Abstract:The study on risk preferences and its potential changes amid natural catastrophes has been subject of recent study, producing 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 Turkeys 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. Regarding higher-order risk preferences for individuals heavily affected by the earthquake, we find that prudence is positively associated with selfprotective 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.
What problem does this paper attempt to address?
### What problem does this paper attempt to solve?
This paper aims to explore the impact of natural disasters (such as earthquakes) on individual risk preferences and attempts to address the following specific issues:
1. **Impact of natural disasters on risk preferences**:
- Investigate whether there is a difference in risk preferences between individuals severely affected by earthquakes and those unaffected.
- Explore whether these changes can be attributed to the impact of the earthquake rather than pre-existing differences.
2. **Impact of realized losses on risk preferences**:
- Analyze the impact of the degree of house damage (as a proxy for realized losses) on risk preferences.
- Pay special attention to the impact of complete house destruction on risk preferences and how different levels of loss affect individual risk attitudes.
3. **Relationship between higher-order risk preferences and post-disaster behavior**:
- Explore the relationship between prudence (a higher-order risk preference) and post-disaster self-protection behaviors (such as internal migration).
- Analyze the relationship between prudence and precautionary savings, especially when considering occupational choices.
### Background and research motivation
Existing research indicates that natural disasters have a significant impact on individual risk preferences, but the direction of these effects is inconsistent. Approximately 40% of studies find that extreme events make individuals more risk-seeking, while the remaining studies find the opposite effect. Imas (2016) proposed a possible explanation that realized losses (actual experienced losses) increase individual risk aversion, while unrealized losses (such as holding losing stocks) increase risk-seeking tendencies. However, this explanation has not been fully validated in natural disaster research.
Additionally, the role of higher-order risk preferences (such as prudence) in post-disaster behavior has not been fully explored. The relationship between prudence and self-protection behaviors and precautionary savings is theoretically controversial but lacks empirical evidence.
### Research methods
To explore these issues, the authors conducted the following experiments and surveys:
1. **Risk preference experiments**:
- Using the risk preference module designed by Falk et al. (2018), they tested the risk preferences of individuals severely affected by the earthquake and a control group.
- Compared data from 2012 to rule out pre-existing differences.
2. **Impact of realized losses**:
- Conducted incentivized experiments on individuals severely affected by the earthquake, using the method of Gneezy and Potters (1997) to assess risk preferences through investment decisions.
- Collected data on the degree of house damage, income changes, and other relevant variables.
3. **Higher-order risk preference experiments**:
- Used the prudence experiment developed by Eeckhoudt and Schlesinger (2006) to assess the higher-order risk preferences of individuals affected by the earthquake.
- Explored the relationship between prudence and internal migration and precautionary savings.
### Main findings
1. **Changes in risk preferences**:
- Individuals severely affected by the earthquake are more risk-seeking than the control group, especially women.
- Data from 2012 shows no significant difference in risk preferences between individuals in earthquake-affected areas and control areas before the earthquake, further supporting the impact of the earthquake on risk preferences.
2. **Impact of realized losses**:
- The higher the degree of house damage, the stronger the individual risk aversion, especially for individuals whose houses were completely destroyed.
3. **Higher-order risk preferences and post-disaster behavior**:
- Prudence is positively correlated with self-protection behaviors (such as internal migration).
- Initial analysis shows no significant correlation between prudence and precautionary savings, but when considering occupational choices, prudence is positively correlated with stable income and savings.
### Conclusion
This study provides new evidence on how natural disasters affect individual risk preferences, particularly in terms of realized losses and higher-order risk preferences. The findings help to understand changes in post-disaster behavior and provide valuable references for policymakers.