literature, of course, particularly for real-world policy making. So there are three key findings. Firstly, a global carbon tax that is too high would imply a much lower welfare cost than one that is too low. And secondly, a carbon tax that differs significantly across regions would also have a large welfare cost. And finally, why there is a resurgence of interest in terms of the importance of green R&D subsidy as a climate policy. The authors of this paper actually show that relying only on green R&D subsidy can worsen global warming. So it's an overall very well crafted paper with careful analysis and many important policy messages. There isn't much we can actually point out here to make this paper even better. So my discussion today would be mostly centered around a few key assumptions. So the first key assumption is the fact that the supply of conventional oil in this paper is actually exogenous. And the reason is that the suppliers of conventional oil are considered to be price-taking. So the price of conventional oil is entirely demand driven. And this together with a log utility assumed in the paper would mean that the supply of oil does not adjust to the price changes. So a carbon tax would only reduce the cold use and unconventional oil supply, such as it's modelled for the US. However, we do know that in practice, conventional oil producers do have market power. For example, the literature on green paradox has pointed out that, for example, the OPEC oil producers may actually accelerate extraction of oil to when they anticipate a higher or tighter climate policy in the future. And this is also partially backed by recent empirical studies such as the 2019 ER paper by Balmeister and Hamilton. And this literature finds that they are actually price stabilizing responses by oil suppliers to demand shocks. So in the paper, the authors do acknowledge and also in the presentation that a model with indigenous oil supply would be more realistic. So my first comment and suggestion would be that it is worthwhile to actually pursue this extension. And if I heard correctly, the authors are actually already going in this direction. Now, the second comment concerns the log utility assumed in this paper. So it is an important step that the authors have taken to consider the impact and welfare benefits and costs when different regions have used different carbon tax rates. However, what is surprising is that a uniform carbon tax is still considered optimal in this setting, even though different regions have different TFP growth rates. And the reason why this may occur is again because of the use of the log utility in the model. So several authors have also pointed out if we use log utility, then the TFP growth rate would drop out of the effect of discount rate and thus would not affect a social cost of carbon. So consequently, also the optimal carbon tax to GDP ratio would not be affected by the different TFP growth rates. But this is not a general result. So if we consider a preference with higher curvature for example, so that the elasticity of intertemporary substitution will be lower than one, then a higher TFP would mean lower social cost of carbon and a lower optimal carbon tax to GDP ratio. So this may have an effect for regions with higher or lower TFP growth rates. So even though we may somewhat argue that log utility is a useful and valid simplification when we are considering the world's economy as a single region, but this is then more problematic when we are actually comparing different regions. And particularly because the exercise here is actually to consider the welfare benefits and costs if different regions use different carbon tax rate. So my second suggestion here would be to actually use a more general utility function, especially for this particular exercise. And then finally, the third major comment that I have today is about the use of global mean temperature as a sufficient statistic for damages. So in the paper, the climate-related economic losses are measured by region-specific damage functions which would map changes to global mean temperature to GDP losses in different regions. Now in the paper, the authors do allow the damage elasticities to global mean temperature to differ across regions, but this may still not fully capture the damage variation around the global mean. Since the climate scientists have pointed out that regions with different base temperature are warming at different paces and in fact the temperature in world's coldest regions is increasing at the fastest speed. So this is also a trend that is predicted to persist as shown by Gadia and Kofalo in their 2020 Journal of Economic Metrics paper. So this means that they are potentially not in your implications and including possibilities of tipping points and so on. So this is just a graph to again drive home the message that temperature change relative to global average is actually vastly different across the globe. So in this graph, in this figure which you see that the different shades of redness captures how strongly a region's temperature is changing relative warming relative to global average and the darkest red areas as we see on this figure are actually regions where they are warming most strongly and these are regions which are traditionally colder regions. So as one implication the act of sea ice is melting rapidly as we know this has potentially very damaging consequences. So to reiterate, it is already one step forward that the authors consider different damage elasticities for different regions but the different regional warming response may itself be an important dimension to still consider. So I have a few other comments which in the interest of time I will leave to after the session and now let me just briefly conclude. So this is overall a very important paper helping us evaluate welfare implications of different climate policies particularly by being able to compare the benefits and costs of suboptimal policies this gives policy makers an important tool to actually act. Now it's also important that the paper take into account original heterogeneities such as TEP growth rate differences and damage elasticities and so on but other heterogeneities such as temperature change around the global mean might still matter. And finally even though more sensitivity analysis might still be needed and my message is very clear so even though there is uncertainty concerning the optimal common tax it is important to know that when in doubt we can afford to be more aggressive with climate policy and I'm done with the discussion and thank you very much. Thank you very much for obeying the type limit Sophie that's very much appreciated and now Connie we have a few questions in the chat they were not sent to everyone so let me give them a loud and if you are not speaking please mute yourself because we can hear papers shuffling. So there is first question from Johannes Strobel what is the purpose of including fracking for the US? Then there is a question from Eric Poek thank you for a clear presentation it feels as if the model is very simple lacking more detailed mechanisms why not use more sophisticated models like that of the IEA? Also when you calibrate how good is your model if you benchmark it against history? And then the final question is from Andreas Breitenfeller often I hear the argument supported by the IMF's WIO that you need to invest in green first thus making it cheaper to have the alternative ready when people change behavior because of brown getting expensive due to carbon tax could your model confirm this line of reasoning? So in other words we need to invest in making green cheaper before we make brown more expensive. So Connie you have a few minutes to take these questions and perhaps react to Sophie's remarks. Okay thank you I mean absolutely I don't have time to you know go into detail with this I thank you so much for these questions and thank you Sophie for this great discussion I wrote down the things. I think I think Sophie to you I would say like you're right you point out and also in the chat you point out some weaknesses and I said upfront that there are things we can and should improve I think many of these things that that people brought up actually are straightforward to to you know incorporate into the model such as changing utility functions that's not a problem for instance. So of course I mean we want to include fracking because it's potentially important like a potential source I mean it's a it's potentially a large source for carbon dioxide emissions that we need to think about. The last questions was about first making green cheaper and if I could confirm that I would say I don't agree with that I would say it's the opposite you have to make brown more expensive that's that's our result it goes against that that intuition that was brought forward so since I don't have much time I think I'll stop there but thank you so much everybody