Research
What has to change for societies to adapt to climate risk? We approach this question from four areas. For each, this page introduces key papers and what we found. The figures were drawn for this page from the estimates reported in the papers.
Disaster risk communication and evacuation
Even when warnings and evacuation orders reach people, relatively few actually evacuate. Using survey experiments built on hypothetical disaster scenarios, we study what happens between receiving information and acting on it, and what kind of information moves people.
Factorial survey experiment with 1,671 residents of flood-prone areas. Class shares from two-class finite mixture beta regression models.
Source: based on Tanaka, Kito, & Tanaka (2026), International Journal of Disaster Risk Reduction, 137, 106075
Key papers
Tanaka, K., Akaishi, K., Yokota, T., & Kusano, F. (2026). Crossing the Behavioral Threshold: Social Amplification and Climate Risk Management in Urban Flood Evacuation. Climate Risk Management, 54, 100901. (Open access) DOI
What we foundIn a survey experiment with 1,324 residents of Hiroshima, the threshold at which evacuation intention turned into action was about 0.7 on a 0–1 scale, higher than the 0.5 commonly assumed. Seeing others evacuate and better privacy in shelters raised the likelihood of evacuating substantially; without such support, the likelihood stayed close to zero even among people who perceived the danger as high.
Tanaka, K., Kito, T., & Tanaka, K. (2026). When Flood Warnings Fail: Psychological Predictors of Information Clarity and Evacuation Intention. International Journal of Disaster Risk Reduction, 137, 106075. DOI
What we foundA responsive majority reacted to how the warning was designed — a specific breach point, the basis for the warning, a large-font display — while a non-responsive minority barely reacted regardless of design. Belonging to the minority was related less to age and other demographics than to distrust of weather information and not usually checking local hazard information.
Climate adaptation and urban resilience
Preparing for extreme heat, floods and major earthquakes involves deciding where to direct limited budgets. We study what residents prioritize, how those priorities relate to support for paying the costs, and how residents’ views change through repeated deliberation.
Best–worst scaling (BWS) with 1,819 residents of Yokohama. Mean parameters from a mixed logit model, relative to “refining heat alert systems” (= 0).
Source: based on Tanaka & Akaishi (2026), Risk Analysis, 46(8), e70315
Key papers
Tanaka, K., & Akaishi, K. (2026). Survival First: How Citizens Prioritize Competing Climate-Health Risk Countermeasures Under Fiscal Constraints. Risk Analysis, 46(8), e70315. DOI
What we foundMeasures that protect lives immediately, such as strengthening emergency medical services, were clearly prioritized over longer-term measures. However, the strength of preference for the top-ranked measure was not statistically associated with support for an additional cost, so priorities and willingness to pay did not necessarily coincide. Burdens on daily life, such as effects on sleep, were more consistently associated with willingness to pay than acute symptoms.
Tanaka, K., Matsumoto, R., Takeuchi, S., & Matsuoka, S. (2026). Disaster Risk Uncertainty and the Process of Perception Change: A Longitudinal Analysis of Six-Wave Panel Data from the Citizen Forum on the Tokyo Inland Earthquake. Journal of Disaster Information Studies, 24(2), 253–263. (In Japanese) DOI
What we foundWe held three citizen forums on a major earthquake directly beneath Tokyo and built six-wave panel data from surveys before and after each session. Emotional responses such as a sense of urgency rose only temporarily, whereas values such as consideration for future generations took hold through repeated dialogue. Facing uncertainty and the limits of government lowered evaluations of government at first, followed by a process of recovery.
Green infrastructure and environmental policy
Green infrastructure — rain gardens, street trees, green spaces and other features that use natural processes to soften the impacts of heavy rain and heat — is one way of adapting to climate change. We test its effects at the regional level and study the conditions under which firms and residents who own land choose to adopt it.
Conditional logit estimates for the 78 firms in Sapporo that completed the choice experiment. Dark bars are statistically significant at the 10% level; light bars (n.s.) are not. Installation cost is a dummy for ¥20,000/m² or more, relative to ¥10,000/m². Labels are translated from Japanese.
Source: based on Wang, Tanaka, & Mukoyama (2026), Journal of the City Planning Institute of Japan, 61(3), 840–847 (in Japanese)
Key papers
Han, J., Kim, Y., & Tanaka, K. (2026). The Role of Green Infrastructure in Enhancing Flood Resilience: A Spatial Perspective. Environmental Management, 76, 169. DOI
What we foundSpatial econometric models for 226 municipalities in Korea show that green infrastructure reduces the combined cost of flood damage and recovery. The effect differs by type of green infrastructure, and no spillover to neighboring areas was found: the effect stays within the municipality.
Wang, Y., Tanaka, K., & Mukoyama, M. (2026). An Analysis of Firm Preferences Regarding the Adoption of Green Infrastructure in Sapporo. Journal of the City Planning Institute of Japan, 61(3), 840–847. (In Japanese) DOI
What we foundWe analyzed 189 firms in Sapporo. Among the 78 that took part in the choice experiment, maintenance cost subsidies and technical support were positively associated with choosing to adopt, and a higher installation cost negatively, while consideration for landscape or ecosystems was not statistically significant. Participating firms tended to be larger and more aware of related programs, showing that participation itself needs to be modeled separately.
Tanaka, K., Nelson, H., McCullar, N., & Parulekar, N. (2022). Citizens’ Preferences on Green Infrastructure Practices and Their Enhancement in Portland, Oregon. Journal of Environmental Management, 318, 115415. DOI
What we foundWe surveyed 666 Portland residents on six types of green infrastructure and their willingness to pay for expanding them. Practices with benefits at or near home, such as rain barrels and street trees, were valued relatively highly, while bioswales (planted infiltration channels) and rain gardens were valued relatively low. Valuations and willingness to pay varied widely across people; in policy simulations, a scheme in which contributions vary with willingness to pay raised more revenue than a flat charge.
Sustainable mobility and urban space
Traffic calming and better sidewalks make streets safer and more walkable, but they often divide opinion over car access and cost. We study which treatments residents prefer and how much of the cost they are willing to bear.
Best–worst scaling (BWS) with 1,717 residents living along 19 corridors in Portland, Oregon. Mean parameters from a mixed logit model, relative to pinchpoints (= 0).
Source: based on Tanaka et al. (2027), Transportation Research Part A, 215, 105293
Key papers
Tanaka, K., Dill, J., McNeil, N., Geller, R., & Kondo, N. (2027). What Are Safer Streets Worth? Bridging Resident Preferences and Policy Design in Traffic Calming. Transportation Research Part A: Policy and Practice, 215, 105293. DOI
What we foundOf eight traffic-calming treatments, median refuge islands, mini roundabouts, shared streets and gateway treatments were preferred, while diverters, speed humps and pinchpoints were rated low. Willingness to pay for a traffic-calming program averaged US$34 per household per month (median $30). Preferred treatments did not necessarily coincide with support for a program that comes with a cost.
Ueda, I., & Tanaka, K. (2026). Public Preferences for Sidewalk Space Enhancements: Insights from a Choice Experiment in Yokohama. Journal of the City Planning Institute of Japan, 61(1), 9–16. (In Japanese) DOI
What we foundA choice experiment in Yokohama showed strong support for increasing the number (density) of street trees rather than their height, and for physically separated bicycle lanes. Car users rated these improvements relatively lower than pedestrians and cyclists, so valuations differed by everyday mode of travel.