Climate Adaptation: Issue 2

The early work on adaptation to climate shocks in developing countries focuses on agricultural households’ management of risk from uncertain or abnormal rainfall through formal and informal financial markets (see Besley 1995 for a review of this work), labour supply decisions (Fafchamps 1993, Kochar 1999) and labour markets (Bardhan 1983), or migration networks and kinship (Rosenzweig and Stark 1989). This work finds that adaptation is costly in two ways:

1. Ex-ante mitigation responses from risk-averse farmers reduce productivity and profitability (Rosenzweig and Binswanger 1993, Rosenzweig and Wolpin 1993, Morduch 1995 for a review).
2. Ex-post adaptation is not complete, implying that they don’t ensure recovery to pre-shock levels of income or consumption (Fafchamps et al. 1998). 

The early papers theorise production functions, consumption decisions, and constraints to derive testable conditions on responses to weather uncertainty. These are corroborated empirically with fixed effects regressions using household panel data or regressions with controls from cross-sectional data. Recent work improves upon these empirical methodologies by employing randomised control trials (RCTs), quasi-random experiments from physical or policy variation, as well as short- and long-term shocks in weather (as described in the previous section) to evaluate a variety of adaptation and mitigation approaches.

This section summarises what we know about the effectiveness of different adaptation responses to climate events by elaborating on three modes of adaptation: (1) managing climate risk through financial products, (2) building resilience through technology adoption, innovation, and improved practices, and (3) the impacts of government policies.

Credit and insurance

The early approach to climate adaptation as a risk management problem naturally lends itself to evaluating the efficacy of financial products and markets. Here, we expand on the role of credit and insurance in building resilience to weather shocks and the importance of product design in addressing frictions to adaptation. Karlan et al. (2014) show that, in Ghana, access to weather insurance overcomes the downside risk in agricultural settings and leads to increases in the scale of cropping and input investments, enhances revenues, and facilitates consumption smoothing relative to cash grants. The authors find positive effects on cropping investments from insurance, even in the absence of negative weather shocks, and show that farmers are able to crowd-in investments when their weather risks are insured. Lane (2023) further highlights that a novel financial product that provides both insurance and liquidity, by guaranteeing access to credit in the event of a flood, improved ex-ante investments in inputs and ex-post consumption outcomes for farmers in Bangladesh. Aggregating findings from eight randomised experiments, Castaing and Gazeaud (2025) find modest average positive effects of crop index insurance on investments (0.06–0.11 standard deviations), but with substantial heterogeneity across contexts, limiting the ability to predict programme impacts in new settings. Recent work also explores the demand for and impacts of new insurance products for other weather shocks, such as extreme heat (Jalal et al. 2025, Jayachandran et al. 2025). Despite these positive effects of weather-indexed insurance, the take-up of such products remains low and highly price sensitive (Ahmed et al. 2020, Elabed and Carter 2014, Cole et al. 2013). Possible explanations for the low take-up of such insurance products include low trust, low liquidity to pay premiums, and limited salience of risk when farmers overweight recent events (McIntosh et al. 2019, Casaburi and Willis 2018, Karlan et al. 2014, Mobarak and Rosenzweig 2013).

Meanwhile, the literature considering non-agricultural firms has focused largely on liquidity constraints and credit interventions. De Mel et al. (2012) find that access to capital facilitates recovery and growth for firms exposed to natural disasters. Firms who randomly received access to cash grants recovered to their pre-disaster profit levels approximately two years before other affected firms who did not receive the grant. Annan et al. (2023) show that mobile money platforms enable firms to smooth over unexpected temperature and precipitation shocks in Ghana. They also show that if a weather forecast correctly predicts inclement weather on the previous day, then the harm from hot weather is substantially reduced for firms, indicating an ability to adapt. However, this financial avenue for climate adaptation is inhibited when firms face high levels of market competition. Jayachandran et al. (2025) elicit managers’ willingness to pay for a heat index insurance through firm surveys in Bangladesh, finding evidence that managers’ loss aversion affect insurance demand, particularly in smaller, family-owned firms. How climate insurance shapes firms’ decisions remains an underexplored area of research.

The benefits of credit in enabling firms to adapt can also be helpful for those in agriculture who are vulnerable to climate shocks. Macours et al. (2022) find that Nicaraguan households exposed to weather variability are able to offset the negative effects of drought shocks when provided with a bundle of conditional cash transfers with vocational skills training. The authors find that both interventions on their own were also able to improve resilience to droughts two years after the shock, although in different ways – CCT’s enhanced investments in businesses and skills training intensified wage-work and urban migration. For natural disasters, evidence from Bangladesh suggests that forewarning vulnerable households through cash transfers announced in advance of an anticipated disaster allows them to prepare in advance, minimises losses and supports recovery (Pople et al. 2023). Furthermore, Burgess et. al. (2017) find that bank expansion in rural India mitigates the mortality effects of heat by relaxing credit constraints.

Beyond direct private insurance, a small emerging literature highlights the benefits of sovereign insurance that provides contingent liquidity to governments. Del Valle et al. (2020) show that Mexican municipalities with access to an indexed disaster fund (FONDEN) experienced faster recovery after a natural disaster. Del Valle (2024) shows that infrastructure reconstruction financed by the fund is a potential mechanism and alleviated mortality effects of disasters. More evidence is needed to understand the determinants and impacts of sovereign and other forms of disaster insurance.

Finally, insurance could generate unintended behavioural responses that lead to socially inefficient levels of private adaptation investments. Annan and Schlenker (2015) find that subsidised crop insurance in the US reduced private adaptation investments and increased local vulnerability of crop yields to extreme heat, pointing to the presence of moral hazard. Gunnsteinsson (2020) finds similar evidence in field experiments with index crop insurance in the Philippines: insured plots experienced 28% higher harvest losses from pests and diseases compared to uninsured plots of the same farmer. However, farmers did not reduce prevention of typhoon and flood damage on insured plots, suggesting that moral hazard responses may depend on how far the insured risk can be controlled. Finally, insurance subsidy design can also shape adaptation incentives over time. Obolensky (2025) develops a dynamic land use model to study how US crop insurance subsidies interact with farmers' adaptation behaviours. Under the current design, subsidies are proportional to premiums and therefore increasingly flow to high-risk regions, reducing farmers' incentives to switch away from climate-sensitive crops. Targeted subsidies that adjust generosity based on regional climate risk trends encourage adaptive crop switching, despite generating distributional tensions across regions that may limit political feasibility.

Technology, practices, and innovation

New technologies and practices are a promising avenue for adaptation in both agricultural and non-agricultural settings. Improved seed varieties that build resilience to excessive rains have been shown to improve productivity and consumption outcomes, especially for more vulnerable groups (Dar et al. 2013). Emerick et al. (2016) show that the use of flood-resistant rice varieties expands cultivation, increases the takeup of modern cropping practices, and boosts the usage of inputs like fertilisers and credit in Eastern India. Eliminating the downside risk with this new technology not only impedes yield losses in the event of a flood but improves yields and farm revenues, even under normal conditions (similar to the results of reducing downside risk from insurance). Glennerster and Suri (2018) illustrate the positive effects of a short-duration, high-yielding rice variety that enables farmers to harvest during the wet season in Sierra Leone. The new rice variety improved yields, consumption, and health outcomes for children below the age of five. However, these positive effects were concentrated among farmers who also received high-touch training that addressed the germination sensitivity of the new seed, highlighting that the upfront costs of learning and adapting practices for new technologies could limit their returns and widespread adoption.

High costs of new technologies (Glennerster and Suri 2018), costly experimentation and learning (Foster and Rosenzweig 1995, Munshi 2004), heterogeneous returns to adoption (Dar et al. 2013, Suri 2011), or the absence of investments in innovations that are appropriate for lower-income countries (Moscona and Sastry 2023), could inhibit the adoption of new technologies that can potentially facilitate climate adaptation. Emerick and Dar (2021) find that bridging the information gap while building farmers’ trust through farmer-field days, where farmers can interact with each other and observe the performance of a new seed, increases take-up by 40%. On the other hand, demonstrating the seed through progressive adopters chosen during participatory village meetings has no effect on adoption. Through a lab in the field experiment, Shu and Zhang (2025) find that providing information on climate risks and adaptation benefits shifts farmers' beliefs and improves belief accuracy, making output-based subsidies (targeting sustainable production) more effective than input-based subsidies. That said, the effectiveness of this approach may be limited if other constraints, like liquidity, bind. For example, Glennerster and Suri (2018) find that providing high-yielding seed varieties for free led to a 97% take up as opposed to full-price offers (21% take up), highlighting the importance of liquidity constraints. Aker and Jack (2023) examine the role of liquidity constraints and information barriers in the take-up of water conserving technologies in rural Niger, finding that the training intervention alone was able to significantly increase take-up, with little additional effects from unconditional or conditional cash transfers.

Practices like crop diversification (Auffhammer and Carleton 2018) or conservation-oriented agriculture involving minimum tillage, mulching with crop residue, and crop rotation (Michler et al. 2019) can mitigate the negative impacts of rainfall shocks. However, their adoption is impeded by lower levels of productivity in normal years making these practices costly. Hultgren et al. (2022) further validate these results and show that practices that adapt to rising temperatures also depress yields during periods of moderate temperatures. They also find that farmers are less capable of adapting to variations in seasonal precipitation, especially in drier and hotter regions. One explanation for the restricted effectiveness of adaptation practices is that weather uncertainty skews farmers’ ability to evaluate and adopt the right practices. Burke and Emerick (2016) show that farmers in regions with lower variance in temperature, who could be more likely to diagnose climate change, also don’t make significant adaptive investments to subsequent temperature increases.

High fixed costs of private adaptation investments may also exacerbate the distributional effects of weather shocks. Abajian et al. (2025b) find that wealthy households were more likely to adopt private groundwater during the 2015 drought in Cape Town, which then eroded municipal utilities’ revenue base and shifted fiscal burdens onto middle and lower-income households. Beyond fixed costs, adaptation may require transitioning to new production technologies that operate at larger scales than existing practices. Haque (2025) finds that higher salinity in Bangladesh causes land reallocation to aquaculture, which is significantly greater in villages with lower land fragmentation, highlighting the role of land market frictions in determining adaptation.

Short-term adaptation approaches may sometimes bear other adverse consequences; for example, Fishman (2018) and Taraz (2017) show that while irrigation can somewhat offset the effects of variable precipitation, depleting groundwater levels leaves agriculture more vulnerable to shocks in the longer run (Hornbeck and Keskin 2014). Bruno et al. (2023) find that groundwater pumping for irrigation in California reduces drinking water access for rural households, highlighting the unequal incidence of agricultural adaptation practices. The returns to adaptive technologies such as irrigation can also be a function of frictions in other markets such as labour markets, leading to the substitution of inputs across space that discount the benefits of the technology as an adaptation strategy, making its adoption inefficient (Jones et al. 2022). Additionally, transitioning to non-agricultural sectors as groundwater wells run dry may be an incomplete and costly form of adaptation (Blakeslee et al. 2020).

Therefore, changing agricultural practices alone is not sufficient, but equipped with the right bundle of innovations, information, incentives, and insurance, new technologies and practices can be a powerful tool to mitigate climate risks. Encouraging locally-relevant agricultural innovations and building a policy space that facilitates their adoption are critical components of building local resilience (Suri and Udry 2022, Lybbert and Sumner 2012).

There is increasing research focus on how the quantity and direction of innovation may directly respond to extreme weather (see Dugoua and Moscona 2025 for a recent review). Moscona and Sastry (2023) find that extreme temperature exposure in the US induces developments of heat-resistant crops. They estimate that heat-induced directed innovation has offset 20% of potential losses in land values due to climate trends since 1960. Auci et al. (2021) similarly find that European agricultural firms redirect innovation in response to increased dispersion in temperature and precipitation. Long and Wang (2025) find that heat exposure by industrial firms in China increased patenting in both climate adaptation and mitigation technologies, with effects being driven by heat exposure of trading partners along the supply chain, rather than own exposure. While this evidence demonstrates the potential for directed innovation to ameliorate climate damages, developing countries may face barriers to adopting such innovations, including limited access to improved seed varieties, credit constraints, and weaker agricultural extension services. Furthermore, importing innovations may be of limited effectiveness if they are not adapted to local ecological conditions: Moscona and Sastry (2025) show that frontier agricultural technologies are systematically adapted to conditions in high-income countries, and that this ecological mismatch reduces technology transfer and agricultural productivity elsewhere.

Recent work has also tested how firms can adapt to weather and climate shocks (see Goicoechea and Lang 2026, Lang et al. 2025, Grover and Kahn 2024 for recent reviews). Adhvaryu et al. (2020) find that adopting LED lights improves worker productivity in garment factories by lowering the ambient temperature, indicating that climate mitigation technologies (e.g. energy efficient lighting) can also aid adaptation in this case. Furthermore, workers themselves may adapt to growing heat by taking more breaks and shifting occupations, which impacts aggregate productivity (Masuda et al. 2021, Rode et al. 2024). Using novel firm-manager surveys from Bangladesh, Pakistan, and India, Lang et al. (2025) provide estimates of the prevalence of firm-level adaptation in developing countries, finding that while 63% of firms report some adaptation, most rely on low-cost measures such as fans and building upgrades, with technology adoption and management practice changes remaining rare. They find that managerial capability, access to finance, and expectations about future weather exposure are strong predictors of adaptation investment. While technological improvements can aid adaptation, innovation in this space remains limited, with the share of climate adaptation inventions in 2015 roughly the same as those in 1995 (Glachant 2020).

Beyond firms, households also adapt to rising temperatures through cooling technologies. Air conditioning is one way for households to adapt to rising temperatures and reduce thermal discomfort, although its high costs imply concentrated adoption among richer households (Randazzo et al. 2021, Pavanello et al. 2021) and long-term consequences on burdening the electricity grid (Davis and Gertler 2015, Abajian et al. 2025a). Barreca et al. (2016) find that the gradual adoption of air conditioning helped explain the decline in mortality effects of extreme heat in the US over the 20th century, but the applicability of this finding to developing countries depends critically on electricity access, reliability, and affordability. Individual cooling may be insufficient when heat generates system-wide health spillovers. Aguilar-Gómez et al. (2025) find that extreme heat in Mexico creates hospital congestion and indirectly increases mortality among patients with non-heat conditions. More evidence is needed on the feasibility and effectiveness of alternative cooling strategies, such as improved housing design, urban greening, or community cooling centres.

Finally, technological progress has enabled more accurate forecasts for adverse weather conditions, with emerging evidence that shows forecasts facilitate adaptation and improve welfare. Molina and Rudik (2023) find that better forecasts in the US reduced property damages and mortality impacts of hurricanes, by enabling evacuation and preparation. Jagnani and Pande (2024) experimentally evaluate a community-based flood early warning system in Bihar, India, pairing Google's flood forecasting technology with local volunteer dissemination. Treatment communities received more alerts with improved overall accuracy, leading to greater trust in warnings, more proactive adaptation behaviour, and a 30% reduction in medical expenditures. Burlig et al. (2025) cross-randomise access to accurate monsoon onset forecast and weather-index insurance in Telangana, India. Treatment effects of forecasts vary by prior beliefs and induce farmers to tailor investments to expected conditions. Forecasts reduce risks at very low costs, with comparable effects to index insurance. While not directly focused on weather shocks, related evidence on forecasting systems for other environmental hazards suggests the broader applicability of findings above. In an RCT in Pakistan, Ahmad et al. (2025) find that short-term air pollution forecasts increase avoidance behaviour on high-pollution days and willingness to pay for defensive masks. Gantois et al. (2024) estimate the economic value of desert locust monitoring systems using data across sub-Saharan Africa, the Middle East and South Asia. They find that in-utero exposure to locust swarms, triggered by monitoring interruptions, reduces child height-for-age and increases stunting probability. More work is needed to understand the most effective strategies to disseminate forecasts and potential complementarities with other instruments to encourage adaptation.

Governance, market design, and socio-economic policy

Greenstone and Jack (2015) categorise four components that determine the effectiveness of climate response, especially in developing countries: low-income levels, high costs of adaptation, political economy considerations, and market failures. Government policy is potentially a formidable tool that can shape and manipulate these components. To this end, Kahn (2005) finds that economic development and higher-quality institutions (as measured by a democracy measure, inequality, ethnic fragmentation, and the World Bank’s good governance index) reduce deaths from natural disasters using cross-country panel data from 73 countries. 

Large-scale infrastructure investments are a major portion of government public good provision in low- and middle-income countries, with over 80% of all infrastructure investments being state-financed and close to half of these being transport expenditures (Foster et al. 2022). Investments in transport and connectivity can mitigate the impacts of extreme weather through trade and market access channels. Burgess and Donaldson (2010) find that the colonial-era expansion of railroads in India curtailed the likelihood of famines caused by low rainfall by slashing transport costs for domestic trade. Seasonal floods in Nicaragua reduce labour market income by 18%, but building bridges to connect areas cut off by the floods eliminates this effect by improving labour market activity, farm investments, and household savings (Brooks and Donovan 2020). Barwick et al. (2024) find that China’s high-speed railway expansion reduced air pollution exposure through increased travel to less polluted destinations, suggesting that transport infrastructure can facilitate individual adaptation by short-term migration.

In addition, recent work has made progress on valuing investments in climate adaptation infrastructure through examining housing market responses. Bradt and Aldy (2025) show that US flood levee protection increases protected home values but that there are negative flood risk spillovers to unprotected nearby areas. Benetton et al. (2025) similarly find that local house prices increase after the unexpected activation of the Venice sea wall in 2019. Hsiao (2025) studies commitment problems in the case of Jakarta’s proposed $60 billion sea wall, using a dynamic spatial model of urban development to evaluate welfare gains from alternative policies that alleviate moral hazard behaviours in private land markets.

Apart from these market channels that facilitate recovery, infrastructure investments can also improve first responses to climate shocks. Del Valle (2024) shows that Mexico’s indexed disaster fund accelerated the reconstruction of damaged infrastructure following a natural disaster, and that this infrastructure provision was central to mitigating mortality effects—areas with accessible medical facilities experienced significantly larger reductions in post-disaster deaths. While the indexed disaster funds provide a novel way to build contingent liquidity for governments to deal with disasters, large-scale infrastructure investments are costly and difficult to implement, especially in rural areas. Given these challenges, simpler and more scalable infrastructure may offer viable alternatives for climate-vulnerable rural populations. Barreto et al. (2025) show that the construction of one million rainwater harvesting cisterns in Brazil’s semiarid region, which saved time for water collection and improved residential water access, reduced individual reliance on cash transfers and improved labour market outcomes over ten years post construction.

Adaptation infrastructure that reduces aggregate damages may do so by concentrating costs on specific populations. Hu and Wang (2024) study China's Flood Detention Basins, which redirect floodwater to designated rural counties to protect urban areas. Through a quantitative spatial model, they find that the policy achieved aggregate welfare gains but lowered welfare in designated counties.

Safety nets provided by social protection policies may mitigate the costs of climate shocks for vulnerable, poorer populations who otherwise lack the wherewithal to cope. Asfaw et al. (2017) find in a field experiment in Zambia that cash transfers mitigated the effects of adverse rainfall shocks on consumption and food security. Garg et al. (2020b) show that workfare programmes, such as the employment guarantee programme in India, can attenuate the negative effect of rising temperatures on children’s learning by supplementing household income. Adhvaryu et al. (2024) similarly find that the Mexican Progresa programme attenuates the negative effects of extreme rainfall on children’s educational and employment outcomes.

Liquidity transfers are, however, not a universal solution to facilitating climate resilience. Banerjee and Maharaj (2020) find that providing last-mile healthcare through community health workers in India was more effective in suppressing the negative impacts of rising temperatures on infant mortality than income gains from the same employment guarantee programme in India. Garg et al. (2020b) also show that even though cash transfers attenuate homicide crimes that are driven by rising temperatures in the short run, the homicide rate reverses back to its original levels within five years of receiving access to a Progresa grant.

Emerging evidence also suggests that asset transfer poverty reduction (‘graduation’) programmes can improve households’ resilience to weather events. Bandiera et al. (2025) leverage the randomised implementation of a large-scale graduation programme in Bangladesh to examine how it affects resilience of extreme poor households to flood and drought events. They find that the programme supports consumption and keeps recipients out of poverty in the face of unexpected weather shocks, stemming from beneficiaries' increased ability to maintain and diversify productive assets and labour activities, rather than drawing down transferred assets when exposed to shocks. Hirvonen et al. (2023) leverage the randomised rollout of Ethiopia's Productive Safety Net Programme and find that recipients experienced much smaller impacts of drought shocks on food security and consumption, although mostly through treated households drawing down buffer savings to smooth consumption. Avdeenko and Frölich (2025) conduct an experimental evaluation of a humanitarian aid preparedness programme in rural Pakistan, finding that participating communities experienced significantly better food security and health outcomes after extreme monsoon flooding in 2016 compared to a control group. More work needs to be done to understand the impacts of transfer programmes over time, when cash transfers are effective, and how social policy can be bolstered to equip vulnerable communities to face climate shocks. 

Beyond direct transfers, market-based mechanisms can also serve as important tools for climate adaptation, though their effectiveness depends on market structure and the absence of distortionary policies (Greenhill et al. 2026). Rafey (2023) shows that water markets in Australia improve welfare during droughts by enabling flexible reallocation of water resources compared to rigid allocation rules. Garcia and Belmar (2025) find that Chilean water markets only achieve allocative efficiency when supported by local enforcement institutions, despite well-defined and tradable property rights. Kochhar et al. (2025) offer a cautionary tale about how government-imposed market distortions can undermine farmers’ ability to adapt to weather shocks. Studying India’s APMC Acts, which restrict farmers to selling only to licensed intermediaries within state borders, they find that farmers in areas with higher competition achieve higher output per degree-day of extreme heat. Higher competition causes prices to rise more after heat shocks, incentivising farmers to increase inputs. The findings demonstrate that the design of market institutions directly shapes farmers’ adaptation behaviours.

A fundamental consideration in discussion of the government’s role in building climate resilience is whether their own incentives are (1) aligned with promoting climate adaptation, and (2) flexible enough to respond to changing climate conditions. Government responses to weather shocks can also be driven by political economy concerns that operate during times of vulnerability from climate shocks (Amirapu et al. 2022, Balboni et al. 2021, Bobonis et al. 2022, Cole et al. 2012, Fitch-Fleischmann and Kresch 2021, Mahadevan and Shenoy 2023, Tarquinio 2023). Given limited climate literacy, public concerns and support for adaptation policies tend to be driven by lived experiences rather than scientific attribution (Grossman et al. 2025). Coordination problems between sub-national jurisdictions persist, leading to continued negative externalities that depreciate essential resources like clean air, water, and forests (Lipscomb and Mobarak 2016, Dipoppa and Gulzar 2023, Bhogale and Khedgikar 2023, Burgess et al. 2012, Wang and Wang 2021).

The recent literature thus highlights the progress in climate adaptation methodologies and sheds light on financial, physical, and policy instruments that are effective. However, a number of these entail short-term costs and uncertainty about long-term benefits. Climate adaptation, along with our understanding of what works, still remains substantially incomplete. Moreover, the studies reviewed so far largely examine adaptation decisions in isolation. The effectiveness of these responses depends critically on interactions across agents, sectors, and regions. Climate shocks can propagate through channels such as trade networks, migration flows, and financial markets, meaning that the impacts experienced in one location – and the returns to adaptation there – are shaped by conditions and responses elsewhere. In the next section, we review recent work examining these issues using quantitative spatial general equilibrium models to study climate change adaptation.

For full reference list see the end of the conclusion chapter.