Conference Agenda

This paper examines the willingness to engage in community protection against risky climate threats. In a framed field experiment in Bougainville, Papua New Guinea, I study whether individuals contribute to between-community climate funds to expand the scope of local climate adaptation. Specifically, I vary in a between-subjects design (i) whether one interacts with a member of one’s own community or with someone from another community, and (ii) whether one is being observed by local authorities or not. I find that individuals are less likely to cooperate with members of another community, but being observed by one’s own community leader increases contributions. Thus, this study shows that observation by in-group leaders reduces in-group bias, which is crucial for expanding the scope of local climate adaptation.

Abstract: Climate change will lead to more frequent and intensive flooding events, calling for effective adaptation measures by all level of society. While there is a growing evidence on how flood risks and damages may unfold in the future, it is not obvious whether such insights lead to more effective household adaptation decisions. Adaptation tends to be suboptimal due to lacking information about expected damages at the household level, which adds to economic and behavioral constraints. While higher expected flood damages would call for stronger household adaptation, the precise link between expected damages and adaptation at the household level is unknown. Here, we investigate how household adaptation measures are related to household vulnerability, and help us to evaluate private adaptation on a spectrum between ‘random’ and ‘optimal’ adaptation. We focus on the province of Limburg, the Netherlands, which can be considered to be a “best case scenario” in terms of information availability due to its recent flood experiences and accurate flood risk information. Using publicly available flood maps, a national hydraulic model, and a large-scale survey, our study links objective flooding data, household perceptions, and adaptation measures. Results reveal spatial patterns of flood risk within Limburg, emphasizing inequalities in exposure and adaptation strategies. By combining flood maps and survey responses, the study sheds light on the determinants of household adaptation, recognizing the potential amplification of climate-driven inequalities. The findings contribute to the understanding of effective adaptation strategies essential for managing floods and mitigating their economic impacts.

Flood insurance schemes are seen as more efficient than flood aid, but they are politically less popular. To test whether expectations of flood aid undermine political support for mandatory insurance schemes, we survey over 600 firms in a high risk flood area, and provide accurate information about recent flood aid. Against a crowding out hypothesis, we find positive, small, statistically insignificant average treatment effects in the full sample. However, both negative (crowding out) and positive (crowding in) effects co-exist; treatment effects are conditional on prior expectations and insurance status. We rationalize these results in terms of selective attention to the information in the treatment, depending on prior expectations.

The study on risk preferences and its potential changes amid natural catastrophes has been subject of recent study, yet produced contradictory findings. An often proposed explanation specifically distinguishes between the opposite effect of realized and unrealized losses on risk preferences. Moreover, higher-order risk preferences and its relation to post-disaster behaviors remain unexplored, despite potential theoretical implications. We address these gaps in the literature by conducting experiments with 600 individuals post-Turkey's 2023 catastrophic earthquake, specifically heavily affected individuals who are displaced, those who are not and a control group. Results indicate higher risk-taking in heavily affected individuals when compared to unaffected individuals. Our results are specifically driven by affected females. We find no pre-existing differences in risk preferences between earthquake and control areas using 2012 data. Within the heavily affected group of individuals, higher house damage—our proxy for realized losses—increases risk aversion, with total destruction of a house inducing even higher aversion. Regarding higher-order risk preferences for individuals heavily affected by the earthquake, we find that prudence is positively associated with self-protective behaviors after the earthquake, specifically internal migration and/or displacement. While precautionary savings shows initially no correlation to prudence, a positive association emerges when considering that prudence is also related to occupational choices, with individuals with stable incomes and who save being more prudent. Our results contribute insights into how disasters influence risk preferences, specifically aiming to address contradictory findings in the literature, while presenting novel evidence on the relationship between prudence and post-natural disaster behaviors.