Heat adaptation in central Asia: Household cooling choices

Sulaimanova, Burulcha; Azhgaliyeva, Dina; Holzhacker, Hans; Øverland, Indra Working Paper Heat adaptation in central Asia: Household cooling choices ADB Economics Working Paper Series, No. 787 Provided in Cooperation with: Asian Development Bank (ADB), Manila Suggested Citation: Sulaimanova, Burulcha; Azhgaliyeva, Dina; Holzhacker, Hans; Øverland, Indra (2025) : Heat adaptation in central Asia: Household cooling choices, ADB Economics Working Paper Series, No. 787, Asian Development Bank (ADB), Manila, https://doi.org/10.22617/WPS250245-2 This Version is available at: https://hdl.handle.net/10419/322384 Standard-Nutzungsbedingungen: Die Dokumente auf EconStor dürfen zu eigenen wissenschaftlichen Zwecken und zum Privatgebrauch gespeichert und kopiert werden. Sie dürfen die Dokumente nicht für öffentliche oder kommerzielle Zwecke vervielfältigen, öffentlich ausstellen, öffentlich zugänglich machen, vertreiben oder anderweitig nutzen. Sofern die Verfasser die Dokumente unter Open-Content-Lizenzen (insbesondere CC-Lizenzen) zur Verfügung gestellt haben sollten, gelten abweichend von diesen Nutzungsbedingungen die in der dort genannten Lizenz gewährten Nutzungsrechte. Terms of use: Documents in EconStor may be saved and copied for your personal and scholarly purposes. You are not to copy documents for public or commercial purposes, to exhibit the documents publicly, to make them publicly available on the internet, or to distribute or otherwise use the documents in public. If the documents have been made available under an Open Content Licence (especially Creative Commons Licences), you may exercise further usage rights as specified in the indicated licence. https://creativecommons.org/licenses/by/3.0/igo/ ASIAN DEVELOPMENT BANK ASIAN DEVELOPMENT BANK 6 ADB Avenue, Mandaluyong City 1550 Metro Manila, Philippines www.adb.org ADB ECONOMICS WORKING PAPER SERIES NO. 787 June 2025 Heat Adaptation in Central Asia Household Cooling Choices This study examines how households in the Kyrgyz Republic, Tajikistan, and Uzbekistan adapt their cooling strategies to power outages and increasing temperatures. It provides insights into the socioeconomic and behavioral dimensions of energy resilience in Central Asia. It notes the importance of a reliable power supply and the potential of solar panels to meet summer energy demands. About the Asian Development Bank ADB is a leading multilateral development bank supporting inclusive, resilient, and sustainable growth across Asia and the Pacific. Working with its members and partners to solve complex challenges together, ADB harnesses innovative financial tools and strategic partnerships to transform lives, build quality infrastructure, and safeguard our planet. Founded in 1966, ADB is owned by 69 members—50 from the region. HEAT ADAPTATION IN CENTRAL ASIA HOUSEHOLD COOLING CHOICES Burulcha Sulaimanova, Dina Azhgaliyeva, Hans Holzhacker, and Indra Overland ASIAN DEVELOPMENT BANK The ADB Economics Working Paper Series presents research in progress to elicit comments and encourage debate on development issues in Asia and the Pacific. The views expressed are those of the authors and do not necessarily reflect the views and policies of ADB or its Board of Governors or the governments they represent. ADB Economics Working Paper Series Burulcha Sulaimanova, Dina Azhgaliyeva, Hans Holzhacker, and Indra Overland No. 787 | June 2025 Burulcha Sulaimanova ([email protected]) is head of Research and Training Department, OSCE Academy, Bishkek. Dina Azhgaliyeva ([email protected]) is a senior economist at the Economic Research and Development Impact Department, Asian Development Bank. Hans Holzhacker ([email protected]) was chief economist and is currently a consultant at Central Asia Regional Economic Cooperation Institute. Indra Overland ([email protected]) is head of the Centre for Energy Research, Norwegian Institute of International Affairs and a research associate at the Oxford Institute for Energy Studies. Heat Adaptation in Central Asia: Household Cooling Choices Creative Commons Attribution 3.0 IGO license (CC BY 3.0 IGO) © 2025 Asian Development Bank 6 ADB Avenue, Mandaluyong City, 1550 Metro Manila, Philippines Tel +63 2 8632 4444; Fax +63 2 8636 2444 www.adb.org Some rights reserved. Published in 2025. ISSN 2313-6537 (print), 2313-6545 (PDF) Publication Stock No. WPS250245-2 DOI: http://dx.doi.org/10.22617/WPS250245-2 The views expressed in this publication are those of the authors and do not necessarily reflect the views and policies ofthe Asian Development Bank (ADB) or its Board of Governors or the governments they represent. ADB does not guarantee the accuracy of the data included in this publication and accepts no responsibility for any consequence of their use. The mention of specific companies or products of manufacturers does not imply that they are endorsed or recommended by ADB in preference to others of a similar nature that are not mentioned. By making any designation of or reference to a particular territory or geographic area inthis document, ADB does not intend to make any judgments as to the legal or other status of any territory or area. This publication is available under the Creative Commons Attribution 3.0 IGO license (CC BY 3.0 IGO) https://creativecommons.org/licenses/by/3.0/igo/. By using the content of this publication, you agree to be bound bytheterms of this license. For attribution, translations, adaptations, and permissions, please read the provisions andterms of use at https://www.adb.org/terms-use#openaccess. This CC license does not apply to non-ADB copyright materials in this publication. If the material is attributed toanother source, please contact the copyright owner or publisher of that source for permission to reproduce it. ADB cannot be held liable for any claims that arise as a result of your use of the material. Please contact [email protected] if you have questions or comments with respect to content, or if you wish toobtain copyright permission for your intended use that does not fall within these terms, or for permission to use theADB logo. Corrigenda to ADB publications may be found at http://www.adb.org/publications/corrigenda. Note: ADB recognizes “China” as the People’s Republic of China. ABSTRACT This study investigates factors influencing household cooling choices in Central Asia, focusing on air-conditioning and fans/sunscreen films. Using data from the 2023 “Household Access to Energy in the Fergana Valley” survey in the Kyrgyz Republic, Tajikistan, and Uzbekistan, the analysis employs a multinomial probit model to examine socioeconomic, environmental, and power supply factors. Across the three countries, it finds that 48% of households use fans or sunscreen films (without air-conditioning), 30% use no cooling, and 22% use air-conditioning, noting significant variations between countries. Cooling degree days (CDD) significantly impact cooling appliance adoption, with higher CDD regions more likely to use cooling solutions. Power outages negatively affect air-conditioning adoption but not fans/sunscreen films, highlighting the importance of power stability. Robustness checks confirm that power supply reliability is crucial for cooling choices. The findings suggest policy implications, including the potential of solar panels to meet summer energy demands. This research underscores the need to address power sector reliability and climate adaptation in vulnerable regions. Keywords: heat waves, environmental extremes, infrastructural adaptations, power outages, cooling technologies, Central Asia JEL codes: Q41, R21, Q54 _______________________ Acknowledgment: Authors are grateful to Altynai Tolenova, a visiting research fellow at Norwegian Institute of International Affairs (NUPI) and Nomin Batsukh, a visiting OSSE fellow at NUPI, both part of the Climate and Energy Research Group, for their excellent assistance with weather data collection. Authors are also grateful to all participants of hybrid public lecture at OSCE Academy in Bishkek on 25 April 2025. Special appreciation to Ayat Ullah, postdoctoral research fellow at OSCE Academy in Bishkek, for his detailed discussion of this paper. 1. Introduction The rapid global temperature rise due to climate change has significantly increased demand for cooling technologies, particularly in regions susceptible to extreme heat waves (Thomson et al., 2019; Attia et al., 2021; Pavanello et al., 2021; Zhang et al., 2020). Central Asia, with its rising summer temperatures, faces growing risks of heat stress (Wang, et al., 2023), highlighting the urgent need for effective adaptation strategies, particularly in households. While much research on the adaptation of Central Asian households has focused on the broader impacts of heat waves on agriculture (Liu, et al., 2020; Li et al., 2020) and public health (Tursumbayeva et al., 2023), less attention has been given to how households in the region adapt to extreme heat, particularly through their cooling choices. This gap is significant since household cooling strategies play a crucial role in mitigating the adverse effects of heat, promoting energy resilience, and reducing the health risks associated with high temperatures (Lee et al., 2024; Mushore et al., 2017). Moreover, Central Asia’s energy access landscape is often characterized by frequent power outages and grid instability, particularly in rural areas (Mehta et al., 2024), although it has 100% electricity access. These interruptions may pose a significant barrier to the adoption of some cooling technologies. Understanding how these factors interact is crucial for informing policies that promote energy efficient cooling solutions while improving household resilience to heat stress. This study investigates the factors that influence household cooling choices, focusing on the adoption of air-conditioning, fans, and sunscreen films, which are the primary cooling options utilized in Central Asia. By analyzing data from the “Household Access to Energy in the Fergana Valley” survey, which was conducted across the Kyrgyz Republic, Tajikistan, and Uzbekistan in 2023, this study investigates the socioeconomic, environmental, and power supply determinants of cooling choices. The study addresses the following research questions: (1) What are the determinants of household cooling choices in Central Asia? (2) How do climatic stress factors, particularly cooling degree days (CDD), influence household cooling technology choices? (3) In what ways does the reliability of the power supply affect households’ decisions to adopt particular cooling technologies? This study makes several contributions to the existing literature on environmental extremes, household cooling choices, and infrastructure in Central Asia and other arid regions. Firstly, while much of the current literature has examined heating choices, this study shifts the focus to cooling technologies, recognizing their increasing relevance in the context of rising global temperatures. To the best of our knowledge, this is the first study on cooling in Central Asia. Secondly, it assesses the role of power outages in cooling technology adoption. While air-conditioning is often seen as the most effective solution for heat adaptation, this research highlights that its adoption is limited in areas with frequent power disruptions. Lastly, the study explores the impact of climatic factors, specifically CDD, on household cooling choices. By incorporating CDD as a key variable, the research provides a clearer understanding of how climatic conditions influence household behavior and energy consumption patterns in response to heat stress. The literature review provides an overview of existing research on the determinants of cooling technology adoption. The methodology section then outlines the data sources and empirical strategy, including the specification of the multinomial probit model. Attention is given to the variables of interest (cooling degree days and power outages), along with a detailed description of control variables. A discussion of empirical findings is followed by a robustness analysis to evaluate the impact of the duration of power outages on household cooling choices. Finally, the conclusion summarizes the key findings and highlights policy implications. 2 2. Literature Review The growing global demand for cooling technologies in response to heat waves is extensively analyzed in the literature, particularly in the context of urbanization and its socioeconomic implications (Thomson et al., 2019; Attia et al., 2021; Zhang et al., 2020; Pavanello et al., 2021; Nematchoua et al., 2019). As cities expand and heat waves become more frequent and severe, the need for effective cooling solutions is becoming increasingly pressing (Mushore et al., 2017; Bardhan et al., 2020; Jia et al., 2024). Hu et al. (2020) conducted a nationwide online survey in the People's Republic of China to explore the changing patterns of cooling in urban households and the factors that influence this transition. Yan and Liu (2020) modeled residential air-conditioner usage, emphasizing the role of historical temperature data in forecasting energy consumption. Zhang et al. (2020) explored the implications of climate change on air-conditioning (AC) usage in rural areas, highlighting the need to promote high efficiency AC units since low efficiency technologies were exacerbating rising energy demand. Pavanello et al. (2021) drew attention to the risks posed by rising global temperatures, particularly for populations in lowand middle-income countries. While AC is viewed as a critical adaptation tool, access remains limited for lower income households. They find that AC adoption follows an S-shaped curve, influenced by socioeconomic conditions. This suggests that many low-income households will remain unable to afford such technologies, thereby creating an “adaptation cooling deficit”. These findings highlight the vulnerability of low-income populations, who often lack access to air-conditioning and thereby become more susceptible to the adverse health impacts of extreme heat (Mushore et al., 2017). Research by Zander et al (2023) on the socioeconomic dimensions of cooling preferences reveals significant variation in behaviors, influenced by factors such as age, household composition, and heat tolerance. Their work suggests that targeting energy efficient cooling solutions based on these demographic factors could enhance energy resilience. Zhang et al. (2020) highlight the role of CDDs in driving air-conditioner use in rural areas, with socioeconomic variables such as dwelling characteristics influencing behavior. In the context of access to electricity and demand for cooling, Falchetta and Mistry (2021) emphasize the necessity for decision-makers to integrate cooling needs into electricity access policies and power generation planning. They argue that recognizing these needs is critical for forecasting future residential electricity demand. Thomson et al. (2019) argue for a reevaluation of energy policy frameworks, particularly in the context of year-round vulnerability to heat. Meles (2020) highlights the often overlooked issue of electricity supply reliability in developing countries. Focusing on urban households, the study critiques the prevailing narrative that equates electricity access with effective electrification. Meles emphasizes that without a reliable electricity supply, the benefits of electrification are not fully realized since frequent power outages result in households incurring additional expenditures to compensate. Power outages present a unique challenge to households in developing countries. Lee et al. (2024) highlight the shortcomings of passive cooling measures under extreme heat conditions and the critical role of air-conditioning to ensure thermal comfort during power outages. Their study underscores the need for both passive and active cooling solutions to protect vulnerable populations from heat-related health risks. 3 Despite the extensive literature on cooling demands and energy resilience, studies on the influence of power outages on household cooling decisions remain limited, particularly with respect to Central Asia. This study aims to fill this gap by examining the interplay between CDDs, power outages, and household cooling choices in the Kyrgyz Republic, Uzbekistan, and Tajikistan. By exploring how households in these countries adapt their cooling strategies to power outages and increasing temperatures, this research offers insights into the socioeconomic and behavioral dimensions of energy resilience in Central Asia. 3. Methodology 3.1 Data This study utilizes data from the 2023 Household Access to Energy in the Fergana Valley survey, which was conducted by the Central Asia Regional Economic Cooperation Institute (CAREC) in partnership with the Asian Development Bank Institute (ADBI) and the Public Opinion Research Institute in Kazakhstan. A total of 1,522 respondents were interviewed in July-August 2023, comprising 522 respondents in the Kyrgyz Republic, 500 in Tajikistan and 500 in Uzbekistan. The survey aimed to generate detailed data about energy access in a relatively compact and comparable area that spans parts of Uzbekistan, the Kyrgyz Republic, and Tajikistan. It gathered detailed information on energy use for various purposes, as well as the sociodemographic characteristics of heads of households. Among other things, the survey included detailed information on household cooling systems and the reliability of power. More information about the survey can be found in Holzhacker et al. (2024); Sulaimanova, Azhgaliyeva and Holzhacker (2024); Azhgaliyeva, Holzhacker, Rahut, and Correia (2024); and Azhgaliyeva, Kodama and Holzhacker (2025). To analyze cooling demand in the context of climate change, CDD data were incorporated into the study. These meteorological data, sourced from the NASA POWER Project, are freely available. 3.2 Model Specification To analyze the factors influencing household cooling choices, we employ a multinomial probit model, which accounts for multiple categorical outcomes. In this study, the categorical outcomes represent the choice of cooling system. The base category is households without any cooling system, against which the likelihood of adopting (i) air-conditioning and/or (ii) fans and/or sunscreen films (without air-conditioning) is estimated. The model allows the estimation of the probability that a household will select one of the available cooling options based on a set of explanatory variables. The general form of the multinomial probit model is as follows: Pr�𝑦𝑦𝑖𝑖=𝑚𝑚𝑚 𝑋𝑋𝑖𝑖,𝛽𝛽,𝑢𝑢𝑖𝑖𝑖𝑖�=Φ�𝑦𝑦𝑖𝑖=𝑚𝑚, 𝑋𝑋𝑖𝑖𝛽𝛽𝑖𝑖+𝑢𝑢𝑖𝑖𝑖𝑖�= exp ( 𝑋𝑋 𝑖𝑖 𝛽𝛽 𝑚𝑚 +𝑢𝑢 𝑖𝑖𝑚𝑚 ) ∑exp ( 𝑋𝑋𝑖𝑖𝑖𝑖𝛽𝛽𝑖𝑖+𝑢𝑢𝑖𝑖𝑖𝑖) 𝐽𝐽 𝑖𝑖=1 (1) where: • Pr�𝑦𝑦𝑖𝑖=𝑚𝑚𝑚 𝑋𝑋𝑖𝑖𝑖𝑖,𝛽𝛽,𝑢𝑢𝑖𝑖𝑖𝑖� is the probability that household i adopts cooling technology j, where m=1 refers to air-conditioning, m=2 refers to fans or sunscreen films (without airconditioning), and m=3 refers to no cooling system (base outcome). • 𝑋𝑋𝑖𝑖 represents the set of explanatory variables for household i. • 𝛽𝛽𝑖𝑖 are the coefficients to be estimated for each cooling technology j. 4 • Φ denotes the cumulative distribution function of the standard normal distribution, used to model the likelihood of different outcomes. The multinomial probit model is estimated using maximum likelihood estimation (MLE), which enables the simultaneous estimation of the probabilities of each household choosing one of the three cooling system options. These probabilities are conditioned on household characteristics, power supply variables, and CDD conditions. The analysis is performed on both the total sample and subsamples, including urban versus rural households and male-headed versus female-headed households. This approach allows the exploration of heterogeneity in cooling technology adoption across different demographic and geographic groups. Sensitivity analyses are conducted to assess the stability of the estimated coefficients in the main model. This is done by incorporating the duration of power outages, thus providing a more nuanced understanding of the relationship between electricity reliability and the adoption of cooling technologies, as well as testing the robustness of the estimated coefficients. Outcome Variable Table 1 presents the descriptive statistics for the outcome variable, summarizing household cooling preferences across the three countries as well as for the total sample. Option 1, airconditioning, is the least common in the Kyrgyz Republic, where only 9.20% of households have selected it as their primary cooling method. Adoption rates for air-conditioning are higher in Tajikistan (35.40%) and Uzbekistan (20.60%). Table 1. Descriptive Statistics of Outcome Variable–Household Cooling Options Outcome Variable Country Household cooling choices Kyrgyz Republic Tajikistan Uzbekistan Total Option 1 - air-conditioning 48 177 103 328 9.20% 35.40% 20.60% 21.55% Option 2 - fan or sunscreen films for windows (without air-conditioning) 264 147 325 736 50.57% 29.40% 65.00% 48.36% Option 3 - no cooling system 210 176 72 458 40.23% 35.20% 14.40% 30.09% Total 522 500 500 1522 100% 100% 100% 100% Note: First row has frequencies and second row has column percentages. Source: Authors’ calculations using data from 2023 Household Access to Energy in the Fergana Valley survey. Option 2, fans and/or sunscreen films for windows (without air-conditioning), is more prevalent, accounting for 48.36% of households covered by the survey. In Uzbekistan, 65% of households have adopted this method, while the proportion drops to 50.57% in the Kyrgyz Republic and 29.40% in Tajikistan. 11 Unlike the case for air-conditioning, power outages (scheduled or rolling blackouts) do not significantly deter households from selecting fans or sunscreen films, reflecting the lower energy dependency of these alternatives. The preference for fans or sunscreen films (without air-conditioning) as a cooling method is influenced by household size, with larger households more likely to opt for this alternative. This may be attributable to the relatively lower costs and energy demands associated with these options, making them more suitable for households with greater economic constraints. Though less influential (only at 10% level of significance) than for air-conditioning (at 5% level of significance), income still plays a role, with middle-income households showing a positive but weaker association with the adoption of fans or screens (as opposed to having no cooling method). Income has no significant impact at 5% level of significance on the likelihood of having fans or sunscreens, probably due to their greater affordability than air-conditioning. An increase in household size is associated with a higher probability of adopting fans and/or sunscreen films as cooling technologies. However, a greater number of children in the household negatively affects the likelihood of choosing these cooling methods, mirroring the pattern observed for airconditioning. In other words, households with more children are more likely to have no cooling method at all. These empirical findings reveal that the adoption of air-conditioning is predominantly driven by higher education and income, along with urban residency and reliable power supply and climatic conditions, while the adoption of fans or sunscreen films is predominantly driven by household size and climatic factors. These findings underscore the importance of addressing the infrastructural challenges such as the unreliability of electricity supply to ensure equitable access for cooling solutions in Central Asia. 12 Table 6. Multinomial Probit Model Coefficient Estimates for Household Cooling Choices (Main Model) Total Sample HH Residence HH Head Gender Urban Rural Male Female (1) (2) (3) (4) (5) Air-conditioning HH head age -0.005 -0.001 -0.008 -0.007 -0.004 (0.004) (0.007) (0.005) (0.006) (0.006) HH head gender (1=male) 0.026 0.109 0.007 (0.116) (0.190) (0.150) Tertiary education level 0.755*** 1.125*** 0.563*** 0.429** 1.188*** (0.128) (0.214) (0.166) (0.174) (0.196) Household size 0.032 0.079 0.020 -0.029 0.099** (0.030) (0.052) (0.038) (0.043) (0.045) Share of children in HH -0.950*** -1.698*** -0.548 -0.822* -1.034** (0.329) (0.524) (0.431) (0.459) (0.486) Share of seniors in HH 0.538* 0.102 0.761* 0.308 0.831* (0.322) (0.536) (0.412) (0.454) (0.461) Middle income level 0.401** 0.667** 0.370* 0.252 0.587** (0.174) (0.302) (0.224) (0.246) (0.254) Highest income level 0.475*** 0.170 0.680*** 0.374* 0.572*** (0.146) (0.235) (0.192) (0.208) (0.212) Residence (1=urban) 0.456*** 0.528*** 0.417** (0.126) (0.175) (0.183) Log CDD 0.381*** 0.614** 0.287* 0.542*** 0.177 (0.145) (0.289) (0.173) (0.197) (0.221) Scheduled or rolling blackouts (1 = power outages) -0.260** -0.219 -0.288* -0.240 -0.302* (0.121) (0.196) (0.157) (0.169) (0.176) Country fixed effects + + + + + Constant -3.942*** -4.940** -3.463*** -4.376*** -3.312** (0.993) (1.975) (1.184) (1.329) (1.512) Fans and/or sunscreen films (without air-conditioning) HH head age -0.002 -0.000 -0.003 -0.004 -0.000 (0.004) (0.007) (0.004) (0.005) (0.005) HH head gender (1=male) -0.001 0.126 -0.057 (0.101) (0.179) (0.124) Tertiary education level 0.068 0.273 -0.007 -0.046 0.191 (0.122) (0.216) (0.151) (0.165) (0.186) Household size 0.079*** 0.116** 0.059* 0.024 0.130*** (0.026) (0.048) (0.031) (0.035) (0.039) Share of children in HH -0.694** -1.438*** -0.288 -0.357 -0.956** (0.287) (0.483) (0.363) (0.413) (0.407) Share of seniors in HH 0.160 -0.464 0.463 0.238 0.124 (0.298) (0.516) (0.367) (0.427) (0.414) Middle income level 0.275* 0.491* 0.255 0.287 0.235 (0.151) (0.277) (0.187) (0.221) (0.209) Highest income level 0.011 -0.253 0.178 -0.026 0.027 (0.131) (0.224) (0.165) (0.191) (0.183) Residence (1=urban) -0.114 -0.044 -0.159 (0.113) (0.160) (0.163) Log CDD 0.435*** 0.181 0.553*** 0.520*** 0.352* (0.130) (0.256) (0.153) (0.181) (0.188) Scheduled or rolling blackouts (1 = power outages) 0.025 0.095 -0.039 -0.010 0.068 (0.107) (0.187) (0.132) (0.148) (0.155) Country fixed effects + + + + + Constant -2.947*** -1.333 -3.777*** -3.198** -2.670** (0.910) (1.791) (1.073) (1.250) (1.328) Continued on the next page 13 Total Sample HH Residence HH Head Gender Urban Rural Male Female (1) (2) (3) (4) (5) Number of obs. 1522 557 965 763 759 Log likelihood -1405.14 -516.69 -875.18 -706.23 -687.05 Chi2 337.83 138.08 199.27 179.18 174.87 Probability 0.000 0.000 0.000 0.000 0.000 CDD=cooling degree days, HH=households. Note: Base outcome is “no cooling system”. Robust standard errors are presented in parentheses. *p<0.10; **p<0.05; ***p<0.010 Source: Authors’ calculations using data from 2023 Household Access to Energy in the Fergana Valley survey. 4.2 Robustness Analysis: Power Outage Duration The multinomial probit model estimates in Table 7 examine the factors influencing households’ choices of cooling systems under varying power outage durations. To test the robustness and sensitivity of our findings, we replace the binary dummy variable for power outages used in the main model with a categorical variable representing different durations: (i) 1–16 hours, (ii) 17–50 hours, and (iii) 50–200 hours. It is important to note that data on power outage duration was incomplete, as 126 households responded with “difficult to answer,” resulting in missing observations. Consequently, the estimation sample is reduced to 1,398 observations. Despite the reduction in the number of observations, the estimated effects of the control variables remain consistent and statistically significant. This suggests that incorporating a more detailed measure of power outage duration does not alter the key relationships identified in the main model, reinforcing the robustness of our results. As in the main model, the duration of power outages significantly and negatively influences households’ likelihood of adopting air-conditioning systems, particularly in rural areas. This finding highlights the practical constraints that prolonged power outages impose on the functionality and utility of energy-intensive appliances, making them less feasible for households that face frequent or extended power disruptions. By contrast, the influence of power outage duration on the adoption of less energy-intensive cooling options (fans or sunscreen films), is less pronounced, as it is in the main model findings. This suggests that households facing prolonged power outages may prioritize alternative cooling methods that are less dependent on a stable electricity supply. The CDD variable also exhibits results similar to those in the main model when the dummy variable for power outages is replaced. CDD remains positively significant, indicating that households are more likely to adopt cooling solutions in response to climatic stress and the immediate need for thermal comfort. 14 Table 7. Multinomial Probit Model Coefficient Estimates (Power Outage Duration) Total Sample HH Residence HH Head Gender Urban Rural Male Female Air-conditioning HH head age -0.006 -0.000 -0.010* -0.005 -0.007 (0.004) (0.007) (0.006) (0.006) (0.007) HH head gender (1=male) 0.121 0.144 0.138 (0.123) (0.199) (0.161) Tertiary education level 0.786*** 1.070*** 0.622*** 0.421** 1.237*** (0.136) (0.223) (0.179) (0.185) (0.208) Household size 0.029 0.021 0.040 -0.028 0.084* (0.032) (0.054) (0.039) (0.045) (0.047) Share of children in HH -0.838** -1.410*** -0.570 -0.552 -1.208** (0.346) (0.539) (0.463) (0.485) (0.503) Share of seniors in HH 0.517 -0.172 0.876** 0.410 0.632 (0.339) (0.548) (0.440) (0.475) (0.481) Middle income level 0.368** 0.660** 0.310 0.222 0.547** (0.186) (0.315) (0.242) (0.263) (0.271) Highest income level 0.318** 0.157 0.467** 0.249 0.415* (0.158) (0.249) (0.212) (0.227) (0.228) Residence (1=urban) 0.463*** 0.466** 0.497** (0.132) (0.183) (0.194) Log CDD 0.414*** 0.538* 0.311* 0.604*** 0.193 (0.154) (0.304) (0.187) (0.204) (0.242) Duration of power outages: 1-16 hours -0.051 -0.209 0.007 0.207 -0.325 (0.164) (0.275) (0.204) (0.228) (0.238) Duration of power outages: 17-50 hours -0.455** -0.119 -0.736** -0.413 -0.544* (0.201) (0.292) (0.302) (0.262) (0.320) Duration of power outages: 50-200 hours -0.132 0.001 -0.221 -0.234 -0.098 (0.187) (0.280) (0.263) (0.264) (0.276) Country fixed effects + + + + + Constant -4.128*** -4.306** -3.583*** -4.879*** -3.045* (1.063) (2.079) (1.287) (1.383) (1.664) Fand and/or Sunscreen Films HH head age -0.005 -0.003 -0.006 -0.005 -0.005 (0.004) (0.007) (0.005) (0.005) (0.005) HH head gender (1=male) 0.066 0.138 0.036 (0.105) (0.186) (0.130) Tertiary education level 0.086 0.262 0.020 -0.082 0.253 (0.128) (0.225) (0.161) (0.174) (0.196) Household size 0.075*** 0.095* 0.055* 0.019 0.119*** (0.027) (0.050) (0.033) (0.036) (0.040) Share of children in HH -0.566* -1.390*** -0.102 -0.144 -0.992** (0.301) (0.504) (0.382) (0.430) (0.432) Share of seniors in HH 0.122 -0.651 0.531 0.294 -0.051 (0.312) (0.525) (0.392) (0.448) (0.430) Middle income level 0.082 0.339 0.035 0.085 0.050 (0.162) (0.292) (0.202) (0.239) (0.223) Highest income level -0.208 -0.390 -0.064 -0.224 -0.204 (0.142) (0.238) (0.181) (0.209) (0.196) Residence (1=urban) -0.112 -0.087 -0.114 (0.118) (0.167) (0.169) Log CDD 0.481*** 0.228 0.564*** 0.608*** 0.363* (0.137) (0.269) (0.162) (0.190) Continued on the next page 15 Total Sample HH Residence HH Head Gender Urban Rural Male Female Duration of power outages: 1-16 hours 0.197 -0.055 0.261 0.318 0.095 (0.151) (0.278) (0.182) (0.220) (0.209) Duration of power outages: 17-50 hours 0.066 0.353 -0.133 -0.073 0.207 (0.162) (0.261) (0.209) (0.225) (0.238) Duration of power outages: 50-200 hours 0.160 0.347 0.013 0.133 0.200 (0.151) (0.258) (0.190) (0.211) (0.219) Country fixed effects + + + + + Constant -3.077*** -1.385 -3.672*** -3.691*** -2.366* (0.956) (1.886) (1.131) (1.312) (1.408) Number of obs. 1398 524 874 704 694 Log likelihood - 1279.55 -484.21 -781.72 -645.67 -622.49 Chi2 314.69 126.73 187.85 170.49 157.59 Probability 0.000 0.000 0.000 0.000 0.000 CDD=cooling degree days, HH=households. Note: Base outcome is “no cooling system”. Robust standard errors are presented in parentheses. *p<0.10; **p<0.05; ***p<0.010. Source: Authors’ calculations using data from 2023 Household Access to Energy in the Fergana Valley survey. 5. Conclusion and Policy Implications This study investigates the factors influencing household choices regarding cooling technologies in Central Asia, focusing on air-conditioning and fans/sunscreen films. Utilizing data from the “Household Access to Energy in the Fergana Valley” survey conducted in the Kyrgyz Republic, Tajikistan, and Uzbekistan, the analysis examines how socioeconomic, environmental, and power supply factors shape household decisions. Employing a multinomial probit model, this study analyzes the determinants for adopting different cooling technologies, highlighting the interaction between climatic conditions and power supply reliability. The research provides valuable insights into the factors shaping household cooling choices in Central Asia, with particular emphasis on the role of power supply infrastructure and climatic conditions. The findings underscore the importance of addressing both power sector reliability and climate adaptation in vulnerable regions. Key results include the significant impact of cooling degree days on cooling appliance adoption and the critical role of power outages in influencing the choice of cooling technologies. First, a significant number of surveyed households (48.36%) rely on fans and/or sunscreen films for windows for cooling (without air-conditioning). In Uzbekistan, 65% of surveyed households have adopted this method, while the proportion drops to 50.57% in the Kyrgyz Republic and 29.40% in Tajikistan. Additionally, 14.40% of surveyed households in Uzbekistan have no cooling system. The corresponding figures in Tajikistan and the Kyrgyz Republic are higher at 35.20% and 40.23%, respectively, suggesting increased household vulnerability to heat in these two countries. Second, cooling degree days significantly impact the adoption of cooling appliances. The cooling season primarily spans from May to September, with CDD ranging from 246 to 1,641 in the Fergana Valley. Approximately one-third of surveyed households live in regions with CDD 1,500 and above, which is considered high. Households in areas with higher CDD are more likely to adopt cooling solutions (air-conditioning, fans, or sunscreen films). The analysis underscores the importance of climatic conditions, particularly CDDs, in influencing cooling technology adoption. This finding highlights households’ adaptive behavior in response to climatic stress, demonstrating that cooling choices are shaped by both socioeconomic factors and environmental constraints. 16 Integrating climatic data such as CDD into the analysis of household energy choices provides critical insights into adaptive behavior under different environmental conditions. Third, power outages negatively impact the adoption of energy-intensive cooling technologies like air-conditioning but have no effect on less energy-intensive options like fans and sunscreen films. Power stability is crucial for adapting to heat using energy-intensive technologies (e.g., airconditioning) in hotter areas during the peak cooling months (May–September) in Central Asia. Fans and sunscreen films, which are less energy-intensive than air-conditioning, are more commonly used in areas where power outages are a significant concern. This result highlights the importance of power reliability for adopting energy-intensive cooling technologies, suggesting that air-conditioning adoption is constrained by economic and infrastructural barriers. It also emphasizes the importance of less energy-intensive cooling technologies (e.g., fans, sunscreen films, and other options not included in this study, such as energy efficiency measures and courtyards in building design) in regions with unreliable electricity access. The study assessed the robustness of its findings by considering the duration of power outages with three dummy variables representing different outage durations. This analysis confirmed that the reliability of power supply plays a critical role in shaping cooling choices. Specifically, households in areas prone to frequent or prolonged power outages are less likely to adopt airconditioning. There are many measures for adapting to heat stress (Shen, Azhgaliyeva, and Baño Leal, 2024). We suggest several based on the study's results. Enhanced power supply reliability is crucial for supporting the adoption of energy-intensive cooling technologies like air-conditioning. Solar panels can help meet summer energy demand, as solar irradiation correlates with cooling needs, particularly in regions with high cooling requirements. Support can prioritize vulnerable households in regions with high heat exposure and unreliable power supply. Based on this study, the following suggestions for future research are proposed. First, this study focused on a limited range of cooling solutions: air-conditioning, fans, and sunscreen films. Future research should explore additional heat adaptation options, such as building codes and passive cooling techniques in building design. These include improved insulation (Sulaimanova, Azhgaliyeva, and Holzhacker, 2024), reflective roofing, and natural ventilation, which can significantly reduce energy demand for cooling and aid in heat wave adaptation. If there is a need for such support, it should be accompanied by training for technicians and other stakeholders on the installation and maintenance of energy-efficient and passive cooling technologies. Additionally, location-specific cooling measures are important. Urban planning strategies, such as incorporating green spaces and water bodies, can reduce urban heat island effects and lower cooling demands. Community cooling centers can also provide immediate relief during heat waves. Second, while we identified the significant impact of power outages on cooling choices, we did not investigate the causes of these outages. Understanding the causes can help provide better policy recommendations to address power outages. Third, we did not take into account electricity tariffs and the impact of their growth. While the paper highlights power outages and grid instability as key barriers to adopting energy-intensive cooling technologies, another critical factor worth considering is the cost of electricity and its growth. The increase in electricity tariffs could significantly hinder the adoption and sustained use of air-conditioning, particularly among lowand middle-income households in rural areas, who are already financially constrained. Many rural households may already be opting for lower energy alternatives—such as fans and sunscreen films—to minimize electricity expenses. 17 APPENDIX Table A1. Matrix of Correlations HH head age HH head gender (1=male) Tertiary education level HH size Share of children in HH Share of seniors in HH Middle income level Highest income level Residence (1=city) CDD Blackouts (1 = power outages) Duration of power outages (hours) HH head age 1 HH head gender (1=male) -0.008 1 Tertiary education level -0.067 0.078 1 Household size 0.042 0.043 -0.067 1 Share of children in HH -0.049 -0.033 0.043 0.389 1 Share of seniors in HH 0.329 0.032 -0.017 -0.043 -0.324 1 Middle income level -0.01 0.006 -0.043 -0.01 0.041 0.003 1 Highest income level -0.023 0.028 0.074 0.18 0.063 -0.097 -0.573 1 Residence (1=city) 0.078 -0.042 0.04 -0.104 -0.04 -0.01 -0.083 0.008 1 CDD 0.074 -0.053 0.011 -0.102 -0.153 0.096 -0.087 -0.132 0.171 1 Scheduled or rolling blackouts (1 = power outages) 0.125 -0.017 -0.029 0.119 0.032 -0.032 -0.004 0.023 0.013 0.044 1 Average monthly duration of power outages (hours) 0.143 -0.008 -0.05 0.051 -0.031 0.003 -0.001 -0.019 0.053 0.181 0.61 1 CDD=cooling degree days, HH=households. Source: Authors’ calculations using data from 2023 Household Access to Energy in the Fergana Valley survey. 18 Table A2. Descriptive Statistics for Cooling Degree Days by Region in 2023 Kyrgyz Republic Tajikistan Uzbekistan 35 regions CDD N 8 regions CDD N 50 regions CDD N Achi 739.14 13 Asht 1,500.06 54 Akhunboboev 1,525.67 10 Ak-Turpak partially 246.07 15 B. Gafurov 585.8 120 Amirabod 1,640.78 10 Andijan 1,183.37 13 Isfara 434.61 17 Andijan 1,183.37 10 Aral 1005.7 13 Istaravshan 1,088.72 89 Aral 1,005.7 10 Asanchek 468.36 13 J.Rasulov 944.34 45 Asaka 1,005.7 9 Batken 434.61 20 Khujand 1,292.56 61 Baghdad 1,640.78 10 Bazar-Korgon 297.58 12 Konibodom 1,500.06 68 Besarang 1,525.67 10 Beget 1,183.37 13 Spitamen 944.34 46 Bozorboshi 1,525.67 10 Byurgendyu 1,183.37 13 Chartak 1,625.48 10 Changyr-Tash 1,183.37 13 Chinabad 1,625.48 10 Dzhany-Abad 1,183.37 12 Chust 1,251.78 10 Jalal-Abad 739.14 20 Ertepa 1,625.48 10 Kara-Seget 468.36 13 Ezgulik 1,251.78 10 Kara-Suu 468.36 20 Fergana 1,525.67 10 Khauz 1005.7 13 Gaiston 1,625.48 10 Kochkor-Ata 1,183.37 25 Garmidan 1,525.67 10 Kolot 1,183.37 12 Girvon 1,625.48 10 Kyrgyz-Kyshtak 1,640.78 15 Karaskan 1,625.48 10 Kyumyush-Aziz 739.14 13 Khojaabad 1005.7 10 Kyzyl-Abad 468.36 13 Kokand 1,500.06 10 Langar 468.36 13 Korasuv 468.36 9 Madaniyat 468.36 13 Korayantok 1,005.7 10 Min-Chynar 1,640.78 15 Kurganteppa 1,005.7 10 Mogol-Korgon 1,183.37 13 Kuva 1,525.67 10 Munduz (Kyzyl-Tuu) 739.14 13 Kuvasai 1,525.67 10 Munduz (Saipidin Atabek) 667.19 13 Loison 1,525.67 10 Naiman 1,251.78 13 Margilan 1,525.67 10 Oktyabr 1,005.7 13 Maslahat 1,625.48 10 Osh city 1,005.7 45 Mirabad 1,183.37 10 Pakhtachi 1,005.7 13 Naiman 1,251.78 10 Telman 468.36 13 Namangan 1,625.48 10 Tepe Korgon 1,005.7 13 Nazarmahram 1,525.67 10 Zar-Tash 1,640.78 15 Okbilol 1,525.67 10 Zarbalik 468.36 13 Okmozor 1,183.37 10 Zhapalak 1005.7 10 Oktosh 1,251.78 10 Olmos 1,251.78 10 Poytug 1,183.37 12 Rishtan 1,640.78 10 . Sarikurgan 1,525.67 10 Shahrikhan 1,525.67 10 Tepakurgan 1,251.78 10 Tinchlik 1,640.78 10 Tulaboi 1,500.06 10 Turakurgan 1,251.78 10 Uchkuza 1,183.37 10 Uychi 1,625.48 10 Yaipan 1,500.06 10 Yangikurgan 1,625.48 10 Yozyovon 1,525.67 10 Zvutkan 1,251.78 10 Total 892.5696 522 Total 1,044.738 500 Total 1,403.894 500 CDD=cooling degree days. Source: Authors’ calculations using the data was obtained from POWER Project's Hourly v2.4.9 on 2025/02/21 https://power.larc.nasa.gov/data-access-viewer/. 19 REFERENCES Attia, S., Levinson, R., Ndongo, E., Holzer, P., Kazanci, O. B., Homaei, S., ... and Heiselberg, P. 2021. Resilient cooling of buildings to protect against heat waves and power outages: Key concepts and definition. Energy and Buildings, 239, 110869. Azhgaliyeva, D., Kodama, W., and Holzhacker, H. 2025. Does awareness and prioritization of environment and health matter for household fuel choice? Empirical evidence from Central Asia. Energy Research & Social Science, 120, https://doi.org/10.1016/j.erss.2024.103898. Azhgaliyeva, D., Holzhacker, H., Rahut, D., and Correia, J. 2024. Empowered Women Prioritize the Environmental and Health Impacts of Fuel Choice: Empirical Evidence from Central Asia. ADBI Working Paper 1482. Tokyo: Asian Development Bank Institute. Available: https://doi.org/10.56506/HDYJ7785 Bardhan, R., Debnath, R., Gama, J., and Vijay, U. 2020. REST framework: A modelling approach towards cooling energy stress mitigation plans for future cities in warming Global South. Sustainable Cities and Society, 61, 102315. Falchetta, G. and Mistry, M. N. 2021. The role of residential air circulation and cooling demand for electrification planning: Implications of climate change in sub-Saharan Africa. Energy Economics, 99, 105307. Falchetta, G., De Cian, E., Sue Wing, I., and Carr, D. 2024. Global projections of heat exposure of older adults. Nature Communications, 15(1), 3678. Holzhacker, H., Azhgaliyeva D., Rakisheva, B., Ibrayeva, A., Kurmangazhinova, A., Yessimova, K., et al. 2024. Household Access to Energy in the Fergana Valley: A Multidimensional Survey-based Assessment in Three CAREC Countries, CAREC Institute. https://opinions.kz/images/books/Project-Report-Fergana-Valley.pdf Hu, S., Yan, D., Dong, B., and Fu, J. 2020. Exploring key factors impacting cooling usage patterns of Chinese urban household based on a large-scale questionnaire survey. Energy and Buildings, 214, 109885. Jia, S., Weng, Q., Yoo, C., Xiao, H., & Zhong, Q. 2024. Building energy savings by green roofs and cool roofs in current and future climates. npj Urban Sustainability, 4(1), 23. Lee, S. H., Hong, T., Le, M., Medina, L., Xu, Y., Robinson, A., and Piette, M. A. 2024. Assessment of energy and thermal resilience performance to inform climate mitigation of multifamily buildings in disadvantaged communities. Sustainable Cities and Society, 104, 105319. Li, Z., Fang, G., Chen, Y., Duan, W., and Mukanov, Y. 2020. Agricultural water demands in Central Asia under 1.5 C and 2.0 C global warming. Agricultural Water Management, 231, 106020. Liu, Y., Geng, X., Hao, Z., and Zheng, J. (2020). Changes in climate extremes in Central Asia under 1.5 and 2 C global warming and their impacts on agricultural productions. Atmosphere, 11(10), 1076. 20 Mehta, K., Ehrenwirth, M., Trinkl, C., Zörner, W., & Greenough, R. 2021. The energy situation in Central Asia: A comprehensive energy review focusing on rural areas. Energies, 14(10), 2805. Meles, T. H. 2020. Impact of power outages on households in developing countries: Evidence from Ethiopia. Energy Economics, 91, 104882. Mushore, T. D., Odindi, J., Dube, T., and Mutanga, O. 2017. Understanding the relationship between urban outdoor temperatures and indoor air-conditioning energy demand in Zimbabwe. Sustainable Cities and Society, 34, 97-108. Nematchoua, M. K., Yvon, A., Kalameu, O., Asadi, S., Choudhary, R., and Reiter, S. 2019. Impact of climate change on demands for heating and cooling energy in hospitals: An indepth case study of six islands located in the Indian Ocean region. Sustainable Cities and Society, 44, 629-645. Pavanello, F., De Cian, E., Davide, M., Mistry, M., Cruz, T., Bezerra, P., ... and Lucena, A. F. 2021. Air-conditioning and the adaptation cooling deficit in emerging economies. Nature communications, 12(1), 6460. Radhi, H., & Sharples, S. 2013. Quantifying the domestic electricity consumption for airconditioning due to urban heat islands in hot arid regions. Applied Energy, 112, 371-380. Shen, B., Azhgaliyeva, D., Baño Leal, A. 2024 Sustainable Cooling: How to Cool the World Without Warming the Planet, ADBI Press, https://doi.org/10.56506/MAGR4101 Sulaimanova, B., Azhgaliyeva, D., and Holzhacker, H. 2024. House insulation: Perceived needs and willingness in Central Asia, ADBI Working Paper, No. 1478, Asian Development Bank Institute, Tokyo, https://doi.org/10.56506/NLWV6429 Thomson, H., Simcock, N., Bouzarovski, S., and Petrova, S. 2019. Energy poverty and indoor cooling: An overlooked issue in Europe. Energy and Buildings, 196, 21-29. Tursumbayeva, M., Muratuly, A., Baimatova, N., Karaca, F., and Kerimray, A. 2023. Cities of Central Asia: New hotspots of air pollution in the world. Atmospheric Environment, 309, 119901. Wang, X., Li, Y., Chen, Y., Li, Y., Wang, C., Kaldybayev, A., ... and Duan, W. 2023. Intensification of heatwaves in Central Asia from 1981 to 2020–Role of soil moisture reduction. Journal of Hydrology, 627, 130395. Yan, L., and Liu, M. (2020). A simplified prediction model for energy use of air conditioner in residential buildings based on monitoring data from the cloud platform. Sustainable Cities and Society, 60, 102194. Zander, K. K., Van Hoof, J., Carter, S., and Garnett, S. T. 2023. Living comfortably with heat in Australia–preferred indoor temperatures and climate zones. Sustainable Cities and Society, 96, 104706. Zhang, X. B., Sun, J., Fei, Y., and Wei, C. 2020. Cooler rooms on a hotter planet? Household coping strategies, climate change, and air conditioning usage in rural China. Energy Research & Social Science, 68, 101605.