Table 2.

Summary of selected publications.

Authors and publication year	Method	Location	Main results (summary)
Aragão and Carvalho (2018)	Quantitative: Regression analysis A regression analysis is used to determine the association between climate and dengue fever incidence using a database containing spatial-temporal information on human health, climate modelling outputs, sea level rise data, population density and topography.	Brazil	The authors estimate 36,134 hospitalizations due to dengue fever in Sao Paulo State between 2008 and 2015, with 40.5% of these registered in 2015. They document a correlation between El Niño/increased temperatures and dengue hospitalizations. Using back-of-the-envelope calculations, they estimate the cost of hospitalizations due to dengue outbreaks. With 300 cases per 100,000 inhabitants in 2010 and assuming a population of 400,000 and a cost of BRL 600 per patient (USD 341), the cost of dengue treatment in 2010 was BRL 720,000 (USD 410,000). These costs are likely much higher because they do not consider other diseases associated with the same vector, such as Zikka and Chicungunha, or emergency actions to mitigate the outbreaks.
Bell et al. (2006)	Quantitative: Economic valuation methods of environmental interventions Economic valuation of avoided health effects using willingness to pay (WTP) and Cost-of-illness (COI) approaches.	Mexico, Chile, and Brazil	The authors estimate ozone and particulate matter concentrations for Santiago, Sao Paulo, and Mexico City under two different scenarios: business as usual and air pollution control policies. They find that air pollution control would have vast health benefits for the three cities, averting numerous adverse health outcomes, including over 156,000 deaths, 4 million asthma attacks, 300,000 children's medical visits, and almost 48,000 cases of chronic bronchitis. The economic value of avoided health impacts is roughly USD 21–165 billion.
Campanharo et al. (2019)	Quantitative: Linear regression and uncertainty propagation methods The total burnt area is estimated using the 2008 biomass map. The loss is directly associated with infrastructure damages, production losses, and costs related to crop reestablishment and future production. The authors find an association between the total burnt area and the total hospitalization cases by municipality and year. They quantify the indirect costs of CO2 emissions from the biomass loss maps and respiratory morbidities.	Brazil	Droughts increase the probability of fire events, which raises the annual estimated cost related to infrastructure damages, agricultural production losses, CO2 emissions, and respiratory morbidities around fifteen-fold in Brazil's Acre state compared to an average climatological year. Estimates represent approximately 7% ± 2.45% of Acre's GDP. Quantifying social fire impacts related to respiratory morbidity is complex. The difficulty to establish a robust diagnostic of cause and effect, with relation to fires and respiratory illness, introduces large uncertainties for defining the proportional contribution of these costs to the total economic loss estimates.
Carreras et al. (2015)	Mixed method: Time series analysis and field research The data were analyzed using Generalized Additive Models (GAM) with a quasi-Poisson distribution link function. Heterogeneous analysis by season, age groups, and socioeconomic status.	Argentina	The paper assesses the impact of the daily temperature range on population morbidity in Cordoba-Argentina. Higher temperatures increase upper and lower respiratory morbidities, especially among the elderly, less-educated individuals, and poor living conditions.
Cromar et al. (2021)	Quantitative: FUND model Using the FUND model the authors evaluate the health-based portion of the existing social cost of carbon. They did a separate analysis for low, middle, and high-income countries. In addition to the base model, three additional experiments assessed the sensitivity of these estimates to changes in the socioeconomic assumptions in the model.	Middle-income countries/South America	Economic impacts from adverse health outcomes represent 4.4% of the current social cost of carbon in middle-income countries, including South American countries. 7.2% of these health impacts are attributable to diarrhoea mortality and morbidity. The results of the socioeconomic experiments show that the health-based portion of the social cost of carbon estimates is very sensitive to assumptions regarding income elasticity of health effects, income growth, and use of equity weights.
Desbureaux and Aude-Sophie (2019)	Quantitative: Linear probability model The paper estimates the effect of droughts on the probability of labour market outcomes using a linear probability model. It exploits random variations in weather to define rainfall and drought shocks and controls for temperature, precipitation, economic activities, and fixed effects for unobserved characteristics and time variations.	78 Latin American cities	Large sustained dry events decrease the probability of being employed, hourly wages, hours worked, and labour income of urban workers, especially for informal workers. The impacts of droughts are higher than the effects of wet events. Workers' health (e.g., increased cases of diarrhoea) and power outages could explain the link.
Dessus and O'Connor (2003)	Quantitative: CGE Uses a CGE model to link policy changes (e.g., a pollution tax) to emissions reductions. Develop several links in a causal chain from climate policy to welfare changes. Dispersion model specified by WHO 1989 to link emissions reductions to changes in ambient. Thus, they can identify exposure and health effects, building epidemiological evidence for some air pollutants in Santiago. These effects are valued in monetary units to welfare gains such as willingness to pay (WTP) for reduced mortality risk and other health improvements (morbidity).	Chile	Results suggest that, even with the most conservative assumptions (low WTP, low elasticities), Chile could reduce CO2 emissions by almost 20% from the 2010 baseline with no net welfare losses. However, a 10% reduction is closer to "optimal". If, instead, Chile was to target a 20% reduction in particulate matter concentrations, a particulate tax would incur slightly lower costs than an equivalent carbon tax to achieve the same health benefits. While the latter is a second-best solution for addressing local pollution, carbon credit sales could fully compensate for the welfare loss of choosing this instrument at a world market price of USD 20/tC.
Ebi (2008)	Quantitative: DALYs and HadCM2 model This study estimates Disability Adjusted Life Years Lost (DALYs) related to the annual number of cases of diarrheal diseases. It uses HadCM2 general circulation climate model to simulate the relative climate change risks.	Latin America	The paper estimates the worldwide costs of treating additional cases of malnutrition, diarrheal disease, and malaria attributed to climate change by 2030 to range between USD 4 and USD 12 billion. They use current treatment costs, assume no population or economic growth, and no adaptation measures undertaken. The authors separately estimate costs for different WHO regions, including Latin America. The three diseases pose significant risks for future populations, particularly in low-income countries in tropical and subtropical regions.
Fishman et al. (2019)	Quantitative: Regression analysis for causal inference The empirical specification exploits weather shocks to investigate the effect of temperature around the time of birth on adult earnings. The model incorporates location-fixed effects to capture the long-term, month-specific expected weather.	Ecuador	Elevated in-utero exposure to high temperatures harms formal sector earnings in Ecuador. Individuals who experience in-utero temperatures that are 1 °C above average are less educated and earn 0.7% less as adults. Extrapolation of estimates suggests that future warming may have additional economic impacts that have not been sufficiently appreciated to date.
Hasegawa et al. (2016)	Quantitative: CGE and DALY The study measures changes in morbidity and mortality due to nine diseases caused by being underweight with changes in the labour force, population, and healthcare demands. It conducts a simulation and assesses the value of lives lost and the willingness to pay to reduce the risk.	Latin America	The paper quantifies the impact of climate change on human health through undernourishment, which impacts nine diseases (diarrheal diseases, pertussis, measles, tetanus, meningitis, malaria, lower respiratory infections, birth asphyxia and birth trauma, and protein-energy malnutrition). They find that the economic value of healthy lives lost to undernourishment as a result of climate change ranged from 0.4% to 0.0% of the world's gross domestic product (GDP) and varied regionally, from 4.0% to 0.0% of regional GDP in 2100. Contrarily, the actual economic losses caused by increased healthcare costs and the decline in the labour force brought on by undernourishment as a result of climate change, respectively, corresponded to changes of 0.1%–0.0% in GDP and 0.2%–0.0% in household consumption at the global level. These changes are close to 0% for Brazil and the rest of Latin America
Markandya and Chiabai (2009)	Mixed method: CE and CB analysis and systematic review The study integrates cost valuation methodologies (i.e., CE and CB) with epidemiological estimates about the proportion of people exposed to malaria (using climatic modelling and clinical evidence of incidence) and projections of population growth rates (unit costs have been calculated in each country for the baseline year).	Latin America	The cost per death avoided through disease control programs focusing on combined health interventions is USD 300–600. The costs per life saved in the case of diarrhoea are considerably lower than those of malaria.
Nakano (2018)	Quantitative: LCA The study uses the life cycle assessment framework for adaptive planning to climate change to evaluate the potential climatic effects on industries throughout the supply chain. To quantify the involvement of workers throughout the supply chain, they use multi-regional input-output tables	Brazil	In a multi-country study, only Brazil from South America, the authors identify the industries more vulnerable to the effects of dengue fever. They find that more than 70% of workers in major industries in Brazil would be at risk in 2030, directly impacting the construction, textile, and hotel/restaurant sectors.
Pattanayak et al. (2009)	Quantitative: CGE model The paper analyses a Brazilian policy to expand National Forests (FLONAS) by 50 million hectares. It measures health impacts in a CGE model (baseline scenario vs FLONAS) via the labour market, including labour-leisure trade-offs and productivity declines in the workplace. The econometric estimates, OLS, and propensity score matching of health effects are translated into reductions in labour endowments by converting additional cases of morbidity and mortality into "healthy" years lost because of disability, represented as a percentage change in labour endowment.	Brazil	The baseline scenario starts with a 0.3 per cent (urban) to 0.6 per cent (rural) reduction in labour stock because of increased climate-related diseases and 3 million hectares of forest cleared annually. FLONAS conservation scenario mediates the health effects while reducing the land available for agriculture and pasture. FLONAS conservation scenario, compared to the "no action baseline" of climate change and deforestation, suggests a relatively small but negative impact on GDP (−0.1 per cent) by reducing agricultural output and other key macro indicators (investment, exports, imports, and earnings). FLONAS scenario improves health but lowers incomes in frontier regions of the Brazilian Amazon.
Rocha and Soares (2014)	Quantitative: Regression analysis for causal inference The paper analyses the health impacts of rainfall fluctuations during the gestational period using a municipality-by-month of birth panel. Heterogeneous effects on infant mortality by GDP, water, and sanitation coverage. Cost-effectiveness analysis focused on mortality before age 1 of expanding water and sanitation services coverage.	Brazil	Adverse rainfall shocks/droughts increase infant mortality, lower birth weight, and shorten gestation periods. An increased incidence of intestinal infections and malnutrition explains mortality effects. They are minimized when the local public health infrastructure is sufficiently developed (municipality coverage of piped water and sanitation). Higher effects are documented during the foetal period (2nd trimester of gestation). Estimates suggest that expansions in public health infrastructure would be a cost-effective way of reducing the response of infant mortality to rainfall scarcity.
Takakura (2017)	Quantitative: AIM/CGE model The paper uses the Asia-Pacific Integrated Model/Computable General Equilibrium (AIM/CGE) model as a core tool to estimate the future macroeconomic cost of workplace heat-related illness prevention.	Latin America	The paper estimates the economic costs for workplace heat-related illness prevention through worker breaks under various climate scenarios and socioeconomic conditions. This action reduces work time and labour productivity, which is economically costly. Under a high-emissions scenario (RPC 8.5), GDP loss is expected to range between 2.6% and 4.0% of global GDP in 2100. In Brazil and the rest of Latin America, these costs range from 0.31 to 3.4% and 0.86 to 1.76% of GDP, respectively. The construction and primary sectors are the most affected. Socioeconomic development and climate mitigation can reduce these costs. Even if temperature increases are limited and socioeconomic development is achieved, costs are non-negligible due to outdoor work. Adaptation measures that can be applied to outdoor work should be quantitatively investigated.
Tol (2005)	Quantitative: FUND model The study makes simulations for nine groups of countries using the Climate Framework for Uncertainty, Negotiation and Distribution model with climate projections from the IPCC.	Latin America	The paper tests if development aid is more effective in reducing the impacts of climate change than greenhouse gas emission reductions. Using a multi-region simulation based on the FUND model, the authors find that investing in development, particularly aid, which targets vector-borne infectious diseases in poorer countries, is a better strategy for reducing the impacts of climate change than greenhouse gas emission reduction. These investments reduce vulnerabilities in some sectors, such as infectious diseases, water resources, and agriculture. Latin America and Africa enter the group of countries in which this is the case.
Zamand and Hyder (2016)	Quantitative: Probit Outcomes of interest: enrolment, three cognition and psychometric development indicators, two anthropometric measures of health and nutritional status (BFA z-score and HFA z-score). Probit estimation using Young Lives Project.	Peru	The paper analyses the impact of self-reported exposure to droughts and floods on human capital, measured by educational and health outcomes, for children aged 14–16 years in Peru. Only a negative association is found between droughts and Peabody Picture Vocabulary Test scores, while negative but non-significant results for other educational outcomes such as school enrolment, mathematics test scores, and health measures such as body mass index or height-for-age z-scores were documented.
Bakhtsiyarava et al. (2022)	Quantitative: Multi-level distributed lag non-linear model Using live birth data from SALURBAR for cities in Brazil, Chile, and Mexico and monthly average ambient temperature from ERA5, the authors use a non-linear distributed lag model controlling for sex, season of conception, and calendar year of child’s birth; controlled for maternal age, education, partnership status, presence of previous births, and climate zone; and included a random term for the sub-city of mother’s residence. They use other findings in the literature to infer losses in birthweight and future earnings, but they do not report specific numbers for cities in the sample.	Brazil, Chile, Mexico	Higher temperatures (relative to the 19 °C reference) during gestation are associated with lower birth weight, particularly in Mexico and Brazil. The cumulative effect appears to be driven by stronger associations during the last months of gestation. Studies have shown that a 10% increase in birth weight is associated with a 0.9% increase in earnings and a 1.2% increase in high school graduation (Black et al. 2007).
Madeira (2022) In Spanish	Literature Review Review estimates from the literature to compare Chile's economic impacts of climate change to other South American and OECD countries.	Chile	The literature shows that climate change will not significantly affect Chile’s GDP and labour productivity compared to other Latin American and OECD countries under different RCP scenarios. In contrast, Brazil could lose approximately 10% of its labour productivity. Climate change, particularly droughts, increases mortality risks, especially in Chile’s central zone. Estimates for Chile indicate that mortality costs of climate change could be equivalent to 3.2% of GDP and 0.9% of health-related costs. Chile is highly vulnerable to air pollution mortality.
Freitas et al. (2020) In Portuguese	Accounting Exercise Analysis of the economic impact of natural disasters on healthcare facilities in Brazil from 2000 to 2015. They use data from 15,950 disaster register forms, categorizing them according to the Brazilian Classification and Coding of Disasters (COBRADE). The study primarily focused on costs (in Brazilian Reais, R$) associated with damages to healthcare infrastructure. Forms lacking healthcare damage data and those with missing values were excluded. The study adopted COBRADE's disaster classification, and the results provide insights into the financial implications of different disaster types on healthcare facilities in Brazil, considering their location and extent of damage.	Brazil	The authors calculate that the total cost of disasters in healthcare facilities between 2000 and 2015 was almost 4 million R$ (Brazilian Reais). Regarding the type of disaster that affected healthcare facilities, climatological disasters (droughts, forest fires, cold spells) were the most common, accounting for 56.2% of occurrences, followed by hydrological (floods, landslides) (34.9%), meteorological (thunderstorms, hailstorms, windstorms) (8%), and geological (landslides) (0.9%) disasters. Despite frequent climatological disasters, they represented only 0.3% of total costs. In contrast, though less frequent, hydrological disasters accounted for 88.5% of the total costs. Hydrological disasters are 3.2 and 3.6 times costlier than other disasters. Notably, while making up 49.3% of events with cost data, meteorological disasters contributed only 9.9% to the total costs. The study also examined costs by region, revealing variations in disaster impact and financial burden across different parts of Brazil.
Da Cruz et al. (2016) In Portuguese	Quantitative: Correlation exercise Ecological time-series study to examine the relationship between healthcare parameters and meteorological factors in São Carlos, São Paulo, Brazil, from 2008 to 2012. Monthly and annual data on all hospital admission authorization numbers (AIH), mortality rates, hospital mortality rates, and spending categorized by disease group (according to ICD-10) from the Unified Health System's Hospital Information System. Meteorological data on temperature and humidity from the National Institute of Meteorology (INMET). Spearman's correlation coefficient was used to evaluate correlations between meteorological factors and healthcare outcomes, classified based on the correlation strength.	Brazil	Between 2008 and 2012 in São Carlos, 7144 hospital admissions were recorded under different disease categories (DR). Correlation analysis showed a moderate negative correlation between hospital admissions and average compensated temperatures and a weak negative correlation with minimum temperatures. Mortality rates showed a weak positive correlation with average compensated and maximum temperatures, while costs had a weak negative correlation with average compensated, maximum, and minimum temperatures.
Pereira et al. (2014) In Portuguese	Quantitative: Cost of illness Partial economic evaluation using a relative cost-of-illness methodology for dengue cases attributed to the disaster in Nova Friburgo on January 11–12, 2011. Dengue and leptospirosis were the most intensified diseases post-disaster and the focus of specific epidemiological control measures. Data sources include restricted-access secondary data from the Municipal Health Foundation of Nova Friburgo, covering epidemiological reports, financial reports, and dengue notification records. The study assesses costs to the health system (hospital and outpatient procedures) and societal costs (loss of productivity).	Brazil	A total of 937 confirmed dengue cases and 419 discarded cases were reported in Novo Friburgo attributed to the natural disaster favoring vector maintenance and circulation. Treatment costs through the healthcare system were estimated at R$58,341.97 (Brazilian Reais) for treating suspected cases, excluding diagnostic tests, of which R$45,791.97 corresponded to hospitalized cases and R$12,550.00 for out-of-patient care. The authors estimate that hospitalization cases translated into 175 days of absenteeism for economically active individuals, resulting in estimated productivity losses ranging from R$5652.60 to R$9673.74, depending on income references. Among confirmed out-of-patient care, productivity losses varied from R$20,820.00 to R$312,300.00, depending on income references. The total cost of the illness, including healthcare and societal costs, ranged from R$48,865.18 to R$580,021.54.
Wen et al. (2023)	Quantitative: Association between ambient temperature and productivity loss Analyzing data on age, sex, date of death, and the primary cause of death for the working-age population in Brazil. All-cause mortality data was classified according to the Tenth Revision of the International Classification of Diseases (ICD-10). Meteorological data, including temperature and relative humidity, were obtained from the European Centre for Medium-Range Weather Forecasts Reanalysis dataset. PM 2.5 data was estimated using machine learning models that combined information from chemical transport models, meteorological records, and air quality stations. The association between ambient temperature and productivity loss was assessed using a two-stage time-series analysis, with a distributed lag nonlinear model (DLNM) applied to evaluate the relationship. Sensitivity analyses were conducted to test the robustness of the results.	Brazil	The study assesses productivity losses due to non-optimal temperatures in Brazil. They use working-age population (15–64 years), all-cause mortality, meteorological data, and PM 2.5 data. The study estimates Productivity-Adjusted Life-Years (PALYs) lost per 100,000 residents from 2000 to 2019. Results show a significant productivity burden associated with non-optimal temperatures, with variations observed across different geographical areas, age groups, and sexes. The findings highlight the need for tailored policies and adaptation strategies to mitigate the impacts of climate change on labour capacity and social development in Brazil.