Policy Research Working Paper 11133 The Macroeconomic Implications of Climate Change Impacts and Adaptation Options A Modeling Approach Kodzovi Abalo Brent Boehlert Thanh Bui Andrew Burns Diego Castillo Unnada Chewpreecha Alexander Haider Stephane Hallegatte Charl Jooste Florent McIsaac Heather Ruberl Kim Smet Ken Strzepek Economic Policy Global Department & Climate Change Group May 2025 Policy Research Working Paper 11133 Abstract Estimating the macroeconomic implications of climate losses in 2050 exceeding 10 percent of gross domestic prod- change impacts and adaptation options is a topic of intense uct in some countries and scenarios, although only a small research. This paper presents a framework in the World set of impact channels is included. The paper also presents Bank’s macrostructural model to assess climate-related estimates of the macroeconomic gains from sector-level damages. This approach has been used in many Coun- adaptation interventions, considering their upfront costs try Climate and Development Reports, a World Bank and avoided climate impacts and finding significant net diagnostic that identifies priorities to ensure continued gross domestic product gains from adaptation opportuni- development in spite of climate change and climate policy ties identified in the Country Climate and Development objectives. The methodology captures a set of impact chan- Reports. Finally, the paper discusses the limits of current nels through which climate change affects the economy by modeling approaches, and their complementarity with (1) connecting a set of biophysical models to the macroeco- empirical approaches based on historical data series. The nomic model and (2) exploring a set of development and integrated modeling approach proposed in this paper can climate scenarios. The paper summarizes the results for five inform policymakers as they make proactive decisions on countries, highlighting the sources and magnitudes of their climate change adaptation and resilience. vulnerability --- with estimated gross domestic product This paper is a product of the Economic Policy Global Department and the Climate Change Group.. It is part of a larger effort by the World Bank to provide open access to its research and make a contribution to development policy discussions around the world. Policy Research Working Papers are also posted on the Web at http://www.worldbank.org/prwp. The authors may be contacted at cjooste@worldbank.org. The Policy Research Working Paper Series disseminates the findings of work in progress to encourage the exchange of ideas about development issues. An objective of the series is to get the findings out quickly, even if the presentations are less than fully polished. The papers carry the names of the authors and should be cited accordingly. The findings, interpretations, and conclusions expressed in this paper are entirely those of the authors. They do not necessarily represent the views of the International Bank for Reconstruction and Development/World Bank and its affiliated organizations, or those of the Executive Directors of the World Bank or the governments they represent. Produced by the Research Support Team The Macroeconomic Implications of Climate Change Impacts and Adaptation Options: A Modeling Approach ∗ Kodzovi Abalo1 , Brent Boehlert2 , Thanh Bui1 , Andrew Burns1 , Diego Castillo2 , Unnada Chewpreecha1 , Alexander Haider1 , Stephane Hallegatte1 , Charl Jooste1 , Florent McIsaac1 , Heather Ruberl1 , Kim Smet2 , and Ken Strzepek2 1 World Bank 2 Industrial Economics, IEc JEL Classification: C6, Q5 Keywords: Climate Change, Economic Impact Assessment ∗ The authors are grateful to Kevin Carey, Diji Chandrasekharan Behr, and Nepomuk Dunz for helpful comments. The authors thank the multi-donor Climate Support Facility for funding this work. 1 Introduction Climate change is a major challenge for economic development and human well-being. Past and future devel- opment gains and improvements in living standards are under threat from rising global temperatures, changes in precipitation patterns, and increased frequency and severity of extreme climate events. Low- and middle- income economies are particularly vulnerable to changes in climate conditions and extreme weather events. Between 2010 and 2019, eight of the ten most affected countries by extreme weather events were classified as low- or middle income economies by the World Bank, with six countries in the lower-middle income bracket.1 Since Dell et al. (2012), many studies have found that the effects of higher temperatures on agricultural yields, industrial output, political stability, and economic growth are especially pronounced for low-income economies (Schlenker and Lobell; 2010; Burke et al.; 2015; Kalkuhl and Wenz; 2020; Damania et al.; 2020; Kahn et al.; anzig; 2024). Within countries, impacts also affect the 2021; Nath et al.; 2024; Kotz et al.; 2024; Bilal and K¨ poorest and threaten poverty reduction, with modeling estimates suggesting that more than 100 million people would be living in extreme poverty in 2030 due to climate change (Hallegatte and Rozenberg; 2017; World Bank; 2020). Developing effective adaptation and mitigation strategies against the most severe effects of climate change poses a major challenge for all economies irrespective of their income levels. However, low- and middle-income economies face particularly difficult challenges, with constrained financial resources to invest in resilience and adaptation (World Bank; 2024b). The World Bank’s Country Climate and Development Reports (CCDRs) address those challenges at the country level. The reports discuss how countries can achieve their development objectives in spite of climate change impacts and minimizing their emissions of greenhouse gases. The reports identify the key climate risks for each country, and explore policies and interventions to move toward a resilient low-emission development path. To do so, the reports combine sector deep dives (e.g., on water availability, agricultural risks, or energy ac- cess) with macroeconomic and poverty assessments. Three CCDR synthesis reports (World Bank; 2022a, 2023, 2024a) summarize the lessons learned from reports covering 72 economies. They show that climate objectives in terms of greenhouse gas emissions and resilience can be achieved without compromising development as long as policies ensure appropriate private sector contributions and protect the most vulnerable populations, and that international support is available, especially for the poorest countries. Achieving development objectives in spite of climate change requires an understanding of how uncertain climate change will translate into biophysical impacts, such as changes in agricultural yields, which in turn will influence economic outcomes. This is not trivial, requiring experts and models from a variety of different dis- ciplines to coordinate efforts to identify relevant mechanisms for effective policy action. Projections of climate variables are based on multiple socioeconomic and emission scenarios (using the Shared Socioeconomic Path- ways or SSP) and converted to biophysical effects—e.g., changes in agricultural yields and labor productivity effects due to heat stress—based on biophysical models. By focusing on the effects of climate variables at a disaggregated spatial level, these models deliver important inputs for distributional and poverty assessments, emphasizing the varying nature and impacts of a changing climate across a country. Ultimately, the projected biophysical effects are weighted and aggregated to a national scale, so that their macroeconomic impacts can be estimated. Different macroeconomic models may be utilized for this task, including Computable General Equilibrium (CGE) models and macro-structural models. Here, we focus on the integration of biophysical models with the World Bank’s Macro-Fiscal Model (MFMod, Burns et al.; 2019), a macro-structural model.2 This paper documents a modeling framework that enables decision-makers to better understand climate change impacts on economic growth and GDP, in support of improved policy for climate and development, 1 See Eckstein et al. (2021) for the ranking of the most affected countries and the World Bank country classification. The World Bank classifies countries based on gross national income (GNI) per capita (Atlas method). Countries with per capita income of $1,315 or less are classified as low-income. Lower middle-income economies exhibit a per capita income level between $1,315 and $4,465, and upper-middle-income countries have income levels between $4,465 and $13,845. 2 During the CCDR process, macroeconomic models are linked to other techno-economic models, such as the World Bank Electricity Planning Model [EPM] (Chattopadhyay et al.; 2018), see Hallegatte, McIsaac, Dudu, Jooste, Knudsen and Beck (2024) for a description of the methodologies. 2 and its application to five African countries. Its mechanism-based approach, modeling various impact chan- nels through which climate change will affect economic growth and GDP, is meant to complement empirical approaches that provide aggregate estimates and may fail to capture future threshold effects or represent counterfactual policy scenarios. The remainder of this paper is structured as follows: Section 2 summarizes key past work on incorporating biophysical climate impacts into macroeconomic modeling that this approach builds on; Section 3 details the methodological framework, including an overview of both the biophysical models used to model different climate change impacts, as well as the macroeconomic models that translate these climate effects into economic impacts; Section 4 presents results from recent CCDR studies for five Sub-Saharan African economies: Guinea Bissau, Madagascar, Malawi, Mali, and Zimbabwe. Section 5 concludes with discussion and recommendations. Appendix A offers further detail on the selection of climate change scenarios as well as the biophysical models summarized in Section 3, while Appendix B provides more detail on the macroeconomic modeling summarized in Section 3.5. Finally, Appendix C provides further detail on the results for the case studies discussed in Section 4. 2 Review of Key Literature The recent literature on the macroeconomic impacts of climate change, especially the impact on GDP, has focused on empirical estimates, using historical data series and past changes in temperatures or precipitations (Auffhammer; 2018; Tol; 2024). Dell et al. (2012); Burke et al. (2015); Kalkuhl and Wenz (2020); Kahn et al. (2021); Nath et al. (2024); Kotz et al. (2024) estimate the effect of a marginal change in local temperature anzig (2024) explores the effect of global temperature, and Damania et al. (2020) show on GDP. Bilal and K¨ the importance of changes in rainfall, provided the analysis is done at high spatial resolution. These empirical estimates are sometimes used to calibrate damage functions, which are usually polynomial fits used to translate changes in temperature and precipitation into economic damages. While these empirical studies are important, they can be usefully complemented with model-based ap- proaches like the one presented in this paper, for multiple reasons. First, model-based approaches can go beyond the aggregated effect to help identify the various channels through which climate change affects GDP, by modeling each mechanism individually. This helps check that we understand the mechanisms behind the detected aggregated impact (and that the sum of various effects is consistent with empirical results). Model- based approaches can also estimate localized impacts, a challenge with empirical estimates that tend to rely on aggregated economic variables. Second, model-based approaches can capture threshold effects and “new-to-the-world” impacts that cannot be estimated using historical data series. This is particularly important for channels such as the effect of sea level rise on the built environment, the effect of unprecedented temperature on health and labor productivity, or the possible impact of ecosystem shifts (such as an Amazon dieback) that cannot be captured in historical data series. Third, empirical and model-based approaches are independent ways of estimating future GDP losses due to climate change. Combining them helps build confidence in results (when they agree) or identify knowledge gaps (when results diverge). Finally, model-based approaches allow for policy counterfactuals and assessments, testing the possible impact of various adaptation interventions that cannot be assessed empirically. For all these reasons, we consider the multi-model approach proposed in this paper as an important complement to empirical approaches. To provide a model-based assessment of the impacts of climate change, this paper describes the linking of biophysical climate impacts, as estimated through so-called impact channels,3 with macroeconomic modeling conducted for CCDRs. Impact channels quantify different ways that changes in climate variables influence biophysical metrics of interest. Modeling the economic damages caused by climate change is complex due to the interaction between physical and social systems, as well as the prevalence of uncertainty, from uncertainty 3 Alternatively referred to as the transmission channel. 3 in the Earth’s climate response to future development and mitigation trajectories, to the occurrence of tipping points. With this in mind, a group of approaches known collectively as biophysical models or enumerative studies are well-suited to providing insights on climate impacts under different possible future conditions at the scales and time horizons of CCDRs (to 2050), and are particularly powerful when coupled with macroeconomic models. Biophysical models are process-based models that translate projected changes in temperature, precipitation, or sea-level rise (among other model inputs), to impacts such as changes in crop yields, worker productivity, hydropower production, or flood depths. These biophysical models are typically sector-specific, allowing the inclusion of important considerations tailored to a particular sector. For instance, biophysical models for water include water mass balance models, wetland models, and pollution transport models, to name a few, with each model including different features (such as the existence of water storage and hydropower infrastructure), and enabling the evaluation of different management decisions (such as alternative reservoir operating rules). Similarly, models focused on nature can consider entire biomes, evaluate species biodiversity, or simulate fish populations under different conditions. Individual biophysical models can be linked, allowing trade-offs and cross-sector linkages to be evaluated. These biophysical impacts are in turn used to estimate economic damages, through so-called “hooks” to a macroeconomic model, allowing the discussion of adaptation and resilient investments in a tractable way. Within the context of coupling biophysical and economic models, the remainder of this section provides a brief overview of the set of past studies, reports, and methodologies that this paper builds on, first as they relate to using biophysical models in estimating climate damages, before then exploring how these impacts have subsequently been incorporated into economic modeling efforts conducted to date. 2.1 Estimating biophysical climate impacts Biophysical climate impact modeling has been a topic of research interest for several decades. Studies of bio- physical climate impacts generally take into account projected changes in climate variables such as temperature and precipitation, translate them into impacts on biophysical processes such as water resources availability and agricultural productivity, and then relate these biophysical impacts to changes in economic measures (e.g., con- sumer surplus, imports, exports) as well as welfare measures (e.g., calories per person, food self-sufficiency). Past work has been conducted both at the global level (such as IMAGE Stehfest et al. (2014), a global integrated assessment simulation model) and at more local scales (such as several studies focused on climate change impacts on the Nile River in the Arab Republic of Egypt, including Strzepek and Smith (1995), Yates and Strzepek (1996), Strzepek and Yates (2000)). In parallel, a strand of the literature has focused on how to model the macroeconomic implications of natural disasters and extreme events, with or without reference to climate change (West and Lenze; 1994; Rose et al.; 1997; Rose and Liao; 2005; Hallegatte et al.; 2007; Hallegatte; 2008; Cavallo et al.; 2025). The study on the Economics of Adaptation to Climate Change applied some of these same principles to estimate adaptation costs globally (Margulis et al.; 2010), as well as for Bangladesh (World Bank; 2010b), Bolivia (World Bank; 2010a), Ethiopia (World Bank; 2010c), Ghana (World Bank; 2010d), Mozambique (World Bank; 2010e), Samoa (World Bank; 2010f), and Viet Nam (World Bank; 2010g). This study evaluated, among others, the impacts of climate change on agriculture, fisheries, consumption, health, water availability, and infrastructure, before then using a CGE model to evaluate the subsequent macroeconomic impacts. One framework that has formalized this coupling of biophysical modeling with macroeconomic modeling is the Systematic Analysis for Climate Resilient Development (SACReD) framework. This integrated model- ing framework was developed by the United Nations University World Institute for Development Economics Research and the International Food Policy Research Institute, through their Development Under Climate Change project - see Arndt et al. (2012). The framework uses individual channels of impact to assess the effect that climate change will have on various sectors. According to the International Food Policy Research Institute (2022), key features of this framework include its focus on a particular country or region rather than a global focus; its bottom-up approach to modeling; its ability to integrate diverse channels of impact into a 4 single, unifying framework; its flexibility to integrate different specific sectoral models, as long as each interacts appropriately with the others; and its focus on uncertainty and variability in the context of weather, climate, and climate change. Since its development, the SACReD framework has seen both continued academic research seeking to improve methodological components, as well as applications of the framework in numerous countries around the world. Key publications related to the framework have looked at Africa generally (Schlosser et al.; 2020), Ethiopia (Kahsay et al.; 2019), Ghana (Twerefou et al.; 2015), Malawi (Arndt et al.; 2014), Mozambique (Manuel et al.; 2020), Malawi, Mozambique, and Zambia (Arndt et al.; 2019), South Africa (Hartley et al.; 2021), Zambia (Ngoma et al.; 2021), the Zambezi Basin (Payet-Burin and Strzepek; 2021), and Viet Nam (Neumann et al.; 2015). 2.2 From biophysical impacts to economic damages The step of translating climate impacts as produced by biophysical models to economic damages has been undertaken in a variety of different ways to date. Similar to the approach taken in many integrated assessment and reduced-form models, early efforts to translate biophysical impacts to economic damages within World Bank projects relied on damage functions. Such damage functions are derived by taking the output of an individual biophysical model and fitting a functional form to the output, thereby relating changes in climate variables to estimated aggregated impacts. For instance, a damage function for climate impacts on crops could relate changes in future temperature and precipitation to changes in crop production. The damage functions derived in this way are subsequently used as input to the relevant macroeconomic model — see the CCDR previously completed for South Africa (World Bank; 2022b) as an example. However, damage functions that estimate aggregated impacts have limitations as climate impacts are typically local in scale. National-level damage functions that use annual average inputs tend to under- estimate local impacts and miss impacts that happen at shorter timescales. For instance, damage functions that rely on average annual changes in precipitation and temperature do not capture the effects of increased numbers of days above a certain maximum temperature threshold, and do not capture the impacts of changing precipitation seasonality or the more frequent occurrence of droughts and floods. Hence, more recent work by the World Bank has seen the transition from using damage functions to translate biophysical impacts to economic damages, to using so-called vectors of shocks. Shocks describe the incremental impact caused by climate change in a given time period and capture local as well as shorter-term effects. For instance, a vector of shocks for crop production impacts under climate change is a time series showing the estimated changes in crop yield per month, as caused by projected changes in temperature and precipitation. While the vector of shocks is typically aggregated to show national-level impacts, it is based on spatially disaggregated modeling that estimates the impacts of changing climate variables on crops within for instance half-degree grid cells. In this way, vectors of shocks are able to capture local, spatially disaggregated and shorter-term impacts in a way that damage functions typically are not. This approach relying on vectors of shocks has been used across more than 30 more recent CCDRs, including those developed for the Democratic Republic of Congo, Zimbabwe, Croatia, the Republic of Yemen, and Nigeria. 3 Methodological Framework Broadly, the process of evaluating different impact channels for a country is made up of the steps shown in Figure 1. First, the modeling team selects representative sets of future scenarios of interest, covering a wide range of possible development, climate, and adaptation policy futures. Biophysical models are then selected, tailored, and/or developed to convert changes in climate variables into biophysical impacts through individual channels of impact. The climate scenarios chosen serve as input to these biophysical models to estimate the impact of climate change on biophysical metrics of interest. These biophysical impacts in turn become input to country-specific macroeconomic and/or poverty models to determine the impacts of climate change relative to a counterfactual 5 reference scenario. Biophysical impacts then serve as input to the economic model, with these same steps repeated to explore the differences in economic outcomes under different climate, development, and adaptation scenarios. Estimated impacts under a constant-climate scenario can be compared with a scenario without impacts to provide insights into current levels of risk. For instance, one can estimate the average reduction in GDP due to tropical cyclone damages in Madagascar, in the current climate. However, such comparison needs to be made carefully to avoid double counting: for instance, the current impact of high temperatures on labor productivity is already included in labor productivity and other economic data used to calibrate macroeconomic models. Estimated impacts under a climate-change scenario can also be compared with the constant-climate sce- nario, to quantify the incremental impacts resulting from climate change. The right reference scenario depends on the policy under investigation (Jafino et al.; 2021). For instance, the design of a policy to reduce the macroeconomic impacts of tropical storms does not need to depend on whether climate change makes tropical storms more likely or more intense. Figure 1: Modeling approach 3.1 Economic development scenarios The future is inherently uncertain. Rather than trying to predict future socioeconomic conditions, our ap- proach considers diverse future scenarios to explore how economic outcomes differ given different possible future conditions. As such, the biophysical modeling conducted for each impact channel considers different development scenarios defined for the country of interest by the relevant country experts. In general, these scenarios consider (1) a business-as-usual trajectory, in which the country continues its current growth, development and/or climate change mitigation and adaptation trends and (2) one or more aspirational de- velopment trajectories in which reforms are undertaken to accelerate development, in the form of faster economic growth, better access to infrastructure, financial and social services, or stronger safety nets and in- surance options. These aspirational scenarios can be informed by national development targets and climate action plans, and are based on countries’ official development goals and priorities. These scenarios generally span a horizon of 30 to 40 years (i.e., the period from 2021 to 2050 or 2060), to match the horizon of Country Climate and Development Reports. The variables considered under each of these development scenarios vary by country but can generally include gross domestic product (GDP) per capita, population growth, urbanization levels, capital and labor expansion, and access to energy, water, and other basic services. Within these different development scenarios, different degrees of investment in climate adaptation interventions may be pursued, with these adaptation interventions seeking to reduce the negative effects of climate change experienced through a particular channel. These different development scenarios are 6 incorporated into each impact channel by modifying assumptions on socioeconomic and development-related variables. Moreover, these scenarios may consider macro-structural changes in the relative size of certain sectors of the economy as certain activities develop over time, which can then impact the overall effect of climate change on the total economy. For instance, many countries are expected to experience a decreasing relative contribution to GDP from agriculture due to the expansion of industry and services, making the economy less vulnerable to agricultural shocks but more susceptible to impacts on other sectors. 3.2 Climate scenarios The inclusion of multiple climate scenarios in the analysis is critical to account for uncertainty in both future greenhouse gas emissions trajectories, as well as uncertainty across different climate model projections. Figure 2 (from Stocker et al. (2013)) shows the relative magnitude of these different sources of uncertainty over time, as compared to the range of natural variability. Climate scenarios are available from the World Bank’s Climate Figure 2: Relative importance of different sources of uncertainty (interannual variability, emissions uncertainty, and climate model uncertainty), and their evolution over time Change Knowledge Portal for 29 General Circulation Models (GCMs) from the Coupled Model Intercomparison Project 6 (CMIP6) suite of models. Each GCM has up to five combinations of SSP and Representative Concentration Pathway (RCP) emissions scenario runs available. For each SSP-RCP combination, a modeled history from 1995 to 2014 and projections from 2015 to 2100 are available, for monthly mean temperature and precipitation at a 1x1 degree grid resolution. Given that GCMs are biased relative to observed climate conditions, we apply a bias-correction and spatial disaggregation technique to disaggregate the projections to 0.5 x 0.5-degree grid cells, and then bias-correct these projections using the observed historical dataset from 1995 to 2000 from the Climatic Research Unit gridded Time Series of the University of East Anglia (CRU TS 4.05). Where climate data is available from local weather stations, these data can be used to validate the CRU data, although differences between station 7 data and CRU data may occur as a result of new stations coming online, flaws in existing local data, or station data representing point values as compared to the gridded average CRU values. The resulting bias-corrected and spatially disaggregated projections are then interpolated from monthly variables to a daily timestep. From among the full set of CMIP6 outputs described above, we select an optimistic and a pessimistic greenhouse gas emission scenario, that allows one to assess the impact of uncertain global greenhouse gas emissions trajectories and climate mitigation efforts. Because the uncertainty in future climate conditions is also driven by the choice of climate model, an additional set of eight climate scenarios is then selected to capture the broadest range of climate change effects across GCMs. This allows exploration of the vulnerability of the economy (and the performance of adaptation options) under different possible scenarios. Typically the range of uncertainty in precipitation projections is much more significant than for temperature projections, and capturing the uncertainty on future precipitation is particularly important. Of these eight scenarios, three are often chosen to represent future dry/hot conditions and three represent wet/warm conditions. To reduce the dependency on single models, the final two scenarios are usually comprised of the mean of these three dry/hot and wet/warm scenarios respectively. Similar to the development scenarios described above, the length of the selected climate scenarios is scaled to match the 30 or 40-year investment horizon used in the CCDRs. The climate shocks experienced under each climate scenario are compared to a climate baseline that is consistent with conditions experienced from 1995 to 2020. 3.3 Biophysical impact channels As introduced above, this approach aims to estimate economic damages from climate change to a country’s economy, with these estimates built around different “impact channels” through which climate change will result in shocks to the country’s economy. In total, 15 different channels of impact have been modeled to date, producing climate shocks to natural capital, agriculture and natural resources, and infrastructure and services. Each individual impact channel relies on stylized biophysical models that take climate information and projections as inputs, and simulate changes in biophysical (e.g., streamflow or infrastructure conditions) and/or socioeconomic (e.g., labor supply hours) variables under these future climate conditions. The biophysical models are typically customized to the country context by using country-specific inputs, obtaining key assumptions from country experts and available literature, and calibrating outputs using lo- cal data. Where locally collected data are not available, global data sources are used. Different possible development futures are included by modifying model inputs and assumptions. The remainder of this section presents a summary of the methodology used to model each impact channel. We do not include extensive methodological detail for each channel here (beyond references to relevant method- ological sources) as the focus of this paper is to explain and demonstrate the overall biophysical-macroeconomic framework, rather than provide recommendations on individual channel methodologies. Ultimately, the choice of which channels to evaluate for a particular country and which analytical tools to rely on for each channel are determined by country experts based on their knowledge of which channels are most relevant for a particular country’s situation as well as the state of existing analytical tools and methods. 3.3.1 Human capital Climate change may reduce human capital through increases in extreme temperatures that result in excess mortality and reduced labor capacity, as well as by facilitating the spread of infectious diseases as larger areas experience climatic conditions favorable for diseases, which in turn cause excess mortality and increased disease incidence in the population. We estimate these effects through a labor heat stress channel which models changes in the ability of labor to perform work as workday temperatures increase in the future, as well as a human health channel which models changes in the incidence and mortality of vector-borne (malaria and dengue), water-borne (i.e., cholera, dysentery, etc.), and heat-related diseases as local temperatures and precipitation levels and patterns change in the future. 8 In addition to impacts from climate change, both the coverage and the quality of basic services can play a significant role in the overall prevalence of certain diseases in the population. While these changes respond to policy decisions rather than to changes in climate, their effect is intertwined with human health effects and can mitigate the effects of climate change or generate additional co-benefits. To this end, we consider a water, sanitation, and hygiene (WASH) channel which models the effects of improved WASH coverage on the incidence of diarrheal diseases. We also include a clean cooking channel which models the effects of improved cooking services on indoor air pollution and associated illness. All of these channels impact labor. Heat stress reduces the productivity of labor by reducing the effective number of hours a person can perform work, while human health and the associated policy channels impact total labor supply by way of changes in the number of deaths and the incidence of diseases. To calculate these effects, we model the total labor hours in the country (see Appendix B for more detail on the labor supply model used), which we then shock through each impact channel, as described in Table 1. Impact Description of impacts evaluated Modeling approach Channel Labor Labor productivity impacts from General approach follows that developed by heat daily heat stress to both indoor and International Labour Organization (2019). stress outdoor workers. For outdoor workers, Moran et al. (2003) is applied, plus a bias correction factor as de- rived by Kong and Huber (2022) from Lilje- gren et al. (2008). For indoor workers, Kjell- strom et al. (2018) is applied, building on Bernard (1999) wet bulb globe temperature formulation. Both impacts to informal and formal labor can be captured, depending on the labor sector data available for use as model input. Human Labor supply impacts from changes Vector-borne diseases: methodology described health in the incidence and mortality of by Ebi et al. (2005), using fuzzy functions vector-borne (malaria and dengue), following Craig et al. (1999). Water-borne water-borne (i.e., diarrheal), and diseases: methodology described by World temperature-related diseases. Health Organization (2014). Temperature- related diseases: methodologies of Romanello et al. (2021) and Honda et al. (2014). Water, Labor supply impacts from changes Methodology developed by Wolf et al. (2019). sanita- in diarrheal incidence and mortality The outputs of this channel can serve as input tion, and due to investments in improved wa- to the human health channel, in terms of esti- hygiene ter supply and sanitation coverage. mating changes to diarrheal disease incidence and mortality under different climate futures. Clean Labor supply impacts from indoor Methodology used in the Clean Cooking Plan- cooking air pollution, including the effect of ning Tool, based on the Energy Sector Man- changes in cooking services and co- agement Assistance Program methodology for benefits in fuelwood use reduction. The State of Access to Modern Energy Cook- ing Services work (Energy Sector Management Assistance Program; 2020). Table 1: Overview of human capital impact channels evaluated 3.3.2 Agriculture and natural resources Natural resources are expected to experience a variety of impacts from climate change. Changes in precipitation patterns can result in reduced water resources for rain-fed agriculture, hydropower generation, and other uses, as well as impact erosion levels that can result in additional downstream effects. Temperature increases are likely to reduce the suitability and productivity of crops, pastures, and livestock, and can have additional 9 impacts on overall water resources availability. We estimate these effects through five different channels. The water supply channel models changes in the availability of water resources for water-dependent sectors of the economy and translates any unmet demand for water into the cost of providing this water from an alternative source such as desalination. The hydropower channel models changes in hydropower generation due to changes in river runoff from altered precipitation regimes. The crop production channel models changes in crop productivity as a function of the availability of rainfall and irrigation water, as well as heat stress effects from increasing temperatures. The erosion channel models changes in soil conditions and topsoil erosion from altered precipitation, and the resulting impact this has on crop productivity. Finally, the livestock channel models changes in the production of main commodities (milk, meat, eggs) due to altered heat stress in animals, as well as from altered feed source availability due to changes in temperature and precipitation. The water supply, hydropower, and crop production channels are typically modeled jointly through a water-agriculture-energy nexus model (such as a WEAP model Sieber (2024)) that evaluates water resources availability for competing water-dependent sectors by conducting a mass balance assessment of available water supply compared to water demands from different sectors, thereby estimating unmet demands. The various sectoral demands for water are tailored to the local context and can include water for municipal use, industry, irrigation, aquaculture, livestock, hydropower, thermal cooling, and/or the environment. While most CCDRs conducted to date have assumed static land use conditions, land use modeling to capture changes in land use and land cover can be conducted, supplying inputs to several channels presented below in the form of, for instance changes in erosion, and changes in sediment loading in water bodies. While the final shocks for each channel are evaluated individually, this joint modeling allows consideration of linkages and trade-offs between individual channels. Table 2 summarizes the approach used for each individual impact channel. Impact Description of impacts evaluated Modeling approach Channel Water sup- Productivity impacts in various Water-agriculture nexus model evaluates un- ply water-dependent sectors. met water demands. Hydropower Impacts to energy generation re- Water-energy systems model estimates sulting from changes in river changes in water available for hydropower. If runoff. outputs from a land use model are available, impacts of increased sediment loading in reservoirs under climate change can also be evaluated. Crop pro- Impacts to crop revenues Uses the Food and Agriculture Organization’s duction through changes in yields. yield response functions to estimate the im- pact of altered water availability (Carr et al.; 2013). Heat stress impacts to yield are esti- mated as per Salman et al. (2021). Erosion Impacts to crop revenues due to Uses the Revised Universal Soil Loss Equa- topsoil erosion and flooding from tion (Renard et al.; 1996), with erosion esti- altered vegetation conditions. mates converted into crop yield losses follow- ing Kassam et al. (1991). Model considers on- farm impacts and does not include externali- ties caused by soil loss. Livestock Impacts to livestock revenues Milk production losses from heat stress were through changes in animal pro- estimated following Mauger et al. (2015), ductivity due to changing heat while meat and egg production losses were stress and feed availability. estimated following St-Pierre et al. (2003). The impact of changing feed availability used methods by Carr et al. (2013) and Salman et al. (2021) to calculate a Feed Conversion Efficiency parameter. Table 2: Overview of agriculture and natural resources impact channels evaluated 10 3.3.3 Infrastructure and services Climate change is likely to impact infrastructure and the services provided by it, by increasing the frequency and magnitude of extreme events that result in damages to assets, as well as by increasing deterioration caused by heat and precipitation levels. We model these effects through six different channels. The inland flooding channel models damages to capital across the country from changes in the magnitude and frequency of fluvial flooding, while the urban flooding channel models damages to urban infrastructure from changes in the frequency and magnitude of pluvial flooding. The sea-level rise and storm surge channel models impacts to coastal capital from increases in mean sea level and storm surge events. The tropical cyclones channel models damages to capital from changes in the recurrence and magnitude of tropical cyclones. Roads and bridges are covered in a separate channel which models increases in the repair and maintenance to road infrastructure due to higher temperatures, precipitation, and flooding recurrence, as well as road disruption effects on labor. Finally, the tourism channel models impacts to tourism revenues due to altered tourism demand for a particular destination as a result of changes in temperature. The estimation of damages through each individual infrastructure channel relies on an infrastructure and capital stock model. The exposure and vulnerability of assets were determined by three key variables: the geospatial location, the type or sector of the asset, and the total value of the asset. Details of this capital stock model are provided in Appendix B. Table 3 summarizes the approach used for each individual impact channel. 3.4 Evaluation of adaptation investments The modeling sequence described above is not only used to explore the possible impacts of climate change on a country’s economic performance, but also to evaluate the reduction in impacts offered by different adaptation investments. This is done by comparing a baseline that assumes a continuation of the current economic, environment, adaptation and resilience policies, with an adaptation scenario that assumes enhanced emphasis on proactive adaptation investments. It is important to flag that even under the baseline, some adaptation is assumed to occur as people and firms will always do something to adapt when they begin to experience negative impacts.4 Generally, greater levels of adaptation can achieve larger reductions in the cost of climate change, but with decreasing marginal benefits for incremental expenses in adaptation (see the left panel in Figure 3, taken from the Intergovernmental Panel on Climate Change’s Economics of Adaptation). In practice, full adaptation to the impacts of climate change is not possible due to technological limitations and other implementation barriers, leaving unavoidable residual costs (as seen in the right panel of Figure 3). Moreover, the returns on adaptation investments are subject to uncertainties regarding future climate conditions. The modeled adaptation interventions are defined in consultation with sector teams at the World Bank, based on knowledge of the country, modeling feasibility, and the degree of expected climate change impacts. A few illustrative examples of adaptation interventions modeled through impact channels for CCDRs to date are presented here, with further details documented in Appendix A. For the labor heat stress channel for instance, adaptations focused on increased air conditioning coverage as well as the shifting of working hours to cooler times of day have been explored. Further adaptation interventions that could be evaluated in future rounds of work could include increased use of passive cooling techniques or the impact of efforts to reduce urban heat island effects. Adaptation costs are estimated based on unit costs obtained from international and, where available, local sources. Adaptation costs can be divided into two categories: upfront investment costs (CAPEX) and operational costs (OPEX) of the measures. The benefits of these measures are calculated based on the reduction in damage shocks for the respective channel. The benefits of different sectoral adaptation interventions are estimated through the impact channel modeling framework, comparing climate change impacts under the baseline with impacts under a scenario with enhanced adaptation. 4 For this reason, we do not refer to the baseline scenario as a ”no adaptation” scenario. 11 Impact Description of impacts evaluated Modeling approach Channel Inland flood- Impacts to capital from changes Models runoff using the TR-20 model devel- ing in the recurrence of peak precip- oped by the United States Department of itation events that result in flu- Agriculture. Estimates floodplain boundaries vial flooding. using Lehner and Grill (2013). River flows use the Muskingum-Cunge method of flood routing (Ponce; 2020). Damages estimated us- ing depth-damage curves from Huizinga et al. (2017). Urban flood- Impacts to urban capital from Drainage and flood depths estimated through ing peak precipitation events that re- the Itzi model (Courty et al.; 2017), with sult in pluvial flooding. depth-damage curves from Huizinga et al. (2017) used for damage estimation. Sea level rise Impacts to coastal capital from Uses sea-level rise projections from Garner and storm changes in mean sea level and et al. (2021) with inundation estimated using a surge storm surge. bathtub approach, with depth-damage curves from Huizinga et al. (2017) for damage esti- mation. Tropical cy- Impacts to capital from changes Modeled by assessing the impact of thousands clones in tropical storm recurrence in- of synthetically generated tropical storm tervals. tracks as per Emanuel et al. (2006) and Emanuel et al. (2008). Roads and Impacts to transportation capi- Uses the Infrastructure Planning Support Sys- bridges tal due to damages to and in- tem model (Chinowsky and Helman; 2021). creased maintenance of roads Model modifications could be made to esti- and bridges, as well as labor sup- mate secondary impacts from transport net- ply effects of road disruptions. work damages, such as trade disruptions, busi- ness interruptions, lack of access to healthcare facilities etc. Tourism Impacts to tourism sector rev- Impacts on leisure revenues due to changes in enues due to changes in climate average climatic conditions estimated follow- variables, which produce changes ing Maddison et al. (2005). in tourism potential. Table 3: Overview of infrastructure and services impact channels evaluated 3.5 Macroeconomic and/or poverty modeling The individual climate shocks developed through each of the biophysical channels described above are ulti- mately combined to simulate the overall economic impacts of climate change on a country. The overall impact of climate change is simulated using a country-specific application of the World Bank’s macro-fiscal model (MFMod, Burns et al. (2019)). While in practice the MFMod model is often linked with more specialized macroeconomic and sectoral models, we limit the description here to the climate application of MFMOD. Note that many of the CCDRs and the links described in this paper are coupled with the World Bank’s dynamic CGE model (MANAGE).5 MFMod is a structural macroeconometric modeling framework (with a global system that contains more than 5,000 estimated equations) designed to support macroeconomic forecasting and policy analysis. The baseline model structure covers a core block of key macroeconomic variables (including GDP, consumption, in- vestment, government spending, exports and imports, wages, exchange rates, and monetary and fiscal policies), as well as modules for major real, fiscal, monetary, external, and financial sector variables. MFMod incorpo- rates expectations and policy reactions through error correction mechanisms, and its semi-structural design 5 See https://thedocs.worldbank.org/en/doc/77351105a334213c64122e44c2efe523-0500072021/related/ MANAGE-WB-Documentation.pdf. 12 Figure 3: Adaptation costs and residual costs of climate change (Source: Chambwera et al. (2014)). enables the model to be extended as needed for customized applications. For a full description of MFMod, including its core equations, key features, and current applications, see Burns et al. (2019). MFMod is updated bi-annually to incorporate the latest economic and policy developments, as well as the most recent macroeconomic and fiscal statistics. To this end, the model captures the transmission channels through which climate shocks impact key economic variables. MFMod is a general equilibrium model, but it differs from static computable general equilibrium (CGE) frameworks. It is similar to macroeconometric models used in institutions such as the Federal Reserve Bank of the US (FRBUS). There is a distinct steady state, where the long run growth path is determined by total factor productivity growth and population growth. The equilibrium transition back to the steady state equilibrium is determined empirically. The model allows for short and long term frictions such as wage and price markups, different tax and subsidies. Thus, while it shares a neoclassical long run feature with CGE models, the short run can show different dynamics. MFMod can be used to explore the long-term economic impact of climate change, but also the short- and medium-term responses to climate shocks (see an example in Hallegatte, Jooste and McIsaac (2024)). The model accounts for a broad range of fiscal and balance of payments variables. A shortcoming of the model is its limited sectoral disaggregation. CGE models account for a large number of sectoral variations to shocks, while MFMod is primarily focusing on aggregate sectors (agriculture, industry, which is split between energy and non-energy, and services). The vector of shocks is effectively expressed as percentage deviations from a reference point (e.g., labor productivity, or crop production in the baseline with no further climate change). Figure 4 is a country example that summarizes the percentage change in labor productivity for agriculture (AGR), industry (IND), and services (SRV) for a dry and wet climate scenario under two different climate scenarios. In the dry world (SSP3-7.0) agriculture labor productivity falls by roughly 12% in 2090 compared to labor productivity levels in the baseline. This shock is much smaller in a wet scenario. The biophysical models produce the inputs of different climate shocks that then need to be mapped to the appropriate economic channels, which can be summarized as: • Supply-side effects: Biophysical climate shocks impact the production of goods and services in the 13 Figure 4: Example of heat shock on productivity by sector economy, either by directly influencing sector outputs (e.g., reductions in crop yields) or by altering factor inputs (e.g., reductions in labor supply due to increased disease incidence). Some assumptions are made regarding aggregation given that we cannot account for each type of capital and each type of labor. As an example, losses in crop production are entered as a productivity shock that shifts the production function. In aggregate terms, the loss in crop production is modeled as a decrease in aggregate total factor productivity (TFP) by approximately the weight of crops in total gross value-added multiply by the shock.6 In terms of labor productivity, the shocks are weighted by the share of each sector’s employment in total employment. Labor productivity shocks directly enter the production function by scaling structural employment. Damages to physical capital are translated into output losses through a modified production function that assumes that the lost capital has a productivity equal to the average (not the marginal) productivity of capital. See Appendix B for details. • Demand-side effects: Climate shocks are primarily supply side shocks (e.g., TFP changes, capital and labor). However, the model accounts for demand responses. The overall economic impacts depend on the sensitivity of household incomes and consumption for example. A decrease in TFP will lower wages that are offered to household and subsequently reduce incomes. Households in our economies can also smooth consumption over time - consumption smoothing is an empirical outcome based on estimated elasticities of habits. Changes in government spending are discretionary and may be used as a policy lever to generate ex-ante or ex-post policies in response to expected climate change or when unknown shocks materialize (e.g., higher spending on health or disaster relief). The supply-side shocks are important for understanding long term changes in the economy. However, supply shocks also have short term implications as they shift prices that consumers and firms face. The overall response to the economy depends on the estimated parameters of the system (e.g., if different types of capital are complements or substitutes, the sensitivity of firms and households to interest rates etc.). Table 4 provides an example of how the various impact channels affect the key variables in MFMod, either through labor, sub-sectoral productivity, capital stocks, or exports. Different applications in different countries consider different impact channels with different connections with the macroeconomic framework. To estimate the overall economic impact of climate change, we simulate the model over a 30 to 40-year horizon under different climate scenarios. The shocks from the biophysical models are introduced into MFMod as changes in key economic variables, such as productivity, investment, and government spending. The model then projects the evolution of the economy under these climate shocks, capturing the dynamic interactions between supply, demand, and price effects. 6 This integration implicitly assumes that Hulten’s theorem holds in the context of a Cobb-Douglas production net- work. Moving from Cobb-Douglas to CES production network, Baqaee and Farhi (2019) developed a framework in which the input-output matrix responds endogenously to shocks, with the resulting nonlinearities shaped by the microeconomic details of the production structure. Future research could explore the introduction of these nonlinear features into the framework presented in this paper. 14 Table 4: Impact channels and their effect on the macroeconomic framework Labor Sub-sectoral Capital Exports productivity stock Human health and development Labor heat stress Human health Water, sanitation and hygiene Clean cooking Sub-sectors Water supply Hydropower Crop production Erosion Livestock Tourism Infrastructure Inland flooding Urban flooding Sea level rise and storm surge Tropical cyclones Roads and Bridges 4 Case Study Results This section summarizes the impacts of climate change for a select group of countries used in the CCDR. The country discussion is summarized in Appendix C. 4.1 Macroeconomic impacts from channel-specific damages A summary of the range of GDP impacts from each channel’s specific damages across different CCDR case studies is shown in Figure 5. The range of GDP impacts within each channel represents the least pessimistic and most pessimistic outcomes for different countries coupled with countries’ different economic development baseline assumptions. On average, climate change is projected to cause negative impacts on GDP across all the considered channels. The magnitude of these effects varies significantly, depending on factors such as the projected climate future, a country’s geographical location, the scale of shocks, and its economic structure. Nevertheless, certain patterns emerge from the case studies, indicating that the most substantial impact channels include heat stress affecting labor productivity due to rising temperatures, cyclone induced damages to capital stock, and damage to roads and bridges. Damage resulting from reduced livestock and rainfed crop yields varies significantly based on the predicted levels of precipitation under different climate scenarios. In the worst case, it can lead to nearly an 8% impact on GDP. The results reflect the fact that agriculture constitutes a significant share of GDP in our five country case studies, and there are a substantial number of outdoor workers who are exposed to the effects of heat. On the other hand, damages to capital stock from flooding and sea level rises are relatively smaller. In our case study, only Madagascar is exposed to tropical storms, and these storms are expected to cause significant losses in the country, reading 2%-3% of GDP damages by 2050. There are a few important caveats to consider when interpreting these results. First, the figure reports expected annual losses, and thus does not capture the concentrated impacts of extreme events. For instance, a flood with a 1 in 200-year probability can occur at any point in the future, but its annual probability is only 0.5% (1/200). Even a major vulnerability for such an event (e.g., a 50 15 Figure 5: Impact of climate change on GDP by 2050, for various impact channels, scenarios, and countries. Source: World Bank CCDR percent loss) would ”only” lead to a 0.25% drop in GDP. In most CCDRs, a separate modeling approach is considered alongside the annual expected loss to capture the vulnerability associated with these extreme events (Hallegatte, Jooste and McIsaac; 2024). Another challenge with extreme events is that — in most countries — they only affect a fraction of the population, making the aggregate GDP impact a poor proxy for welfare losses, especially in large countries. Second, these losses are for 2050, and may not fully capture the long-term effects of climate change (see Figure 2). For instance, sea-level rise is expected to occur over the long term, with growing impacts even past 2100. Third, and maybe most importantly, the model-based approach captures only the impact channels that are explicitly modeled, leaving very important risks unquantified. This is the case, for instance, of the link between climate change and violence and conflict, which has been identified as a key driver of losses in historical data series (Burke et al.; 2024), but is not represented here. Another example is the possibility of rapid ecosystem shifts, such as in the Amazon, with implications for rainfall and agriculture productivity at the continent scale. These non-modeled risks — some of them considered as low-probability high-impact events — should be considered in the assessment of climate policies. Table 5 provides the detailed results from the case studies. The GDP impacts of climate change are presented based on various economic baseline assumptions, encompassing both pessimistic and less pessimistic climate scenarios. Channel Country Damage Inputs (IEc) GDP impacts (MFMod) Damage to labor productivity (inputs = reduction in labor productivity) Heat stress Guinea-Bissau A: -12.5% (-7.3%) -2.5% (-1.5%) L I: -9.8% (-6.9%) -2.5% (-1.5%) M S: -8.7% (-5.7%) 16 Heat stress Madagascar Constrained Growth (CG): -0.7% (-0.4%) CG A: -4.7% (-2.6%) -0.5% (-0.2%) SR I: -0.7% (-0.4%) S: -1.2% (-0.7%) Structural Reform (SR): A: -2.9% (-1%) I: -0.8% (-0.4%) S: -1.2% (-0.7%) Heat stress Zimbabwe A: -11.0% (-6.6%) -2.3% (-1.2) BAU I: -7.2% (-4.5%) -2.0% (-1.0) ASP S: -7.1% (-4.3%) Heat stress Malawi A: -8.3% (-4.6%) -4.6% (-2.1) BAU I: -5.5% (-2.8%) -4.1% (-1.8) ASP S: -1.3% (-0.6%) Heat stress Mali A: -10.6% (-6.3%) -6.1% (-4.3%) M I: -8.9% (-5.5%) -6.0% (-4.2%) H S: -4.0% (-2.4%) Health Guinea-Bissau -0.37% (-0.35%) -0.5% (-0.5%) L -0.6% (-0.5%) M Health Zimbabwe -0.22% (-0.14%) -0.2% (-0.1%) BAU -0.2% (-0.1%) ASP Health Malawi -0.9% (-0.4%) -1.3% (-0.6%) BAU -1.3% (-0.6%) ASP Health Mali -0.7% (-0.5%) -0.8% (-0.6%) M -0.9% (-0.6%) H Damage to agriculture productivity (inputs = loss in yields) Rainfed crops Guinea-Bissau -5.3% (-2.6%) -0.8% (-0.1%) L -0.7% (-0.1%) M Rainfed crops Madagascar -5.3% (-2.6%) -0.53% (-0.29%) CG -0.47% (-0.26%) SR Rainfed crops Zimbabwe -6.1% (+2.8%) -1.3% (0.9%) BAU -0.8% (0.6%) ASP Rainfed crops Malawi -11.4% (-5.0%): BAU -1.2% (-0.6%) BAU -3.3% (3.8%): ASP -0.2% (0.2%) ASP Rainfed crops Mali -4.4% (3.0%) -0.7% (0.2%) M -0.6% (0.2%) H Erosion (crops) Guinea-Bissau 0.8% (-1.9%) 0.3% (-0.7%) L 0.3% (-0.6%) M Erosion (crops) Madagascar 0.1% (-1.5%) 0.1% (-0.1%) CG 0.1% (-0.1%) SR Erosion (crops) Zimbabwe 0.2% (-1.2%) 0.0% (-0.2%) BAU 0.0% (-0.1%) ASP Livestock Madagascar -15.2% (-5%) -0.4% (-0.1%) CG -0.4% (-0.1%) SR Livestock Zimbabwe -8.1% (-7.9%) -0.2% (-0.1%) BAU -0.1% (-0.1%) ASP Livestock Malawi -7.2% (-1.6%) -0.3% (-0.2%) BAU -0.2% (-0.1%) ASP 17 Livestock Mali -13.0% (+19.0%) -1.8% (2.2%) M -1.5% (1.8%) H Fisheries Madagascar -9.5% (-5.5%) -0.4% (-0.4%) CG -0.4% (-0.4%) SR Damage to the country’s capital stock (Inputs = capital loss) Inland flooding Guinea-Bissau -0.1% (-0.12%) -0.9% (-1.1%) L -0.8% (-1.1%) M Inland flooding Madagascar -0.05% (-0.03%) -0.2% (-0.1%) CG -0.2% (-0.1%) SR Inland flooding Zimbabwe -0.06% (-0.03%) -0.3% (-0.2%) BAU -0.0% (-0.0%) ASP Inland flooding Mali -0.4% (-0.2%) -0.4% (-0.2%) M -0.2% (-0.2%) H Cyclones Madagascar -0.7% (-0.58%) -3% (-2.4%) CG -2.6% (-2.1%) SR SLR Guinea Bissau Repairable: -0.01% (-0.009%) L -0.0085% (-0.007%) -0.1% (-0.1%) M Non-repairable: -0.008% (-0.0037%) SLR Madagascar -0.007% (-0.005%) -0.0% (-0.0%) CG -0.0% (-0.0%) SR Roads and bridges Guinea-Bissau Capital damage: -1.3 (0.4) -0.0% (-0.7%) L (bn 2020 USD, 2021-50 to- -0.0% (-0.6%) M tal) Labor productivity: average 5.6 (-0.2) million hours de- layed. Roads and bridges Zimbabwe Labor productivity: -0.80% -0.3% (-0.1%) BAU (-0.31%), Capital damage -0.0% (-0.1%) ASP (BAU): -3.6 (-3.8): BAU (bn 2020 USD, 2021-50 total), Capital damage (ASP): -5.8 (-8.4) (bn 2020 USD, 2021- 50 total) Roads and bridges Malawi Labor productivity: -0.8% -5.6% (-5.1%) BAU (-0.69%), Capital damage -1.6% (-1.5%) ASP (BAU): -3.0 (-3.8): BAU (bn 2020 USD, 2021-50 total), Capital damage (ASP): -6.8 (-8.4) (bn 2020 USD, 2021- 50 total) 18 Roads and bridges Mali Labor productivity: -0.13% -1.1% (-3.6%) M (-0.09%), Capital damage -1.1% (-3.6%) H (BAU): -8.8 (-36): BAU (bn 2020 USD, 2021-50 total), Capital damage (ASP): -8.8 (-37) (bn 2020 USD, 2021-50 total) -0.68 (-0.45) H Specific sector impacts Tourism Madagascar -13.4% (-8%) reduction in -1.2% (-0.6%) CG tourism revenues -1.1% (-0.6%) SR Tourism Mali -11.3% (-6.3%) reduction in -0.0% (-0.0%) M tourism revenues -0.0% (-0.0%) H Hydropower Zimbabwe -7.0% (6.0%) -0.5% (0.4%) BAU Reduction in generation ca- -0.4% (0.4%) ASP pacity Hydropower Malawi 0.0% (-0.0%): BAU 0.0% (0.0%) BAU -1% (0.3%): ASP 0.0% (0.0%) ASP Reduction in generation ca- pacity Table 5: Note(s): The GDP impact results for the year 2050 are based on a pessimistic climate scenario. To more accurately reflect short-term trends, the damage inputs represent the average impact in the 2040s, as these shocks have a carryover effect on GDP response. The numbers in parentheses represent the results from the least pessimistic climate scenarios. Both the inputs and GDP impacts are differences compared to baselines without further climate change. A = Agriculture, I = Industry, S = Services. Economic Development Baselines: Guinea-Bissau, Mali: L = low growth, M = Medium growth, H = High growth. Madagascar: CG = Constrained Growth, SR = Structural Reform. Malawi, Zimbabwe: BAU = Business as Usual, ASP = Aspirational Growth. Source(s): IEc and MFMod-CC. 4.2 Macroeconomic impacts from climate change (combined channels) un- der the dry/hot and wet/warm futures This subsection presents GDP impacts from combined damage channels considered in each CCDR case study under two consistent climate futures: dry/hot mean and wet/warm mean. Further climate change is projected to result in a reduction of real GDP, ranging from 0.4 to 15.7 percent by 2050, if no additional adaptation interventions are implemented. The extent of GDP damages is influenced by various factors, including climate futures, damage channels, and economic development baselines assumed in each case study. Country GDP impacts in GDP impacts in Channels considered dry/hot mean wet/warm mean Guinea-Bissau -7.7% (L) -6.4% (L) 1,2,3,4,7,9,10 -7.3% (M) -6.0% (M) Madagascar -5.8% (CG) -4.4% (CG) 1,3,4,5,6,7,8,9,11 -5.1% (SR) -3.9% (SR) Zimbabwe -4.9% (BAU) -0.9% (BAU) 1,2,3,4,5,7,10,12 -3.5% (ASP) -0.4% (ASP) 19 Malawi -12.8% (BAU) -8.4% (BAU) 1,2,3,5,10,12 -7.4% (ASP) -3.8% (ASP) Mali -10.7% (M) -6.4% (M) 1,2,3,5,7,10,11 -10.2% (H) -6.7% (H) Table 6: Note(s): 1= heat stress, 2 = health, 3 = rainfed crops, 4 = erosions, 5 = livestock, 6 = fisheries, 7 = inland flooding, 8 = hurricanes, 9 = SLR, 10 = road and bridges, 11 = tourism, 12 = hydro power. Economic Development Baselines: Guinea-Bissau: L = low growth, M = Medium growth, H = High growth. Madagascar: CG = Constrained Growth, SR = Structural Reform. Source(s): MFMod-CC using IEc inputs. The overall GDP impacts, shown in Figure 6 for a larger set of countries covered by CCDRs, reflect both the direct and indirect effects of the initial shocks. The total impact depends on each shock’s magnitude, but also on the weight of the affected sector in the economy and the estimated knock-on effects. For example, heat stress lowers labor productivity more for agricultural workers than for those in industry and services. The overall impact on GDP is therefore lower when agriculture accounts for a relatively small share of GDP, or when agriculture is more capital intensive. This effect is visible in Figure 6, which shows that damages are highest in lower-income countries and those that have a large share of agriculture in GDP. Countries that have diversified towards the industry and services sectors appear less vulnerable to climate change (i.e., lower damages relative to output). Figure 6: Damages and country incomes. Source: World Bank CCDR As already noted, it is important to note that these GDP estimates should be considered conservative, as they do not capture all potential damage channels, or do it imperfectly. The lower vulnerability of higher- income countries may be partly explained by the fact that the modeling capture better the channels affecting low-income countries (e.g., effect of heat on agricultural workers) than those affecting higher-income countries (e.g., infrastructure-related disruptions to complex supply chains). Additionally, these estimates do not account for interactions with non-climate-related crises. To fully un- derstand macroeconomic risks, a study should also assess compounding impacts, which occur when climate and non-climate-related shocks happen simultaneously, leading to greater damage than if each occurred separately 20 (see example Hallegatte et al. (2022); Giuliano et al. (2024)), and cascading impacts, where climate shocks trigger a chain reaction of disruptions across interconnected systems (see example Pollitt and Petrauskaite (2024)). 4.3 Macroeconomic impacts from adaptation investment Investing in adaptation measures can help reduce the extent of climate damages in the future. However, in certain cases, these measures may require significant financial resources to be allocated by countries. It is important to note that due to limited data availability, the benefits of specific adaptation measures and their associated costs can only be quantified for certain channels. Table 7 provides a summary of some adaptation costs and benefits derived from the case studies. Country Channel Adaptation costs (an- Reduced shocks nual average) Guinea-Bissau Heat stress $14m CAPEX, $49.8m From 9.8% loss in labor pro- OPEX ductivity to 3.7% (agricul- ture), 6.1% to 4.6% (Indus- try), and 3.9% to 2.7% (ser- vices). Crop erosion Cost neutral From to 3.0 % loss in crop yields to improvement in yield by 1.4%. Rainfed crop $4.2m CAPEX, $0.4m From to 16.8% loss in crop OPEX yields to 9.8%. Inland flooding Cost neutral From 0.14% of capital stock damage to 0.07%, Road and Bridges $4.2m CAPEX, no OPEX From $176m of damage to cost capital stock and 23.6 mil- lions hours delay to $9.4m and 5.4m hours improve- ment in traffic, Madagascar Heat stress $32.9m CAPEX $144.7m From 7.5% loss in labor pro- OPEX ductivity to -4.3% (agricul- ture), 1.3% to 0.7% (Indus- try), and 2% to 1.4% (ser- vices). Rainfed crops $5.7m CAPEX, $1.4m From 9.1% loss in crop yields OPEX to 6.2%. Livestock $0.06m CAPEX, $6.3m From 15.2% loss in livestock OPEX yield to 9.2%. Cyclone 0.15% of capital stock From 0.7% of capital stock CAPEX, No OPEX damage to 0.2% Zimbabwe Heat stress $31.6m CAPEX and OPEX From 6.0% loss in labor pro- ductivity to 0.8% (agricul- ture), 2.3% to 1.7% (Indus- try), and 1.9% to 1.3% (ser- vices). 21 Road and Bridges $95m CAPEX, No OPEX From $24m of damage to capital stock and 9.6 mil- lions hours delay to $6.2m and 1.3m hours improve- ment in traffic. Malawi Livestock $20.7m CAPEX, No OPEX From -50% loss in livestock yield to -44%. Rainfed crops BAU: $68m, ASP:$14m BAU: from -11.7% loss to CAPEX, No OPEX +23% gain. ASP: from - 3.2% loss to 24.6% gain Hydropower $1.7m CAPEX, No OPEX BAU: generation capacity increases from 0.0% to 1.9%, ASP: generation capacity in- creases from -0.7% to 1.3%. Road and bridges BAU: $72.5m, ASP:$147m Lost labor hours: gains from CAPEX, No OPEX -0.9% to -0.4% Destroyed capital: from a cumula- tive loss of $3 to $1.7 2021 USD billions (BAU), and $6.8 to $1.5 2021 USD bil- lions (ASP). Mali Rainfed crops $130m CAPEX, $35.3m Crop loss reduces from -3% OPEX to a gain of +7.5%. Livestock $96.8m CAPEX, No OPEX Livestock loss reduces from -31.8% to -20% for the most impacted year. Road and bridges $42.6m CAPEX, No OPEX Lost labor hours: gains from -0.15% to -0.08%. Destroyed capital: from a cumulative loss of $8.8 to $4.4 2021 USD billions (BAU). Table 7: Note(s): 1: CAPEX = capital cost, OPEX = operating expenses. All numbers are in 2021 real USD unless otherwise stated. 2: Reduced shocks are based on damages in 2050 under each channel’s specific most pessimistic scenario. Results are based on impacts by 2050. Source(s): IEc. The economic impacts of adaptation are modeled by considering two key inputs: the costs associated with implementing adaptation measures and the resulting reduction in damages. The funding for adaptation investments can vary depending on the assumptions made for the specific CCDR studies. The standard approach is to assume that any additional investment in adaptation measures (CAPEX) is financed through an increase in public debt. This additional investment in adaptation measures is not considered productive and does not contribute to a country’s capital stock formation. For example, if an adaptation investment is made to strengthen the foundation of roads and bridges to withstand future flooding, it will still result in roads and bridges that can accommodate the same number of vehicles as those without the adaptation measures. There are other assumptions of adaptation investment funding that can be considered, such as fiscal reforms to fund additional investment in adaptation, or the role of international climate finance from donor countries. Spending on adaptation operation costs (OPEX), such as higher electricity bills for air conditioning units or increased livestock feed costs, are assumed to lead to higher prices for households. As a result, the overall 22 benefits of adaptation at the macro level may be reduced due to these additional adaptation investment and operational costs. The modeling results show that benefits of stepped-up adaptation outweigh the costs in all cases because the GDP impacts after adaptation are less negative than in the BAU with climate damage case (see Figure 7). Despite this, it is important to consider alternative funding sources for these expenses. Financing adaptation through public debt can exacerbate the debt-to-GDP level (e.g., by 2050, Mali’s debt-to-GDP ratio is projected to increase by 14.6 points under the hot/dry scenario and by 30 points under the wet/warm scenario, compared to a scenario without adaptation measures), especially considering the existing debt distress in many countries. Note that debt is a stock in the model, so it cannot be directly compared with GDP (which is a flow variable). Another related dynamic at play is that the inflation resulting from damages without adaptation tends to be higher than with adaptation. This higher inflation reaction, in turn, decreases debt-to-GDP levels through its inflationary effect. Figure 7: Benefits from adaptation interventions recommended in the CCDRs. Note: In countries with net gains from adaptation, the benefits of the interventions are higher than the considered climate and disasters impacts, either because they reduce disaster losses below the baseline, or because they generate other development gains. Source: World Bank CCDR 5 Discussion and Recommendations The integration of biophysical and macroeconomic models provides a comprehensive approach to assessing the impacts of climate change on economic outcomes and evaluating the impact of different possible adapta- tion strategies, complementing the insights from empirical studies. The case studies presented in this paper demonstrate the application of this framework to different country contexts, highlighting the importance of considering multiple impact channels and diverse adaptation options. 5.1 Policy implications Results include a few key findings with direct policy implications, based on robust results from the analyses. Protect workers from heat stress and preserve labor productivity: The impact of heat stress on labor productivity is consistently the most significant impact across case studies (keeping in mind that not all possible risks are included in the analysis). Reducing labor exposure to future heat can be achieved through 23 various measures, with the most important driver being structural change, with workers transitioning from low-probability farming (with a lot of physical labor) to more mechanized agriculture or toward manufacturing and services. And because workers in manufacturing and services are also affected by heat, expanding access to cooling technologies — including passive options and air conditioning — is also essential, especially in hot climates. Address large tail events with a layered approach: The short-term impact of extreme events (like severe floods) is not fully captured in our results, as the analysis focuses on annualized damage. Given the unpredictability and scale of losses from such events, they require anticipated planning, using metrics going beyond average annual losses. For instance, insurance products can be appropriate to manage risks with low probability and large losses, while prevention makes more sense for frequent events with small losses. It is recommended to use a ”layered approach”, in which hazards with different probabilities of occurrence are managed using different tools, in the context of an integrated risk financing strategy.7 Identifying adaptation strategies for unaddressed or highly uncertain damage channels: Some damage channels lack clear adaptation measures in the current analysis, or are missing from the assessment. For example, tourism –— a critical sector for some countries —– faces significant climate-related risks, yet effective adaptation strategies are not well established. Efforts should focus on identifying and implementing measures to mitigate climate change-induced losses in these sectors. In other sectors, deep uncertainty is making it harder to estimate likely impacts and design adaptation strategies, but it does not mean that nothing should be done. From data collection and research to the implementation of no-regret interventions, governments can start preparing even for low-probability outcomes. What-if analyses can be used to explore these scenarios, identify vulnerable sectors, and design no-regret strategies. Adapt to climate change through resilient development Overall, managing the impact of climate change often relies more on resilient development strategies than on targeted adaptation interventions (World Bank; 2024b). For instance, reducing a population’s vulnerability to high temperatures will require structural change (toward mechanized agriculture and manufacturing and services), not only specific heat management interventions. And deploying modern cooling technologies require higher quality building and widespread access to electricity, which in many countries remains a development challenge. And is important to note that there is no scenario without significant adaptation needs: Even in an optimistic scenario with a wetter and warmer climate rather than a hotter and drier one, significant damages are still expected, and many economic activities and infrastructure systems will have to be adapted or upgraded. 5.2 Knowledge and methodological gaps This paper also highlights important knowledge and methodological gaps, with a set of key recommendations. Strengthen data collection and modeling capabilities, in academia and governments. Enhanc- ing data collection and modeling methodologies is essential to improving the accuracy and reliability of climate impact assessments. Strengthened capabilities will enable more robust projections and support evidence-based policy decisions that better address climate risks. The work of the Coalition of Finance Ministers for Climate Action8 is an example of initiatives to promote capacity building within governments. Capture cascading climate impacts: The impact channels analyzed in this study do not fully account for the cascading effects of climate change on macroeconomic damages. While factors such as labor heat stress, human health, water, sanitation, hygiene, and clean cooking are considered, they do not encompass all pathways through which climate change affects human capital. While the effects of heat on learning outcomes are well known (Park et al.; 2021), there is no global assessment of their magnitude. Similarly, the effect of infrastructure disruption on the rest of the economy is well documented (Rentschler et al.; 2021), but its modeling remains a topic of research (see an example on the impact of transport disruptions in a firm-level economic model in Colon et al. (2021)). Future modeling efforts should expand impact channels to incorporate cascading effects across sectors and systems for a more comprehensive assessment of climate-induced damages. And current 7 https://www.globalshieldfinancingfacility.org/sites/default/files/2020-04/DRF%20Primer_5.pdf 8 https://www.financeministersforclimate.org/ 24 estimations do not sufficiently capture interactions between different impact channels. For example, the labor supply effects of infrastructure damage from floods and cyclones are not directly incorporated into damage calculations. A more integrated modeling approach is needed to better represent these interconnections and their broader economic consequences. Quantify compounded, complex, and systemic risks: While integrating climate and non-climate shocks is feasible, see Ranger et al. (2021), it is often overlooked in modeling exercises. Expanding impact assessments to include non-climate shocks, such as economic downturns or geopolitical instability, would provide a more holistic understanding of macroeconomic risks. However, certain effects—such as ecological shifts and social disruptions—are inherently more challenging to quantify due to their complex, nonlinear nature. Future research should focus on advancing methodologies to incorporate these difficult-to-measure risks, ensuring a more comprehensive assessment of climate change’s long-term consequences. Expand to global-scale assessments: The current analysis is conducted at the country level, which may underestimate the broader climate damages. Expanding the modeling framework to a global scale could anzig; 2024), as climate impacts transcend national borders. yield more comprehensive estimates (Bilal and K¨ A global assessment would help capture trans-boundary risks and economic spillovers that are often missed in country-level studies. Extend the time horizon beyond 2050: The case studies presented in this study are conservative in scope, as the modeling is limited to 2050. However, biophysical modeling suggests that climate-related damages may intensify significantly beyond 2050, even though results vary greatly depending on the methodology used (Tol; 2024). By restricting the projection period, the models may not capture the full extent of long-term risks. Extending the time horizon would provide policymakers with critical insights to design forward-looking adaptation strategies and scale up resilience investments in anticipation of escalating climate impacts. To conclude, this paper highlights the value of model-based approaches to complement empirical analyses, and notes that its GDP estimates should be considered conservative due to deep uncertainties on key potential impact channels. Continued efforts to model additional risks and impact channels, combined with empirical work to quantify these channels when they are already active, will be critical to better inform adaptation policies, as well as the design of resilient development strategies. 25 References Arndt, C., Chinowsky, P., Fant, C., Paltsev, S., Schlosser, C. A., Strzepek, K., Tarp, F. and Thurlow, J. (2019). Climate change and developing country growth: the cases of malawi, mozambique, and zambia, Climatic Change 154: 335–349. Arndt, C., Chinowsky, P., Robinson, S., Strzepek, K., Tarp, F. and Thurlow, J. (2012). Economic development under climate change, Review of Development Economics 16(3): 369–377. Arndt, C., Schlosser, A., Strzepek, K. and Thurlow, J. (2014). Climate change and economic growth prospects for malawi: An uncertainty approach, Journal of African Economies 23(suppl 2): ii83–ii107. Auffhammer, M. (2018). Quantifying economic damages from climate change, Journal of Economic Perspectives 32(4): 33–52. Bakkensen, L. and Barrage, L. (2018). Climate shocks, cyclones, and economic growth: bridging the micro- macro gap, Technical report, National Bureau of Economic Research. Baqaee, D. R. and Farhi, E. (2019). The Macroeconomic Impact of Microeconomic Shocks: Beyond Hulten’s Theorem, Econometrica 87(4): 1155–1203. URL: https://ideas.repec.org/a/wly/emetrp/v87y2019i4p1155-1203.html Bernard, T. E. (1999). Prediction of workplace wet bulb global temperature, Applied occupational and envi- ronmental hygiene 14(2): 126–134. anzig, D. R. (2024). The macroeconomic impact of climate change: Global vs. local temperature, Bilal, A. and K¨ Technical report, National Bureau of Economic Research. Burke, M., Ferguson, J., Hsiang, S. and Miguel, E. (2024). New evidence on the economics of climate and conflict, Handbook of the Economics of Conflict 1: 249–305. Burke, M., Hsiang, S. M. and Miguel, E. (2015). Global non-linear effect of temperature on economic produc- tion, Nature 527(7577): 235–239. Burns, A., Campagne, B., Jooste, C., Stephan, D. and Bui, T. T. (2019). The world bank macro-fiscal model technical description, Policy Research Working Paper 8965, The World Bank, Washington, DC. URL: https://openknowledge.worldbank.org/handle/10986/32217 Burns, A., Jooste, C. and Schwerhoff, G. (2021). Macroeconomic modeling of managing hurricane damage in the caribbean. Cantelmo, A., Melina, G. and Papageorgiou, C. (2023). Macroeconomic outcomes in disaster-prone countries, Journal of Development Economics 161: 103037. Carr, M., Steduto, P., Hsiao, T., Fereres, E. and Raes, D. (2013). Crop yield response to water., Experimental Agriculture 49(2): 311–311. omez, S., Noy, I. and Strobl, E. (2025). Climate change, hurricanes, and sovereign debt in the Cavallo, E., G´ caribbean basin, Climate and Development pp. 1–10. Chambwera, M., Heal, G., Dubeux, C., Hallegatte, S., Leclerc, L., Markandya, A., McCarl, B. A., Mechler, R. and Neumann, J. E. (2014). Economics of adaptation. Chattopadhyay, D., De Sisternes, F. and Oguah, S. K. W. (2018). World Bank Electricity Planning Model (EPM): Mathematical Formulation World Bank Electricity Planning Model, Technical report, World Bank. Chinowsky, P. and Helman, J. (2021). The infrastructure planning support system (ipsstm), Technical report, Resilient Analytics, Inc. 26 Colon, C., Hallegatte, S. and Rozenberg, J. (2021). Criticality analysis of a country’s transport network via an agent-based supply chain model, Nature Sustainability 4(3): 209–215. na, A. and Bates, P. D. (2017). Itz¨ Courty, L. G., Pedrozo-Acu˜ ı (version 17.1): An open-source, distributed gis model for dynamic flood simulation, Geoscientific Model Development 10(4): 1835–1847. Craig, M. H., Snow, R. and le Sueur, D. (1999). A climate-based distribution model of malaria transmission in sub-saharan africa, Parasitology today 15(3): 105–111. Damania, R., Desbureaux, S. and Zaveri, E. (2020). Does rainfall matter for economic growth? evidence from global sub-national data (1990–2014), Journal of Environmental Economics and Management 102: 102335. Dell, M., Jones, B. F. and Olken, B. A. (2012). Temperature shocks and economic growth: Evidence from the last half century, American Economic Journal: Macroeconomics 4(3): 66–95. URL: https://www.aeaweb.org/articles?id=10.1257/mac.4.3.66 Ebi, K. L., Hartman, J., Chan, N., Mcconnell, J., Schlesinger, M. and Weyant, J. (2005). Climate suitability for stable malaria transmission in zimbabwe under different climate change scenarios, Climatic Change 73: 375–393. unzel, V. and Sch¨ Eckstein, D., K¨ afer, L. (2021). Global climate risk index 2021. who suffers most from extreme weather events? weather-related loss events in 2019 and 2000 to 2019, Technical report, Germanwatch. Emanuel, K., Ravela, S., Vivant, E. and Risi, C. (2006). A statistical deterministic approach to hurricane risk assessment, Bulletin of the American Meteorological Society 87(3): 299–314. Emanuel, K., Sundararajan, R. and Williams, J. (2008). Hurricanes and global warming: Results from down- scaling ipcc ar4 simulations, Bulletin of the American Meteorological Society 89(3): 347–368. Energy Sector Management Assistance Program (2020). The state of access to modern energy cooking services, World Bank. Garner, G. G., Hermans, T., Kopp, R. E., Slangen, A. B. A., Edwards, T. L., Levermann, A., Nowikci, S., ottir, G., Drijfhout, S. S., Palmer, M. D., Smith, C., Fox-Kemper, B., Hewitt, H. T., Xiao, C., Aoalgeirsd´ Edwards, T. L., Golledge, N. R., Hemer, M., Krinner, G., Mix, A., Notz, D., Nowicki, S., Nurhati, I. S., ee, J.-B., Yu, Y., Hua, L., Palmer, T. and Pearson, B. (2021). Ipcc ar6 sea-level rise projections. Ruiz, L., Sall´ Giuliano, F., Navia, D. and Ruberl, H. (2024). Macroeconomic impact of climate shocks in uruguay, Policy Research Working Paper 10740, The World Bank, Washington, DC. URL: http://hdl.handle.net/10986/41298 Hallegatte, S. (2008). An adaptive regional Input-Output model and its application to the assessment of the economic cost of katrina, Risk Analysis 28(3): 779–799. Hallegatte, S., Hourcade, J.-C. and Dumas, P. (2007). Why economic dynamics matter in assessing climate change damages: illustration on extreme events, Ecological economics 62(2): 330–340. Hallegatte, S., Jooste, C. and McIsaac, F. (2024). Modeling the macroeconomic consequences of natural disasters: Capital stock, recovery dynamics, and monetary policy, Economic Modelling 139: 106787. Hallegatte, S., Lipinsky, M. F., Morales, P., Oura, M. H., Ranger, N., Regelink, M. G. J. and Reinders, H. J. (2022). Bank stress testing of physical risks under climate change macro scenarios: typhoon risks to the Philippines, International Monetary Fund. Hallegatte, S., McIsaac, F., Dudu, H., Jooste, C., Knudsen, C. and Beck, H. (2024). Macroeconomic impli- cations of a transition to net zero emissions, The Green Frontier: Assessing the Economic Implications of Climate Action p. 233. 27 Hallegatte, S. and Rozenberg, J. (2017). Climate change through a poverty lens, Nature Climate Change 7(4): 250–256. Hallegatte, S. and Vogt-Schilb, A. (2019). Are Losses from Natural Disasters More Than Just Asset Losses?, Springer International Publishing, Cham. Hartley, F., Gabriel, S., Cullis, J. and Arndt, C. (2021). Climate uncertainty and agricultural vulnerability in south africa. Hashimoto, R. and Sudo, N. (2022). Transmission of flood damage to the real economy and financial interme- diation: Simulation analysis using a dsge model, Technical report, Bank of Japan. Honda, Y., Kondo, M., McGregor, G., Kim, H., Guo, Y.-L., Hijioka, Y., Yoshikawa, M., Oka, K., Takano, S., Hales, S. et al. (2014). Heat-related mortality risk model for climate change impact projection, Environmental health and preventive medicine 19: 56–63. Hsiang, S. M. and Jina, A. S. (2015). Geography, depreciation, and growth, American Economic Review 105(5): 252–256. Huizinga, J., De Moel, H. and Szewczyk, W. (2017). Global flood depth-damage functions: Methodology and the database with guidelines, Technical report, Joint Research Centre (Seville site). International Food Policy Research Institute (2022). Sacred: Systematic analysis for climate resilient develop- ment. Accessed: 5 September 2023. International Labour Organization (2019). Working on a warmer planet: the impact of heat stress on labour productivity and decent work. Islam, N. (2003). What does a dry season mean to the western australian economy? a cge investigation, Technical report, Australian Agricultural and Resource Economics Society. Jafino, B. A., Hallegatte, S. and Rozenberg, J. (2021). Focusing on differences across scenarios could lead to bad adaptation policy advice, Nature Climate Change 11(5): 394–396. Kahn, M. E., Mohaddes, K., Ng, R. N., Pesaran, M. H., Raissi, M. and Yang, J.-C. (2021). Long-term macroeconomic effects of climate change: A cross-country analysis, Energy Economics 104: 105624. Kahsay, T. N., Arjoon, D., Kuik, O., Brouwer, R., Tilmant, A. and van der Zaag, P. (2019). A hybrid partial and general equilibrium modeling approach to assess the hydro-economic impacts of large dams–the case of the grand ethiopian renaissance dam in the eastern nile river basin, Environmental Modelling & Software 117: 76–88. Kalkuhl, M. and Wenz, L. (2020). The impact of climate conditions on economic production. evidence from a global panel of regions, Journal of Environmental Economics and Management 103: 102360. Kassam, A. H., van Velthuizen, H., Sloane, P. H., Fischer, G. W. and M., S. M. (1991). Agro-ecological land resources assessment for agricultural development planning: A case study of kenya., Technical report, Food and Agriculture Organization of the United Nations (FAO) and International Institute for Applied Systems Analysis (IIASA). Kjellstrom, T., Freyberg, C., Lemke, B., Otto, M. and Briggs, D. (2018). Estimating population heat exposure and impacts on working people in conjunction with climate change, International journal of biometeorology 62(3): 291–306. Kong, Q. and Huber, M. (2022). Explicit calculations of wet-bulb globe temperature compared with approxi- mations and why it matters for labor productivity, Earth’s Future 10(3): e2021EF002334. 28 Kotz, M., Levermann, A. and Wenz, L. (2024). The economic commitment of climate change, Nature 628(8008): 551–557. Lehner, B. and Grill, G. (2013). Global river hydrography and network routing: baseline data and new approaches to study the world’s large river systems, Hydrological Processes 27(15): 2171–2186. Liljegren, J. C., Carhart, R. A., Lawday, P., Tschopp, S. and Sharp, R. (2008). Modeling the wet bulb globe temperature using standard meteorological measurements, Journal of occupational and environmental hygiene 5(10): 645–655. Maddison, D., Hamilton, J. and Tol, R. (2005). Climate change and international tourism: A simulation study, Global Environmental Change 15: 253–266. ao, E., Vilanculos, O., Mandlhate, G. and Hartley, F. (2020). Economic implications of climate Manuel, L., Tost˜ change in mozambique, Southern Africa-Towards Inclusive Economic Development (136). Margulis, S., Hughes, G., Schneider, R., Pandey, K., Narain, U. and Kemeny, T. (2010). Economics of adaptation to climate change: Synthesis report. e, E. (2015). Impacts of climate change on milk production Mauger, G., Bauman, Y., Nennich, T. and Salath´ in the united states, The Professional Geographer 67(1): 121–131. Moran, D., Pandolf, K., Laor, A., Heled, Y., Matthew, W. and Gonzalez, R. (2003). Evaluation and refinement of the environmental stress index for different climatic conditions, Journal of basic and clinical physiology and pharmacology 14(1): 1–16. Nath, I. B., Ramey, V. A. and Klenow, P. J. (2024). How much will global warming cool global growth?, Technical report, National Bureau of Economic Research. Neumann, J. E., Emanuel, K. A., Ravela, S., Ludwig, L. C. and Verly, C. (2015). Risks of coastal storm surge and the effect of sea level rise in the red river delta, vietnam, Sustainability 7(6): 6553–6572. Ngoma, H., Lupiya, P., Kabisa, M. and Hartley, F. (2021). Impacts of climate change on agriculture and household welfare in zambia: an economy-wide analysis, Climatic Change 167(3-4): 55. Nordhaus, W. D. (1992a). An optimal transition path for controlling greenhouse gases, Science 258(5086): 1315–1319. Nordhaus, W. D. (1992b). The ‘dice’model: background and structure of a dynamic integrated climate-economy model of the economics of global warming. Park, R. J., Behrer, A. P. and Goodman, J. (2021). Learning is inhibited by heat exposure, both internationally and within the united states, Nature human behaviour 5(1): 19–27. Payet-Burin, R. and Strzepek, K. (2021). Interrelations between the water, energy and food systems and climate change impacts in the zambezi river basin. Pollitt, H. and Petrauskaite, M. (2024). The potential cascading impacts of climate change in cambodia, Policy Research Working Paper 10983, The World Bank, Washington, DC. URL: http://hdl.handle.net/10986/42468 Ponce, V. M. (2020). Muskingum-cunge method explained. Ranger, N., Mahul, O. and Monasterolo, I. (2021). Managing the financial risks of climate change and pan- demics: What we know (and don’t know), One Earth 4(10): 1375–1385. 29 Renard, K., Foster, G., Weesies, G., McCool, D., Yoder, D. et al. (1996). Predicting soil erosion by water: A guide to conservation planning with the revised universal soil loss equation (rusle), Agriculture handbook 703: 400. Rentschler, J. E., Kim, E. J., Thies, S. F., De Vries Robbe, S. A., Erman, A. E. and Hallegatte, S. (2021). Floods and their impacts on firms: Evidence from tanzania, Technical report, The World Bank. Romanello, M., McGushin, A., Di Napoli, C., Drummond, P., Hughes, N., Jamart, L., Kennard, H., Lampard, P., Rodriguez, B. S., Arnell, N. et al. (2021). The 2021 report of the lancet countdown on health and climate change: code red for a healthy future, The Lancet 398(10311): 1619–1662. Rose, A., Benavides, J., Chang, S. E., Szczesniak, P. and Lim, D. (1997). The regional economic impact of an earthquake: direct and indirect effects of electricity lifeline disruptions, Journal of Regional Science 37(3): 437–458. Rose, A. and Liao, S.-Y. (2005). Modeling regional economic resilience to disasters: A computable general equilibrium analysis of water service disruptions*, Journal of Regional Science 45(1): 75–112. ıa-Vila, M., Fereres, E., Raes, D. and Steduto, P. (2021). The AquaCrop model–Enhancing Salman, M., Garc´ crop water productivity: Ten years of development, dissemination and implementation 2009–2019, Vol. 47, Food & Agriculture Org. Schlenker, W. and Lobell, D. B. (2010). Robust negative impacts of climate change on african agriculture, Environmental Research Letters 5(1): 014010. Schlosser, C. A., Sokolov, A., Strzepek, K., Thomas, T., Gao, X. and Arndt, C. (2020). The changing nature of hydroclimatic risks across southern africa, SA-TIED Working Paper 101. Sieber, J. (2024). Water evaluation and planning system, Technical report, Stockholm Environment Institute (SEI), USA. URL: https://www.weap21.org/WebHelp/index.html St-Pierre, N. R., Cobanov, B. and Schnitkey, G. (2003). Economic losses from heat stress by us livestock industries, Journal of dairy science 86: E52–E77. Stehfest, E., van Vuuren, D., Kram, T., Bouwman, L., Alkemade, R., Bakkenes, M., Biemans, H.and Bouw- uller, C. and Prins, A. (2014). Integrated man, A., den Elzen, M., Janse, J., Lucas, P., van Minnen, J., M¨ assessment of global environmental change with image 3.0. model description and policy applications, Technical report, PBL Netherlands Environmental Assessment Agency, The Hague. URL: https://www.pbl.nl/uploads/default/downloads/pbl-2014-integrateda ssessmento fg lobale nvironmentalc hangew ithi mage307 Stocker, T., Qin, D., Platter, G., Tignore, Mand Allen, S., Boschung, J., Nauels, A., Xia, Y., Bex, V. and Midgley, P. (2013). Climate Change 2013: The Physical Science Basis. Contribution of Working Group I to the Fifth Assessment Report of the Intergovernmental Panel on Climate Change, Cambridge University Press. Strzepek, K. M. and Smith, J. B. (1995). As climate changes: international impacts and implications, Cam- bridge University Press. Strzepek, K. M. and Yates, D. N. (2000). Responses and thresholds of the egyptian economy to climate change impacts on the water resources of the nile river, Climatic Change 46: 339–356. Tol, R. S. (2024). A meta-analysis of the total economic impact of climate change, Energy Policy 185: 113922. Twerefou, D. K., Chinowsky, P., Adjei-Mantey, K. and Strzepek, N. L. (2015). The economic impact of climate change on road infrastructure in ghana, Sustainability 7(9): 11949–11966. 30 Von Peter, G., Von Dahlen, S. and Saxena, S. C. (2012). Unmitigated disasters? new evidence on the macroeconomic cost of natural catastrophes. West, C. T. and Lenze, D. G. (1994). Modeling the regional impact of natural disaster and recovery: a general framework and an application to hurricane andrew, International regional science review 17(2): 121–150. Wittwer, G. and Griffith, M. (2010). Closing the factory doors until better times: Cge modelling of drought using a theory of excess capacity, 2010 Conference (54th), February 10-12, 2010, Adelaide, Australia 59263, Australian Agricultural and Resource Economics Society. URL: https://EconPapers.repec.org/RePEc:ags:aare10:59263 uss-Ust¨ Wolf, J., Johnston, R., Hunter, P. R., Gordon, B., Medlicott, K. and Pr¨ un, A. (2019). A faecal contamination index for interpreting heterogeneous diarrhoea impacts of water, sanitation and hygiene interventions and overall, regional and country estimates of community sanitation coverage with a focus on low-and middle-income countries, International journal of hygiene and environmental health 222(2): 270– 282. World Bank (2010a). Adaptation to climate change vulnerability assessment and economic aspects - plurina- tional state of bolivia (english), Technical report, World Bank, Washington, DC. URL: http://documents.worldbank.org/curated/en/642891468162260845/Adaptation-to-climate-change- vulnerability-assessment-and-economic-aspects-BR-Plurinational-state-of-Bolivia World Bank (2010b). Bangladesh - economic of adaptation to climate change : Main report (english), Technical report, World Bank, Washington, DC. URL: http://documents.worldbank.org/curated/en/841911468331803769/Main-report World Bank (2010c). Ethiopia - economics of adaptation to climate change : Main report (english), Technical report, World Bank, Washington, DC. URL: http://documents.worldbank.org/curated/en/310391468340238724/Main-report World Bank (2010d). Ghana - economics of adaptation to climate change (eacc) : Main report (english), Technical report, World Bank, Washington, DC. URL: http://documents.worldbank.org/curated/en/278431468337213682/Main-report World Bank (2010e). Mozambique - economics of adaptation to climate change (english), Technical report, World Bank, Washington, DC. URL: http://documents.worldbank.org/curated/en/978481468178764388/Mozambique-Economics-of- adaptation-to-climate-change World Bank (2010f). Samoa - economics of adaptation to climate change (english), Technical report, World Bank, Washington, DC. URL: http://documents.worldbank.org/curated/en/849221468326982416/Samoa-Economics-of-adaptation- to-climate-change World Bank (2010g). Vietnam - economics of adaptation to climate change (english), Technical report, World Bank, Washington, DC. URL: http://documents.worldbank.org/curated/en/563491468149078334/Vietnam-Economics-of- adaptation-to-climate-change World Bank (2020). Poverty and shared prosperity 2020: Reversals of fortune, Technical report, The World Bank, Washington, DC. URL: https://openknowledge.worldbank.org/server/api/core/bitstreams/611fc6f2-140b-551e-9371- 468eec64c552/content 31 World Bank (2022a). Climate and development : An agenda for action - emerging insights from world bank group 2021-22 country climate and development reports, Technical report, The World Bank, Washington, DC. URL: https://openknowledge.worldbank.org/server/api/core/bitstreams/611fc6f2-140b-551e-9371- 468eec64c552/content World Bank (2022b). South africa country climate and development report, Technical report, World Bank, Washington, DC. URL: http://hdl.handle.net/10986/38216 World Bank (2023). The development, climate, and nature crisis : Solutions to end poverty on a livable planet - insights from world bank country climate and development reports covering 42 economies, Technical report, The World Bank, Washington, DC. URL: https://openknowledge.worldbank.org/entities/publication/c9d962c9-5796-48c4-afc7-9feac8216ab8 World Bank (2024a). People in a changing climate: From vulnerability to action - insights from world bank country climate and development reports covering 72 economies, Technical report, The World Bank, Wash- ington, DC. URL: https://hdl.handle.net/10986/42395 World Bank (2024b). Rising to the challenge: Success stories and strategies for achieving climate adaptation and resilience, Technical report, The World Bank, Washington, DC. URL: https://hdl.handle.net/10986/42326 World Health Organization (2014). Quantitative risk assessment of the effects of climate change on selected causes of death, 2030s and 2050s, World Health Organization. Yates, D. N. and Strzepek, K. M. (1996). Modeling economy-wide climate change impacts on egypt: A case for an integrated approach, Environmental Modeling & Assessment 1: 119–135. 32 A Impact Channel Methodologies This appendix provides further detail on the climate scenario selection process as well as the methodologies used to model each impact channel evaluated in the biophysical models described in Section 3. An illustrative overview of adaptation interventions by channel is also included. A. Climate scenario selection Table 8: Overview of climate scenario selection process ff B. Climate impact channel methodologies (Guinea Bissau example) 1. Heat impacts on labor productivity. Increasing workday temperatures affect labor productivity by decreasing the hours an individual can work. The labor productivity shocks are different among sectors and are based on the sector’s labor hours by occupation and the extent to which these occupations are exposed to outdoor work. Generally, heat stress intensifies for outdoor labor types and with more intense physical work often associated with the agriculture sector. In the most pessimistic scenario, the productivity of Guinea-Bissau’s workers is expected to reduce by 13%, 10%, and 9% in agriculture, industry, and services, respectively. It is worth noting that in more favorable climate scenarios, heat impacts on labor productivity are still significant (around 6%). 2. Vector-borne, water-borne, and temperature-related diseases impact human health. Climate change can impact the labor supply through increased incidence and death rates of various diseases such as malaria and dengue (vector-borne), diarrhea (water-borne), and heat-related illnesses. This results in working days lost due to absenteeism or increased deaths. For Guinea-Bissau, these health impacts due to climate change can cause up to -0.5% reduction in labor supply in 2050 in the most pessimistic scenario. 3. Changes in precipitation and temperature impact on crop yields. Changes in precipitation patterns can reduce water resources for rain-fed agriculture, while temperature increases are likely to reduce the suitability and productivity of crops and can produce additional impacts on overall water resource availability. Impacts are varied significantly for different crops, with coconut, plantain, and cassava being the most affected crops, with up to 31% yield loss, from the dry/hot conditions. On average, yields will reduce by 12% across all crops from 2040 to 2050 in the most pessimistic scenario. 33 4. Erosion impacts on crop yields. Changes in rainfall can also change soil conditions and topsoil erosion, resulting in crop productivity changes. Yield impacts vary significantly across crop types. For example, in the most pessimistic erosion case under a wet climate scenario, cashew nuts and millet yields reduce by almost 8%, with an average loss across the crops at 2% between 2040 and 2050. Under a dry future climate scenario, crop yields are expected to improve from the current situation between a range of 0%-2.5% with an average improvement of just under 1% across the crop types. 5. Inland flooding impact on capital stock. Inland flooding channel considers damages to capital stock across the country from changes in the magnitude and frequency of riverine (fluvial) flooding. Considering all future inland flooding events, the expected damages (i.e., the sum of the probability of each event times its magnitude) from inland flooding is estimated to be 0.15% of Guinea-Bissau’s capital stock. 6. Sea-level rise and storm surge impacts on capital stock. This channel considers shocks on coastal capital stocks from increases in mean sea level and changes in the frequency and magnitude of storm surge events. Damages are split into repairable and permanently destroyed coastal capital stocks, where the latter results in a permanent loss in the country’s total capital stock, and the former requires a divergence of public investment to non-productive spending on repairs. The change in mean sea level relative to baseline conditions is expected to increase throughout the period. Relative changes across selected scenarios are similar, with sea levels expected to increase by 0.1 meters by 2030. Damages from sea-level rising are becoming noticeable in the 2050s. Given the big jump in the shocks between the 2040s and 2050s, it is expected that damages from sea-level rising and storm surges in Guinea-Bissau will become more severe toward the end of this century. The spread between the scenarios increases slightly by mid-century with ssp3-7.0 resulting in a shock of -0.07 percent incremental capital losses. 7. Road and bridges impact on capital stock and productivity. Road and bridge infrastructure are at risk of damage from higher temperatures, precipitation, and flooding recurrence, as well as road disruption effects on reduced working hours. By 2050, it is estimated that additional annual damages relative to the baseline under a wet future will be around $60 million, approximately 0.5% of Guinea- Bissau’s projected capital stock. In addition, extra delay hours will range from -0.25 million to 13.5 million hours (about 0.5% reduction in overall productivity). 34 ff C. Illustrative adaptation interventions by channel Table 9: Overview of adaptation interventions by channel ff 35 B Macroeconomic Modeling This appendix provides further detail on the macroeconomic modeling framework used to translate biophysical impacts into economic outcomes, as summarized in Section 3.5. There is a growing literature on mapping natural disasters to economic models, starting with the seminal work of Nordhaus (Nordhaus; 1992a). The first macro models, or integrated assessment models did not distinguish between different types of natural disasters on the economy. In these models (see Nordhaus (1992b)), a generic damage function was estimated using temperature. Temperature is a function of emissions. The modeling specifies a production function, where output (Y ), depends on capital (K ), labor (N ), total factor productivity (A) and (Ω), which captures the economic losses from climatic impacts. Assuming a Cobb-Douglas production function yields: α 1−α Yt = Ωt At Nt Kt (1) To derive Ωt , Nordhaus (1992b)) starts by specifying a damage function (d), which captures the loss in output: dt = Yt θ1 Ttθ2 (2) This specification assumes that output losses from climate change is a function of the scale of damages (θ1 ) and a parameter that captures the nonlinear effects with regards to temperature change (θ2 ). The production function then becomes: 1 α 1−α Yt = At Nt Kt (3) (1 + θ1 Ttθ2 ) This allows for an endogenous damage setting, where temperature is related to emissions changes, and emissions are related to economic activity. While Nordhaus (1992b)) attempts to estimate the damage function for several sectors (e.g., agriculture, coastal erosion from sea level rise and energy), they are still aggregated, missing important spatial effects. As an example, they derive estimates of sea level rise vulnerability (coastal area divided by total land area) and regress this on log per capita GDP to calibrate their model. In MFMod we remove the damage function and replace it with a vector of computed damages from the biophysical models. This helps us to circumvent any issues pertaining to statistically fitting data, and enables to capture the different effects of various impact channels, as well as the opportunities for adaptation and resilience. Below, we describe the variables through which MFMod can capture the biophysical impacts from climate change.9 Building on the literature, we discuss below how these damages are mapped to macroeconomic models. There are four key types of impact channel: • Labor impacting: climate changes or events which impact on the productivity or supply of labor 1. Labor productivity: climate changes or events which impact on the productivity of labor, such as heat stress. 2. Labor supply: climate changes or events which impact on the supply of labor either through the participation of workers in the labor force, through the hours that it is possible to work, or through the size of the working aged population. • Sub-sector total factor productivity: climate changes or events which impact on the overall productivity of specific sub-sectors in the economy, such as reduced agricultural outputs. • Capital stock damages: climate changes or events which damage the capital stock, such as flooding, sea level rise or cyclones. • Export impacting: climate changes or events which impact on exports specifically, such as damages to export oriented sectors like tourism. 9 Werefer the reader to Burns et al. (2019) for the full derivation of MFMod. This paper focuses on the main equations to connect with biophysical outputs. 36 Some climate shocks impact on the economy through multiple channels - for example, damages to roads and bridges are modelled as a destruction of the capital stock and a reduction in labor supply due to lost work hours. Labor impacting To capture the impacts of climate change on labor productivity or labor supply, labor is augmented with a vector (Λt ). This can be interpreted as a productivity or labor supply shifter and is an exogenous input into the model. 1−α Yt = At (Λt Nt )α Kt (4) Sub-sector total factor productivity To capture the impacts of reduced (or increased) sub-sectoral output due climate change on the aggregate economy, we introduce a sub-sectoral productivity shifter (Γi,t ) on Total Factor Productivity, which is weighted by the share of the relevant sector (i) in total value added (ωi ). 1−α Yt = ωi Γi,t At (Λt Nt )α Kt (5) Capital stock Many macroeconomic models have one aggregate capital stock measure that depreciates at a fixed rate over time. A standard methodology for describing the evolution of capital stock is the perpetual inventory method, which states that capital today (Kt ) is equal to capital yesterday, net of depreciation (δ is the capital depreciation rate), plus any new investments (It ): Kt = (1 − δ ) Kt−1 + It (6) Mapping asset damages to the equation above can be done in several ways. Hsiang and Jina (2015) estimate how depreciation rates at a spatial level are affected by cyclones, and ultimately measure how growth is affected using a standard Solow growth model. They show that the average cyclone exposure for countries that face cyclones, leads to an asset depreciation of roughly 1.2 percent per year. Cantelmo et al. (2023) define the asset loss due to natural disasters as a stochastic process where capital, net of damages are: ∗ log(Kt ) = log(Kt−1 ) − dt θt (7) ∗ Where Kt is capital stock in the previous period without damages, dt equals 1 in the case of a natural disaster and zero otherwise and θt is the amount of loss caused by the disaster: ¯) + ρ log(θt−1 ) + σθ ϵθ,t log(θt ) = (1 − ρ) log(θ (8) ¯) and a volatility scale parameter σθ . Thus, output loss is a stochastic process equal to the expected loss (θ The authors scale their model to ensure that disasters reduce capital and aggregate output by the same factor (dt θt ). However, there is a growing literature that suggests that output losses are larger than what a reduction in the capital stock would suggest (Von Peter et al.; 2012; Bakkensen and Barrage; 2018). These scaled-up effects can be modeled in a macroeconomic framework, by combining an impact on the stock of capital with an impact on its productivity (Hallegatte and Vogt-Schilb; 2019). For instance, using a DSGE model for Japan, Hashimoto and Sudo (2022) introduce floods by modifying the depreciation rate of capital. To account for losses larger than the marginal productivity of capital, the authors also adjust total factor productivity to account supply chain disruptions and include expected losses in the financial sector’s balance sheets. Using MFMod, Hallegatte, Jooste and McIsaac (2024) modify the production function to separate infrastructure and non-infrastructure impact and include the effect of the misallocation of capital after a disaster. 37 The modeling at the World Bank of capital losses considers the above but follows Burns et al. (2021) and Hallegatte, Jooste and McIsaac (2024), where capital stock equals: 1 DS 1−α ˜ K= 1− K (9) K˜ In this formulation, destroyed capital has approximately average product of capital, while new capital has a marginal productivity. We keep track of destroyed capital, separately from new incremental capital. Note that the implication of this formulation is that it is optimal to direct investments into reconstruction or replacement of damaged capital rather than investing in new assets (because replacing damaged asset has higher returns). For capital that is not destroyed, the model follows a simple perpetual inventory accounting that accounts for the rate of capital depreciation and new investments: ˜ t−1 + It ˜ t = (1 − δ ) K K (10) The stock of destroyed capital (DSt ) equals the previous stock of destroyed capital, plus any new residual R damages (RDt ) less investments to reconstruct (It ): R DSt = DSt−1 + RDt − It (11) It is important to note that while it is always optimal to direct reconstruction investments into reducing destroyed capital, agency issues and capacity constraints limit this approach. Specifically, reconstruction capital is the minimum of destroyed capital or a fixed share of total investment, typically calibrated as the size of the construction industry in value added: R It = min(DSt , ϕIt ) (12) MFMod also models adaptation capital. Adaptation investment reduces the damages that enter the model by making capital more resilient to climate change. Adaptation capital also accumulates according to a perpetual inventory method. Furthermore, it is assumed that adaptation capital is not productive, i.e., it does not enter the production function (even though adaptation investment enters GDP). A A A Kt = (1 − δ ) Kt−1 + It (13) Note that K Amax is the investment into adaptation equal to the average expected damage. Successive amounts of adaptation capital declines with additional projects – i.e., level of protection per unit of adaptation capital is not the same. We assume that this protection function is concave as a share of adaptation capital and that maximum adaptation capital: γ2 KA P = γ1 A (14) K max γ2 indicates the extent of diminishing returns to protection, while γ1 indicates the marginal protection from one dollar of adaptation capital. Residual damage (RDt ) is then defined as one less the protection function multiplied by gross damages: RDt = (1 − Pt )GDt (15) We derive K Amax by assuming a constant share of GDP is invested in adaptation capital. If GDP grows at A A a rate g , then in the long run we have Kt = (1+ g ) Kt −1 . Inserting this in the equation for the accumulation of (1−δ ) A adaptation capital, we have Kt = Kt (1+g ) A A + I A . Solving for adaptation capital, we obtain Kt = (1+ g) (g +δ ) I A. If the equivalent of the average gross damages is invested into adaptation, I A = GD ¯ , we can express the maximum adaptation investment a government would want to make as (1 + g ) ¯ K Amax = GD (16) (g + δ ) 38 Exports When climate changes or events impact on an export-specific sector of the economy (for example, tourism), it is important that this is reflected in export volumes in addition to reduced productive capacity. In addition to using a sub-sectoral productivity shifter in the production function, we also scale aggregate export demand (Et ) to reflect the reduction in export demand for the sector (Φi,t ), weighted by the share of that sector in total exports (γi ). new Et = γi Φi,t Et (17) This new level of export demand is then exogenized in the model. Droughts A lot of the literature has focused on the impact of droughts on the agriculture sector (Islam (2003) and Wittwer and Griffith (2010)). Most of drought modeling is typically done within a computable general equilibrium (CGE) framework. The approaches in CGE models are followed and adopted by MFMod. Potential output, Yt , is endowed with Cobb-Douglas technology ˜t K ˜t1−α ˜tα Yt := A −1 N , (18) ˜t is the effective labor, and K ˜t is the total factor productivity, N where A ˜ t−1 is the effective capital stock available for production at time t. Effective labor is a function of time devoted to work, Γt , human capital, ∗ ∗ ht , structural unemployment rates, Ut , structural participation rates, P Rt , and the working age population W , Nt P , so that, ∗ ˜t = Γt ht (1 − Ut ∗ W N )P Rt Nt P. (19) Effective capital is a function of embedded technology in machines θt , so that ˜ t−1 = θt Kt−1 . K (20) ∗ ∗ W If one can only measure total employment, (1 − Ut )P Rt Nt P and capital, Kt−1 , then (empirical) residual TFP will combine many concepts, 1−α ˜t (Γt ht )α θt At = A . (21) With this setup, we can model damages in the utilization of capital as a result of leaving capital idle due to natural disasters, or model the damages to human capital or the number of hours one can devote to work following a natural disaster. Given the aggregate nature of MFMod and the inability to model each sector’s production function, we ˜t ) into a weighted function of sectoral TFP: can further disaggregate TFP (A 1 ˜t = A ˜t (i)di A (22) 0 This allows us to model the destruction of crops and livestock as an example. Specifically, if we think of value-added from crop agriculture as functions of land, capital and labor, and only measure capital and labor, then here again the residual for crop production will contain land. It is important to note that the sectoral shares in MFMod are endogenous. Consequently, shifts in agricultural productivity will affect relative sectoral prices and hence the sectoral shares and finally aggregate productivity. When data allows us to estimate potential value-added functions by sector, requiring data on investment and wages, then the approach collapses to a CGE modeling format with assumptions being made on the degree of substitution across factors of production. The above climate change impacts on capital, labor, and productivity affect the supply side in MFMod. The model then captures dynamic interactions between price and demand effects in both the short term (immediately after the shock) and the long term (as the output gap closes). 39 Macroeconomic uncertainty In MFMod we focus on three sources of uncertainty. First, shock uncertainty — i.e., the stochasticity of the weather — using the derived inputs from the biophysical models. Second, model uncertainty derived from the model properties, or the model’s variance-covariance matrix. And finally parameter uncertainty derived from the model’s coefficient standard errors. For shock uncertainty, we use a standard Monte Carlo approach, where we draw randomly (using the inverse probability transform) from the fitted return period curves. Alternatively, we approximate the return periods with a known-probability distribution (e.g., Poisson for flood events). For model uncertainty, we use a bootstrap procedure (with replacement). First, the model is hit by the climate shock (shock iteration 1), and then the model uncertainty is incorporated (e.g., 1,000 iterations). This is then completed for 1000 climate shocks (per channel). An important caveat is that agents in the model do not perceive the shifts in climate. Agents are myopic in the sense that they only anticipate historical climate shocks. There are ways to extend the model so that the intertemporal solution accounts for uncertainty (e.g., using risk aversion using second moments). This modeling extension is left for future research. 40 C Case Study Results This appendix provides the country discussion for the results presented in Section 4. This section presents results from recent CCDR studies for five Sub-Saharan African economies: Guinea- Bissau, Madagascar, Malawi, Mali, and Zimbabwe. Each case study demonstrates the application of the biophysical and macroeconomic modeling framework described in the previous sections to estimate the potential impacts of climate change and the benefits of adaptation investments. C.0.1 Guinea-Bissau Guinea-Bissau is highly vulnerable to climate change due to its low-lying coastal geography, reliance on agri- culture, and limited adaptive capacity. The CCDR for Guinea Bissau focused on seven key impact channels: rainfed crops, crop erosion, human health, inland flooding, heat impacts, sea level rise and storm surge, and roads and bridges. Biophysical models projected significant reductions in labor productivity due to temperature changes. Crop production was also severely impacted by the lack of rainfall, while the incidence of vector-borne diseases was expected to rise due to favorable climatic conditions for disease vectors. The macroeconomic model projected significant reductions in GDP growth and increases in poverty rates under these climate scenarios. Adaptation measures can be expensive and provide limited benefits given the country’s existing debt level and lack of cheap electricity access (for air conditioning). C.0.2 Madagascar Madagascar faces multiple climate risks, including cyclones, droughts, and sea level rise. The CCDR for Mada- gascar focused on nine impact channels: tropical cyclones, crop production (erosion and rainfed), livestock, fisheries, heat impacts on labor productivity, inland flooding, and tourism. Biophysical models projected an increased frequency and intensity of tropical cyclones, leading to extensive damage to infrastructure and dis- ruptions in economic activities. Crop production was projected to decline due to increased temperature and reduced rainfall, while labor productivity was expected to decline due to changes in temperature patterns. The macroeconomic model projected substantial reductions in GDP growth and increases in poverty rates un- der these climate scenarios. Investments in cyclone-resistant infrastructure, climate-resilient agriculture, and improved heat management were projected to significantly mitigate these impacts. C.0.3 Malawi Malawi’s susceptibility to climate change is notably high, stemming from its reliance on rain-fed agriculture with low productivity and diminishing natural resources, pervasive poverty, and substantial gaps in infrastruc- ture. The CCDR for Malawi examined six key areas likely to be most affected: livestock, crop yields (both irrigated and rain-fed), labor productivity, human health, roads and bridges, and hydropower production, across five different climate projections. The biophysical models indicated a significant reduction in agricul- tural output and livestock productivity due to rising temperatures and shifting rainfall patterns, which could lead to a decrease in food security and an uptick in poverty levels. Labor productivity was expected to take a major hit primarily due to escalating temperatures. Human health could also deteriorate as climate conditions become more conducive to the spread of diseases. The variability of water resources, exacerbated by fluctu- ating precipitation, was anticipated to impact hydropower generation. Malawi’s transport infrastructure was particularly vulnerable to climate-related damages. The macroeconomic model suggested that these impact channels could lead to a marked decrease in GDP growth and a rise in poverty under all five climate scenarios. However, strategic investments in climate-resilient agricultural practices, sustainable management of natural resources, healthcare improvements, robust energy and transport systems, and efficient water management could play a significant role in mitigating these adverse effects. 41 C.0.4 Mali Mali is confronted with a range of climate-related challenges, such as persistent droughts, advancing desertifi- cation, extreme heat events, and unpredictable rainfall. The CCDR for Mali assessed seven critical areas likely to be impacted: rainfed agriculture, livestock, labor productivity, labor health, flooding, roads and bridges, and tourism. Biophysical models suggested a likely decrease in agricultural and livestock productivity from rising temperatures and altered rain patterns. Labor productivity and health were expected to deteriorate in response to these climatic changes. Additionally, the potential damage to infrastructure from flooding could significantly disrupt the movement and supply of goods and services. The tourism sector’s impact was minimal due to its limited contribution to the economy. According to the macroeconomic model, these climate scenar- ios could lead to sharp declines in GDP growth and heightened poverty levels. However, targeted adaptation investments had the potential to substantially reduce these negative outcomes substantially. C.0.5 Zimbabwe Zimbabwe is confronted with a variety of climate-related threats, including recurrent droughts, acute water shortages, loss of biodiversity, occasional storm damage, and unpredictable rainfall patterns. The CCDR for Zimbabwe explored the consequences of climate change through nine different impact channels: labor productivity under heat stress, health, crop yields affected by erosion, agricultural income, livestock production, transportation infrastructure, inland flooding, hydropower generation, and water, sanitation, and hygiene (WASH) systems, under a range of climate scenarios. The projections from biophysical models suggested a significant drop in labor productivity and health due to increased temperatures and altered precipitation patterns. Crop and livestock production were also expected to suffer notable declines as a result of temperature rise, changes in rainfall, and soil erosion. Zimbabwe’s infrastructure was likely to experience substantial damage, with the transport system and hydropower facilities being particularly vulnerable to temperature and rainfall changes, as well as to inland flooding. The macroeconomic model projected considerable reductions in welfare, highlighting the country’s economic dependence on agriculture and natural resources. Nonetheless, investments in climate-resilient agricultural methods, enhanced physical and health infrastructure, and better water management could significantly alleviate these impacts. 42