 Well, hi everyone. Thanks for having me here today with this terrific panel. Before we get started I'd like to introduce myself. For those of you who don't know me I'm Adele Morris, I'm a senior advisor at the Federal Reserve Board of Governors here in DC. I'm in the Division of Financial Stability and my work involves incorporating climate considerations into the Fed's financial stability framework. So, we're, we're working hard to think through how, how climate can feature in the mandates of financial regulators in the United States. I'm joined by two terrific panelists. Our first speaker is Lars Peter Hansen, who's the David Rockefeller Distinguished Service Professor and the director of the Becker Friedman's macro finance research program at the University of Chicago. I'm Peter Wilcoxson, who's a professor at the Syracuse University's Maxwell School, and he's a professor in the Department of Public Administration and International Affairs. He's also a non resident senior fellow at the Brookings Institution. And that's how I know Pete from many years of collaborating on our research at Brookings together. I want to kind of, sort of, first of all, thank you Jim Stock for the wonderful description of how economic models feature in various parts of climate making endeavors, or climate policy making endeavors. I got to give the disclaimer. I'm still getting used to this. So, like I said, I'm at the Federal Reserve Board of Governors. Everything I'm going to say is my opinion, not to be construed as the views of the Federal Reserve or its Governors. And that's like the same disclaimer every Fed person ever in a setting like this. So, just so you know these are my, my viewpoints. So, so I want to thank Jim for the terrific description on the applications of economic models in the climate policy making context and that was a terrific presentation. I'm going to kind of give you the research edition version of that like how our academic researchers using economic models in this space and there is a tremendous amount of overlap, because, you know, economists and others go into climate and energy economics, in part because of the compelling policy questions. So there are many overlaps in the applications of economic models, and the same principles apply you're using a model to ask a question and explore potential outcomes. And these are all, you know, projections of the future forward looking analyses, which is really important as we know in the climate realm where past is not necessarily prologue. And what I want to do is kind of relate what you're going to hear from Lars and Pete about their work into that kind of nice chart that that Jim gave us, but sort of the research edition of it. So at a very high level. We can use economic models to explore really basic fundamental questions about how society allocates resources. So, one of the things you're going to hear from Lars is to, is it in the category that I would describe as how should society allocate resources across the different parts of the climate challenge. So you, there's so many parts there's the market failures of the greenhouse gas emissions that are unpriced, and the public goods associated with fundamental technological research and development for low in low greenhouse gas technologies. You've got issues around adaptation and reducing the damages from climatic disruption, maybe raising roads for example. And then you've got questions that relate to other parts of social science, including diplomacy, and how to manage the distributional outcomes of climatic disruption and climate related policies both domestically and internationally to leverage US action into action internationally. So what Mars Lars is going to talk about is using those models at the very highest levels to think through how various dimensions of uncertainty in the climate challenge can affect the optimal mix of climate change in terms of batement and research and development. So it's not, it's not like Congress necessarily does that explicitly but implicitly society does make those tradeoffs, and, and it also depends like how do you think systematically about these issues, and how does your answer depend on the quality and, and magnitude of risk aversion, and how much do we care about those risks. So, now let me relate these things kind of into the rest of the workshop that you've, that you some of which you've already seen but we'll continue today and tomorrow. Yeah, we've heard about how the, the, the question of how does climate effect the most likely projection of GDP and there it's a forecasting application. This is a special kind of projection where we're trying to decide what's more likely than something else. And what does that depend on. And it's kind of a challenge here because climate hasn't necessarily featured in that literature very much yet. And if you're interested in kind of that, that literature and papers adjacent to that I do recommend that the paper that OMB and CEA produced, and it has a really nice lit review in there. Session three today, the question's going to look at the, the question's going to feature, what are the potential monetized damages from a disrupted climate and acidified ocean I put that in that category. I'm not necessarily going to look at an ocean acidification but those social costs of climatic damages are a well and currently actively being developed direction in the research. Some of the policy applications are quantifying those benefits of avoided damages when we're analyzing the net economic benefits from a particular emissions abatement policy. And this work can also inform the optimal policies for adaptation. How do we allocate across the different climatic damages to optimize how much we can avoid damages in the future by various investments. So there are many dimensions of that research and different kinds of models that go into it. We'll hear a little bit about that this afternoon. Of course, there's other aspects of those costs, not just their aggregate, but also how they're distributed across households by income class and other demographics by regional desegregation and the, the, both the spatial and the temporal dimensions of those climatic damages and all of those are interesting to researchers and to policymakers. Before perhaps the operative question would be, how might the economic and environmental outcomes of different policies emerge depending on what those policies and other assumptions that apply so for example, you know the key application that we talked about, or that Jim talked about in the keynote are, you know the macroeconomic outcomes of potential policies. And as we know there are many other outcomes that policymakers care about and the researchers in this literature generally want to be supportive of quality policy design and implementation right. What policies are going to be most cost effective, most equitable. How do you strike trade offs across those two different policy objectives. How do you design a policy so that it can address, you know, something that might be a concerned particular stakeholders, or some challenge to the policy design maybe you're concerned about trade exposed industries or you're worried about low income households. So researchers like to contribute policy design ideas that can support a functional policy discussion both in Washington at the state and at the state level, but also in international agreements. So some of the outcomes that feature in those models that are used for policy and analysis include obviously the macroeconomic variables, but also those distributional outcomes, sectoral outcomes, trade, labor market outcomes and outcomes that relate to the co benefits such as their quality and outcomes that can reveal things like unintended consequences, where you may be thinking you're helping one industry but you're inadvertently hurting another industry, or something to that effect. So in general the way we use those models is we're comparing a simulation of a, of a future that doesn't have a policy in it with a simulation that does have the policy in it so you need to have a model that can simulate the relevant policy, and the variables of interest. So the research, I mean there's so many aspects to the research in this area. There's a big and growing literature on different kinds of policies, and it's really been one of the highlights of my career to participate in that literature and to be supportive of members of Congress as they design their legislation, and hopefully achieve some of our climate objectives. Another application of this research is as an input to other kinds of analysis. And so this is where we get into some of the financial sector outcomes. If you simulate for example a transition future, where maybe there's an ambitious climate policy, and maybe you layer onto that some technology assumptions that perhaps add economic stress to the, to the transition policy. You get these projected variables that that you can use in the analysis of financial outcomes that maybe inform, well what would happen to the balance sheets of a particular financial institution, or what might happen through the financial system, if you follow those impacts from one part of the financial system to another. And that's something that's kind of a growing field within my space in terms of looking at climate scenario analysis and some of the research and policy projections that are capable for some of these models are important inputs to that kind of work. So finally, so this brings me to Pete's presentation and his work is, I think, solidly in that class of models, he's going to talk about the G cubed model that can be used for some of these policy analysis, analysis and also as an input potentially to some kinds of financial scenario analysis. There's a separate class of models that we haven't talked about too much which are very specialized and they focus on a particular sector, maybe it's land use or transportation or just the power sector, or the energy sector maybe more broadly but not necessarily the macro economy. And these are really important to analyze, you know, specialized policies, for example, you know you want to know the outcomes of a refrigerator energy efficiency standard. Well you're not going to use a macro model for that but you're going to need a specialized model, and researchers contribute to that literature, as well as the federal agencies that might be involved in those kinds of policymaking. So, now how do we're going to bring that back to the topic of the macro models for today. So I think what you're going to hear from Lars and Pete is going to is going to help us think through how long, how long run economic growth can can depend on how we address the climate challenge, like these models and their insights can be inputs to some of the factors that go into macro modeling. And from Lars we can think through how how we address the climate challenge can also influence long run factors, especially given the kinds of uncertainties that we face. And in addition, I think the kinds of model that Pete's going to talk about can help us understand the range of potential transition outcomes. We know that long run growth is going to depend on on our transition policies. And because those have great uncertainty, it helps us to evaluate a range of potential scenarios. And, and these are the illustrative of the kind of tools that can do that. So, with that I'm going to turn to Lars and you can load up his slides. Thank you Adele. I'm about to do something load up the slides. Oh, there they are. Good. Thanks. I appreciate this opportunity to to speak and talk about research topic that I've been very fascinated with for me for several years now, and it's, and it's carried over to a variety of contexts including in the last few years climate economics. I'm just, I'm going to be talking across multiple papers and from the climate side, come economic side collaborators. Mike Bardette and Buzz Brock. I also, there's inputs from this from decision theory and and decision theory collaborators from multiple sources, which are also listed here decision theories a way to think about what prudent decision making is and complicated uncertain environments and ways that are possible are prudent. I'm letting to use the word rational because it's loaded, especially when you get very complicated environments but anyway, it's, it's, it's that those tools turn out to be very very valuable. So let me quote from a person in the audience here in a recent paper. The consequence of many of the complex risk associated with the climate change cannot however currently be quantified. He's unquantified poorly understood and awfully deeply uncertain risk can and should be included in economic evaluations and decision making processes. Certainly we're very sympathetic with that goal where we're trying to address it it's a it's a it's certainly a hard on a hard goal and but but I think there's things that can be said about it. So let me, this is something that keeps me up at night. I quote from Hayek's Nobel address. That is, there's this pretense of knowledge if you that that shows up often in policy making circles and that can be harmful. And the role of you know this is supposed to be a committee that's put together by the National Academy of Sciences as scientists were supposed to be producing credible scientific inputs into into the process. And that means that we have to figure out what the credibility is of the inputs we're putting into it. We can do if we pretend we know everything as you know it's maybe that can impact policy but it's not but it's not true to the scientific underpinnings. So there's a real challenge here like for quantitative policy analysis how should we acknowledge the limits to our understanding. So, I'm going to be talking about uncertainty, but I'm going to be talking about our broader terms I'm going to use a different language than the other Jim use for reasons that will become evidence subsequently. I'm going to reserve risk for something very special. And I'm going to call uncertainty more general uncertainty more broader concepts which I'll say more about as we go forward. In policy making setup there's some important trade offs economists always like to think of terms of trade offs. There's important trade offs which we have to confront here. When you have multiple models are on the table. How do we assign, how do we assign weight to best guesses versus potentially bad outcomes. If we just go for best guesses we missed the possibility of something bad happening. If you focus only on what's bad, then you don't get up in the morning you just give up. So you know that there's this trade off about how you kind of worry about both of these things and that's very important. The other one is one that's been studied a little bit less formally within this climate economics literature and that although it's discussed all the time react now or do we wait until we learn more. One concern when we talk about uncertainty is it's just going to lead to inaction, but there's nothing about decision theory under uncertainty it tells you that you should not not act possibilities of bad outcomes are enough to make you want to act now. You don't have to be 100% sure of them. And so there's an interesting trade off though do we act now or do we wait until we learn more and some of the calculations I'm going to be talking about are meant to get at this in a stylized way. So what's a challenge here so in a very general setting. There's limits to our understanding about the potential impact of climate change. And I'll be talking about three different sources today and some in some illustrative computations. One is from the geoscientific part of it CO2 emissions today impact future outcomes. There's the economic component climate change in the future alters opportunities and social well being. And the third piece of this is research and development invested today may eventually lead to economically viable clean technologies. Now all of these are uncertain. And so one of the questions we want to face is, you know, which ones really matter for designing different policy questions. And part of what you want are ways to think about that. That challenge and and in fact if we can isolate the ones that are most important maybe there's ways we can actually close knowledge gaps and along those dimensions. Now, the literature this uncertainty literature I've been talking about is really aims for a very broad perspective on uncertainty. It's kind of useful to put into three categories. Right. I'm going to use a term risk for something special. It's kind of what we teach in our micro economics classes typically is unknown out. It's what's in rash expectations models and in terms of what we give economic agents risk is unknown outcomes but known probabilities coin flips to roll the dice and the like. So like when we build dynamic models we put shocks in those models with distributions that that that's going to be a source of risk inside those models. I'm going to use the term ambiguity to mean, I've got multiple models on the table. I always like to think, you know, beware the person of one model, especially when there's situations of uncertainty, we want to look across models, how much weight do we assign to each of these. From a standpoint of statistician there's a form of prior uncertainty. We can't resolve this just from historical evidence typically it requires subjective inputs. How do we confront prior uncertainty. There's a, there's a, you know, a wonderful contribution back in the 50s by, by, you know, by Jimmy Savage, what is that the University of Chicago that produces axiomatic system that just makes that subjective probabilities that are part of the decision making process. But even savage in his own writing acknowledged the fact that these subjective inputs we may only know with very crude amounts of uncertainty. Now, in a lot of an econometric situations where we teach we imagine sample size and data just swamps priors we don't have to think about this within the climate region we know I think that is just not not the correct perspective. One is potential mis-specification unknown ways in which the model might give flawed probabilistic predictions. So in this, and this context does it for statistical language there's kind of likelihood uncertainty. I'm going to write down this model I'm going to write down this parameterized family of likelihoods, or, or some such short sort. We use models as simple. We know their simplifications. We know that there are a lot of some dimensions necessarily wrong. The simplifications make them transparent and understandable. But then how do we use them in sensible ways. And the three components of uncertainty I often think of this one as the one that is hardest to wrestle with, but, but, but perhaps in many respects the most important one. So certain there is going to include form, you know, formulations that are explicitly dynamic recursive can be implemented with dynamic programming type methods, because if the methods that come out aren't tractable they're just not going to be very valuable to us. And the way I think about this is it suggests better ways for doing uncertainty quantification for dynamic economic models used for private sector planning and government policy assessment. So we can go through and characterize the uncertainties and their magnitudes, but the end of the day, not all certainties are going to matter. And so what you want is a framework allows you to figure out which uncertainties are really critical to the question hand, and which ones don't matter very much. So basically what we're aiming for is these tractable methods and we want to isolate the impact of uncertainty on climate policy and outcomes, assess the assess it and then isolate the form that really matter matters. So roughly speaking, I would love to give you all the mathematics behind this, but I think in 20 minutes I'd be a bit extreme. It's a way to navigate this uncertainty. We want to allow these models there's ambiguity there's going to be in practice mis-specified. We want to use them in sensible ways. And so what we're going to set up for ourselves is basically a very extensive sensitivity analysis sensitivity over quote, quote, these ambiguity or these prior sensitivity over likelihoods and the like. We still want to use the tools of probability and statistics to limit the type of an amounts of uncertainty. I'll put the following question on the table, which is not usually talked about on policy settings, but anytime you open the door uncertainty, it's I kind of has to be there. Aversion. How much do you dislike the uncertainty about probabilities over future events. You can't really address a prudent policy problem without, without putting this on the table. And we as scientific researchers, the thing that we can do is trace through the impacts of these different aversions to get to pause the outcomes. It's not our job to tell society exactly how a verse they should be. We can visit our job to explain what the consequences are of those aversions. And finally the implementation is you got to target these components which are most adverse that comes out of the whole kind of computational solutions of the models. And, and, and part of an outcome of this is I is is a robust adjust it or uncertainty adjusted probability measures. Basically, we have uncertainty about the probabilities we're going to produce these uncertainty adjusted probability measures that are going to be most that that they're pertinent for valuation, along with the design of robust decision rules. Along with working out these decision rule, we're going to back out these probabilities, these certainly adjusted probabilities are the ones that are pertinent for valuation. So, so they help us think about uncertainty quantification to address two questions. How much uncertainty aversion should we impose there we can do trace through this. This sensitivity I talked about, and which one matters the most. So, social valuation, there's in the, there's too often discussions of in kind of cost benefit analysis looking at local valuation have these discussions what should the, what should the discount rate be. If you're in a stochastic environment, you need to do stochastic discounting, in order to do these margin valuations and the same is true if you want to even compute off all Pelluvian taxes. Asset pricing theory for a long time has insights and how to address this, although it's, you know, the other was designed for a different problem it was a design for market valuation. Now the valuation of assets like financial, physical, human and organizational environmental capital. But each of these assets is a prospective sequence of net payoffs or investments. We're going to apply the same tools to social instead of market valuations, because we think about the COT, you know, if we're going to emit carbon in the atmosphere today there's going to be this social adverse cash flow that comes out. If we invest in new green technologies today, there's going to be uncertain payoffs of that in the future. So we can use the same type of constructs, but to think about social valuation, instead of market valuation, and extend them to say to include these uncertainty components which I talked about, and that's the aim. So if you want to do social valuation under this these types of a versions. Yes, you want to do discounted expected values of social cash flows as is typical, but the expectations are constructed from these kind of minimizing probabilities which we compute as part of the solution. Again we want to do stochastic discounting but under a probability measure inferred as part of the solution. There's some, a very illustrative economy with some calculations. This is going to be a highly stylized aggregate economy. I've been doing other research, which I would love to talk about to what time doesn't permit on the Brazilian Amazon this is that that that's a spatial dynamic models with data on up to up to 1000 different locations in the Amazon and your productivity data on agriculture and carbon absorption alike. But I'm going to give you this more aggregate example today. So there's many calls for immediate kind of policy implementation to say limits or understanding of the timing and magnitude. How do we confront this. So we're going to put imagine this policy maker in situation which at the future going to know more about the damage severity. We're uncertain about it now substantially so as we start damaging substantially the environment, which things are going to get get revealed to us or or. But at that point in time it may be very costly to act. So this is when we trade off between acting now versus waiting. We're going to have research and development, and this can hasten the uncertain discovery of green technologies as part of a part of the policy toolkit. And then finally we have this risk component and this is typical and in so called stochastic equilibrium models that typically focus on risk, and we're going to have both kind of so called Brownian motion or kind of local Gaussian local locally normally distributed shocks, as well as Poisson type jump risk or much more dramatic shocks that can also occur. So, so I will just give you a quick schematic of the model. It's kind of a fairly standard model with with with climate change put on top of it we're going to have capital it's going to produce output output then for that can be divided into investment or consumption of missions then alter the output that comes in. Without climate change the more you admit then you have to do with more energy input into production and that's going to increase the output. So now we add climate and emissions also do damage the also impact the climate and that damages the economic well being. So this is the externality so economies, if I took to just took the model on the right hand side. Oh, I could imagine the invisible hand works just fine and there's a no intervention is necessary. Now I've got the externality, and now this is where the market failure shows up. Finally, we want to add in. Oops, the possibility of this kind of R&D investment some of the outputs use for R&D investment and down the road that could generate economically viable clean technologies, which are which are then improve economic well being. We've got these different multiple channels that we're uncertain about, and we want to kind of put them on the table at once. So this will be emissions impact on climate climate impact on damages returns to investment in new green technologies as three sources. So what I'm going to show you is we're hungry for better inputs, and we're hungry for richer inputs we just have to do this in an attractable way. What I'm putting here is in the red histogram is with baseline kind of climate sensitivities. But if you look at pulse experiments from climate models, you roughly speaking you to get a peak impact from emissions after about 10 years and then things start flattening out this is kind of where things flatten out, looking across models and so this this looks over about 144 of the models. In terms of mapping, and then and then the red histogram just gives you some initial statement about about that dispersion across models. So the histogram priors, if I assume all models are equally likely or equally plausible, you may not want to be convinced that that's that that's the relevance subjective prior to put into this and you want to may allow for tilting or okay. And so once you have ambiguity over that waiting, there's going to be this notion of shifting the other distribution. And since we're looking for a theory of caution, you can shift things to the right so that's going to take this blue, the red histogram and make it the blue one. That brings up a little bit more to the right. And I've got a case where less aversion and more aversion the more aversion I have the more I'm going to shift that distribution. So, so that minimizing distribution use for valuation is given by is given by the blue histogram. That's a cautious adjustment because we really don't know how to wait these 140 for different models. Next one I'm going to put a stylized model of damages. So the temperature anomaly at point five. As in carbon budgeting, we're not going to fall off a cliff here. Instead, their damages are going to be realized. Now I'm showing you the whole trajectory of possible damages from the upper bound which is the more typical North house numbers to more extreme ones that are more like Weissman style numbers, but we're going to be unsure about that curvature. Once we cross this threshold of there's going to be some plus on event that gets triggered, then, then we're going to know about that curvature was to start damaging the environment much more. So, so this is the counterpart to a kind of combination tipping point between the kind of climate and economic systems, so fully highly stylized description of the uncertainty where again we're happy to put in Richard type specifications. So we don't know where that anomaly is going to be a 1.5, or it could be a two, it could start at two sort of uncertain and certainly about where that takes place and we've got this kind of local personal intensity that increases with temperature that that's meant to capture that uncertainty. So, so this, so there isn't a big event in this model this realization of this damages and there's and the other big events going to be a realization of a new green technology based on R&D investment in part. So, we don't know what the rich, how much damage curvature there is that could be good news that could be bad news once we get to once he crosses threshold. The red line is the case we're treating all these different we've got a, you know, like 20 different curvature curves retreat them all equally likely. And then the blue one is once we start thinking about well I'm not so sure about that. I'm not sure about that waiting under less a version you see the modest adjustment under more a version that bit then that becomes more extreme. There's a bigger push to the right as to be expected. And then we also have calculations I don't have time to show you about what happens. Once that gets realized that it could be good news and then and so versus bad news so there's kind of heterogeneous responses once the information is realized. So we've got a few of the outcomes here. We've got two assets, social assets here. One is the social value of the stock of R&D. So we're investing in R&D every period or instance in this particular model. There's going to be. So, so the there's going to be a stock of R&D, and that implicitly becomes valued. So it shows you the adjustments that this is in a log scale the adjustments for for the for the social value of R&D neutrality says that I'm going to basically return to the baseline I'm just going to ignore. I'm just going to go with the baseline probabilities. And then if I have less, less a version, I'm going to move it up some and then more a version, even more. So in the log scales we're thinking about kind of proportional adjustments here, but these look quite modest but I'm going to, but, but I'm going to get back to that in a moment. Now I can put in these three sources of uncertainty. Climate damage and technology for the, for this calculation under the modest aversion, really what matters here is the, what matters the most is clearly the technology aversion. And the two are much less consequential. Now, these should be stochastic simulations we have to figure out the right way to do them report them. We know how to compute them, but yet, but here I'm running it under a baseline that two things happen, there's no to be closed down the shocks along this computer trajectory, not in the solution. And this is also conditioned on the, on the technology event the new technology that event not being realized, which is also there's some probability that happens as well so we have probabilistic adjustments for all that. Now these adjustments as what's the big deal, you know this seems like it's nothing. What turns out that even these seemingly small amounts of changes in the social value of R&D have a big impact on the, on the robust R&D policy. So now we're, as a, as a percent of GDP, if you compare the green line under neutrality to this blue line, you see a substantial increase in the amount of R&D you're doing, based on the uncertainty inputs. Yeah, I'm going into it. And again was driving all this is the, is the, is the technology uncertainty process special to the to the inputs we give into it but this helps us think about, you know, which one really matters for this calculation. So we'll look at the kind of social cost of global warming. Again, these are a log scale. These adjustments, you know, again look somewhat modest in terms of the valuation. But, but again the thing that matters here is the technology aversion. So, so the big deal so far is in the R&D investments and that they're very sensitive to the uncertainty. Finally, carbon emissions. So here the implied carbon emissions and, you know, on trajectories and again these are conditioned on there do not realizing the new green technology which of course would, you know, drop down to zero. And again, just kind of, it's the, so the, I'm so under more aversion then you're going to admit you're going to admit less carbon in the atmosphere. And again, kind of technology aversion is kind of the big deal on these particular calculations. So, so basically putting technological considerations on the table are very important to these calculations, and the uncertainty inputs into those also matter in a big way. So, let me just close by saying, on certain matters for policy tools like the social cost of global warming the social investment and green technology and development research and development. And understanding the sources broadly conceived by use by the private sector and governments will make economic policy all the more effective in our views. There's a variety of other things I'm happy to talk about, but times up about computational challenges, how we can push this, this other simple model and the ones with much more heterogeneity across, across space, which we've kind of done in there in our Amazon process that project and like but that'll be well, reserve those conversations for some of their time. Thank you. Thanks so much Lars. We're going to go straight to Pete so write down your questions for Lars and we'll start the Q&A. After we've looked under the hood of the G cubed model. Okay, thanks very much Adele. And thank you everyone for being here and for the privilege of letting me fill you in on some work that I really enjoy talking a lot about. We're talking about a lot. Okay, it's afternoon. So, my topic is going to be talking about multi sector climate energy models. These are generally from the class of computational general equilibrium models. I'm going to use examples from my own work, including the G cube model on which Adele and work McKibben are close collaborators. And also my work with Dale Jorgensen, Munsing Ho and Dick Gettle on the iGen model, which is a model of the United States. But having said that many of the things that I'm going to point out today are true of the broader class of models. This is just to make things a little bit more concrete. I'm also going to draw somewhat on a study that the science advisory board conducted for the EPA on economy wide modeling for air regulations. And this is a very detailed scientific review of the landscape of computational general equilibrium models. And it has a lot of helpful insights for today's task as well. Okay, so the general plan is that I'm going to talk a little bit about the structure of the models, so that people who are not familiar with them will know how they work. I really recommend the white paper that has come up several times in the conversation today. It is excellent, very well written, and this will just fill you in a little bit on the details of how the CGE models that are discussed their work. I'll show you some illustrative results. And then I will do my own version of Lars is what keeps him awake at night speech and talk a little bit about the parameter uncertainty and how when we think about these models we have to be modest about exactly how precise they are. And then finally I'll conclude with some thoughts on challenges and research needs. So the way these models are put together. The hallmark of them is that they divide the economy up into lots and lots of individual sectors or agents. G cubed, which is the model that's shown up here is an international model so it breaks the world up into a bunch of regions. Each region is broken up into a bunch of actors, producers household sectors, the government. And then each of the produce producing sides of the economy broken up into sub sectors. G cubed has about 20 the ones that are highlighted or connected to the energy side of the economy. When you look at the broader landscape of CGE models, they all put their emphasis and their detail in different places so there are some models and sage which is an EPA model is an example of this which have much more detail on geography within the United States but aren't international models. There are other models which have much more detail on the energy sector, but less detail on parts outside the energy sector. So it's a big landscape. But, and, and I think the thing that the Science Advisory Board report for EPA landed on that I'm going to repeat here is there is no single model for all purposes, different models have advantages for different uses. All right, when we think about what goes on inside one of these models. Oh, so, so for people are online. I'm sorry I'm going to do this to you but I'm going to use a laser pointer briefly in this to point some stuff out to people in the room. So on the left side of the diagram here is a list of markets so the models divide the predicting producing side of the economy up into a bunch of individual individual markets, each represented by a row in the table. The top part of the table are the markets for which there are private sector producers, and there are 20 sectors in G Cube so they're 20 rows one for each product. The things at the bottom for people who don't do this all the time are primary factors, labor and capital, those are not produced by individual firms instead they're supplied by households. Then there's one column for each agent in the economy, somebody who buys things so they're 20 industries in G cubed, they all buy things intermediate goods, and then their final demand sectors consumption for the households, the investment sector government spending exports, and then imports. The key characteristic of all of these CGE models is that there are the agents interact in all these markets. And the models task is to find a set of prices that cause supply and demand to eat to balance in each of the markets. Okay. Getting down into more detail each of the agents in turn is represented by mathematical model of its behavior. The top part, the sort of overarching piece of the model is the definition for what objective the agent is trying to satisfy so it could be minimizing costs. It could be maximizing utility depending on whether you're looking at a household or a producing sector. And then the agent buys a lot of inputs and they capital labor and also individual products electricity natural gas transportation durables. And the hallmarks of these models are that there are they represent how agents are willing to substitute one input for another. So those three green circles in there represent substitution elasticities, which are estimated from historical data and try to capture how each of these agents is willing to trade off goods in the bundle, one against the other. So we're trying to pick up how they react, particularly to price changes. The parameters the green parameters are estimated from historical data and then also from historical data we get inputs. We get information on what people's patterns of purchases actually are right so what fraction of people's budget do they spend on energy how much do they spend on agricultural products, durables and so on. These models are also where climate comes into the picture. So, so these models that I'm talking about here have been used for studying lots of different kinds of environmental and climate policies. And when we want to introduce climate into them. One of the ways that it's done is by introducing reductions in productivity in the industry so if we think that agriculture will be harmed by lower rainfall. We can reduce the productivity of the agricultural sector by changing parameters within this model. We can also we've also these models have also been used for looking at other kinds of environmental impacts so looking at changes in conventional air pollution which changes mortality rates which changes population and effective labor force. We can also look at how morbidity changes labor force availability as well. And then finally, there's less work that has been done on this but some of these models have also had environmental amenities introduced into the utility function for the agent so that improving the economy or improving the environment is actually valued by the individual agents and increases their benefits. Okay, a critically important thing for thinking about climate is the behavioral links between periods. So, so the model so, so many but not all CG models that are used for studying climate policy have forward looking agents, forward looking firms that think about how their investment decisions are going to affect their future cash flows for looking households, we think about how their consumption and savings decisions today are going to affect their future. Their future utility their future ability to consume or climate modeling this is extremely important. These, these particular decisions investment and saving our prima facie inter temporal decisions. So it's important to have that baked into the model, but also because a lot of what we're interested in here is thinking about how agents react to anticipated changes in climate conditions or anticipated policy changes. And then finally we're also going to be interested over the course of thinking about macroeconomics and climate and thinking about agents respond to policy risks. So this is something that is involved in what Lars was talking about but I don't think he mentioned explicitly but when you think about putting incentives in place like the IRA and important thing to keep in mind is whether agents believe you're going to stick around and actually continue to provide the incentives that the legislation that the legislation puts forward because if they don't. If it's something like a production tax credit and people don't believe it's going to be there they're not going to invest, because they're looking forward and they think the policy may not last. Okay, in these models, I want to circle back to the main topic of the workshop today and talk about macroeconomics. The GDP growth. So a lot of models that are used Adele mentioned that are used for studying, for example, the energy sector take GDP growth to be exogenous it's just put in that by assumption. The CGE models build GDP up from components, either on the income side by adding up what the returns are to labor and capital, we're on the expenditure side by adding up with the value is of what people spend on final demand. So we build these up from the individual behavior, the components of the model, and the things that typically drive GDP growth there are things that we've talked about a lot today, but just to emphasize how they interact into them with the model is of course growth is very important. That's usually exogenous. Although not always capital formation is is is critically important that's usually endogenous in the model productivity growth. The CGE models treated as exogenous and base productivity assumptions, sometimes at a very granular level across industries on historical data. Other models have endogenous sub models that determine productivity growth as a rate of as an outcome of growth that are on R&D, again, touching back on what Lars was talking about. In the short term, these models can also have GDP growth, boosted or hindered by terms of trade effects or international models, or by employment or unemployment in the short run which is unusual in in G models but I will show you how it works in a minute. So productivity is also important and thinking about GDP consequences so factor mobility is jargon for saying that in the models, labor, for example, may or may not be able to migrate, and where migrate can mean between regions, but it can also be related to occupations and industries. If labor is not very mobile, if it's fixed in industries occupations and locations, then the consequences of policies can often have, they can often have very long lasting costs because people can't move to new industries very easily. Physical capital can also be a mobile, usually it is pretty immobile, it can be specific to regions or specific to sectors. So this is really important because it's very hard to uproot physical capital equipment machinery and so on from one industry and put it into another industry. Except for a few things like it there's just huge losses in the value of the capital when you try to do that. So these are important things to think about it when you're trying to model the short run impacts on GDP of policies. So I'm going to show you just to give you a concrete feel for how the model works. A simple applicant or an example application. This is from an EPA workshop on transitional labor dynamics held a few years ago. And the thing that I'm going to show you is what happens in G cubed when we simulate a decline in productivity in durable manufacturing. And so what happens, what's going to happen in the simulation to illustrate several points is that I'm going to show you what the results are from reducing the level of productivity and durable good product production five years in the future. So agents are going to know in years, the productivity is going to be the same through year five, and then it's going to go down by 1% and be permanently lower. And the reason that this is of interest today is that this is both how regulations are often simulated in these models by saying, look, we're going to impose a regulation on industry it's going to increase their cost, reduce their productivity. So you could think of this as anticipating a direct regulation rather than a price regulation, but it could also be thinking about a future climate impact that's going to reduce productivity in the industry. So, so I'm going to just skip over this mostly, but I'm going to just mention that most of what I said so far applies to general equilibrium models as a class, what I'm about to show you now really applies specifically to GQ because it has a bunch of features that are unusual for general equilibrium models it was built initially to try to glue together key features of macroeconomics with general equilibrium and house. It has a bunch of unusual features so one of the features that it has which is not universal in general equilibrium models is that it has some agents who have foresight and some that do not. And there are a bunch of reasons for that but it helps us do a better job of simulating the rate at which people adapt to actual policies in the world. It has sector specific capital stock so it's hard to turn a coal plant into a solar farm, for example, which not all believe it or not, not all models have that feature. It's a very detailed treatment of financial markets so we treat we track equity in each sector government debt, international debt foreign currencies money supply central bank policies, and the risk premium and all the assets, which has allowed us in other studies to examine questions like what happens if people suddenly become more anxious about the rate of return on household equity. Right, so if you're worried about people's owns losing value because they are in flood prone regions. Also nominal wages adjust slowly so that this model is not fully in equilibrium at all times the labor market can be out of equilibrium in the short run, and it has full bilateral trade. Which is very important for thinking about climate policy because the US is not a small island economy and climate is a global problem so what goes on in the policy world in other countries matters in the US. Okay, so what you see in this. So this is a tiny slice of the results that you see after this study so this is showing you some in some outputs of the model in your seven of the simulation so this is two years after the decline in productivity so this is a snapshot in time. The graph that's on the left shows you the impact on output by the 20 industries in GQ and the durable sector is affected the most because we reduced its productivity and minutes price go up people buy less of it it harms the US terms of trade and the durable is an important export. It also reduces demand by a bunch of things that are upstream and downstream of the durable market, including for example the services sector. The services sector doesn't see very much output decline, but the services sectors where most people in the US economy work. The graph on the right hand side shows you what happens to employment. It's a little unusual it's a waterfall graph so so the height of each individual bar shows the change in employment in that industry. But each bar is stacked at the end of the bar that was to its left so you can read it as a cumulative reduction as you go across the page. So the value of their shock reduces employment and durables, it reduces employment and services. These are measured as proportions of baseline employment so the index the size of these things are similar numbers of people. So there are a lot of people who are lose their jobs in the services sector even though the services sector isn't affected very much and it's because that's where most people work. The service sector which is where construction is in our model is also hit by this as well that construct investment and because durable goods are part of the investment good. One of the reasons I wanted to put this up here is that we are here to talk about macro but a thing that I think we all have to always remember is that macro is not where the action is for climate. Where the action is with the individual sectors, it's with individual places. And so these, so these things are going to average out to small impacts on GDP, but they're big impacts for people who work in the durables and services sector. Okay. Okay, this is what happens. If you look over time the model is solved at an annual frequency and so the thing on the left hand side shows the impact of this particular policy on total employment. So total employment is not very much affected by anticipation of this shock there's a little blip upward as people in the model produce a little extra durable goods just before the productivity in the sector goes down right this would be producing a little bit in anticipation of an onerous new regulation. So total employment goes down a lot for the reasons that we just talked about through year seven, and then it result, then it returns as the wage adjusts and by your 15. We're back to approximately full employment, although what's happened is that the mix of where people work has changed right people have moved from durables into services. And I remember that even though there's no net employment change there, there've been a bunch of costs of people having potential, having to change sectors potentially having to change careers, and maybe locations in the economy, geographic locations. Okay, what does this do to GDP and its components. The thing on the left shows GDP. Well, it shows the family left shows consumption investment in that exports. It doesn't show government spending which is an important part of GDP but here itself constant in construction of the simulation. So what you can see is that this particular policy causes investment to dip. That's the dash line because we just made investment goods a lot more expensive. And it eventually stays level ish, and then it eventually starts to dip and the reason it starts to dip is because the decline in consumption reduces future capital stock, which reduces future income and consumption has to fall as a result. Also this determined this deteriorates us terms of trade so it actually pushes net exports up as a result of that. The impact on real GDP is shown on the right. There's a big overshooting the overshooting in this model comes about because we allow for there to be unemployment without unemployment you'd see a smooth trajectory going from up top over to the long term outcome on the right hand side. All right. I'm almost done here but I want to connect back to something that's really important and also that came up. I mentioned what came up. Lars brought it up and I mentioned that this is also my fear. It's really important to remember that these models are shot through with uncertainty, shot through with uncertainty. Even when you try to build them really, really carefully using all the possible data and the best techniques in econometrics so these are these are parameter estimates from another model the I gem model I mentioned earlier which is a 35 sector model the US. This is just part of the parameters in the model, there is a part that has to do with modeling household behavior, household consumption behaviors and a bunch of parameters over on the left. And you can see there's a value of parameter, that's the, or there's a column for parameter values minus point five to two, and then a standard error point 0028. All of your graduate students would be so excited right like that. That's super significant right. So let's get this out. Well it turns out first of all that's a, in this context that's a dumb question to ask because this is the share of something that people buy in the budget and there's zero chance of it being non positive. But the standard error is actually bigger than you think. So it's the tip of an iceberg of the covariance matrix between the parameters in the model which is shown on the right. So what we can do is we can take those covariance matrices and we can push them all the way through the model to calculate the uncertainty in the models results to calculate standard errors for the results. These two graphs are diagrams from a study that we did looking at how uncertain the models results are as a result of parameter uncertainty, and this is setting aside all kinds of other uncertainty uncertainty about the exogenous variables. The residual uncertainty and the equations that because they don't fit very well than the specification Lars mentioned climate uncertainty. This is just the parameters the easy piece. So on the left shows you how much uncertainty there is in the long term level of output in different industries. And in case you can't read it the bottom, these are percentages of the base case value. So the axis on the bottom says 40. That's 40%. That means the standard error, just from the parameter estimates in this model for our ability to predict the level of crude oil output in the United States. So this is 40% of what we think that value is right if you imagine just drawing this as an error bar, that is huge. Most of it comes from uncertainty on the production side of the model we did a decomposition into how production and consumption uncertainty manifests itself in different variables, the household uncertainties are in green and the production uncertainties are in yellow. On the right hand side are uncertainties in the macro variables. So consumption the our ability to predict the level of consumption. The standard error on that from just the parameter uncertainty is 4% of the baseline value value our ability to predict carbon emissions is worse right the uncertainty and that is more than about 13% of the level of output so all these trajectories that we're talking about for one trajectory I understand administrative and politically why, why the answer has to be one trajectory but the bands are big. The bands are big on these things. Nonetheless, there is a little glimmer of hopeful light at the end of the tunnel. So the thing that we were really concerned about in this particular analysis was, do these uncertainties mean that we cannot say for sure whether climate policies actually help. And it turns out the answer to that is, I'm going to have to phrase this carefully just to make sure it isn't misunderstood. The answer to that is, despite having all of these uncertainties when you track them all the way through the model, you can still say with authority that climate policy would work. And so climate policy, it's carbon tax with a particular recycling assumption run through the model. And so these are, these are policy derivatives it's how much the level of the variable changes as a result of climate and you can see the error bars on there. And so, so everyone here will probably be not surprised but hopefully a little relieved to know that we can say with authority that climate policy is bad for the coal industry, right. There is no chance that the error bar does not include zero coal is going down. Okay, now I think as Jim pointed out it already has, you know, but it's good that the mock and reproduce that. That's not true of all things agriculture this could be bad news a carbon tax or not bad news. It depends on how people substitute to or from agriculture as energy prices change on the carbon tax. On the macro side we can say definitively that carbon will go down in spite of all these uncertainties, and that the capital stock will actually go up this was a capital stock recycling experiment. Okay, so I am now over time but I am at the conclusion. And so this is just, I'll just go over this quickly. Part of my brief was to talk about challenges and research needs. So for modeling transition policies, one of the difficulties is that we need a whole lot of energy sector detail, and it's difficult to get enough data on all of the pieces that you need of the model to have as much energy sector detail as we'd like on modeling climate impacts we need a high degree of geographic detail, because that's where climate impacts really show up they affect different people in different locations. So we would ideally like to have a lot of geographic detail. To skip to the bottom. One of the conclusions that came out of this API SA B study is that it is going to be necessary to link models and to develop better protocols for doing that so we know exactly what happens if we try to link. To a more geographically detailed model of climate impacts to a more aggregate general equilibrium or macro model of the US economy. And then finally, we need to not oversell this stuff. So these I put this percentages of baseline values up there. Just as a reality check so when we're thinking about some of these changes that might occur to individual industries. It's important to think are they going to be in the noise or bigger than the noise right and how we can do this modeling. So that is the end and I will stop there. Thanks. Thank you so much Pete. So we're going to open it up to questions. And while I load up your, the chat, I'm going to see. Or no, here, sorry. That's not what I want. Let's see. I think. Okay, so one real quick question for you Pete is on the on the uncertainties and in your parameters. Tell us a little bit about where you think those uncertainties come from. And one of the things that I hear about, you know, kind of skepticism about projections from from the CGE models is, you know, what we really don't know is what new technologies are going to emerge and what their costs are going to be and how and when they're going to penetrate at scale. And I'm just curious, like how do you respond to those kinds of questions and over what timeframe you think technology uncertainties become important. Thanks. That's, that's a great set of questions. I think there are at least two questions there. And the first one is about where do the uncertainties come from that I mentioned and so those, those are just the, they just come from the covariance matrix in the parameters in the model so they don't have anything to do with technology, they just have to do with the fact that we have, you know, in the US we have data since basically World War two, which is not a super long data set we have a lot of parameters that we want to do. And in some cases we don't have a lot of variation the prices that went in, like for many years, energy prices were pretty stable in the United States so we have limited variation, and we just don't have precise estimates of a lot of those parameters and and part of the reason we did the study is that you might think well there's a whole bunch of imprecision in different sectors maybe it all averages out. And the answer is no we're not that lucky it doesn't all average out right some uncertainties matter a lot, depending on what you're looking at, and you can't get away from them by just hoping that they average out. Now, then the, the second question you asked is about technology. And so in the models that I've worked on and, and most familiar with, we put in productivity growth based on estimates from the historical record so there's, you know, a huge literature on people estimating productivity. Careers have been devoted to that and so we typically take estimates from the historical literature, people often say well, this time is different. And so, if, if you think it might be different, and that productivity growth is not really driven by historical record, you can do scenario analysis and people often do that sensitivity analysis and look at what happens. So, did you want me to say something or not. Yeah, if you want to address the question sure. I can talk very briefly about the technology side of things so for us as a very, very specific type of technological advance we're looking at and that's a development of a new green technology that dramatically changes the whole production technology. And so for us, it's a rather abrupt discovery with lots of R&D going into it. There's other ways to have modeled it and and which would be much more smoother trajectories and and and those would, it's certainly a place where we had it. We're looking forward to doing further work, but we have to input into that things like these probabilistic specifications about the success of the technology and then we introduce the possibility that's misspecified and see what the consequences of that are. Okay, so I'm going to start on the, I think, Heather and then Galena. Okay, great. And so thank you what a great set of presentations thank you so much. So I have two questions. And I think they're both for Pete but I'd love to hear you both on them. Pete you made a, you made a comment near the end that the action in the transition is really about sectors, places, jobs. I think one of the things that I am, you know, we struggle with in this in all of this work is that because our tools are designed for a different kind of macroeconomic and buy, you know, a different set of macroeconomic questions, where we're not able to use them to actually unpack what policymakers really I think need to know to do both fiscal and monetary policy. Right, if you have inflation that's caused by supply chain challenges due to a clean energy transition that might, you might act in a different way than if it's caused by something else, for example. And so I just wanted to hear, like, how do we think about that. I mean I'm coming at this from the policy perspective but I'm curious as modelers how you think about that question and how we could do better to unpack that more. Are there ways to show that better so I just I wanted you to say a little bit more about that. My second question is one of the things that we've struggled with in the Biden administration is that the policies that we're doing aren't what modelers model. And so when you say that that it could be that this policy is effective. I'm like well yeah but we're not actually doing a carbon tax so I'm just curious. So you all are thinking about incorporating the much more complicated set of policies and it's not just that there's more of them. They're more complicated, but they are being done in concert. And I think that is something that we have been really thinking about and I'm curious on your thoughts on that as well that it's not just one policy but we are doing these suites of policies sector by sector. So those are my two big questions. Thank you. Great. So, so I'll, I'll try to tackle them in reverse order so. So the, the, the very fine grain nature of the regulations and policies are the reason at the end I concluded by saying that that there's always going to be an important need for more sectoral detail and more technological detail in the model because that's going to have the hooks for the policy to fit in. So, so more detail is always useful in the short term though if you're working with a model which doesn't have as much detail as you want. So we'll have lots of back. So, so typically what people have done in the past for energy sector policies like that is to work with energy engineers engineering studies from the energy sector to look at what would happen to energy companies under some policy like how fast new technologies would be adopted. And then we can map those into inputs that could be used in a reasonably detailed CGE model. I just want to give a shout out to the EPA folks who are behind the sage model because EPA has a very, a very nicely done generally cool remodeled which is now for the first time being used for regulation regulatory purposes and they've had to grapple with a lot of those issues in building month. So you never have as much detail as you'd like but there are ways to approximate it. The, the first question I think of as, as a, as a reason I guess why I feel relieved to be an academic and to not be one of the people in the hot seat who have to respond to the executive order. So, so the executive order says we want to know what the risks of climate change are for the macroeconomy and really if we go out in the hall will all say well the risks are to the microeconomy right it's the people who live on the coast, or near a lake or the property that's now going to be dry. And, and the, and actually from a career and working in generally glaring mall I can tell you if you don't already know this that a lot of that stuff will average out. There are going to be the reason I put up the thing about the changes in employment is that there are winners and losers from all these policies right they're going to be people who lose their jobs and other people who get jobs because of this. The net impact is going to be small, but the micro impact is big and so going forward I think we just need to always be sure wherever possible to communicate that the macro average is not the micro level of detail. Yeah, if I could respond to the last point, I think it's often the case that the macro impacts are smaller than the micro impacts, but in the case of climate change I don't, I don't even know that we know the macro impacts are going to be small. I think down the road they could actually be quite substantial, but but I absolutely agree that said that that the way to go is to look at more kind of heterogeneous effect across regions locations countries and the like. On the other part parts of the question. In terms of monetary versus fiscal policy, I've written for the journal monetary economics so somewhat skeptical essay on the potency of monetary policy to address climate change and what, and the limits limitations what central bank policy is I think really the important challenges are more in the fiscal side. The financial stability is a non trivial issue potentially, but that's going to play out over much longer time period and truth, the financial stability concerns. Right now, that firms face are based on policy uncertainty so it's like you're asking the monetary policy authority to address uncertainty that comes from other sources which is an interesting, it's an interesting gamer in its own in its own way. I think of which which I wanted to emphasize and I should emphasize this is my talk that the first thing that we're doing well to most uncertainty quantifications as we're putting it inside the policy problem instead of outside, lots of studies to kind of do it outside and we're put inside. Now the point I should have emphasized and I'm glad you raised this question is, but when we're solving our most prudent policy that is only a baseline against which we compare ad hoc policies. And so, so we have to take the next step to say what's, you know, suppose you start imposing ad hoc policies how close are they getting us to do something as prudent and and and kind of where the big gaps. And so what's missing in my calculations here was the assessment of the ad hoc policies which is absolutely vital for this research program would be successful and stuff that we have to work on. That's also, you know, very important as we're studying the Brazilian Amazon to we can figure out what the prudent policy is there but there's lots of potentially ad hoc solutions that might get you at least part of the way there and so then we can at least try to measure how kind of what's left over. But anyway, thanks for that question. Sorry, Lars reminded me that I missed part of Heather's question so the just one tiny thing to add to that is that one thing that comes out of our work is that the different piece of the government need to talk to one another because a lot of climate policy is about making things more expensive it's explicitly about raising the cost of energy in places. And so, so if the Federal Reserve doesn't handle it properly that could be mistaken for regular price inflation right when really it's a designed in relative price change in the economy. But so, so if the Fed mistakes it for inflation when you're doing this, then the reaction would be to slow the money to money supply growth and possibly create a recession at the time when you're trying to implement the carbon tax so coordination is important. I think we have a paper on that don't we Pete. Okay. Okay, so I'm going to move to the other questioners I'm going to just ask folks to, so we can get enough questions in to keep your questions kind of succinct. So we can, we can take, let's take, let's take Galena and Fockrey together. I know you go first. Yeah, thanks. I have a question for Pete I guess it's a little bit really two questions about agricultural sector so we, you know, in majority of macro models we don't have it but for climate I think it's the most vulnerable sector to climate so we really need to worry about that. And one question I have is for the US how important it is in terms of the impact on crop yields, how that translates into the inflation and therefore consumption and so just quantitatively is that important enough for the US. And in that same context how much of an adjustment are you allowing in the model for crop switching for example because maybe crop yields go down to zero so I've got to grow something else I can't grow rice in California anymore. And another question is about the emissions from the agricultural sector. So it's especially if the policies we're considering our CO2 specific, the share of agriculture emissions will start going up if we're reducing CO2 from the energy sector, and potentially the adaptation of agricultural sector to the physical risks can also be affecting those emissions so is that something that your model can incorporate at this stage. Yeah, so again. I'm sorry Galina, do you mind introducing yourself. I should have. Oh, I'm sorry I'm going to have UC Santa Cruz. Okay, great. Okay, and fuckery let's take your question as well and please introduce yourself. Okay, yeah. Thank you very much for the presentation. I'm from King Abdullah petroleum studies and research center located in Riyadh, Saudi Arabia. So, my question is related to Peter's presentation, specifically speaking related to GQ model. So, one of the slides showed that in the model we have labor sector has this equilibrium represented by unemployment. However, the other sectors like let's say real sector fiscal sector monitor and etc. They are all in equilibrium. So I'm trying to understand I much appreciate if you, if you can help me to understand. So if one sector can have this equilibrium. And the other set while other sectors are in equilibrium then this equilibrium from the labor sector, again represented by unemployment. How can be accounted for other sectors because they have implication for other sectors for example if you have high employment then government spending or fiscal spending can go up or real sector can go down. It can go to unemployment in certain sectors. So how can be disagreeing emerging from the labor market can be properly accounted in the other sectors but they are in equilibrium set up. Thank you very much. Okay, thanks so so on the agriculture. I have a pair of questions. I had my colleagues haven't done work on looking at the impacts of climate on agriculture at the level of detail you asked about but if we were going to do something like that I think we would reach out to one of the agriculture models to see what they have done like and see what impacts they expect to happen and then we would translate those into productivity shocks to feed them into the model and see what happens to the rest of the economy. In terms of emissions, these models are models but also many other CG models have had emissions coefficients attached to all of the inputs and outputs from all the sectors so they're able to track. Emissions of CO2 from agriculture but also emissions of nitrous oxide and and the whole gamut of pollution, both emissions in process and emissions from combustion. And in terms of switching gears to the labor market. What happens in our model is is basically pretty straightforward even though everyone, there are people who are who would like to work but can't at the current wage. We track everyone's income and spending and spending throughout the economy, and we look for the equilibrium set of prices that makes supply and demand balance and all the other markets, given those sorts of income so. There's an equilibrium in say the durable goods market that is not the same as the one that we would have if there were no unemployment, but we do get one where all the accounting holes. Everyone's expenditure is someone else's income no one gets any income that doesn't come from someone else's expenditure. And we end up with a with basically an equilibrium that's conditional on some of the people not being working. Great. Thanks Pete. I'm Laurie and then Bob. I was stuck I wasn't allowed to unmute but thanks for the authority to unmute. So hi everybody I'm Laurie Hunter I'm a sociologist and demographer from the University of Colorado Boulder and related to that I had a question for Pete, especially the the micro economy model of human migration a couple of times the spatial spatial kind of human migration. I was curious how that shows up in the model. You know, what any sort of movement in that realm is based on and does climate is climate able to impact any of it as yet in the model. So, so. Yeah, let's go ahead and take Bob as well. So, I think this is going to be your question to wrap up on probably but you guys highlighted several things that were not in the models we heard about this morning sectoral detail geographic detail uncertainty and a set of questions for which you need those to answer where, what do you think can be done with the existing models and where is the answer simply that the models being used or not up for the questions being asked of them. That is definitely a question to end on but deal back to Laurie's question. That's an excellent question. The models that I know of our have not yet been used to model regional migration due to climate change as an endogenous process. It could be done. It could be added into my but right now most models treat population as as immobile between regions. And that's true even if there is climate change. In terms of moving forward I think I think the white paper actually lays out a pretty good plan for how to move forward on some of these questions. I think it is not an easy thing to do in in some sense GP is just not the right thing for measuring this problem right and earlier question came up we talked a lot about infrastructure and whether spending to raise highways is wasteful and actually wasteful at all. It's just not very well included in GDP right the reason it isn't including GP as we spend money building a highway, and we don't charge anyone anything to use the highway we give it away right so none of the benefits of highways ever appear in GDP, but not because they're not there it's because the accounting doesn't include them. They're directly in productivity growth so they're there and then environmental benefits don't appear in GDP and investment is maybe not what you want if you're trying to measure welfare and GDP so it's so moving, moving toward more detailed models and moving toward a more detailed representation of the consequences of policies and just looking at GDP growth is important but it's going to be a big lift. Bridget do we need to stop there could we have just like a yeah go ahead. First I want to make an amendment about the spatial models that my colleague Esteban Rossi and various co authors have been doing models in which you and dodgingly move people around the world in order to address climate change I worry that their cost structure isn't quite right. My ideas are going to get always move people to Canada or Siberia in order to address climate change what's nice about those models is the endogenous responses and whether we got the whether we got the migration cost right yeah I think it was really remains a very open issue. The same thing shows up in our models of the Brazilian Amazon and which were reallocating production activities over time, and into different locations and ways that are socially prudent. And so there's a very important spatial component to those two. So, there's lots of interest in this kind of spatial part of it but it's, it's, I think there's much more important work to be done. Let's say, Robert, I'm not sure I remember your question it was about. Basically what you guys identified and your focus was on an uncertainty several things that we're really not present in the models this morning so right. How far can we push those models and where do we need to help you tell our folks that. You've kind of two responses to that I think this idea about thinking about uncertainty and broader terms. All the way from this specification ambiguity can, can, can help frame uncertainty discussions more generally across a variety set of models and that's not really confined to our setup here. Now, now, now the thing that the challenge we're facing is to is to put a lot of richness both geographical and elsewhere. We're facing computational challenges in order to go forward, in order to integrate the full range of uncertainties that we want so I don't want to claim that's a fully trivial thing to do. We're having to use various machine learning method, you know, in order to expand state variables we're doing a variety of other to try to find computational tricks to make the models much more richer along the spatial dimension, multiple sectors multiple locations and the like so I think there's much important work to be done there and we're anxious to pursue it. Thanks. Okay, so we're going to have this like a lightning round. George, Jim, and Steve, just kind of keep your question to like. No, okay. Alright, maybe you can. One more question. Okay. All right. All right, George, George, you're it. Hi, I'm George Kovortov I'm an economist at the OCC. I was wondering about kind of your thoughts on the materiality and availability of tools on the interaction between macroeconomic midterm scenarios and the climate scenario so this is coming from a financial stability perspective. And I could imagine like, you know, there's we have stress testing which is like a business cycle shock and we have our climate scenarios and, and, and, you know, I think people are interested about the impact of them in both ways it could be that climate scenarios, you know, the physical hazards or transitions could cause increased vulnerability and some sort of non one year away. And then, and then on the other side, you know, a business cycle shock could cause like a delayed transition so so are the existing tools kind of up to the task or their additional kind of avenues forward. So, I've written rather skeptically on the, on the use of stress test with 30 year climate scenarios that central banks have been doing. First of all, I'm not sure the answers to the questions they asked make a whole lot of sense. If you allow an affirm to tell you, under this climate trajectory for 30 years. Here's what I would do that I can address that problem but that's not that that's not the real problem they address they face fundamentally uncertainty, and that's going to unfold over a very long long time period. That's very different than the financial crisis about you know suppose something's going to happen tomorrow, are you prepared for that. So I really don't see how that the scenario stuff it seems to me has to be pushed into realms under which we put some type of probability bounds on things or plausibility across the scenarios or it's just not going to work very well. And one, one of my concerns about scenarios as we go kind of put multiple inputs into it is there's ought to be interdependence in those probabilities and how we put those together on the table becomes quite becomes quite important like imagine that we have uncertainty about conditions, but we also have uncertainty about damages. As we start damaging the environment more and more there's going to be an endogenous policy response that's going to start connecting those things and so. So, so even there you have to start thinking about the co dependencies of these things that that in ways that are that that could be somewhat challenging. So I, I think there's some interesting stuff to be done there but it requires a lot more thought. Thanks for the question. So, please join me and I'll just add a little bit that if you, if, if you had to go forward with doing this now with existing models. I think there's a lot that can be done but I think the key for understanding financial stability in particular is understanding asset holdings and having data on who owns what because the financial stability. The question is, are, are the losses going to be concentrated in in organizations or sectors, which, which might then be subject to margin calls or something like that, and then fail when if the if everyone was diversified there wouldn't be a problem. So, so, so I, so the Fed I think has this data, whether whether other people can get access to it to understand who holds what is less clear to me. I think it can be done. Great, please join me in thanking our truth panel. And we'll now open the Slido tab again for this session for folks to kind of discuss another ideas additional questions or feedback on on the sessions topic. So again, we encourage people to discuss amongst themselves and I think we'll open the breakout rooms for our virtual participants, but then we'll be breaking until 315. We'll move into our last panel.