Abstract
Climate impact assessments inform climate change discussions. Integrated use of biophysical and economic models made it possible to move from assessments exclusively focused on the physical impacts, to assessments that incorporate the prospective effects on human welfare. Effects on poverty and livelihoods are better understood. However, even though structural inequalities exacerbate exposure and vulnerability to climate change, the nexus between climate change and inequality remains underresearched. We suggest ways to feature inequalities prominently in climate impact assessments hoping to encourage new research. We suggest how to use modeling capability to explore how existing inequalities may worsen in the face of climate hazards, through perturbation of natural resource systems, unemployment of production factors, a lack of access to human capital and basic services, and socioeconomic attributes that place people at a disadvantage. We also point to the policy analysis that one can develop and areas to improve it going forward.
Keywords
The discussion of impacts of climate change remained for quite some time focused on the physical impacts. It took some time for researchers across different disciplines to broaden the focus of the analysis to include socioeconomic impacts.
The World Health Organization (WHO; 1990) addressed the potential health impacts of climate change as early as 1989. During the timeframe of the first assessment reports of the Intergovernmental Panel on Climate Change (IPCC) in the 1990s, there was also the recognition of distributional concerns in climate research (Rao, Van Ruijven, Riahi, & Bosetti, 2017).
Broader evidence on the prospective effects on human welfare emerged more clearly until the 2000s. Studies began to combine the use of different modeling tools to look at the biophysical and socioeconomic impacts together. The literature interchangeably calls these studies integrated climate impact assessments, climate impact assessments, or integrated assessments.
Within this literature, the effects of climate change on poverty and livelihoods receive special attention (e.g., Brainard, Jones, & Purvis, 2009; Carvajal-Velez, 2007; Hughes, Irfan, Moyer, Rothman, & Solórzano, 2012; Skoufias, 2012; Stern, 2006; United Nations Economic Commission for Africa, 2010; World Bank, 2002, 2008; among others). Recent research focuses on how the mechanisms through which climate change affects poverty and livelihoods work in practice. Hallegatte et al. (2014) identify prices, assets, productivity, and opportunities as four critical channels through which households may move in and out of poverty in the presence of climate change. In their contribution to the Fifth Assessment Report of the IPCC (AR5), Olsson et al. (2014) provide an extensive review of evidence, from cross-sectional statistical analyses as well as case studies, which demonstrates there is a dynamic interaction between climate change, poverty, and livelihoods. Hallegatte et al. (2016) examine the magnitude of future climate change impacts on poverty as these are channeled through food prices and production, natural disasters, health, and labor productivity.
Despite the emerging literature, the AR5 identifies gaps with regard to the distributional concerns that are still seen today (Rao et al., 2017). The nexus between climate change and inequality within countries remains relatively underresearched or inadequately studied. This is quite surprising, given the strong association among inequality, poverty, and the exposure and vulnerability of individuals and societies to climate change. This gap in the literature is all the more worrisome, given the long-term and continuous increase in the within-country component of global inequality (Lakner & Milanovic, 2013).
According to IPCC (2014), exposure to climate change refers to the presence of people (including their livelihoods), ecosystems and species, or economic, social, or cultural assets in places that could be adversely affected by climate hazards. Vulnerability is defined as the propensity or predisposition to be adversely affected, which encompasses sensitivity or susceptibility to harm and a lack of capacity to cope and adapt. Conforming to these definitions, exposure and vulnerability are determined implicitly by the conditions of poverty, marginalization, and social exclusion as they affect specific population groups.
Only recently has the role of inequality per se gained visibility in the climate discussion. Skoufias (2012) had noted the regressive nature of climate change impacts in the context of Brazil, where these impacts tend to affect relatively more the poor than the rich. Olsson et al. (2014, p. 796) argued that “socially and geographically disadvantaged people exposed to persistent inequalities at the intersection of various dimensions of discrimination based on gender, age, race, class, caste, indigeneity, and (dis)ability are particularly negatively affected by climate change and climate-related hazards.” Disproportionate erosion of physical, human, and social assets as a result of climate change and climate-related hazards exacerbates inequalities and places these people even at more of a disadvantage (Olson et al., 2014).
The recognition that inequality should be a key element in the climate discussion warrants an additional discussion on the methodologies that are available to address the issue of how climate change and climate-related hazards exacerbate inequalities. This article argues that, despite the methodological advances, integrated climate impact assessments are not significantly addressing inequalities within countries, even if these inequalities influence the exposure and vulnerability of people to climate hazards. The article discusses ways to integrate existing modeling frameworks for more prominent and adequate exploration of within-country inequalities in integrated climate impact assessments.
The remainder of the article contains four sections. The second section discusses how the link between climate change and inequality began to be established using quantitative analyses and points to their limitations. The third section briefly describes the standard use of modeling tools in integrated climate impact assessments and discusses the critical areas where the scope of these assessments needs to be expanded for improving the analysis of within-country inequalities. This section lays the groundwork for a deeper understanding, in the fourth section, of ways to explore different dimensions of inequality using existing modeling frameworks and examples. The final section provides the conclusions and sets out key research challenges going forward.
Methodological Advances and Limitations in Analyzing Inequality
Initial studies addressed equity considerations on the basis of an analysis of the social cost of carbon—the expected present-value of damages arising from carbon dioxide emissions (Anthoff, Hepburn, & Tol, 2009). These studies provide estimates for socially desirable mitigation policies. However, these policies are difficult to implement in practice because of two restrictive assumptions of these studies: (a) People who benefit from the policies will be better off if they compensate those negatively impacted by them, and (b) a dollar given to a poor person is the same as a dollar given to a rich one. Equity weights were introduced to relax the second assumption, significantly changing the results of calculating the social cost of an incremental emission (Anthoff et al., 2009). Some studies have posited simple linear functional relationships between the fair share of emissions and its determinants, including an income term whose coefficient is weighted by the proportion of poor within any given country (Mattoo & Subramanian, 2011). Weighing for equity has been an important step toward accepting the suggestion that equity should be a prime concern in climate policy. However, a limitation is that because of data restrictions, equity weights are generally not constructed using average per capita income of individuals (Anthoff & Tol, 2010). The strong assumption of static present-day subnational income distributions is another limitation in some of these analyses (Rao et al., 2017).
Other relevant studies rely heavily, albeit not exclusively, on the use of general equilibrium frameworks and household survey data. These studies analyze the distributional impacts of, for example, carbon pricing in the United States (Raush, Metcalf, & Reilly, 2011), emissions taxation in Brazil (Da Silva Freitas, De Santana Ribeiro, De Souza, & Hewings, 2016) and the United Kingdom (Kehlbacher, Tiffin, Briggs, Berners-Lee, & Scarborough, 2016), and emissions trading schemes and allocation of emission permits across China’s regions (Zhang, Springmann, & Karplus, 2016), among others.
These studies have, undoubtedly, contributed methodologies and new empirical evidence to inform the climate discussion as it relates to inequality. An important limitation is that these studies do not focus on the impacts of climate change and policies on the specific population groups that are particularly vulnerable to climate change and climate-related hazards. Olsson et al. (2014) and other contributors to the AR5 have raised the concern that few assessments examine how inequalities shape differential vulnerabilities to climate change.
Part of the problem is that, by focusing mostly on mitigation, there has been a tendency to leave out issues such as adaptation and resilience that are most relevant for understanding the nexus between climate change and inequality. And even the fewer adaptation studies have paid insufficient attention to inequality. Chambwera et al. (2014) cite 13 different economic assessments of adaptation options, spanning the period from 2006 to 2013. We thoroughly reviewed these assessments and corroborated the observation that inequalities do not feature prominently in these assessments. Some of them address health issues or focus on food security and the livelihoods of the rural poor and farmers; however, they make no explicit reference to inequality. Only one study considered inequality but only contextually. What can be done, analytically, to broaden the focus of climate assessments and bring inequality into the picture is a question that deserves deeper study.
Deployment of Modeling Tools in Integrated Climate Impact Assessments
Rao et al. (2017) have recently reviewed the history and the state of the art of models used in climate research, including integrated assessment models (IAMs) and national studies, and those that model mitigation and climate-change impacts. Their conclusion is that current models used in climate research have a limited ability to represent the poor and vulnerable, and there is much scope to improve the representation of income distribution and poverty.
Our article follows naturally from Rao et al.’s study. While these authors describe the usefulness (or lack therefore) of each modeling approach individually, our article focuses more on how models can be integrated to bridge the gaps between them and improve the analysis of inequality. Our article also pays more attention to biophysical modeling.
The integration of modeling tools, generally—but not exclusively—in a “top-down manner,” has become a widely accepted practice to estimate socioeconomic impacts of climate change (Mimura et al., 2014). The long-term climate change impacts and mitigation policies have been at the center of most assessments, but there is also modeling capability to assess risks associated with other climate hazards as well as policies for adaptation and climate resilience.
The cascade of analytical steps that is generally followed in comprehensive climate impact assessments is represented in Figure 1—of which only the continuous arrows need to be considered at this point. Representative Concentration Pathways (RCPs) are used from the outset to represent greenhouse gas (GHG) emissions and concentration pathways for the world, under different levels of mitigation. The RCPs are the product of an innovative collaboration among integrated assessment modelers, climate modelers, terrestrial ecosystem modelers, and emission inventory experts (Van Vuuren et al., 2011). The RCPs are used in analyses with global climate models, also known as global circulation models (GCMs).
Representation of the deployment of modeling tools in integrated climate impact assessments.
GCMs are numerical climate models that apply known physical, chemical, and biological principles to simulate the interaction of the atmosphere, oceans, land surface, snow, ice, and permafrost in determining the earth’s climate (McCluskey & Qaddami, 2011). They describe climate changes over relatively large spatial and temporal scales under the RCP scenarios; for example, changes in temperature or in precipitation (McCluskey & Qaddami, 2011). GCMs are useful to simulate a common set of scenarios where GHG emissions evolve according to their key drivers (population, energy technology, land use, and so on) and to describe broad storylines of alternative, stylized future climate paths under these scenarios.
Information from these scenarios is downscaled to generate new scenarios, using global biophysical models. These new scenarios are critical for understanding how projections in temperature or precipitation, under given GHG emission and concentration assumptions, may affect a particular area or sector within an ecosystem. For example, crop models—a type of biophysical model—are often used to analyze the endogenous response of agriculture yields and area under different climate-change scenarios (Nelson et al., 2014). The ultimate purpose is to assess the biophysical impacts of changes in climate to arrive at some sort of socioeconomic evaluation.
The socioeconomic evaluation is more thoroughly carried out adding two additional components. First, there needs to be a global economic model through which the biophysical impacts are channeled and simulated to quantify socioeconomic impacts. Global economic models that are well integrated with biophysical models are generally called IAMs, and they are amply overviewed in Tol and Fankhauser (1998). These models represent the complex cause–effect relationship between biophysical processes that can be climate-driven and economic growth. Second, so-called Shared Socioeconomic Pathways (SSPs) proposed by Kriegler et al. (2012) are being used to add socioeconomic details in the analysis. SSPs include three elements: storylines, which are descriptions of the state of the world; quantitative variables from the IAMs such as climate paths, population, gross domestic product (GDP), and technology availability; and other variables, not included in IAMs, such as ecosystem productivity and sensitivity or governance index (Burkett et al., 2014). The RCP scenarios presenting both GHGs emissions and concentration pathways generally correspond with the SSP scenarios.
The cascade of scenarios from integrating climate, biophysical, and socioeconomic models at the global level is further downscaled to assess impacts, vulnerabilities, and policy options at lower geographical levels. In this case, biophysical and economic models for a country, a region, or a sector (or for all three levels) are needed. Information from global scenarios could directly be downscaled to regional models, without necessarily using national models; in either case, the analysis can be extended to inform sectoral economic models. In some cases, scenario results may even be scaled up, following a bottom-up approach, if what happens at the sectoral-level analysis is expected to affect biophysical or socioeconomic systems at the regional or country levels or for consistency purposes (Valdivia et al., 2015).
Areas for Broadening the Scope of Assessments
This article argues that some of the modeling tools identified up until now can be deployed to explore different dimensions of inequality within countries. This, however, requires expanding the focus of integrated climate impact assessments—and, as further explained later, choosing a suitable economic modeling framework. As noted earlier, assessments have mostly focused on mitigation (vis-à-vis adaptation and resilience) and on the long-term impacts of climate change (vis-à-vis climate hazards arising from climate variability and extreme weather events).
Mitigation is relatively easier to measure because GHG emissions are quantifiable. On the other hand, adaptation and resilience have no common reference metrics comparable with the ones that exist for mitigation and measuring them requires a larger number of indicators relevant to each country and specific local context (Noble et al., 2014). By their very nature, adaptation and resilience are interwoven with broad development goals because reducing exposure and vulnerability to climate hazards requires livelihood improvements, food security, improved health systems, infrastructure development, and better educational services (Berkes, 2007; Moss, Malone, Engle, de Bremond, & Delgado, 2012). Any analysis that integrates such goals and examines potential policies to achieve them is multimetric in nature.
We argue that tools used in integrated climate impact assessments make it possible to integrate different facets of development and to analyze adaptation and resilience in the context not only of long-term climate change but also of climate hazards resulting from climate variability and extreme weather events. There is ample evidence of the severity of impacts from climate extremes and variability on people and livelihoods at national and subnational levels (Smith et al., 2014). This evidence provides an order of magnitude of potential shocks inflicted on natural resources and socioeconomic systems. Such information can be used in designing scenarios. The sequence of analytical steps may begin with imposing an exogenous change (i.e., a shock) on the national, regional, or sectoral models, without necessarily linking these models with global models (see Figure 1, dotted arrows). This makes it possible to estimate the sensitivity of outcomes to climate variability and extreme weather events and explore policy options, as we shall see later.
The recent recognition that inequality should be part of analyses that use the SSPs is also relevant for the implementation of a comprehensive climate impact assessment. Van Ruijven et al. (2014) recommend that SSPs incorporate not only the standard variables of population and GDP but also indicators such as income distribution, spatial population, human health, and governance in order for scenarios that rely on them to become more useful to climate change impacts, adaptation, and vulnerability research. Van Der Mensbrugghe (2015) argues that assumptions about changes in the within-country income distribution are important for better interpretation of the SSP story lines. He highlights the need for a more nuanced view of the evolution of the distribution of income as a necessity to alter the rather mechanical, although potentially rich, exercise of projecting different futures.
Choice of Economic Modeling Framework
The choice of the economic modeling framework is a critical aspect of the assessment. It is important to be aware that, in practice, prices in the different markets of the economy change over time, particularly in contexts characterized by changing climatic conditions, in response to which economic agents may allocate resources differently. The prices of internationally traded food commodities interact with climate change (Porter et al., 2014), and changes in these prices tend to have a greater effect on the welfare of households that use a large income share to purchase staple crops (Olsson et al., 2014). As a consequence, these households may adapt by shifting their consumption habits, which has implications for their vulnerability and well-being. Allowing for resource allocation effects in the assessment is also important for a better understanding of the macroeconomic and financial feasibility of policies.
The macroeconomic and financial feasibility of policy options is usually overlooked in the climate literature. There has been a tendency to use economic models that are aggregate and simple in terms of their data requirements and estimation techniques, because typically the objective is to estimate the costs and benefits for a single project or intervention. In any case, it is important to broaden the scope of the economic modeling analysis to encompass not just a simple cost–benefit analysis of a single policy but also the economy-wide repercussions and macroeconomic feasibility of multisectoral development policies, which requires the use of more comprehensive modeling approaches. This is particularly important if there are significant gaps in the financing of adaptation and multiple investments needed to build climate resilience.
Reduced-form econometric models are often used to estimate long-term effects of climate change. They are based on the notion that adaptive responses to climate change can be represented by equations that relate climate variables directly to economic outcomes. These models are estimated econometrically using cross-sectional or panel data (pooled cross-sectional and time series) and are then simulated using projected future climate variables to determine the impacts of climate change on the dependent variable in the model (Mendelsohn & Dinar, 2009; Mendelsohn, Nordhaus, & Shaw, 1994). These models, however, are unable to trace resource allocation effects because they take prices as given.
Microeconomic structural and land-use models do allow for changes in resource allocation as the economic decision-making is more explicitly represented in them. There is plenty of evidence emanating from these types of models that farmers adapt to climate change via switching crops or livestock species (Seo, 2010; Seo & Mendelsohn, 2008; Seo, Mendelsohn, Dinar, & Kurukulasuriya, 2009). By construction, however, these models lack details on how prices are determined in different markets.
Market equilibrium models, such as computable general equilibrium (CGE) and partial equilibrium (PE) models, are better suited to simulate shocks caused by changes in productivity, policy, or other exogenous factors, such as climate on various economic outcomes, including market prices, production, productivity, consumption, trade, and land use. The differences between these two types of models are well demarcated (Rivera, 2003). CGE models have better capability to trace resource allocation effects throughout the economy in response to shocks; in this case, the effects are transmitted through the different markets of the economy (e.g., factors, commodities, and foreign exchange) under macroeconomic constraints. Because of these features, CGE models seem to be a better fit than PE models to help in assessing the wider costs and benefits of climate policies, including their economic and financial feasibility. If need be, a CGE model and a PE model can be part of the same integrated climate impact assessment. Nine of the major PE and CGE models used for climate impact assessment have been intercompared (Nelson et al., 2014; Von Lampe et al., 2014).
Analytical Dimensions of Inequality in Modeling Frameworks
Broadening the scope of integrated climate impact assessments in order to deepen the analysis of inequalities within countries requires deploying the adequate modeling frameworks. This section discusses the strengths and weaknesses of existing modeling frameworks in exploring inequalities in the face of climate hazards and policies to address them and build climate resilience. We draw upon existing model-based analyses as well as a modeling analysis specifically developed for this article.
Livelihoods and Climate-Sensitive Natural Resources
Large groups of people and communities affected by poverty and structural inequalities whose members secure a living in climate-sensitive environments are particularly exposed and vulnerable to climate hazards. Understanding how their exposure and vulnerability to climate hazards is exacerbated by inequalities begins with an analysis of the impacts of climate hazards on climate-sensitive natural resources such as land, water, or energy.
The analysis relies on natural resource systems models which are biophysical in nature and allow for assessment of adaptation options. For example, Bhave, Mittal, Mishra, and Raghuwanshi (2016) downscale regional scenarios of future climatic change through a water systems model in order to estimate impacts on water availability in India’s Kangsabati river basin. In assessing policy options, they find that increasing forest cover is more suitable for addressing adaptation requirements than constructing check dams. Different studies in Cervigni, Liden, Neumann, and Strzepek (2015) rely upon an energy systems model to channel the impacts of a wide range of future climate scenarios on hydropower and irrigation expansion plans in Africa’s main river basins. These studies suggest that hydropower infrastructure needs to be developed irrespective of the scenario for water availability, so as to avoid significant losses of hydropower revenues and increases in consumer expenditure for energy (driest scenarios) or foregoing substantial revenues from not expanding hydropower production (wettest scenarios).
Each natural resource systems model (whether for land, water, or energy) is useful in its own right. However, a more holistic approach whereby systems models are integrated is better suited to facilitating an understanding of how changes in one resource resulting from a climate hazard may impact other resources as well as how natural resources can be allocated more efficiently to meet the demands for crops, water, and energy services or to achieve a broader form of adaptation to improve livelihoods. A number of studies speak of the advantages of using frameworks such as the Water–Energy–Food Security Nexus or the Climate, Land, Energy, and Water Systems (CLEWs), which integrate different natural resource systems models.
International Renewable Energy Agency (2015) reports the noteworthy findings derived from a number of exploratory case studies following the Nexus approach. One study shows that half of China’s proposed coal-fired power plants, which require significant water for cooling, are located in areas already affected by water stress, leading to potential conflicts between power plant operators and other water users. Another study demonstrates that, in India, where nearly 20% of electricity-generation capacity is used for agricultural water pumping, lower-than-usual rainfall accompanied by decreasing water tables is putting tremendous stress on the electricity system during peak seasons. These studies underscore the functionality of the Nexus approach in leading to important policy insights centered on the fact that water, which is constrained by climate change, faces competing allocations between energy generation and other uses such as in farming. The scarcity of water can hamper farmers in their pursuit of a livelihood, and it may not be easy for them to find alternative means of coping with these changes, leaving the poorest farmers behind.
Important policy concerns have also been addressed for the island of Mauritius using the CLEWs framework (Howells et al., 2013; Welsch et al., 2014). Facing the recent loss of the sugar industry’s export competitiveness, the government considered two policy objectives: developing bioethanol production to reduce GHG emissions and cutting energy imports. Achieving these objectives entails diverting sugarcane production away from export markets toward the domestic processing of bioethanol on an island where sugarcane plantations cover 80% to 90% of cultivated land. The CLEWs analysis shows that the two policy objectives could be achieved but not without trade-offs. In recent years, lower rainfall has led to water shortages, which under scenarios of climate change imply that the water needed for sugarcane production would have to be supplied through irrigation so as to maintain bioethanol production. This would ultimately lead to a gradual drawdown of storage levels in reservoirs, and if the demand for more energy needed to desalinate water for irrigation is met with coal-fired power generation, as intended, then the GHG-related benefits of the bioethanol policy will be eroded by increased emissions from the power sector. Higher coal imports would also have a negative impact on the energy security. In this case, the benefits of the policy are, therefore, vulnerable to the impacts of climate change. As a result, the island faces two possibilities: Either sugarcane producers will have to scale back production (which would jeopardize the livelihood of populations that rely on that production) or they will have to resort to expensive water desalination (which would have detrimental environmental repercussions).
This holistic approach to natural resource systems analysis offers a first entry point into the analysis of inequalities in integrated climate impact assessments. It allows for an initial understanding of how climate-sensitive natural resources are affected by climate hazards—with and without adaptation policies—and provides information on how livelihoods depending on those resources may be affected. Additional socioeconomic analysis is nonetheless needed to estimate the distributional impacts of the changes in climate-sensitive livelihoods and to investigate policy options to offset adverse impacts.
In the CLEWs analysis for Mauritius, for example, under the hypothetical scenario where sugarcane producers scaled back production owing to climate change, unemployment, welfare, and perhaps income distribution would likely be affected. The population that owns production factors employed in the bioethanol industry, whether labor, capital, or land, could be adversely affected in the process. However, these impacts are not quantifiable by simply applying the CLEWs framework or even the Nexus approach for that matter; the assessment needs to be expanded using an economic modeling framework.
Ownership of Production Factors and Income Distribution
Channeling the physical impacts of climate hazards on natural resources throughout the economy requires establishing some sort of soft- or hard-link between biophysical and economic models. A discussion of how this link can be implemented in practice is beyond the scope of this article. 1 We are more interested in showing the advantages of economic modeling in integrated climate impact assessments, particularly the useful insights these models provide with regard to income gains and losses of people, depending on the ownership of production factors.
Climate hazards have disproportionate impacts on the limited assets of disadvantaged population groups when they cause a disruption of economic activity and unemployment of production factors. For disadvantaged groups, a small but adverse change in the employment of production factors that they own, generally labor or land, will likely exacerbate their exposure and vulnerability to climate hazards. The impact of climate hazards can also propagate throughout the entire economy: Poverty and distributional impacts will be the result of the multiple direct and indirect effects of the initial climate shock. The use of a CGE model—alone or combined with other economic models—makes sense because the direct impact of climate hazards makes its way through myriad transmission mechanisms, leading to a multiplicity of indirect effects.
We developed an assessment of climate shocks to illustrate the functionality of CGE modeling. The assessment is based on a CGE model for Bolivia (Plurinational State of). A baseline scenario was built to project, for the 2017–2080 period, a continuation of the economy’s trajectory and policies seen in the country during the 2010–2016 period.
2
Scenarios of climate shocks, with and without policy responses, were subsequently built and compared with the baseline scenario. The climate shocks imposed are as follows and their magnitude falls within a range of possible impacts according to empirical studies:
Productivity shock: Labor productivity gradually decreases in all production sectors, starting in 2017, until falling by 10% by 2080, presumably as a result of rising temperature and its effects on workers’ health.
3
Infrastructure shock: Fifty percent of public infrastructure (such as roads, bridges, and so on) is lost in 2020, presumably as a result of an extreme weather event.
4
Food price shock: Food prices rise gradually starting in 2017, until they are 70% higher by 2080, presumably as a result of climate change.
5
productivity shock and an increase in government health spending by 15%, aiming at boosting labor productivity to offset the shock, and infrastructure shock and an increase in government investment in order to fully recover the infrastructure stock by 2024.
The first two shocks are combined with a policy as follows:
In the last two cases, the shock and the policy response are assessed assuming that one of the following sources at the time finances government spending/investment: foreign borrowing, domestic borrowing, or direct taxation. This gives a total of six policy scenarios under analysis and a total of nine scenarios altogether. The main results of this scenario analysis for Bolivia as well as other studies’ results are discussed next.
Economy-wide and factor income distribution effects
The three climate shocks—without policies—hurt GDP growth and trigger important changes in factor income distribution (Figure 2). The negative impacts of lower labor productivity through the first shock are particularly seen through effects in the labor market: wages and employment decrease. The average wage relative to the average capital return decreases by approximately 18% in 2080—compared with the baseline. Capital becomes relatively scarcer as a result of the abrupt destruction of infrastructure through the second shock, which transitorily affects the distribution of factor incomes with wage earners losing out more compared with capital owners—until the economy converges to the preshock situation. In the first two scenarios, there is a reduction in labor wages, both in absolute terms and relative to capital. Households whose livelihoods rely mostly on labor income, who generally belong to vulnerable groups, lose out in the process.
Bolivia: Labor wage/capital return and labor wage/land rent ratios, under climate shock scenarios (deviation from baseline scenario values).
The food price shock simulation is interesting because it shows that there may be not only losers but also winners. Bolivia is a net exporter of agriculture products and food and the sectors that produce them grow and gain with food price rises. Because the growing sectors are relatively labor intensive, the food price shock results in a reduction of unemployment. In terms of income distribution, the labor wage/land return ratio falls, which is not observed in the previous two climate shock scenarios. The labor wage/capital rent ratio tends to improve, first rapidly, then less rapidly until converging to the preshock situation. Unskilled nonsalaried workers are most favored by the increase in the world price of food because of the large presence of unskilled small farmers and self-employed workers in Bolivia’s food production. In the face of such a situation, however, policy options would have to be considered for reducing the burden imposed by the price shock on vulnerable consumers.
Once the effects on the distribution of factor incomes have been estimated, the next step is to translate them into changes in the distribution of income across households. This can be done within the contours of a CGE model analysis. Andersen et al. (2016), for example, estimate the economy-wide and distributional impacts of crop-yield losses of 10% to 30% over the next half century owing to the impact of climate change. They find that such a significant shock would not necessarily translate into proportional income losses for farmers or the population in general, if farmers were to find ways to adapt autonomously. Should this be the case, the final effects of climate change would tend to be smaller than that of the initial crop-yield shock, and the net effects on income of different household groups would be modest in either direction. The authors analyze the impacts on 80 different household types for Mexico, differentiated by gender of household head, agro-ecological zone, and income decile, and find that the impacts are very similar for them all (i.e., there are tiny losses in welfare between 0.1% and 0.3%). Interestingly, this small effect on income across income deciles is robust to the choice of climate model (Andersen et al., 2016).
The analysis of income generated through employment of production factors and its distribution across different household groups is useful but insufficient. It is useful because households located at the lowest deciles of a distribution are those that tend to exercise relatively less ownership over production factors and assets in general. These households are generally vulnerable and understanding how their income changes in the face of climate hazards is critical. Changes in the income of these households can be compared with changes in the income of households located in higher income brackets to draw conclusions on income distribution. However, this approach is still highly aggregative, even if household groups are classified according to income decile and other attributes, and misses out on the details of income distribution within household groups (Vos & Sánchez, 2010), which can ultimately affect the well-being of vulnerable households. Nor is economy-wide analysis alone well suited for addressing other forms of inequality, including those that are determined by certain configurations of socioeconomic characteristics such as gender, age, race, religion, and ethnicity. Analysis at a level that is more micro in nature helps surmount these methodological limitations, but before describing this form of analysis—including results emanating from the Bolivian analysis introduced earlier–it is important to understand other useful features of the CGE modeling approach that are relevant for the analysis of inequalities in integrated climate impact assessments.
Macroeconomic impacts and policy feasibility
As noted earlier, climate shocks simulated for Bolivia harm GDP growth and labor incomes. When the climate shocks are combined with increased government spending/investment, we find that the final effect on GDP growth will depend on the source of finance of the new spending/investment. Consider for simplicity only the infrastructure shock combined with new public investment in infrastructure. In this case, output recovers most—albeit not fully—when foreign borrowing finances the investment (Figure 3). The inflow of foreign resources, however, results in a real exchange rate appreciation as a result of which export growth is slower and import growth faster. There are winners (nontradable sectors) and losers (tradable sectors), but, overall, foreign financing allows increasing household consumption and GDP more than in the other two financing scenarios. In fact, if financing alternatively comes from domestic direct taxation, there is less disposable income, and consumption is depressed as well as savings to some extent. The least favorable policy scenario is with domestic borrowing, though, as this source of finance implies allocating private savings to financing public investments rather than private ones. Even if public investment is replacing private investment, an unintended consequence of the policy is that GDP grows less compared with the other two financing scenarios.
Bolivia: Real GDP under infrastructure shock scenarios, without and with new public investment and financing (percentage deviation from baseline scenario).
There are additional considerations to make on macroeconomic feasibility. Domestic borrowing or foreign borrowing may lead to debt sustainability issues, or it is unrealistic to believe that tax revenues can suddenly grow by a few points of GDP—all of which need to be part of a rigorous assessment in practice. Indeed, in Bolivia’s example, public borrowing or tax revenues set above baseline scenario levels by a few points of GDP when the climate shocks are simulated in tandem with the policy responses (not shown here). In practice, of course, policy makers will have to assess scenarios where policies are financed using a mix of options, depending on fiscal and debt sustainability considerations.
Human Capital and Public Services and Resources
In coping with climate hazards, poor and disadvantaged groups often face the difficult choice between protecting their human capital (health and education) and preserving their physical capital or even their consumption levels. These groups face such choices, as they have to make ends meet under an income constraint, and possibly have insufficient access to basic public services and resources. The choices they make may eventually affect their exposure and vulnerability to climate hazards, so it is worth exploring those human development policy options available for increasing the climate resilience of these groups.
The effects of climate change on human development are mainly estimated for the long term using reduced-form econometric models. Studies find that climate change reduces life expectancy in Peru (Andersen, Suxo, & Verner, 2009), depresses people’s incomes in Chile (Andersen & Verner, 2010), and encourages within-country migration in Bolivia (Andersen, Lund, & Verner, 2010). Long-term econometric estimations constitute a means of capturing the various economic adjustments or adaptations that occur in response to climate change and can be interpreted as reflecting a type of “analog” approach to climate impact assessment (IPCC, 2016, Chapter 7). However, reduced-form econometric models are not designed to analyze resource allocation effects and the macroeconomic feasibility of human development policy options.
Human development policy options are better explored using a CGE model. This is particularly the case when the model specifies human development indicators and their determinants, including, for example, household income, private and public spending on basic services (education, health, water, and sanitation), and public investment in infrastructure. 6 The Maquette for Millennium Development Goal (MDG) Simulations (MAMS), for example, is a dynamic–recursive CGE model that has been used extensively to analyze financing strategies for meeting the MDGs (Lofgren, Cicowiez, & Díaz-Bonilla, 2013). However, only few CGE analyses with these characteristics examine climate policies.
Sánchez and Zepeda (2016) use MAMS to explore the space to impose a carbon tax in order to raise revenue and reduce carbon emissions in Bolivia, Costa Rica, and Uganda. Their scenarios show that the carbon tax hinders economic growth, but this unintended consequence would be offset if the newly collected revenue is used to step up public investments in transport and electricity infrastructure or, alternatively, in schools. Moreover, these investments would improve primary completion rates and reduce child mortality rates. This type of analysis is useful to explore coherent policy options for reducing GHG emissions while building climate resilience through a more equal access to basic services.
Other MAMS-based analyses show by how much 27 developing countries should have increased public spending to have achieved on time, by 2015, the MDGs in primary education, health, and water and sanitation (Sánchez & Cicowiez, 2014; Sánchez & Vos, 2013; Sánchez, Vos, Ganuza, Lofgren, & Díaz-Bonilla, 2010). These analyses also underscore the importance of analyzing financing options of public spending under existing macroeconomic constraints.
Socioeconomic Characteristics at the Household Level
Analyses at the microlevel using household surveys add value to integrated climate impact assessments. These analyses are useful for identifying households whose exposure and vulnerability are determined by specific socioeconomic characteristics such as gender, race, ethnicity, religion, or others. They need not be complex: They can rely on a single household survey or a panel of surveys and they may also be nonparametric to avoid complex econometric estimations.
Andersen and Cardona (2013) use the household survey for 2011 of Bolivia to construct indicators of vulnerability (and resilience) on the basis of level and diversification of income. They observe that the households that are particularly at risk of being vulnerable are young households with high dependency burdens, large households, urban households (given that in Bolivia income is relatively more diversified in rural areas) and households in indigenous communities. Using a panel of data from the Ethiopian Rural Household Survey (1994–2004), Dercon, Hoddinott, and Woldehanna (2005) find that female-headed households are particularly vulnerable to drought-induced shocks.
Bolivia: Effects of Policies on Per Capita Income, Vulnerability, and Resilience Under Microsimulation Scenarios.
Note. Vulnerable households have low levels of income and of income diversification. Resilient households do not live in poverty and their income is diversified. The thresholds that determine if a household is highly vulnerable or highly resilient are defined in Andersen and Cardona (2013).
Source. Reproduced from microsimulations based on the household survey for 2011 of Bolivia and the vulnerability methodology of Andersen and Cardona (2013).
Albeit useful, these microsimulation analyses do not consider the economic feasibility of the policies being simulated. To scrutinize this aspect, the analysis may be supplemented with a CGE model. In this case, the analysis usually begins with an understanding of the macroeconomic repercussions of the policy and its financial and macroeconomic feasibility using the CGE model. Key information on employment and income changes emanating from the CGE model is subsequently imposed on the household survey, in top-down manner, through microsimulations, in order to determine distributive impacts (Vos & Sánchez, 2010). The strength of this approach is that impacts are quantified for the full income distribution rather than across different types of representative household groups, as would be the case if a CGE model was used alone. Van Ruijven, O’Neill, and Chateau (2015) review existing methods for including income distribution in economy-wide models for long-term climate change research, including different microsimulation approaches.
In following this top-down approach, we took labor market results from the climate shock scenarios developed with Bolivia’s CGE model and imposed them on the 2013 household survey carried out by Bolivia’s National Statistical Institute. We find that income poverty rises as labor wages are hit adversely as a result of the climate-related productivity and infrastructure shocks, both in absolute and relative terms vis-à-vis other factors, particularly capital (Figure 4). Because the infrastructure shock is introduced abruptly, poverty increases much more in 2020 in this scenario compared with the productivity shock. The impacts on income distribution at the household level, as measured by the Gini coefficient (not shown here), are small because the shocks are designed to similarly affect all production sectors. The results are different in the case of the food price shock though. Because unskilled nonsalaried workers are most benefited by the food price hike, the poverty headcount ratio (Figure 4) as well as income inequality (not shown here) record a modest reduction. The adverse poverty effects from these shocks are ameliorated by public policy options provided that financing them triggers no adverse macroeconomic trade-offs (not shown here).
Bolivia: Poverty rate (US$2 per day) under climate shock scenarios (deviation from baseline scenario values).
The importance of analyzing the economic feasibility of policies is also underscored in other studies. Cicowiez and Sánchez (2011) apply a similar macro–micro modeling framework to simulate the impacts and feasibility of cash transfers targeting poor households in Latin American countries, in the face of economic shocks, including a hike in food prices. They find that while these transfers lead to a reduction in poverty and income inequality, financing and sustaining them largely depend on achieving sustained economic growth. Microsimulations were also run as part of the MAMS-based studies for the 27 developing countries cited earlier.
Sources of Inequality in Modeling Frameworks.
CGE = computable general equilibrium.
Source. Author’s construction based on the analysis presented.
Final Considerations and Areas for Further Research
It is critical to address the very core of the climate change adaptation challenge: The fact that the exposure and vulnerability of disadvantaged population groups to climate hazards is shaped by inequalities. Integrated climate impact assessments can play a critical role in this direction.
This article has systematically described the ways in which existing modeling frameworks can be integrated to more prominently feature within-country inequalities in assessing climate-related impacts. In doing so, it has pointed to the necessity of broadening the scope of assessments in order to make the most of existing modeling frameworks. It suggests to incorporate issues of adaptation and resilience, not only mitigation; assess climate hazards that afflict people in the short-term, not only in the long run; and consider the economy-wide feasibility of development policies for climate resilience, not just the cost–benefit of a single mitigation policy. In this way, a suite of modeling tools can be more readily deployed to improve the understanding of the following transmission mechanisms by which climate hazards—and policy responses to offset them—may affect vulnerable population groups:
climate-sensitive natural resource systems—using biophysical models; ownership and employment of production factors (land, capital, labor)—using economy-wide models; access to human capital, basic public services and resources (education, health, sanitation, infrastructure)—using economy-wide models; and a configuration of socioeconomic attributes that puts people at disadvantage—using household survey and microsimulation analysis, preferably in combination with economy-wide models.
An important area for further research relates to the engagement of relevant stakeholders. People and communities can provide information and share knowledge regarding existing socioeconomic factors that, in their belief, may be shaping inequalities that exacerbate their exposure and vulnerability to climate hazards. This feedback is critical for improving the capability to assess policy options that address inequalities so as to increase the resilience of people and communities.
Regional analyses of adaptation strategies point to the great potential of generating valuable information in this regard. A research team in Zimbabwe’s Nkayi region applied the regional integrated assessment framework developed by Agricultural Model Intercomparison and Improvement Project (AgMIP), with the aim of generating information on adaptation strategies for crop–livestock systems (Homann-Kee et al., 2015; Masikati et al., 2015). Researchers interacted with stakeholders including farmers in exploring and designing alternative sets of plausible future scenarios and climate change adaptation packages for integrated modeling. Scenario results show that the ownership of cattle affects the exposure to climate change of those farmers who possess it, whereas farms without cattle are poorer and more dependent on a single source of farm income and are thus more vulnerable to climate change. These results point to the importance of engaging with stakeholders, particularly communities (in this case, farm communities), to uncover the aspects of poverty and inequalities that are relevant for modeling analysis and for consideration of policy options. Although important steps are being taken in this direction, more research is still necessary.
Moreover, important data and statistics gaps need to be bridged to facilitate the identification of the most vulnerable population groups and communities whose feedback on inequalities may be critical to improving integrated climate impact assessments. Information to help identify characteristics of vulnerable populations at the local level in developing regions, where adaptation is most needed, is lacking. There is no systematic data available on the size of the populations groups most vulnerable to climate hazards, including their demographic characteristics and their livelihoods. 7 The regional studies developed by the AgMIP rely on their own farm surveys in different regions precisely because that type of information is not collected under standardized processes. There is also limited access to other important sources of information (e.g., global climate projections, geographic information systems, visualization of sea level and forest coverage). The international statistical community can play a fundamental role in helping to assess existing data and statistical capacity and in bridging the existing gaps to make it possible for the international community of natural and social scientists to develop integrated climate impact assessments for the most vulnerable populations at the local level in developing regions.
Further research to overcome these challenges is necessary at a critical time where countries have pledged efforts to meet the sustainable development goals that replaced the MDGs, as part of the 2030 Agenda for Sustainable Development that heads of State and Government adopted in the United Nations Headquarters, in New York, on September 25, 2015. One of the major challenges in the implementation of this global agenda is integrating the various facets of the environment into development policies. Based on the experience of the past decades, there is better understanding of the links between the economic and social dimensions of development. However, environmental concerns, in general, and the impact of climate hazards on people’s livelihoods, in particular, need to be better understood. Integrated climate impact assessments have an important role to play in improving this understanding.
Footnotes
Acknowledgments
I am grateful to Martín Cicowiez with whom I teamed up to assess climate scenarios for the Plurinational State of Bolivia whose results are analyzed in this article. I am thankful to the Economic Policy Unit of the Ministry of Development Planning in the Plurinational State of Bolivia with whom we collaborated to build an economy-wide model to assess climate scenarios. I am grateful to Lykke Andersen for her substantive inputs from a microeconomic analysis of vulnerability and resilience for the Plurinational State of Bolivia. I also appreciate the research assistance from Joanna Felix Arce and valuable comments from Diana Alarcón, John Antle, Alex Julca, Marcelo Lafleur, Friedrich Sotau, and Nicholas Sitko. The views and opinions expressed are those of the author and do not necessarily reflect those of the Food and Agriculture Organization of the United Nations.
Declaration of Conflicting Interests
The author declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The author received no financial support for the research, authorship, and/or publication of this article.