 So we'll be talking about environmental justice, climate air pollution as a decision in the U.S. power sector. And really this is at the intersection of the research that I do and my group does, which focus on engineering and economic feasibility of low-carbon, sustainable, affordable, and socially just energy systems via the efficient use and supply of energy. And of course as we think about the future in sustainable energy systems, it becomes important to consider also the role that policy plays. So part of the group also focuses on understanding the implications of different designs of regulations with policies, both in the context of what they achieve and whether they are meeting the intended goals, as well as understanding and quantifying unintended consequences that may emerge from policy design. And finally we're looking at these transitions in the context of people that make decisions, that use technology, that select technologies. And so a third stream of research really has to do with the way people make decisions and how they behave. And the emphasis in much of the work from my group has been to quantify environmental justice associated with transitions in the energy system. Now this is really and the work that is being generated thanks to a phenomenal group of graduate students from different programs here at Stanford, as well as a couple of folks at Carnegie Mellon University across a series of departments from energy resources engineering, the YAKRA program, MSNE, CE, and just to name a few. And so let's look at the big picture first. This is news we need energy to have the types of systems that we have today. We have enormous benefits arising from energy systems in terms of heading where we'd like to go, in terms of having a warm or cold environment to live in, in terms of storing food and so on and so on. And all those enormous benefits come also at some costs and costs to climate damages, as well as health damages from breathing polluted air, if we continue to rely on fossil fuels, and then just to the environment of the ecosystems. So we have this sort of two dimensions that have raised a lot of importance in the policy design spectrum. One is climate change impacts associated with our energy systems. And the other one is health damages from air pollution. And those two things are quite different building that treatment and their implications that come for all aspects. And who will suffer from the damages associated with climate change and health damages. So when I think about the production of electricity, we'll have different steps from mining to fuel extraction and transport to the emissions onsite that we generate in power. And we'll have two different streams of pollutants of interest to this research. One is greenhouse gases like CO2 and methane. And those who have a global dispersion will stay for a very long time in the atmosphere. And so then treatment is really on consequences of a very large time horizons. And then I have health damaging air pollutants and those are namely SO2 NOx and PM2.5. And so we have different streams of pollutants. The fuel combustion will emit both PM2.5 directly, which is called the primary PM, as well as emissions of sulfur dioxide NOx, which may react to ammonia in the atmosphere and form secondary PM. So throughout this seminar, as I mentioned PM or PM2.5, I'm really thinking about this combination of secondary PM formation and primary PM that have both impacts on health damages. So and then are we really concerned about just where does the concentration of this pollutant increase? Not really work. We're really worried about impacts associated with PM2.5. And so that means understanding where the concentration is increasing, but also who is exposed. So think about this very dirty cold parked lamp that is in the middle of nowhere and where the dispersion is such that it doesn't affect any densely populated or mildly populated areas will have a much lower impact than potentially vehicles that are emitting damages, pollutants at ground level source, but in a very highly densely populated area. So just a quick question. Yes. Isn't the concentration of ammonia in the atmosphere pretty minimal? The concentration is minimal, but it has important implications in the reaction and the formation of PM2.5. So the question was whether the concentration of ammonia would matter in the overall outcomes. And so we would want to understand the health damages from air pollution, which will depend on where the concentration of PM will be increasing and how many people are impacted by that increase of pollution. And in contrast to GHGs, these types of pollutants will have a very short time in the atmosphere. Meaning if we turn off the knob and stop emitting SO2 knocks and PM2.5, we could quickly get a handle on the damages that are associated with air pollution. In contrast, even if we turn off the knob for greenhouse gases, we're already committed to a massive problem even that persistence in the atmosphere. Not only that, but the dispersion of those pollutants will depend on the type of industrial or emitting facility that we're talking about. So things like power plants will emit primarily SO2 and knocks and will have a wide dispersion. And by wide dispersion, I really mean crossing state boundaries, as we'll see in a little bit. And then you have other sources of emissions like industrial facilities that will emit SO2 and other hazardous air pollutants, which have medium to wide dispersion. And finally, things like trucks and cars, which will primarily emit knocks and those will be at ground level. And again, truck may be more highly polluting than a car, but if it is in a highway with no one around and there is no congestion, the impact will possibly be lower than a vehicle in a densely populated area. So all of those tradeoffs come into play and those are things that we've tried to quantify. And so we looked at this issue of trying to understand the demographic and environmental justice implications associated with the U.S. electricity sector. This is what that has been done in collaboration with a phenomenal group of researchers who are very fortunate to collaborate with them, Manader Thien, Chris Thesson, and Julian Marshall. And so folks had looked previously at the premature mortality associated with electricity production in the United States and elsewhere. But what hasn't been the focus of attention was, okay, how did these impacts differ by demographics, by income, by race, and ethnicity? So we cannot add a depth layer. Zooming just out, sorry, just a second as I cannot see. I have things on my own slides. So in terms of the broad scope of the problem, PM 2.5 is the largest informal health problem in the U.S. and globally. So overall, in just looking at the U.S. and at PM from all sources, it accounts for more than 100,000 fatalities per year. And the electricity generation is still a significant contributor to the problem, primarily owing to the production of electricity from coal war plants. Even if that contribution has declined over time, so owing to environmental regulations that require the addition of air pollution control technologies on the stack of power plants, and the fact that we've been transitioning away from coal and adding natural gas, thanks to very cheap natural gas prices, and then to a lesser extent also, thanks to the increased share of renewables, the two other factors kind of dominate the decline on the damages from the power sector that we've been seeing. And so as I hinted at previously, there were estimates of PM 2.5-related mortality from electricity production. They have a wide range from 10,000 to 52,000 premature deaths per year. And this really depends on which year is being modelled, which model is being used, and all sorts of other details and designs, in particular the assumptions that the modeller uses for the relationship between an increase in concentration of PM and the impacts to health or those response functions associated with politics. And to date, the demographic distributions of resulting exposure is still largely unknown. We try to contribute to that by looking at the exposures to and health impacts of PM 2.5 from electricity generation in the US. We looked at the seven regional transmission organizations, or RTOs, and for each of the state and from and to a state, as well as by income and race. And it's just like, you know, if someone is complaining about sound on the ventilator. Now, how do we compute those damages? We use what are called integrated assessment models. And so what those models do is that they couple spatially resolved emissions data. We reduce complexity models of the atmosphere, and they compute first and foremost the baseline concentrations of PM 2.5 at each location. And several groups of colleagues have been pushing forward the reduced form models. And folks like myself are users of those models and generally contrasts against different AIMs that have been developed to quantify the damages from air pollution. And so the way these models turn is that first and foremost, you really need a very detailed inventory of emissions. So this is like annual emissions at every single power plant, industrial facility, or ground level emissions in the case of transportation. As to capture all emission sources that we may have in geographical location like the United States and with a resolution that is predefined by location. And so that gives you both on the emissions by pollutant and location. And then by running this model with all sources of pollution, you have the simulated PM 2.5 concentration at different locations. And from that, we can see, okay, what are the premature deaths associated overall with these levels of pollution. But now how do we attribute the damages to different sources and different locations of pollution? So what we do is that after the baseline run, start doing the perturbation at the time for every single source of pollution. So we go to a power plant located in Allegheny County and we increase the amount of emissions for a specific pollutant by a ton. We run the model again and compute the difference in concentrations of PM across the entire country that are due to that point source emission of one additional marginal ton. And you go on and do that for all the facilities that we have information for or for ground gas sources. So that's the next step in terms of understanding the change in concentration of pollution. But as I mentioned, we're not as much worried about the concentration issues but the implications of that concentration. So the next step is to understand what that means in terms of premature mortality. So it's a function that relates the change in concentration at the location with the additional mortality that induces. So that's the dose response function and multiply that by the population that is exposed to that pollution. Once we have those sorts of information, the layer that we added was census information on self-reported racism between 65 population by law group as well as household income data and with that information we started looking at, okay, who is impacted and how by the different sources of pollution. And so I'll jump to the very high level results. First, our own estimate is that more than 60,000 people die prematurely every year due to electricity generation related emissions. This amounts to an average of four premature deaths per tear-with-hour of electricity generated. And 85% of those premature deaths are located and attributable to electricity generating units that are between an RTO and I'll show the maps of RTOs in just a little bit. And 90% of the premature mortality is due to electricity generation from cold electricity generated units. So even if there is a decline in emissions overall from cold electricity, they are still the dominant source of premature mortality. A little bit about these figures. So on the right, we have the RTO boundaries just to understand where those major entities are, their names and the share of cold in their overall electricity generation. And in the table on the left, we have a few details related to those units. So we see that MISO, which is this sort of dark blue and PGM, the navy blue, where coal is still pretty prevalent are also markets that are very large. So they produce the largest amount of electricity and providing that service. Those are also the RTOs that have the largest amount of premature mortality both in absolute terms and in terms of premature mortality per tear-with-hour of electricity produced. The other noteworthy one is SPP. So SPP is this olive green region here in the needle, which is a smaller market, but the share of coal in the overall production is still fairly large. And the premature mortality per tear-with-hour is also on the top three, actually second to just MISO. So that's in terms of big picture. And now let's look at the issue of how do these results differ by race and ethnicity. So here in the vertical axis, I'm showing premature mortality for 100,000 people for premature mortality related to the emissions from electricity generating units for both primary and secondary PM. On the horizontal axis, we show the groups of race and ethnicity. And so across the United States, on average, air pollution from electricity generation has a consequence of 5.3 premature deaths per 100,000 people. But when we start looking at this in the perspective of different race and ethnicity, we see that black African-American have higher rates of premature deaths per 100,000 people followed by white non-Latino with the other groups having lower exposure and premature mortality. And one may wonder, okay, how is this differentiated by income? Are we all breathing the same air? Or is it the case that there are segments of the population that are breathing more less polluted air than others? So here in the vertical axis, we have the same thing premature mortality for 100,000 people related to emissions from electricity generating units. And on the horizontal axis, we have household income groups in $1,000. And first, we'll look at the average across the US for different income group segments. And what we see is that there is a slight decline. The kink is just an issue of my animation. There is no kink here in the lines. So that people, households with higher income are located in regions where there are less exposed to air pollution than low income households. But it's a mild decline. In contrast, when we look at layered different race and ethnicity self-reported groups, we see that across all income segments, black African-American are more exposed than the other groups to the consequences of air pollution followed by the white and non-Lutino. And where the bubbles correspond to the number of people that are in that income bracket in the United States. Now, that's one layer of distributional effects. And now we'll move on to another one. And that one has to do with where the pollution is generated and who suffers the damages from pollution across states. So in this first panel, I'm showing the premature mortality in the state related to emissions from electricity generation, regardless of where those emissions originate. So it may be that the portion of the damages here in Texas are coming from the nearby states and so on and so forth. So we see that Pennsylvania, Texas and Ohio are the states that have the largest amount of premature mortality occurring between their state boundaries. And so the next layer is trying to disentangle where do those damages come from and who is opposing damages to home. So the second panel shows the premature mortality inside each state that is associated with emissions that occur between the state boundaries. And so we see that states like Ohio and Pennsylvania still have a large number of premature mortality, but much smaller than the number that occurs between the state boundaries, meaning the difference between these and what we've seen in panel A are damages that are important from somewhere else. So it may be that in Pennsylvania they come from Ohio and so forth and so on. But at the same time, we're able to see who is imposing damages to others. So this third panel shows the premature mortality that occurs outside the state boundaries due to emissions that occur between the state boundaries. So once again, and kind of calling out on Ohio and Pennsylvania, there are very large numbers here, meaning that these states are imposing a lot of premature deaths. They're simply not occurring between the state boundaries. They're imposing that elsewhere. So this begs really for cooperation and interstate policy when thinking about how to address these problems. And finally, D, which is the net effect of these damages. So this net effect is the damages cost to self plus the damages to cost to others minus the deaths that occur between state boundaries. And you see states like New York, which emit very little pollution, but which are amongst the largest ones in terms of the net effects going to pollution that occurs in neighboring states and that lands on the New York state. What is the negative value? The negative means that you're importing that amount of damages. So on that you're suffering those damages between state borders. So some of the key findings from this work is that the average exposures are the highest for Black, African American people followed by non-white and the exposures for the remaining groups are somehow lower. That the disparities between by race and ethnicity are observed for each income category, indicating that the racial and ethnicity differences hold even after coming for differences in income. And the levels of the disparity differ by state and RTO. And we observe that exposures are higher for lower income than higher income, but disparities are larger by race than income. And geographically, we observe that there are large differences between where electricity is generated and where people experience the resulting PM2.5 health consequences, with some states being net in these quarters of health impacts and other net quarters. And indeed for 36 states, we find that most of the health impacts are attributable to emissions from other states. Yes. How do you quantify large differences in where people are exposed and where people are getting admitted? Like how large is that? If you go to the previous slide, like the third last point, like large differences. Yeah. So in addition to the amounts that I'm showing here, which indicate where the, whether the damages are to self or the damages that are imposed to others, in the paper itself, we include a matrix on sources and receptors of pollution across the state. So I'm happy to share that. I don't have it in the slides, but happy to share that through a way. I got a quick question. Yeah. How do you decouple many, many other things that cause people to die early from PM2.5? I'm just curious if there are studies done on the medical side in conjunction with yours that do a little more of that. I understand it's a model and there's a lot of assumptions that probably go into the model. That's a great question. Just to repeat in case folks didn't hear online, the question is how do one disentangle the premature mortality that is associated with air pollution versus all other causes of premature mortality? And so let me start by saying that there is definitely uncertainty there. And in the second part of the talk, I'll show the importance of those response functions on the differences of increasing in pollution to premature death. How important they are actually on the outcomes that we can see. And I am a user rather than a developer for that research. But the way researchers have gone about this is to study courts across different locations and with different levels of exposure, try to disentangle the factors that are other causes of premature mortality across those courts and identify where there are changes really just attributable to the different concentrations. So through the statistical econometric studies that follow those courts, there are a few all-weekly patients on the edification of the courts or other things that may be going on. But that's what we rely on and that's definitely an important uncertainty associated with this. I think that the research has established with a very high level of confidence that there is an association between the level of air pollution and premature mortality from these causes. The magnitude of those effects is still a question. So in the second part of the work, we are bringing together what was in my initial slides in terms of climate and air pollution. So up until now we talked just about the air pollution consequences. One of the other questions that we had was really more on the policy grounds. A lot of attention has been devoted in the past in issues like the clean power plant on the decarbonization of electricity sector and accounting for air pollution consequences as a co-benefit or an added damage, but really not integrating that into the policy design itself and the goals. And we were wondering how that could change the types of technologies and their locations that one would incentivize to be built up. So this paper was also a collaborative effort with my former student Brian Sturdy who is now at NREL, several faculty members Peter Adams, Nick Neuler, Alan Robinson, Stephen Davis and Julian Marshall. And as I mentioned, generally climate policies often treat air quality and al-qaq co-benefits and despite the linkages between these two realms, which are tied to sources of emissions like the transportation sector or power plants, there are few policies that have exclusively considered both when designing improvements about those dimensions and the choice of which power plants are replaced by low emissions alternatives can dramatically alter the health benefits from a given reduction in system-wide emissions. And there has been some previous work but very limited exploration on how especially granular co-optimization of climate and health benefits could affect the emissions reductions across U.S. power plants. So we did that. And our ongoing approach was to explore the optimal locations for emissions reductions from U.S. power sector and including this integrated treatment of climate and health damages associated with air pollution. And so we did this at a county level. We developed a model for the capacity requirement of the experiential model to export the implications of location while integrating those two dimensions, climate damages and health damages from air pollution explicitly also in the objective function on decisions about new build-ups and requirements. And as a test that we use a policy that exhaustively specifies the need to meet the 30% due to reduction in emissions and targets, which is kind of the number floating around in several existing and proposed policies. And we run this across different scenarios where we aim at minimizing social costs, but in some instances the social costs explicitly account for the externalities associated with air pollution, whereas in other instances we just include the costs of the level of the system to meet this policy as well as the costs imposed by CU2 in action using the social cost of carbon. And we make several simplified assumptions. The first one is that the new capacity needs to be built in the same county where an existing plant is being retired or replaced. And the reason for this assumption is to avoid the need of additional electricity transmission infrastructure. So if we relax this assumption, maybe we could have better sites or better wind sites, but then we'll need to add the transmission cost assumptions too, and then we'll have for future work. A little bit more on data and assumptions. So we did detailed electricity fleet characterization. So we used the 2017 SEMS data, which has both emissions and generation for all fossil fuel units larger than 25 megawatts. This is a fantastic data set that is put together by the Environmental Protection Agency, and one that we're always worried that will stop being published at some point, but really tracks hourly basis generation and emissions for all large fossil fuel based units. And we have unit level data on C2 as C2 enox emissions as well as generation, the fuel and unit type of the facility location, including its coordinates, which would then use the air quality modeling portion. And regarding the greenhouse gases emissions and damages for the climate change damages, we multiply the C2 emissions by the social cost of carbon, famous and infamous social cost of carbon for which we assume that $40 per ton of C2 in 2017 dollars. Just wondering what's the source for that, that gets $40? Yeah, so this is from the interagency working group. I think it was the previous lab to last million close to median estimate, not IDCC. So there is an interagency working group from the United States agencies that over the years has done several studies on the social cost of carbon. There is a more recent one than the number that I have posted here. They actually don't post a single number. They post a distribution of numbers and they're high or low discount rate assumptions. They have an equivalent assessment even for other greenhouse gases like methane that makes you to account the lifetime of the atmosphere. So it's definitely pointing to those resources. We did say the range is when you look at the interval range from really very low numbers all the way to over $130 per ton of C2. All of this depends really on the assumptions. And so in our work specifically, we generally do the sensitivity analysis over a wide range of social cost of carbon values given that uncertainty. We also consider the climate impacts associated with meeting leakage from the natural gas extraction and distribution. And we assume that MREL study to back calculate the leakage rate, which was 3%. And we convert methane to CO2 equivalent using a 100 year global warming potential. The meat and leakage is not including the objective function as something that would need to be achieved is just tracked separately as a consequence of increasing the amount of natural gas that we would add to the system. Now for air pollutants, we use also the existing fleet data on the emissions inventory as previously I explained. And then for the directures, we use something very similar to what I explained already for the previous paper. So we use the integrated assessment models and do first the baseline run and then a perturbation for the emissions at each power plant. But we add on a few layers that I didn't discuss in the previous work. So the first one is this issue of the importance of the selection of the dose response or concentration response function. We use the two main ones that have been developed in the literature, the ACS study or American Cancer Society and the Harvard CCT study. And as you'll see the estimates on premature mortality kind of double when you use one versus the other does the importance of specifying and showing both. The researchers are willing more and more to the use of the ACS one as the state of heart. And so we use that actually in some of the baseline results. We also were interested about modeling uncertainty on these integrated assessment models. So the co-authors in the study have been actually the developers of several of these models in that easier and AP2. And so we use all three of them and contrast the results that emerge across these different integrated assessment models. And just a little bit about those different models, they vary a little bit in their approach. AP3 uses a Gaussian pool modeling with very red, the rudimentary chemistry. Easier is the regression based approximation to a full chemical transport model or CTM. And ENAB has a modeling structure very similar to CTMs, but simplified temporarily in terms of chemistry and physics. So all of them are there different just to do it. And they all also have different grid levels. So that's the other layer. And all of these reduced form models are less precise than full-scale chemical transport models. The previous work by the researchers involved in the development of those models and others have shown that estimates for annual PM2.5 emission concentrations and resulting health damages are quite in good agreement across these three models. So we went over the existing fleet, the consequences for the CO2 and emissions damages. And now the third piece of the puzzle is what are we replacing the existing capacity with? Because that will matter on the cost of achieving those reductions, as well as on the performance of the amount of emissions avoided. So we limit the alternatives to three main ones, NGCC, so natural gas and rice cycle, or wind or utility-scale solar. So a few more details on the assumptions here for NGCC plants. The good thing about those is that they meet about half of the CO2 in combustion phase than coal power plants. And they have far less CO2 and not emissions than coal. So they will also result in fewer health consequences. They also have the benefit of being very easily dispatchable. And so as we're replacing coal power plants, they will provide the option to have firm capacity. And we assume that they are able to meet the same loads as the thermal loads that were always providing before. Sorry, did you have a question? Yeah, that CO2 is an equivalent of CO2 or it's just pure CO2. Because there will be quite a bit of methane emissions when it comes to it. Yeah, so it explains CO2. This is really use-based combustion. And then we track the methane leakage and methane implications separately, not as part of this function. Yeah. And so we estimate the amount of plant capacity that is needed to be built up to replace predominantly coal. And we assume capacity factors are 56% and it creates that are based on NREL databases. And we assume that the plants need to be built in increments of 150 megawatts. So we'll increase those until they need to be a generation from the equivalent coal plant that was onsite. And to your question, we do account for the upstream emissions from methane leakage, but not in the energy function. So that's it for NGSTC for plants. And then we also provide the option of either wind or utility-scale solar. And so for that to rely, in the case of wind, we use capacity factors from NREL that are location-specific and we use average value by county. And in solar, a very similar strategy has been put together by NREL. So we have the irradiance levels and convert that into capacity factors that will be average at the county level. And now we will also make this important assumption, which is if we were to just add wind or solar, and those are mistake, they wouldn't necessarily meet the amount of generation that is provided by coal in the site. So we impose the storage requirement too. So solar and storage or wind and storage are combined to be able to provide that firm power, and we include the cost of that added storage too. And we have, we assume that $1,500 per kilowatt of installed storage. The cost of storage is declining very rapidly as well as those of solar. So I think in the near future we'll be interested in reviewing these analysis of the new assumptions of those costs and maybe new insights will emerge. Now, next we can get the cost of mitigation. So this will be the total annual mitigation costs that come from the optimization scenario versus the baseline. What do we include in the objective function? We include all the annualized capital expenditures for all new capacity that need to be built, as well as all the sum of the annual fixed costs, O&M variable costs and fuel costs. And we subtract from that the reduced O&M and fuel costs from existing coal or natural gas plants that would stop operating or shut down. And for all the assumptions on capital O&M and fuel costs, we rely on NREL's database. For the results that you'll see that we did then extensive sensitivity analysis, we assume a 20-year lifetime at 7% of this country. And then these series of sensitivity analysis. So we designed this linear optimization model that will include the total costs of the system when achieving the 30% CO2 emissions reductions. Plus the social cost of carbon targets the emissions of CO2 for each source of emissions. And then we either include or not this additional factor which is the marginal damages from a pollutant at the specific sites times the emissions of that pollutant as a site. And we look at the difference between including or not including the health damages portion exclusively here. So a little bit about results. In the vertical axis, I'm showing the annual damages and we'll look at the annual damages both from climate change and from air pollution. And the baseline is indeed the system as we have it today with the system composition and the mix of coal, natural gas and renewables that we have today across the US. And so if we do so and then there is the assumption of $40 per ton of CO2 for the climate damages, the climate damages associated with CO2 emissions are roughly near $70 billion per year. And the air quality related damages then have a wide range. So I'll go over this slowly. The light blue corresponds to the dose response function from the Harvard City study which is older than the ACS study which is shown in sort of naked blue. So you can see already over here how striking it is that the dose response function may make it such that the damages from air pollution are actually more important or less important than the climate change damages alone. And the different symbols, so the circle corresponds to AP3, triangle to EMAP and square to E0, so the three different air quality models. So we also see the importance of the different air quality models on overall damages quantification. And across all these three, we're using the same value of a statistical life and that they're under wide assumption. Now what happens when we impose a policy that ends up reducing CO2 emissions by 30% and where in the objective function we just include the costs to the system plus the climate change damages? Well, we see that the optimal strategy then leads us to emissions reductions of 30% and so an equivalent reduction also on the climate damages by nearly 30%. And as we pursue that climate only policy, we still have a co-benefit in terms of the health damages being reduced. So this is good as we transition to a decarbonized system would have the co-benefit indeed to air pollution reductions. So the annual health damages decrease from being between 34 and 120 billion dollars in the baseline to being between 13 and 50 billion in the climate only scenario. So achieving a 30% CO2 reduction target using the climate only approach still yields annual health benefits that range between 21 and 68 billion relative to the current baseline. So that's the difference between these scenarios in the same case. Now what happens if we account explicitly for the health damages when making the decisions of where to retire prior plans and what to build? And so what we see is that once again we assume we need the constraint of reducing CO2 emissions by 30%. But we have this additional reduction in health damages that occurs by explicitly accounting for the health damages in objective fashion. So when health is explicitly considered as a co-objective, the range of annual damages now falls from 4 to 14 billion dollars, meaning the annual health benefits associated with the decreasing pollution range from between 30 and 140 billion dollars per year. And now economists in the room if there are any would start asking me how much does this cost? Because right now we're just looking at health damages and the associated benefits. So let's look at that next. In this slide and the vertical axis we have the annual benefits in billion dollars. C corresponds to looking at the climate policy alone and tracking the health damages as co-benefit. Gage plus C means optimizing jointly for health damages and climate damages. And then we have the series of scenarios for each of the air quality models. And the orange bars correspond to the annual benefits in terms of Gage emissions reductions. The blue bars correspond to the health benefits from this air pollution. And the green bars correspond to the mitigation costs that are associated with building new park lands and retiring part of the existing capacity. The diamonds correspond to the net benefits which are all positive meaning that society overall if aiming to internalize the damages from air pollution from climate change would be better off by pursuing these investments versus the status quo. And a little bit about some of the key comparisons. We do see that the costs increase by about one billion two in some cases two billion dollars per year when explicitly including health as part of the portfolio. But overall the net benefits are also so the diamonds are also increased by explicitly incorporating health into this process. So yeah. Did you also model who is seeing those health benefits demographic was in the optimization? Yeah so we'll get the question a little bit. Not to the same extent as the previous one but I'll give you a little bit on how that's applicable. So the question was about who benefits from the health benefits and so we'll get to that thing just a little bit. You also talked more about what those benefits are specifically. So the benefits are premature mortality reduction due to exposure to air pollution. That's exactly what's in this blue bar. That's translated into a cost. And that's translated into an avoided cost as you reduce the pollution. So this whole thing is annual benefits in terms of reducing premature mortality that you otherwise have had if you sustain high pollution levels. So the healthcare system that would be staying? Yes so this is generally so we use the value of statistical life which is the value that is used by national agencies and others as a proxy for the value that we have associated with taking jobs that have different risk levels for example. It doesn't include the cost to the healthcare system in terms of treatment and so on. It doesn't so it's kind of a different perspective. That could be done too and that would be an added analysis. So under the baseline assumption the model builds primarily natural gas with the modest amount of wind concentrated primary locations where the wind has a really easy way to be resourced. But the mitigation costs are very sensitive to natural gas prices. There are many scenarios from ranging from current prices to tripling the gas prices even in certainty on future prices and that would of course translate to increased cost augmentation. And as we increase the price of natural gas then the selection of technologies between the models starts to be feeling more of wind storage rather than natural gas. A little bit about the health damages where they occur under the baseline and the different locations. So getting first at your question and then we'll see other lenses of this. On the top left we have the baseline so this is where the premature mortality is incurred regardless of where it is emitted under the current conditions. And the circles correspond to where coal power plants are located and the amount of generation from those coal plants which are shared in terms of power. And the second panel shows what happens under the climate policies so we see that several of the coal power plants in these regions are retired as part of that climate change mitigation goal. And the parent too is that you have benefits in terms of the health damages associated with pollution in those locations. And when including the climate and air policy first not we see that the power plants that get retired and removed are not necessarily the same ones. So we still see some prevalence of these power plants that had disappeared in the climate only policy. And we see even a further increase in reductions from health damages from air pollution. Now a little bit on equity and environmental justice. Lucy feels as being critical to continue when thinking about the locations of where plants are going to be removed and where new infrastructure is going to be built up. And the worry here is that the policy that will optimize for net benefits at the expense of specific groups may not be desirable specifically if some of those groups are low income racial minorities elderly or other at risk populations which already tend to experience poor air quality and higher health damages from air pollution as we've seen in the previous work. And so here our analysis are computing things at county level so we can say some things but we'll need much more detailed data to establish further evidence. But to give you a sense of some of the findings in this plot we showed the median annual health benefits and we look at those distributed across the nation by income quantile from the lowest quantile to the highest one. And we're looking at the health benefits from reduced air pollution when you have policies just focused on climate that's the dark blue bars or climate plus health where you have added benefits. So a good thing is that we see health benefits from these policies across all income quantiles and we see added benefits across all income quantiles when we help with climate. So yeah I'll just mention that and that the the largest benefits actually occur on the low income income quantiles rather than the high ones. Now we've looked at this also in terms of the share of minority people in a county from lowest amount of minority in the county to the highest quantile with population minority in a county. And in contrast to income we find that counties with lower shares of minority population are the ones that create the most health benefits regardless of the optimization strategy point. Now again in terms of who would need to bear the burden of those replacements the policy definition whether it's just climate or climate for plus health would have tremendously different impacts across the states. So over here to the left we have the new capacity that would need to be installed in gigawatts. The orange bars correspond to the climate only scenario the blue bars correspond to the health plus climate health damages plus climate change damages being explicitly included and you have a list of a few of the states. The axis on the right represent the share of existing capacity in the state that would need to be replaced and that corresponds to the black diamonds over here. So you see that some states really go from a very low amount of capacity that would need to be replaced about 12 percent to more than 50 percent of the state capacity needing to be replaced if we think about health plus climate scenarios. So this has tremendous implications for the design of policy and before the state's own investment in new infrastructure. So I'll move to summary so we see that improved air quality and human health are often discussed as co-benefit or site issue in climate change mitigation yet they are rarely considered when designing and implementing climate policies and so we analyze the implications of integrating health and climate when determining the best locations for replacing power plants. We wind solar natural gas and to meet an example of a CO2 a nationwide reduction target of 30 percent. We developed a capacity expansion model we coupled that we integrated assessment models of damages and compared portfolios. I'll say that all of this has been established as open source code so that if any of you are interested in exploring other aspects associated with these issues you're more than welcome to do so. There's plenty of work that could be developed in addition to what we've done to date. We explicitly tackled uncertainty on model formulation by using three integrated assessment models and a lot of concentration response functions by using the two major ones and really find that 262 by 30 percent would yield benefits to health too ranging from 21 to 68 billion per year and additional benefits between nine and 36 billion per year being possible when co-optimizing climate and health benefits and the health benefits can equal or exceed the climate benefits over a wide range of modeling assumptions. Again this will depend on what you assume for the social cost of carbon, what you assume for a value of a statistical life on the health damages portion and so on. And additional benefits include in prioritizing emissions reductions in counties with high population exposure. This is kind of given from the introduction slides. And then finally really suspects for considering both those attributes in policy design as well as thinking about interstate cooperation and thinking about infrastructure plans. So I'll end with that and we're going to look for discussion or questions. Thank you so much for being a person or online. We're going to have like two minutes I guess for questions or comments. We're just curious about the necessity of a co-benefit model because it seems like currently the U of S already has robust policies when it comes to evolution. We have the Q and A at several program regulations on PM. So it seems like that is already being handled and we need to kind of like look at the climate site so that overall we are optimizing local social environment. Yeah, so the question was do we really need to look at the air quality portion given that we already have some air quality policy in places through the clean air acts whereas in the climate we still have a long way to go. And so hopefully the message that I convened was not a lack of urgency in tackling the climate change policy problem and getting those regulations into play. It is though that air pollution damages are not yet a minor issue even with the Clean Air Act and with the progress that we have pursued. This is still an issue that has a large magnitude and very explicitly how it can play a role. Not only that but where the interaction between gap and other climate driven or sustainable energy policies will further get an increase in importance such as vehicle electrification, vote on the light duty and start trying to electrify all potential end users even if the grid is cleaner it will still constitute a large majority of damages. So my group for example is looking at the consequences of electrifying vehicles across the nation and whether that's a good thing versus a bad thing when compared to gasoline or diesel or hybrid vehicles depending on where you are and where you charge your vehicle with. And the story again may be a little different if you look at just GEDs versus GEDs plus air pollution. But to your point even more important than the case study that we observed here for the United States we're developing a lot of work in India and hopefully we'll have also focus in China in the future where those tradeoffs are even less clear in India the air pollution control technologies are still not there out of the stack so the premature mortality is of at most importance in China it's a little bit more underway but still the damages are going to be large. So this is also kind of a framework that can be applied to other locations across the nation where the air pollution consequences are even more damaging than what you see in the United States. Very long answer which brings us to 30 but there is one more pressing question if having to take that on otherwise we'll conclude and vote for folks online and here in the room. Oh there is one pressing question. It might be a quick mail but I was wondering it's like when you were getting all the data from the different bits of different demographics it might have happened or if it's even a thing or not it's like some of the different demographics are not as reported and you're out of them. Yeah so in a nutshell we rely on the census data so whatever is on the census data we're able to use but we are not looking at other sources of information to complement the results of of the census but happy to talk more then offline about that on the details and any suggestions that you may have in terms of complementing those data sets too and to everyone feel free to reach me via email if you have any additional comments or questions. Okay thank you. Thank you. Thank you. Thank you very much. The next seminar is next week same time 1.30. The speaker is Professor Wilmaia from UC Berkeley. She's going to talk about the project in Auckland.