气候冲击、气候政策和气候风险评估变化对全球经济的影 响.docx

《气候冲击、气候政策和气候风险评估变化对全球经济的影 响.docx》由会员分享，可在线阅读，更多相关《气候冲击、气候政策和气候风险评估变化对全球经济的影 响.docx（77页珍藏版）》请在淘文阁 - 分享文档赚钱的网站上搜索。

1、GLOBAL ECONOMIC IMPACTS OFCLIMATE SHOCKS, CLIMATE POLICY ANDCHANGES IN CLIMATE RISK ASSESSMENTMARCH 27, 2021ROSHEN FERNANDOThe Australian National University andAustralian Research Centre of Excellence in Population Ageing ResearchW曰FENG LIUThe Australian National University andAustralian Research C

2、entre of Excellence in Population Ageing ResearchWARWICK J. MCKIBBINThe Australian National University,Australian Research Centre of Excellence in Population Ageing Research,The Brookings Institution andCentre for Economic Policy Research LondonThe authors did not receive financial support from any

3、firm or person for this article or from any firm or person with a financial or political interest in this article. They are currently not an officer, director, or board member of any organization with an interest in this article.mid-century (IPCC 2018). So far, 58 countries have communicated net-zer

4、o carbon emissions by mid-century, including significant carbon emitters such as China, Japan, Korea, and the United Kingdom. The European Union has proposed to make the bloc carbon neutral by 2050, and US President Biden and Canadian Prime Minister Trudeau have agreed to work towards net-zero-emiss

5、ions by 2050. This deep decarbonization will significantly affect the world economy with heterogeneous impacts across countries and sectors. The World Economic Outlook (Bang et al. 2020) simulates the effects of achieving global net- zero-emissions via carbon taxes.Most recently, a few studies have

6、focused on the transition risks from a financial perspective. Carney (2015) warns that the energy transition could give rise to financial risks. Some organizations, such as the European Systemic Risk Board, have recommended stress tests of financial sectors related to climate transition risks. Some

7、central banks have proposed or conducted such tests (Vermeulen et al. 2018). van der Ploeg (2020) reviews pre-requisites to ensure a smooth transition to a carbon-free economy. He also reviews the empirical evidence for the effects of anticipated green transitions on asset returns and argues that th

8、e macro-financial policies should support the green transition. McKibbin et al. (2020) explore the interaction of monetary policy and climate change. They conclude that climate policy responses can have important implications for monetary policy. Monetary policy can also significantly affect the eco

9、nomic outcomes of climate policies. In light of ambitious climate actions urgency, the policy spheres should be brought together more explicitly, and more appropriate macroeconomic modeling frameworks should be developed.3. ESTIMATION OF PHYSICAL CLIMATE SHOCKSClimate scenariosIn this study, we firs

10、t assess the global macroeconomic effects of climate risks up to 2100 under various climate scenarios. We use the four Representative Concentration Pathways (RCP) introduced by van Vuuren et al. (201 I), namely RCP 2.6, RCP 4.5, RCP 6.0, and RCP 8.5. The pathways* names indicate the additional radia

11、tive forcing levels achieved by the end of the century compared to the pre-industrial times due to greenhouse gas concentrations in the atmosphere. Table I summarizes the definitions of the RCP scenarios. Hereafter, we refer to RCP 2.6, RCP 4.5, RCP 6.0, and RCP 8.5 as the climate scenarios.It is wo

12、rth noting that we use these scenarios (particularly RCP 8.5) to obtain a range of estimates about the economic consequences of physical climate risks. We do not attribute any likelihood to any of the scenarios and do not assume any scenario to be Ubusiness-as- usual”. Hausfather and Peters (2020) p

13、rovide a detailed discussion on how best to interpret RCP scenarios in line with the most recent developments. We follow the literature to interpret RCP 8.5 as an upper bound of the estimates.Table I: RCP ScenariosScenarioDescriptionRCP 2.6The peak in radiative forcing at 3 W/m? (490 ppm CO2 eq) bef

14、ore 2100 and then decline (the selected pathway decreases to 2.6 W/m by 2100).RCP 4.5Stabilization without overshoot pathway to 4.5 W/m (650 ppm CO2 eq) at stabilization after 2100RCP 6.0Stabilization without overshoot pathway to 6 W/m (850 ppm CO2 eq) at stabilization after 2100RCP 8.5Rising radiat

15、ive forcing pathway leading to 8.5 W/m (l 370 ppm CO2 eq) by 2100.Source: van Vuuren et al (201 I). Approximate radiative forcing levels were defined as 5% of the stated level in W/m relative to pre-industrial levels. Radiative forcing values include the net effect of all anthropogenic GHGs and othe

16、r forcing agents.Climate variablesWe use maximum temperature, minimum temperature, mean temperature, and precipitation as climate variables to determine the impact of climate risks. We obtain the historically observed climate variables and the projected climate variables under the climate scenarios

17、from the Intersectoral Inter-model Intercomparison Project (ISIMIP) data portal (202l)The projected climate variables under the climate scenarios are available from 2006 to 2100 from four different models (the model ensemble): GFDL-ESM2M, HadGEM2-ES, IPSL- CM5A-LR, and MIROC5. ISIMIP, led by the Pot

18、sdam-lnstitute for Climate Impact Research, facilitates comprehensive, consistent, and comparable simulations from different climate impact models regarding the global impact from various climate scenarios by providing the international modeling community with a coherent framework. The models have b

19、een developed respectively by the Geophysical Fluid Dynamics Laboratory (GFDL), the Met Office Hadley Centre, the Pierre Simon Laplace Institute (IPSL), and the University of Tokyo Centre for Climate System Research, National Institute for Environmental Studies, and Japan Agency for Marine-Earth Sci

20、ence and Technology Frontier Research Centre for Global Change. We use the daily projections for the climate variables from the model ensemble to account for uncertainty in the model results. After aggregating the 0.50 x 0.5 gridded data across 193 countries, specified by the Database of Global Admi

21、nistrative Areas (GADM), we average the daily data to obtain the monthly means from 2006 to 2100.Chronic climate risksThere is a broad range of long-term effects of climate change and an extensive body of literature discussing these effects. However, the availability of damage functions, which map t

22、he physical impacts of climate change onto economic variables, is minimal. Roson and Sartori (2016) review the literature on the damage functions and compile six damage functions for economic modeling assessments. These chronic risks include rising sea levels, variation in crop yields, heat-induced

23、impacts on labor productivity, changes in the occurrence of diseases, changes in tourism, and changes in household energy demand. Out of these, we focus on the first four chronic risks.Roson and Sartori (2016) express the damage functions related to the chronic risks using climate variables* changes

24、 compared to a benchmark level. The damage functions then use the relative changes in the climate variables compared to the benchmark to derive the economic shocks. The benchmark variable primarily used in the damage functions is the average value of the climate variables from 1985 to 2005.The damag

25、e functions we consider in this paper primarily use temperature and precipitation as the climate variables, and we use the projections for the climate variables under the climate scenarios from the model ensemble from 1979 to 2100 to derive the necessary benchmarks and the variations of the future c

26、limate variable from the benchmark. We then average the variations across the models for a given scenario for a given country. Using these variations, we use the damage functions to develop various economic shocks (see Section 4.4).Climate extreme shocksThe International Disaster Database, maintaine

27、d by the Centre for Research on the Epidemiology of Disasters (CRED), classifies disasters mainly as natural and technological disasters. Natural disasters are further classified as geophysical, meteorological, hydrological, climatological, biological, and extra-terrestrial disasters. The definition

28、s of these natural disaster groups and the types of disasters falling under each group are presented in Table Al in Appendix A.Based on the classification, meteorological and climatological disasters are caused by short and long-term variability in the climate. This study focuses on two climatologic

29、al disasters: droughts and wildfires, and two meteorological disasters: extreme temperature events and storms. In addition, despite being classified as a hydrological disaster, we also focus on floods due to the influence of climate variability on hydrological cycle. These five extreme climate shock

30、s collectively account for 73% of extreme climate shocks reported by CRED. A historical summary of these extreme climate shocks categorized by the model regions is presented in Table A2.As CRED reports, extreme events historically have led to 32.5 million lives lost and affected over 8 billion peopl

31、e in various forms (excluding deaths) from 1900 to 2019. The extreme climate shocks considered in this study have contributed to over 20 million deaths and affected almost 8 billion lives (excluding deaths). The breakdowns of historical fatalities and numbers of people affected by the extreme climat

32、e shocks aggregated across the model regions are presented in Table A3 and A4, respectively.Furthermore, the extreme climate shocks considered in this study collectively account for $US 912 billion of insured losses (88% of total insured losses from all extreme events), and almost $US 4 trillion of

33、total insured and non-insured indamages (74% of total insured and non-insured damages from all extreme events). Tables A5 and A6 present the historical breakdown of insured losses and total damages for the extreme events across the model regions.Modeling and predicting weather and extreme climate sh

34、ocks remains a challenge for the research community. Identifying the favorable initial state, large-scale drivers, local feedback processes, and stochastic processes are the underlying reasons for its complexity (Sillmann et al. 2017). However, an extensive literature survey demonstrates the possibi

35、lity to use various monitoring tools to identify the occurrence and duration of weather and climate- related extreme conditions as close approximations for extreme climate shocks. These tools, discussed in depth below, require climate variables, specifically precipitation, maximum temperature, and m

36、inimum temperature.Using the projections for the climate variables from the model ensemble for the climate scenarios and various approaches drawn from the literature, we approximate the frequency and duration of extreme climate shocks. Table 2 summarises the climate variables and the approaches. A d

37、etailed discussion of the estimations follows.Table 2: Approaches to Identifying Extreme Climate ShocksExtreme EventApproachClimate VariablesDroughtStandardized Precipitation IndexDaily PrecipitationFloodStandardized Precipitation IndexDaily PrecipitationExtreme Temperature (Heat waves & Cold waves)

38、Heat/Cold Wave Magnitude IndexDaily Maximum/Minimum TemperatureStormsProbabilistic econometric modelsDaily Maximum TemperatureWildfiresProbabilistic econometric modelsDaily Maximum Temperature & Daily PrecipitationSource: Developed by the Authors.(1) Droughts and floodsThe Standardized Precipitation

39、 Index (SPI) by McKee et al. (1993) is a widely used indicator to identify droughts and extreme precipitation events. The SPI uses observed precipitation data to quantify a point observation s standardized deviation from a probabilistic distribution of historical precipitation data. Thus, the SPI va

40、lues demonstrate the anomalies from the long-term mean and, based on the reference period at a given point of time, the SPI could be calculated for periods from I -36 months. The index value could then be interpreted, as indicated in Table Bl in Appendix B.A few recent studies using SPI to predict d

41、roughts and/or floods include Ekwezuo et al. (2020) for West Africa, Ali et al. (2020) for Pakistan Bhunia et al. (2020) for India, Golian et al. (2015) for Iran, Wang and Cao (201 I) for China, and Manasta et al. (2010) for Zimbabwe.Using the precipitation data obtained from ISIMIP for the model en

42、semble, we calculate 12- month SPI values from 2007 to 2100. We assume that 12-month SPI values below -2.00 for three or more months consecutively identify the future occurrence of droughts. We also assume that the occurrence of 12-month SPI values above 2.00 for three or more months sequentially id

43、entify a future occurrence of floods.We then aggregate the frequency and duration of the droughts and floods across the model regions. For model regions containing more than two countries, we use the proportion of GDP of a given country in 2019 compared to the region to weigh the frequencies and dur

44、ations of the events. By aggregating climate shocks using GDP weights, we better understand the relative vulnerability of different model regions to extreme climate shocks. We then obtain the average number of climate shocks across the model ensemble. Tables CI and C2 in Appendix C presents the cumu

45、lative frequency and duration of droughts and floods, respectively, from 2020 to 2080 under the climate scenarios.(2) Extreme temperature events: heat waves and cold wavesWe follow the approach by Russo et al. (2014) to identify the possibilities of heat waves under the climate scenarios. Accordingl

46、y, we first construct a maximum temperature sample for a given day in a given year using the maximum temperature of the day up to 15 years before and after, and then, compare the maximum temperature with the ninetieth percentile of the sample. If there are six or more days consecutively recording ma

47、ximum temperatures above the ninetieth percentile, those episodes are identified as possible heat waves.To identify cold waves, we use daily minimum temperatures instead of maximum temperature and use the tenth percentile of the sample as the threshold to compare the minimum temperature of a given d

48、ay in a given year. Six or more days of consecutive records of minimum temperature below the threshold are recognized as possible cold waves.After identifying heat and cold waves, we take the GDP-weighted average of their frequencies and durations to obtain the occurrence of the events in the model

49、regions. We then average the results from the model ensemble. Finally, we aggregate the averages to get the frequencies and durations of extreme temperature events under the climate scenarios. Table C3 presents the cumulative frequency and duration of extreme temperature events from 2020 to 2080 under the climate scenarios.(3) Storms and wildfiresA growing body of literature demonstrates the impacts of climate change, or mo