WEBVTT 00:00.000 --> 00:05.000 You have any questions? 00:05.000 --> 00:08.000 All right, okay. 00:08.000 --> 00:10.000 Thanks very much for putting me on the program. 00:10.000 --> 00:12.000 It's an honor to be here. 00:12.000 --> 00:15.000 I should self disclose I'm an economist. 00:15.000 --> 00:17.000 Is there any other economist in the room? 00:17.000 --> 00:20.000 That's what I fear. 00:20.000 --> 00:26.000 So I'm going to talk to you about what economists do about climate change and how we figure out if it's a serious problem. 00:26.000 --> 00:33.000 The very serious problem, not such a serious problem, which tends to make economists amongst the most hated people in the space. 00:33.000 --> 00:35.000 But don't fear. 00:35.000 --> 00:37.000 I think we have good things to contribute. 00:37.000 --> 00:44.000 And today I want to talk to you about how we incorporate tipping points in the climate system into economic impact estimates of climate change. 00:44.000 --> 00:46.000 And my co-first and I, we've built a model. 00:46.000 --> 00:50.000 We call a meta stands for model for economic tipping point analysis. 00:50.000 --> 00:55.000 And I'm going to show you what this model does, how it works, and maybe get you interested in using it or working with it. 00:55.000 --> 00:58.000 And it's in Julia. 00:58.000 --> 01:00.000 Okay, so it's a, it's a collaboration that's not just me. 01:00.000 --> 01:03.000 The core team has seven deets in James Rising Simon. 01:03.000 --> 01:05.000 I would say it's the core architect of the model. 01:05.000 --> 01:08.000 James is the reason we could manage to put this into Julia. 01:08.000 --> 01:10.000 And we've partnered with Frustian Delphor project. 01:10.000 --> 01:11.000 He's a methane scientist. 01:11.000 --> 01:12.000 And we've gone on blackness. 01:12.000 --> 01:13.000 A climate economist. 01:13.000 --> 01:17.000 It's kind of between climate science and climate economics largely speaking. 01:17.000 --> 01:20.000 So tipping points are climate systems. 01:20.000 --> 01:22.000 What are they? 01:23.000 --> 01:32.000 Well, there's different definitions around that the definition we like to use is that you do a little small push in terms of emissions or temperature change and something large this continuous happens. 01:32.000 --> 01:35.000 Or it doesn't even have to be discontinuous. Something very large happens. 01:35.000 --> 01:38.000 So there's this proportionality of impact to output. 01:38.000 --> 01:40.000 Other people have a more strict definition. 01:40.000 --> 01:44.000 But if you take this wider definition, you get a lot of things that you might have heard about. 01:44.000 --> 01:46.000 Let's just look at one. 01:46.000 --> 01:48.000 For example, the green and I sheet here. 01:48.000 --> 01:50.000 It's a big ice sheet. 01:50.000 --> 01:53.000 That as the earth warms melt. 01:53.000 --> 01:56.000 And as it melts, it adds to sea level rice. 01:56.000 --> 01:59.000 And so for additional warming, you get additional sea level rice. 01:59.000 --> 02:02.000 More than you would have gotten just from thermal expansions. 02:02.000 --> 02:05.000 And there are the tipping points that add emissions. 02:05.000 --> 02:06.000 For example, the permafrost. 02:06.000 --> 02:13.000 If the permafrost unfreezes top zone Siberia, that leads to organic matter coming to contact with surface air. 02:13.000 --> 02:17.000 You get fluxes of methane and fluxes of emissions, the warming accelerates. 02:17.000 --> 02:19.000 So these are the kind of things we think about. 02:19.000 --> 02:27.000 Perhaps the most pressing one or the most lowering one that keeps the scientists up at night today is the Atlantic marion overturning circulation. 02:27.000 --> 02:30.000 That transports heat around the Atlantic. 02:30.000 --> 02:32.000 And there's signs that slowing down. 02:32.000 --> 02:34.000 But we don't really have longstanding time series. 02:34.000 --> 02:38.000 So we don't quite know if that's just a fluctuation or is that the start of a collapse. 02:38.000 --> 02:41.000 And there's no good way of putting probabilities to that. 02:41.000 --> 02:45.000 So if you risk averse, you should say we should probably not play around with this. 02:45.000 --> 02:48.000 Because if it does collapse, we have a big problem. 02:48.000 --> 02:50.000 On the other hand, maybe it doesn't. 02:50.000 --> 02:52.000 And then what should we do with that? 02:52.000 --> 02:55.000 So what does economics have to say about this? 02:55.000 --> 02:58.000 What can we learn from that in economics? 02:58.000 --> 03:02.000 Well, we back in 2017, we started work that came out in 2021. 03:02.000 --> 03:04.000 And we did a literature survey. 03:04.000 --> 03:08.000 We thought, okay, what do economists think about these stepping points? 03:08.000 --> 03:14.000 And we discovered to our surprise that there were more than 50 papers that modeled or estimated some kind of tipping point. 03:14.000 --> 03:17.000 And economic impacts of climate change, which is very good. 03:17.000 --> 03:20.000 But then we realized many of these were at-hawks studies. 03:20.000 --> 03:24.000 And it sends that, you know, you have a study that's carefully modeled. 03:24.000 --> 03:26.000 And then you assume tipping points exist. 03:26.000 --> 03:27.000 So something bad happens. 03:27.000 --> 03:29.000 But that's something bad. 03:29.000 --> 03:32.000 Was not calibrated to anything in the natural sciences. 03:32.000 --> 03:35.000 So it's basically what economists thought might happen. 03:35.000 --> 03:39.000 It wasn't any quantitative calibration and then underlying those. 03:39.000 --> 03:44.000 So we thought these studies aren't really useful for getting policy relevant quantifications. 03:44.000 --> 03:49.000 But the good thing is, there were about 20 studies that did have some kind of geophysical realism. 03:49.000 --> 03:53.000 So if you think about the green and ice sheet, they would do a small ice sheet model, 03:53.000 --> 03:58.000 emulator model or an emulator model that would emulate the larger ice sheet models. 03:58.000 --> 04:01.000 And you couple that with an economic model and you have some kind of geophysical realism. 04:01.000 --> 04:04.000 You push on, on emissions, you push on temperature and something happens. 04:04.000 --> 04:07.000 And that's something is calibrated to the science fully. 04:07.000 --> 04:12.000 And there were about 21 studies, which are about, often they're about the same tipping points. 04:12.000 --> 04:16.000 About eight, nine tipping points, half in modeled and climate economics. 04:16.000 --> 04:20.000 Now the problem we have is that these studies come from different research teams. 04:20.000 --> 04:22.000 They come from different models. 04:22.000 --> 04:26.000 Even if they come from the same model, they run under different parameters for different scenarios. 04:26.000 --> 04:29.000 So there's no way to do a normal meta analysis, 04:29.000 --> 04:33.000 where you take the individual estimates and then you do something to some of them into one big estimate, 04:33.000 --> 04:35.000 which is what we wanted to do. 04:35.000 --> 04:41.000 Instead, we quickly realized that you really needed to produce what we call an integrated assessment model. 04:41.000 --> 04:44.000 It's a model that has a climate model and an economic model, 04:44.000 --> 04:48.000 where we can emulate each single one of these studies precisely. 04:48.000 --> 04:51.000 And then put them into the same framework into the same structure, 04:51.000 --> 04:54.000 so that we control the assumptions that go into that. 04:54.000 --> 04:57.000 So we can run that under consistent assumptions and the different assumptions, 04:57.000 --> 05:03.000 and see what the overall effect of all these tipping points is on economic impacts, estimates of climate change. 05:03.000 --> 05:06.000 The model we created is called the model for economic tipping point, 05:06.000 --> 05:08.000 the analysis meta, which is why I put this in the talk. 05:08.000 --> 05:10.000 It's freely available on GitHub. 05:10.000 --> 05:12.000 You can do whatever you want with it. 05:12.000 --> 05:17.000 And it belongs to something economists, a climate economist, 05:17.000 --> 05:20.000 called Social Cost Integrated Assessment Models. 05:20.000 --> 05:24.000 The bigger ones out there are the gift model by resources for the future 05:24.000 --> 05:27.000 that's been used to advise the US government, 05:27.000 --> 05:31.000 and the climate impact labs de-SIM model. 05:31.000 --> 05:34.000 I think it's dynamic, spatial, climate impacts models, 05:34.000 --> 05:35.000 I'm feeling like that. 05:35.000 --> 05:38.000 And both of these use are used to inform policy makers. 05:38.000 --> 05:41.000 So if someone in government wants to know what's the economic impact of climate change, 05:41.000 --> 05:44.000 in 2050, they probably use one of these two models. 05:44.000 --> 05:49.000 There are the ones out there, but that's the latest generation. 05:49.000 --> 05:53.000 It's modular, they tend to be quite good. 05:53.000 --> 05:56.000 It's been calibrated to recent climate signs, 05:56.000 --> 06:01.000 and they've had a features, but for an economist, it's not as interesting. 06:01.000 --> 06:03.000 So a work on software. 06:03.000 --> 06:08.000 So when we created meta in the paper in 2021, I'm very embarrassed, 06:08.000 --> 06:10.000 but it wasn't Excel. 06:10.000 --> 06:12.000 It was not just an Excel. 06:12.000 --> 06:14.000 It used to commercial plug-in, because we needed to run 06:14.000 --> 06:15.000 onto color simulations. 06:15.000 --> 06:17.000 And I'm very sorry. 06:17.000 --> 06:18.000 It's very embarrassing. 06:18.000 --> 06:19.000 We knew that at the time. 06:19.000 --> 06:21.000 It makes it very hard to check code. 06:21.000 --> 06:23.000 I mean, we had to put it on GitHub. 06:23.000 --> 06:25.000 But it makes it much harder to check code. 06:25.000 --> 06:26.000 It makes it much harder to access. 06:26.000 --> 06:27.000 You notice this plug-in? 06:27.000 --> 06:29.000 I think it costs a thousand euros to run it. 06:29.000 --> 06:30.000 I mean, it's... 06:30.000 --> 06:32.000 When we released it, the company contacted us and said, 06:32.000 --> 06:33.000 how are you using a plug-in? 06:33.000 --> 06:35.000 Would you like to do a webinar? 06:35.000 --> 06:37.000 So more people subscribe to our services. 06:37.000 --> 06:42.000 And we thought, no, we'd rather work on putting this into open software. 06:42.000 --> 06:46.000 But the reason we had it in Excel is because climate economists had worked 06:46.000 --> 06:48.000 with this software before. 06:48.000 --> 06:49.000 So it's just a legacy issue. 06:49.000 --> 06:50.000 You start. 06:50.000 --> 06:51.000 And then you have some skills in that. 06:51.000 --> 06:53.000 And you put that into the next version. 06:53.000 --> 06:56.000 But we made a choice to part it into open software. 06:56.000 --> 07:00.000 I should also say there is a person Dominic from the Fundman's 07:00.000 --> 07:03.000 program, I think, who just took the whole thing from Excel and 07:03.000 --> 07:06.000 ported it fully into Gams, which is incredible. 07:06.000 --> 07:08.000 I mean, Gams is not my preferred software. 07:08.000 --> 07:09.000 But he checked everything we did. 07:09.000 --> 07:11.000 And he found very few code errors. 07:11.000 --> 07:14.000 Of course, there always is something in there that was 07:14.000 --> 07:16.000 influential for the results in any way. 07:16.000 --> 07:18.000 So we are very thankful for him for checking all of that. 07:18.000 --> 07:22.000 But still we thought it was a lot of value in putting that into 07:22.000 --> 07:23.000 Julia. 07:23.000 --> 07:26.000 The reason we went for Julia and not Python, say it's because 07:26.000 --> 07:28.000 there is a big community of climate economists that live in 07:28.000 --> 07:29.000 Julia. 07:29.000 --> 07:32.000 And that's thanks to the Mimmy framework, or Mimmy JL. 07:32.000 --> 07:36.000 That's the effect or standard in climate economics. 07:36.000 --> 07:39.000 Now, it comes out of a Berkeley group. 07:39.000 --> 07:41.000 Mostly David, Anta's group. 07:41.000 --> 07:43.000 These are Randall's, Frank Garrick's, and many more people 07:43.000 --> 07:44.000 involved. 07:44.000 --> 07:46.000 And they're very, very helpful, very nice people. 07:47.000 --> 07:51.000 And the reason this started to be created is that there was 07:51.000 --> 07:55.000 some standardization in the United States as a result of a report 07:55.000 --> 07:58.000 by the National Academy of Sciences of the United States on 07:58.000 --> 08:02.000 how to best advice government on an economic impact of climate change. 08:02.000 --> 08:05.000 And Mimmy JL takes some of these recommendations, or is the basis 08:05.000 --> 08:09.000 to making some of these recommendations more readily usable. 08:09.000 --> 08:13.000 For example, this is a framework that allows you to build 08:13.000 --> 08:16.000 integrated assessment models in a way that they can talk to each other, 08:16.000 --> 08:19.000 in a way that you can swap out different modules and components. 08:19.000 --> 08:23.000 And I show you in a second what I mean with that. 08:23.000 --> 08:26.000 So this structure of meta is twofold. 08:26.000 --> 08:30.000 There's a climate model, and then they're economic impacts of climate change. 08:30.000 --> 08:34.000 For economists who might watch this, maybe this is an economist. 08:34.000 --> 08:36.000 It's not an optimization model. 08:36.000 --> 08:37.000 It's a model that runs scenarios. 08:37.000 --> 08:41.000 You feed the models scenarios to have emissions of CO2, 08:41.000 --> 08:44.000 of methane, and of other forces. 08:44.000 --> 08:47.000 And CO2 and methane are explicitly modeled. 08:47.000 --> 08:50.000 So what you do is you take emissions, you convert it into concentrations. 08:50.000 --> 08:51.000 Greenhouse gas is near atmosphere. 08:51.000 --> 08:53.000 You convert it into radiated forcing, 08:53.000 --> 08:57.000 that's to change in Earth's energy balance as a result of that concentration. 08:57.000 --> 09:01.000 And then you convert that radiated forcing change into global means of 09:01.000 --> 09:02.000 a temperature change. 09:02.000 --> 09:05.000 So you go from emissions to concentrations to radiated forcing to warming. 09:05.000 --> 09:06.000 Very standard stuff. 09:06.000 --> 09:09.000 So far, this is basically only climate science. 09:09.000 --> 09:12.000 And the neat thing is, thanks to the Mimmy framework, 09:12.000 --> 09:15.000 we can take the model that was created to standardize 09:15.000 --> 09:19.000 how economists and other emulator models use climate science 09:19.000 --> 09:23.000 to convert emissions to concentrations ready at the forcing. 09:23.000 --> 09:25.000 That's called fair. 09:25.000 --> 09:29.000 And people involved with the Mimmy framework built Mimmy fair, 09:29.000 --> 09:33.000 which is an exact version of fair, but in Mimmy. 09:33.000 --> 09:35.000 So instead of us having to re-program that, 09:35.000 --> 09:38.000 we can just take that off the shelf and put it into our model, 09:38.000 --> 09:42.000 which is great because it minimizes errors. 09:42.000 --> 09:44.000 It minimizes different climate modules, 09:44.000 --> 09:46.000 because the economists should agree on the climate science. 09:46.000 --> 09:49.000 We should disagree on the economics and check what that changed. 09:49.000 --> 09:51.000 So that's very good. 09:51.000 --> 09:53.000 So what we do is, as I said, we take emissions. 09:53.000 --> 09:55.000 We convert it into global means of a temperature change. 09:55.000 --> 09:57.000 But this is about tripping points. 09:57.000 --> 09:59.000 So how do the tripping points come in? 09:59.000 --> 10:01.000 Well, you can add the tripping points. 10:01.000 --> 10:02.000 You can switch them on or off. 10:02.000 --> 10:04.000 Each of these modules is independent. 10:04.000 --> 10:05.000 You can switch them on. 10:05.000 --> 10:06.000 You can switch them off. 10:07.000 --> 10:09.000 I discussed the permafrost current feedback before. 10:09.000 --> 10:12.000 So what happens when you switch that on is that 10:12.000 --> 10:15.000 global means surface temperature change gives you 10:15.000 --> 10:18.000 more permafrost melting. 10:18.000 --> 10:21.000 And then you have more trumps all in Siberia and other places in the north. 10:21.000 --> 10:23.000 It comes into contact with the surface air. 10:23.000 --> 10:26.000 And from organic decomposition mostly, 10:26.000 --> 10:29.000 to get fluxes of methane and to get fluxes of CO2. 10:29.000 --> 10:35.000 And they go back up there and generate additional warming estigo back through the climate model. 10:36.000 --> 10:37.000 The same happens. 10:37.000 --> 10:39.000 First one, dissociation of ocean emission hydrates. 10:39.000 --> 10:42.000 You can think of those as kind of submerged permafrost 10:42.000 --> 10:44.000 on the continental slopes and the oceans. 10:44.000 --> 10:46.000 But this is a very uncertain tipping point. 10:46.000 --> 10:49.000 So for the most part, we don't include it in the standard runs. 10:49.000 --> 10:51.000 We used to do that, but we decided not to do it anymore 10:51.000 --> 10:54.000 because the science has become less clear rather than more clear. 10:54.000 --> 10:56.000 And there are other ones, like the Amazon forest, 10:56.000 --> 11:00.000 rainforest, dive back, but no time to go through the mob. 11:00.000 --> 11:04.000 So as I said, we produce global means surface temperature change. 11:04.000 --> 11:07.000 This is the climate change variable. 11:07.000 --> 11:12.000 The drives and damages and impacts in this model. 11:12.000 --> 11:17.000 And so on the damage aside, we start with global means surface temperature change. 11:17.000 --> 11:21.000 And there's two damages channels, one from temperature. 11:21.000 --> 11:23.000 And one from sea level rise. 11:23.000 --> 11:25.000 Now you could say that's incomplete. 11:25.000 --> 11:26.000 There are might be other ones. 11:26.000 --> 11:29.000 You know there might be things like ocean acidification. 11:29.000 --> 11:31.000 Also it's a thing. 11:31.000 --> 11:35.000 But economists haven't been able to create 11:35.000 --> 11:39.000 mappings from global means surface temperature change to these outcomes. 11:39.000 --> 11:40.000 So that's working progress. 11:40.000 --> 11:43.000 I think these people working on biodiversity loss as well. 11:43.000 --> 11:46.000 And this is just a snapshot of very our economics. 11:46.000 --> 11:49.000 It's not us saying other channels are important. 11:49.000 --> 11:50.000 So keep this in mind. 11:50.000 --> 11:54.000 Missing channels usually means your underestimate damages of climate change. 11:54.000 --> 11:57.000 Because most of the time, these things tend to be negative. 11:57.000 --> 11:59.000 Not always, but any time is there. 11:59.000 --> 12:00.000 All right. 12:00.000 --> 12:02.000 So we take global means surface temperature change. 12:02.000 --> 12:05.000 We do something called statistical downscaling or pattern scaling. 12:05.000 --> 12:08.000 That gets us national means surface temperature change. 12:08.000 --> 12:12.000 And then we have damages at the national level that calibrated 12:12.000 --> 12:16.000 to a study in climate economics, which we think is still the best out there. 12:16.000 --> 12:17.000 Despite a lot of flaws. 12:17.000 --> 12:21.000 So the difficult part is in climate economics is to map geophysical change 12:21.000 --> 12:25.000 like temperature change to economic outcomes. 12:26.000 --> 12:29.000 We do this, and we think it's better than not doing it at all. 12:29.000 --> 12:31.000 But it has a lot of caveats. 12:31.000 --> 12:32.000 It's very uncertain. 12:32.000 --> 12:33.000 So keep that in mind. 12:33.000 --> 12:36.000 We don't think a precise quantification is possible at this point. 12:36.000 --> 12:41.000 But a quantification, we believe, is better than no quantification. 12:41.000 --> 12:45.000 Similarly, on the sea level rise damages channel. 12:45.000 --> 12:48.000 So you have without any tipping points, you have thermal expansion. 12:48.000 --> 12:52.000 So the oceans just become larger as the water heats, needs more space. 12:52.000 --> 12:55.000 And then you have melting of small ice caps and glaciers. 12:55.000 --> 12:58.000 That contributes to water and sea level rise. 12:58.000 --> 13:00.000 Now you can add tipping points to this. 13:00.000 --> 13:05.000 So if you have the disintegration of the large ice sheets in Greenland and in the Antarctic, 13:05.000 --> 13:09.000 these melt and give you additional sea level rise. 13:09.000 --> 13:10.000 Very simple. 13:10.000 --> 13:11.000 That's implemented here. 13:11.000 --> 13:13.000 That gets us total sea level rise. 13:13.000 --> 13:15.000 That gets us national sea level rise damages. 13:15.000 --> 13:18.000 Which in the end affect consumption per capita. 13:18.000 --> 13:21.000 That's the variable economists like to use to think about 13:21.000 --> 13:26.000 how much do people suffer or how much do people gain from climate change. 13:26.000 --> 13:31.000 Aimock, as I said before, is the Atlantic meridional overturning circulation that 13:31.000 --> 13:33.000 transports heat across the Atlantic. 13:33.000 --> 13:39.000 If that were to fully disappear, the climate of the Sun Europe would be very, very different. 13:39.000 --> 13:42.000 And not very nice. 13:42.000 --> 13:45.000 And then we have a tipping point that just affects India. 13:45.000 --> 13:47.000 So it's a bit outside of the channel. 13:47.000 --> 13:50.000 There is a change to India and some amongst the reliability. 13:50.000 --> 13:54.000 So these are all the tipping points we can model at the moment in climate economics. 13:54.000 --> 13:57.000 But this is by no means an exhaust of list of tipping points. 13:57.000 --> 14:01.000 It's also by no means us saying this is the right implementation. 14:01.000 --> 14:04.000 This is a metal study for the moment for the most part. 14:04.000 --> 14:07.000 We have taken what people in climate economics thought was the right thing to do. 14:07.000 --> 14:10.000 In some cases we change the calibration a little bit. 14:10.000 --> 14:13.000 But mostly this is trying to assess what climate economists believe. 14:13.000 --> 14:14.000 And you might disagree. 14:14.000 --> 14:17.000 You might say, well, maybe this is a component on aimot. 14:17.000 --> 14:18.000 It needs a different calibration. 14:18.000 --> 14:19.000 It needs a different implementation. 14:19.000 --> 14:21.000 And there is indeed work being done. 14:21.000 --> 14:24.000 So there is a really good work by two PhD students in Hamburg. 14:24.000 --> 14:30.000 They took the aimock module here and added a different channel and refined the calibration. 14:30.000 --> 14:31.000 Which improved the estimates. 14:31.000 --> 14:34.000 And they used the model for this purpose. 14:34.000 --> 14:35.000 So they forked the model. 14:35.000 --> 14:38.000 They changed this aimock component and they were in everything. 14:38.000 --> 14:42.000 With this change component, which I think is exactly why we wanted to put this on 14:42.000 --> 14:44.000 the first place. 14:44.000 --> 14:46.000 Okay. 14:46.000 --> 14:50.000 I think that's on the model structure and on what we have done now. 14:50.000 --> 14:54.000 I'll show you a few results before we get to the Q&A. 14:54.000 --> 15:03.000 So the first thing, the main result of this paper we published in 2021 was that if you look at damages that you come 15:03.000 --> 15:07.000 from a meeting one ton of CO2 today. 15:07.000 --> 15:11.000 We want to know what's the additional damage you get if they're tipping points in the climate system. 15:11.000 --> 15:15.000 So what if no percent of this ton on economists? 15:15.000 --> 15:20.000 So the social cost of carbon is the damage from a meeting one ton of CO2 today. 15:20.000 --> 15:23.000 You sum up all the damage over the lifetime of this ton of CO2. 15:23.000 --> 15:26.000 It might be near atmosphere for hundreds of years. 15:26.000 --> 15:33.000 And you take the present value of that in terms of how much it affects humans around the world. 15:33.000 --> 15:38.000 Very complex but basically think of this as damages from a meeting CO2, but just for one ton. 15:38.000 --> 15:39.000 Okay. 15:39.000 --> 15:42.000 But I didn't even tell you what the absolute level is. 15:42.000 --> 15:49.000 It's just an estimate of we estimate these damages without tipping points and with tipping points. 15:49.000 --> 15:53.000 And this graph what it tells you is what's the difference between the estimate with tipping points. 15:53.000 --> 15:56.000 And the estimate without tipping points and the different model configurations. 15:56.000 --> 15:58.000 That's why you have this distribution. 15:58.000 --> 16:02.000 There's a lot of parameters to some scenarios we run there. 16:02.000 --> 16:10.000 And so a positive number means that tipping points run has a higher damage than the non-tipping points run. 16:10.000 --> 16:11.000 Okay. 16:11.000 --> 16:14.000 And so you can see that there's basically no run in which tipping points help us. 16:14.000 --> 16:17.000 They almost always make damages worse. 16:17.000 --> 16:21.000 Our best estimate is about 25% increase in damages. 16:21.000 --> 16:23.000 But we think this is quite conservative. 16:23.000 --> 16:26.000 So we try to be under conservative side of things. 16:26.000 --> 16:31.000 If you take the full uncertainty you get something like a 40% increase in damages from having tipping points. 16:31.000 --> 16:34.000 From considering that they exist. 16:34.000 --> 16:37.000 And we calculate about a 0.1 probability of them doubling. 16:37.000 --> 16:39.000 So you just sum up everything here. 16:39.000 --> 16:43.000 Or a 0.02 probability of troubling. 16:43.000 --> 16:45.000 So it's more than 200%. 16:45.000 --> 16:47.000 That's the first result. 16:47.000 --> 16:50.000 So they make damages worse and the way that we can measure. 16:50.000 --> 16:55.000 The second is they're almost worse for almost everyone around the world. 16:55.000 --> 16:58.000 So no good news here. 16:58.000 --> 17:03.000 And this is work not by us, but again by the people who've used our model. 17:03.000 --> 17:07.000 And they used it to quantify future impacts and stock indices. 17:07.000 --> 17:10.000 And they find that climate change. 17:10.000 --> 17:16.000 They create a stock market returns and tipping points worse and that. 17:16.000 --> 17:19.000 And lastly that's the paper we're working on now. 17:19.000 --> 17:22.000 It's its submission at the moment where we asked okay. 17:22.000 --> 17:26.000 What can we learn from tipping points in terms of climate policy? 17:26.000 --> 17:31.000 So if we do ambitious climate policy does this decrease the likelihood of tipping points occurring. 17:31.000 --> 17:34.000 And we've done a paper off on global methane action. 17:34.000 --> 17:36.000 So cutting methane around the world. 17:36.000 --> 17:39.000 And quickly does that change tipping points? 17:39.000 --> 17:42.000 And indeed we find that if you do ambitious a methane action. 17:42.000 --> 17:46.000 You decrease tipping point intensity across all the tipping points. 17:46.000 --> 17:51.000 Something between depending on where you are and 2050 between two and 10%. 17:51.000 --> 17:53.000 Depending on which tipping points you're looking at. 17:53.000 --> 17:54.000 That's unpublished. 17:54.000 --> 17:58.000 So numbers might still change, but roughly that's where we are. 17:58.000 --> 17:59.000 Okay, let me conclude. 17:59.000 --> 18:05.000 So we've made quite a lot of progress in economics and in on how we model tipping points. 18:05.000 --> 18:07.000 There's people who do different work in this. 18:07.000 --> 18:12.000 It's starting to be integrated in conversations with US authorities. 18:12.000 --> 18:15.000 So they're taking part of meta at the moment to put that into the other model. 18:15.000 --> 18:17.000 Stay using. 18:17.000 --> 18:21.000 And we think quantification is useful because people do do benefit cost analysis. 18:21.000 --> 18:25.000 And if you don't quantify tipping points, the estimate that's used is zero. 18:25.000 --> 18:30.000 Other people widely disagree to say, well, if you're trying to do a partial quantification. 18:30.000 --> 18:34.000 Because you might not have the ocean acidification or you might underestimate worst case scenarios. 18:34.000 --> 18:39.000 That's dangerous because you make it seem like not such a big problem and it's really a gigantic problem. 18:39.000 --> 18:40.000 That's a fair criticism. 18:40.000 --> 18:41.000 So you're in this tension. 18:41.000 --> 18:42.000 Should we quantify? 18:42.000 --> 18:43.000 Should we not quantify? 18:43.000 --> 18:44.000 I'm large. 18:44.000 --> 18:50.000 And certain she's remain in particular about how economists think about impacts of certain physical change. 18:50.000 --> 18:54.000 And what has happened recently is that we have improved calibration of some of these tipping points. 18:54.000 --> 18:55.000 Likely hoots. 18:55.000 --> 19:00.000 The Antarctic model is a recent paper by Macau Frisam and the son of his PhD students. 19:00.000 --> 19:02.000 Sorry if one of his master students I think. 19:02.000 --> 19:04.000 We've recorded this thing in Julia. 19:04.000 --> 19:07.000 So it's accessible to community now. 19:07.000 --> 19:11.000 And the different things you could do if you interested. 19:11.000 --> 19:14.000 First you just fork it and run it yourself under different scenarios. 19:14.000 --> 19:15.000 You write your own research paper. 19:15.000 --> 19:17.000 That's perfectly fine. 19:17.000 --> 19:24.000 You could try to maintain or contribute or improve calibration of some of the components or add a new component. 19:24.000 --> 19:28.000 Or you could even change other parts of the model. 19:28.000 --> 19:30.000 You could add different damages functions. 19:30.000 --> 19:33.000 So you want to study not national consumption per capita, 19:33.000 --> 19:36.000 but maybe financial stability or something like that. 19:36.000 --> 19:38.000 So you could produce a component that does that. 19:38.000 --> 19:40.000 So if you're interested, do reject. 19:40.000 --> 19:42.000 Very happy to chat. 19:42.000 --> 19:44.000 And you find everything on GitHub. 19:44.000 --> 19:45.000 Thanks. 19:45.000 --> 19:47.000 Thank you.