Environmental governance is critical for mitigating human displacement due to weather-related disasters

Abstract

The impact of climate change is increasingly evident through intensified and more frequent weather-related events, resulting in a higher likelihood of disaster-induced displacement. In response, effective adaptation and resilience-building measures implemented through environmental governance are crucial. Here, we investigate the relationship between environmental governance (measured by the Notre Dame Country Index) and weather-related disaster displacements. Using a panel dataset of 92 countries from 2010 to 2020, Granger non-causality tests reveal a strong bidirectional relationship between environmental governance and disaster displacement. The relationship is primarily driven by governance readiness in the higher-income group, while both governance and economic readiness are crucial in the middle-income group. Notably, all three readiness components appear significant in the lower-income group, emphasizing the imperative for comprehensive disaster management strategies that address economic development, governmental effectiveness, and social conditions. These findings underscore the importance of effective environmental governance for climate change adaptation and sustainable development goals.

Introduction

Disaster displacement, as defined by the Nansen Initiative Protection Agenda¹, refers to “situations where people are forced to leave their homes or places of habitual residence as a result of a disaster or in order to avoid the impact of an immediate and foreseeable natural hazard”. It is a crucial humanitarian and development challenge, affecting millions of people and impeding sustainable development efforts.

Disastrous events like earthquakes, hurricanes, floods, and droughts displace approximately three times more people annually compared to conflicts and violence². From 2008 to 2018, these events caused around 263 million internal displacements, resulting in greater economic and social insecurity for those affected³. Weather-related events, such as storms, floods, and wildfires, were responsible for 98% of recorded disaster displacement in 2020⁴. Although geophysical events like earthquakes are a lesser cause of disaster displacement, their consequences are equally detrimental to those who are displaced. The disaster-prone regions are often more vulnerable and lack the resources to reduce risks or adapt to them^5,6. As climate change is expected to increase the frequency and intensity of weather-related disasters, the risk of displacement increases, highlighting the urgent need to better understand disaster displacement to reduce the risk and enhance adaptation capacity⁷.

Social and economic factors play a crucial role in shaping the preparedness and response to disasters, affecting the resilience of affected communities⁸. As opposed to migration, disasters mostly lead to internal displacement, meaning people are forced to move within their own countries’ borders⁹. This provides national and local actors, such as governments, NGOs, and communities, with the opportunity to take a comprehensive approach to reducing disaster risk and promoting the resilience of internally displaced persons.

According to the International Displacement Monitoring Centre¹⁰, three key ingredients are necessary for durable solutions to displacement: reinforced political commitment, strengthened capacity, and improved evidence. National policies and investments reflect political commitment, while strengthened capacity involves designing and implementing effective disaster risk reduction programs. Improved evidence refers to data quality and availability to monitor risks, disasters, and displacement. While there is no one-size-fits-all solution, these factors are likely essential elements.

International treaties and agreements, such as the Paris Agreement, the United Nations Framework Convention on Climate Change’s Warsaw International Mechanism on Loss and Damage, and the Sendai Framework for Disaster Risk Reduction, as well as the Sustainable Development Goals, are helping to facilitate efforts in disaster risk reduction and climate change adaptation at national and local levels. However, integrating disaster displacement into national policies and programs has mostly occurred as a crisis response in countries with high rates of disaster displacement and increased risk, such as Fiji, Kiribati, and Vanuatu^9,10. Despite the growing risk of displacement, the preparedness and capacity of national governments appear inadequate.

Capacity building in environmental governance is crucial to reduce the risk and increase preparedness for disaster displacement. Disasters induced by climate change lead to a vicious cycle of displacement, which, in turn, reduces adaptation capacity due to economic, social, and political costs. To break or weaken this cycle, climate adaptation action is required by the state, market, and social actors to decrease vulnerability and enhance readiness to adapt and mitigate the risk. In this study, we aim to test this proposition by examining the link between environmental governance and disaster displacement. To do so, we use the ND-GAIN readiness index to measure environmental governance, which provides individual scores for governance, economic, and social readiness. This approach allows us to explore the linkages between the capacity of state, market, and social actors in environmental governance and displacement.

There is a growing body of literature examining the relationship between disasters, climate change, and internal displacement^3,11,12. Kam et al.¹³ found that every degree of global warming increases global displacement risk by 50% and global flood displacement by 150% by the end of the century. Silva Rodríguez de San Miguel et al.¹⁴ similarly concluded that water-related disasters are the primary cause of internal migration due to environmental factors in the Americas. Investigating the link between disasters and internal displacement is most relevant in the Asia-Pacific and Sub-Saharan Africa regions, where over 85% of disaster-induced displacement occurs¹⁵. Saha and Ahmed¹⁶ analyzed the trends and impacts of population displacement due to disasters and climate change, laying out the impacts on education, health, and income in Bangladesh, a country highly vulnerable to hydrometeorological disasters. They found a negative correlation between socioeconomic conditions and internal displacement, which also supports the findings of Islam and Khan¹⁷ in Southeast Asia. The high dependence on climate-sensitive livelihoods and frequent occurrence of extreme events in South Asia makes it particularly vulnerable to disaster-induced displacement¹⁸. Kaenzig and Piguet¹⁹ highlighted the social and economic fragility dependent on environmental vulnerability, catalyzing displacement and migration in Latin America. Ibrahim and Mensah²⁰ studied the effects of floods, sea-level rise, and droughts on displacement and migration due to the impact on agriculture-based livelihoods in African countries. Finally, Borderon et al.²¹ revealed the complexities underlying the link between environmental change and human mobility in Africa, taking into account various sociodemographic, economic, and political factors.

Overall, the connection between disasters and internal displacement is highly complex and heavily reliant on the vulnerability of livelihoods to environmental factors and adaptability, which involves a combination of social, political, and economic factors embedded in environmental governance. While the above studies collectively contribute to our understanding of the relationship between disaster displacement and essential social and economic factors, none of them investigated such a relationship from a global perspective. To our knowledge, this study presents the first attempt to uncover the linkages between environmental governance and disaster displacement using a unique panel dataset, covering 92 countries over an 11-year period. Existing analyses have focused more on the local or regional level, which is limited in identifying commonalities across regions and drawing overarching conclusions. This study aims to emphasize that the issue of disaster displacement extends globally. Confronting such global challenges necessitates adopting a global perspective, and this study is geared toward bridging this critical gap. Moreover, the domain of global disaster studies and capacity building has traditionally relied on measures such as fatalities and economic losses, leaving a gap in quantitative research from a humanitarian standpoint, particularly concerning displaced populations. Hence, the relationship established in this study and the statistical evidence provided contribute to enriching our understanding of disaster studies with a global outlook, particularly in the face of imminent climate change threats.

Results

Country comparisons and temporal trends in disaster displacement and environmental governance

Figure 1 presents a bar graph that shows the percentage of the displaced population for all-scale and small-scale weather-related disasters by aggregating estimates from all countries for each year. Additionally, we provided a list of the top 10 (weather-related) disastrous events based on the total number of internal displacements for the study period inside the figure. As shown, the majority of internally displaced people were affected by large-scale extreme hydro-meteorology events, such as floods and typhoons, which predominantly occurred in developing countries with large populations.

Fig. 1: Temporal trends in global weather-related disaster displacement.

Percentage of weather-related disaster displacement globally during 2010–2020 and top 10 events by the number of people displaced.

Figure 2 displays a spatial visualization of the 92 countries included in our study and their ND-GAIN readiness scores in 2010 and 2020. The maps illustrate that Sub-Saharan Africa and South Asia have the lowest readiness scores, coinciding with their high disaster-related displacement rates. Regional inequality is particularly apparent in terms of economic readiness. Notably, the economic readiness scores of Russia, China, India, the United States, Canada, and Brazil decreased in 2020 compared to 2010. While there was a slight improvement in governance and social readiness across the African continent, the intervals utilized in the spatial representation of readiness revealed no significant changes.

Fig. 2: ND-GAIN economic, governance, and social readiness (2010 & 2020).

Mapping of the ND-GAIN readiness scores for the years 2010 and 2020, including economic, governance, and social readiness. The scores range from 0 to 100, with 0 indicating low country readiness and 100 indicating high country readiness.

Table 1 presents a comprehensive summary of the variables for all countries and four subgroups classified by income level, based on the World Bank’s gross national income (GNI) per capita. The table reports that the average percentage of internal displacement caused by all-scale weather-related disasters was 0.27%, with a maximum of 9.26%, while the average percentage of internal displacement caused by small-scale weather-related disasters was 0.13%, with a maximum of 8.29%. The readiness score ranges from 0 to 100, with a higher score indicating a greater level of readiness. The table highlights that overall readiness varies widely across countries, with the lowest readiness score observed in the Central African Republic at less than 15 and the highest readiness score in New Zealand, with scores exceeding 80.

Table 1 Summary statistics of all countries and subgroups by income

Table 1 also shows clear variations in overall readiness and three individual components by income level. In high-income countries, the mean readiness levels range from 53 to 70, with the greatest variation found in social readiness (SD = 14.99). Upper-middle-income and lower-middle-income countries have mean readiness levels between 26 and 44, with a larger variation in economic readiness (SD = 14.20 or 14.59). Lastly, low-income nations have mean readiness levels between 23 and 32, with relatively little variance across all three components. Based on the conceptual framework discussed previously, we consider the ND-GAIN readiness index to be a robust measure of environmental governance, encompassing the capacity and effectiveness of the three primary actors in environmental decision-making and behavior.

To gain a quick overview of the panel data for internal displacement and readiness index, we also explored the general trend of the two variables by plotting the time series data using the country mean. As seen in Fig. 3, the global percentage of the displaced population globally ranged between 0.2% and 0.4%, with some notably high numbers in 2010 and 2016. Concurrently, the ND-GAIN readiness score has shown ups and downs, with a clear decline in 2013. This decline was particularly evident in lower-middle-income and low-income countries. The majority of the fluctuations seemed to be driven by economic readiness. Governance readiness exhibited a steady increase from 2010 to 2014, followed by a subsequent decline. This trend was more pronounced in higher-income and lower-income countries, while it was less apparent in middle-income countries. Conversely, social readiness has exhibited continuous growth globally and across all three income groups. Additionally, we observed a negative correlation between internal displacement and ND-GAIN readiness with lags, indicating that a growing readiness score might be followed by a lower percentage of internal disaster displacement. This correlation was more pronounced for economic and governance readiness and less so for social readiness.

Fig. 3: Disaster displacement and readiness index over time.

Line graphs show the relationship between disaster displacement and the readiness index for all countries and subgroups by income over time. The black dashed line represents the percentage of the displaced population. The blue solid line represents the overall readiness index. The dark orange, purple, and dark green lines show the economic, governance, and social readiness indexes, respectively.

To ensure the validity of the Granger non-causality tests, we next conducted panel unit root testing since non-stationary panels are required. Three commonly used tests were employed to confirm the stationarity of our variables: (1) the Levin-Lin-Chu (LLC) test²²; (2) the Im-Pesaran-Shin (IPS) test²³; and (3) the augmented Dickey-Fuller (ADF) test^24,25. The null hypothesis for all three tests is that the variables contain a unit root. The optimal lags according to the AIC criterion were used for the LLC and IPS tests, while three lags were assigned for the ADF test. To accommodate the dynamic nature of country income classification over the past decade and ensure an adequate number of groups in the subsample, we combined the four classifications into three income groups in our analysis. Specifically, the three subsamples were categorized as high and upper-middle-income, upper and lower-middle-income, and lower-middle and low-income countries. Overall, the results of all three-panel unit root tests presented in Table 2 are consistent and satisfactory. Thus, we can conclude that all variables are level-stationary and can proceed with the Granger non-causality tests.

Table 2 Panel unit root tests

Bidirectional relationship between disaster displacement and environmental governance

We conducted a series of Granger non-causality tests to investigate the causal relationship between internal disaster displacement and environmental governance as measured by ND-GAIN readiness and its three components. The null hypothesis stated that readiness does not Granger-cause displacement, and displacement does not Granger-cause readiness. In a non-technical term, disaster displacement from previous lagged years does not contain useful information for predicting readiness, and readiness levels from previous lagged years do not contain relevant information for predicting internal displacement.

Table 3 presents the results of the Granger non-causality tests for weather-related disasters of all scales. The top panel examines if displacement caused by these disasters Granger-causes readiness scores, while the bottom panel examines whether readiness scores, in turn, Granger-cause disaster displacement. Column (1) indicates whether the null hypothesis can be rejected, while Columns (2)–(4) report the feedback coefficients with varying optimal lags. Of particular interest are the negative coefficients in the bottom panel, which provide empirical evidence on the effectiveness of environmental governance in mitigating internal disaster displacement.

Table 3 Granger causality tests: disaster displacements & readiness (all-scale)

We found that disaster displacement has a Granger-causal effect on environmental governance, as measured by the ND-GAIN readiness index. The significantly positive signs in the two feedback coefficients indicate that a higher percentage of displacement predicts a higher readiness score one and three years later. This effect is reflected through individual readiness components: a higher percentage of displacement leads to higher economic readiness in lag one and three, and higher governance and social readiness in lag one and two. These findings are reasonable, as disasters can increase public and private investments needed for immediate reconstruction and recovery, thereby stimulating economic activities. Additionally, post-disaster recovery efforts provide an opportunity to improve social readiness by addressing vulnerabilities exposed by the disaster²⁶. Past disaster experiences play a critical role in preparing for future disasters by increasing awareness, building resilience, and enhancing information sharing^27,28. These findings suggest that the impact of displacement can extend beyond immediate recovery, fostering long-term improvements in governance and social structures.

Conversely, we found that ND-GAIN readiness Granger causes disaster displacement, indicating a bidirectional relationship between the two variables. The significantly negative feedback coefficient of readiness on displacement in lag two suggests that more effective environmental governance leads to a lower percentage of disaster-induced displacement in the second year. This finding is mainly driven by economic and governance readiness, with significantly negative coefficients also observed in lag two. Although we observed negative feedback coefficients for social readiness in lag two and three, these coefficients are not statistically significant.

The same Granger non-causality tests were repeated to examine internal displacement caused by small-scale weather-related events, with the results presented in Table 4. In the top panel, we observed consistent positive feedback coefficients of displacement on overall readiness, as well as on governance and social readiness. However, a notable difference lies in the feedback coefficients for economic readiness, which shifted from positive to negative in lags one and three. This indicates that a higher percentage of displacement predicts a lower economic readiness score one and three years later. This finding suggests that the positive impacts of disaster recovery and reconstruction are primarily driven by large-scale disastrous events. For small-scale events, the economic damage and disruptions cannot be offset by the limited economic investments that occur, unlike large-scale events where substantial infrastructure projects are often implemented.

Table 4 Granger causality tests: disaster displacements & readiness (small-scale)

Furthermore, unlike the results observed for all-scale disasters, a bidirectional relationship was only evident for overall readiness and governance readiness (see Table 4 bottom panel). Although negative feedback coefficients were observed in lag two or three for the three readiness components, they were not statistically significant. These findings suggest that the bidirectional relationship between disaster displacement and environmental governance might be less strong when considering only small-scale events. This is likely due to small-scale disasters being predominantly managed within the local context, where they receive comparatively less attention or investment from the central government and larger community. The lack of targeted central intervention or any intervention at all can lead to less pronounced improvements in mitigating disaster displacement.

Overall, we identified a strong bidirectional relationship between environmental governance and disaster displacements caused by all-scale weather-related disasters. Improved economic and governance readiness, in particular, revealed beneficial effects by reducing the percentage of disaster displacements. However, in the case of small-scale disasters, localized management and limited resources resulted in a less significant relationship between the two variables. Small-scale events may not generate the same level of urgency or resource allocation as large-scale disasters, leading to less comprehensive recovery efforts and fewer opportunities for governance and readiness improvements. Policymakers should consider these dynamics when planning disaster response and recovery strategies, ensuring that even small-scale events receive adequate investment for the long-term benefit of reducing displaced populations.

Bidirectional relationship between disaster displacement and environmental governance by income groups

Next, we examined the causal relationship between internal disaster displacement and environmental governance across three different income groups: (1) the higher-income group, which includes high-income and upper-middle-income countries; (2) the middle-income group, which includes upper-middle and lower-middle-income countries; and (3) the lower-income group, consisting of lower-middle-income and low-income countries. The results of the Granger non-causality tests for all-scale weather-related disasters are presented in Table 5, while Table 6 displays the results for small-scale weather-related disasters.

Table 5 Granger causality tests: disaster displacement & readiness by income (all-scale)

Table 6 Granger causality tests: disaster displacement & readiness by income (small-scale)

The results presented in Table 5 indicate that (all-scale) disaster-induced internal displacement Granger causes an increase in overall readiness across all income groups, as reflected by the significantly positive feedback coefficients in lagged years. The causal relationships between displacement and the individual components of readiness are also consistent across income groups. In the higher-income group, positive feedback coefficients were observed for displacement on economic and governance readiness, while a negative coefficient on social readiness was observed in lag one, followed by a positive coefficient in lag two. In middle-income and lower-income groups, positive and significant coefficients on individual readiness components were also observed, indicating that a higher percentage of disaster displacement could predict a higher level of economic, governance, and social readiness, with varying lags following the shock.

In the opposite direction, we discovered that overall readiness Granger causes displacement in all income groups. However, significantly negative feedback coefficients were only observed among middle-income and lower-income groups, highlighting the crucial role of enhanced environmental governance in reducing the percentage of future disaster displacement in these countries. For the higher-income group, a significantly negative feedback coefficient was observed only in governance readiness in lag three, underscoring the critical importance of enhancing governance quality, such as the stability and capacity of government institutions, in mitigating disaster displacement in these nations. Conversely, significantly negative feedback coefficients were observed in economic and social readiness in the middle-income group in lag two, suggesting the necessity of prioritizing improvements in economic and social conditions. Lastly, negative feedback coefficients were observed in all three individual readiness components in the lower-income group, emphasizing the imperative for comprehensive disaster management strategies that address economic development, governmental effectiveness, and social conditions to reduce the number of displaced populations caused by disasters.

The results for small-scale weather-related disaster displacement are presented in Table 6. Consistent with the aggregated findings in Table 4, a bidirectional relationship is observed between overall readiness and disaster displacements, as well as governance readiness and disaster displacements across all income groups. However, the relationship between economic and social readiness and disaster displacements is not bidirectional. Moreover, significantly negative feedback coefficients of displacement on economic readiness were consistently observed in all three income groups, further confirming the previous finding that post-disaster economic investment and opportunities are primarily generated by large-scale disastrous events. Furthermore, we also found significantly negative coefficients of displacement on social readiness in the higher-income group and on both governance and social readiness in the lower-income group. These findings suggest the disruptive nature of small-scale disasters on government capacity and socioeconomic conditions, particularly among countries with lower income levels. These small-scale but frequent weather events can exacerbate issues of social inequality in lower-income countries, warranting policy attention.

In the opposite direction, our analysis revealed a significant negative feedback coefficient of governance readiness on disaster displacement in both higher-income and middle-income groups (in lag 3). This finding consistently underscores the critical role of stable and capable government institutions in effectively managing extreme weather events, irrespective of their scale. For the lower-income group, significantly negative coefficients were observed for the overall readiness index in lag two, as well as for individual readiness scores in either lag one or lag two. These findings illuminate great potential and opportunities for responding to and mitigating disaster displacement in lower-income countries. Strengthening readiness across any or all areas within these nations could substantially contribute to reducing the impact of disasters on displacement and enhancing overall disaster resilience. By investing in readiness measures, such as infrastructure development, community preparedness, and institutional capacity building, lower-income countries can better equip themselves to withstand the adverse effects of both large and small-scale disasters.

Overall, the analysis identified governance readiness as the most crucial component for mitigating weather-related disaster displacements in higher-income countries. Conversely, the middle-income group relied on both economic and governance readiness, while all three readiness components were deemed important for the lower-income group. These findings hold particular significance in light of the escalating threat posed by climate change, which is leading to more frequent and severe weather-related disasters through alterations in the earth’s climate patterns, such as rising global temperatures, sea levels, and changes in precipitation patterns²⁹. The results offer valuable insights for policymakers in developing effective climate adaptation plans and strategies to combat the escalating threat of weather-related disasters.

Discussion

The bidirectional Granger causality between disaster displacement and ND-GAIN readiness implies mutual predictive capacity. The percentage of people displaced in previous years can help predict the readiness level, and vice versa. Governance readiness is found to be the primary driver of this causal relationship, followed by economic readiness, emphasizing the critical roles of state and market actors in environmental governance to mitigate the consequences of internal displacement caused by weather-related disasters. Furthermore, there is always a time lag in the feedback process, suggesting that effective environmental governance takes time to unfold. Given our empirical analysis, negative coefficients are found in lag two for overall readiness and mostly in lag two or three for individual readiness. Nevertheless, our findings provide compelling evidence supporting the need for enhanced environmental governance and resilience to minimize the human impact of disaster shocks.

We recognize two primary limitations in our study. First, the global scope of our research inherently neglects local-national conditions and contexts, encompassing complex cultural, economic, social, and environmental factors, as well as within-country differences. This global-level approach also restricts our analysis to examining weather-related disasters collectively. While we investigated the relationship by considering both small-scale and large-scale disastrous events and categorizing countries based on income levels, alternative perspectives could be explored. This might involve analyzing the relationship based on different types of disasters or categorizing countries according to geopolitical context or human development levels. Given that specific types of disasters are concentrated within particular regions, such as storms in coastal countries with warm climates and earthquakes in nations situated along fault lines, regional studies could offer deeper insights into the relationship between environmental governance and displacement caused by specific types of disasters.

Secondly, it is important to note that the ND-GAIN readiness index is not universally utilized as a measure of environmental governance. While our study introduces a novel approach by employing the ND-GAIN readiness index to assess both overall environmental governance and its individual components, the interpretation of our findings carries strong epistemological implications. When utilizing the findings from this study for policy and practical applications, careful attention should be paid to how the ND-GAIN readiness index is defined and calculated, ensuring a comprehensive understanding of its implications and limitations.

Despite the limitations, our research highlights several key policy implications. First and foremost, our findings emphasize the critical importance of capacity-building in governance readiness to reduce the risk of internal displacement from weather-related disasters across all income groups. Governance readiness is measured by the World Governance Indicators (WGI), which include political stability and nonviolence, corruption control, regulatory quality, and the rule of law. Political stability and nonviolence are fundamental challenges to sustainable development and presuppose government capacity. Corruption control, regulatory quality, and the rule of law offer more feasible opportunities for improvement. As we previously noted, financial capture and inadequate law enforcement mechanisms impede the effectiveness of adaptation initiatives. Given how the WGI indicators measure government effectiveness, improvements in governance readiness would likely lead to less internal displacement through the general enhancement of government effectiveness and efficiency across all sectors.

Improved economic readiness appears to have a significant impact on disaster displacement in middle- and low-income countries, according to our analysis. The World Bank’s Doing Business indicators evaluate the investment climate for adaptation efforts as part of a more comprehensive evaluation of economic development and business. Similar to governance readiness, enhancing economic indicators is likely to lead to better-financed and regulated efforts in adaptation and sustainable development initiatives.

Our findings also suggest that there is a threshold in economic readiness that precludes further improvement in disaster displacement correlated with economic readiness at the country level, as seen in the absence of Granger causality in the higher-income group. In other words, while high-income countries may be economically equipped to adapt to increasing disaster displacement risk, they may have already achieved peak adaptation outcomes in terms of economic readiness. Therefore, focusing on enhancing economic readiness in middle- and low-income countries may yield more significant results in reducing disaster displacement.

Enhancing social conditions by addressing social inequality, enhancing ICT infrastructure, promoting education, and fostering innovation, as measured by the social readiness score, has been found to be particularly effective in reducing disaster displacement in lower-middle and low-income countries. Disasters often hit vulnerable communities the hardest, including those with low income, poor education, and inadequate access to resources and services. When disaster strikes, these communities face challenges in responding and recovering, often leading to displacement. By improving socioeconomic conditions, communities can become more resilient to disasters and reduce their risk of displacement. For example, access to education can provide people with the skills and knowledge necessary to respond to disasters and reduce their vulnerability.

For middle-income and high-income countries, while disaster displacement can provide an opportunity to improve social readiness levels, it does not necessarily lead to a reduction in displacement through Granger causality. This may be due to differences in public awareness and civil society involvement in environmental governance. In upper-income countries, civil society is typically more involved in environmental governance, and there is greater awareness of environmental degradation and climate change. This greater involvement of social actors in governance can lead to more tangible results. Conversely, low-income countries may have room for social actors to take over government responsibilities in providing public goods such as adaptation efforts due to a lack of government effectiveness.

Conclusion

Based on the theoretical framework and the empirical results, we can conclude that there exists a strong link between environmental governance, as measured by the ND-GAIN readiness index, and internal disaster displacement. This bidirectional relationship holds true for all-scale and small-scale weather-related events, as evident from the analysis of both combined country data and subsamples by income group. Governance readiness emerged as the most significant factor across all income groups, while middle- and low-income countries show a particular need for enhancing economic readiness to effectively reduce disaster displacement.

In light of these results, we recommend that policymakers and researchers focus on exploring the underlying mechanisms of the correlation between governance readiness and displacement, such as identifying the vulnerability of governance readiness to displacement shocks and understanding the role of governance readiness in reducing disaster displacement. Moreover, it is crucial for countries to continue to prioritize economic development to adequately respond to disaster displacement. This entails investing in adaptation initiatives like building resilient infrastructure, formulating emergency response plans, and extending financial support to affected businesses and individuals. These investments not only reduce the risk of displacement but can also create new opportunities for economic growth and development.

Furthermore, it is worth emphasizing that social readiness remains an inseparable component of the environmental governance system. Achieving social readiness demands a collaborative approach among governments, organizations, and communities to empower vulnerable populations to respond and recover from disasters. With the increasing threat of weather-related disasters as a result of climate change, the involvement of market, social, and state actors in collective action is crucial to developing more targeted and informed climate action, ultimately reducing the incidence of displacement.

Finally, although the impact of climate change varies tremendously across countries, empirical studies from a global perspective remain valuable. Such studies allow for a comprehensive understanding of the patterns, trends, and factors influencing disaster displacement on a large scale, aiding in the development of international policies and cooperation strategies, as climate change and its associated risks do not adhere to national boundaries. By highlighting the magnitude and complexities of disaster displacement on a global scale, this study also raises awareness and advocates for increased attention and resources to address the issue beyond the local or regional context. Ultimately, this study aims to provide empirical evidence to support the creation of comprehensive strategies that not only address immediate disaster responses but also foster long-term resilience and sustainable development across diverse regions.

Methods

Conceptual framework

Environmental governance refers to the interventions aimed at changing environment-related incentives, knowledge, institutions, decision-making, and behaviors^5,30,31. These interventions involve three categories of participant stakeholders: state actors, market actors, and social actors^30,32. Good governance requires collaboration among these three actor groups.

For example, preparedness involves policies and regulations issued by the government, such as disaster-resistant building codes, land-use regulations, and funding allocation to institutions responsible for pre-disaster warning systems and strategies. The legislative branch of the government issues regulations, while the executive and judicial bodies, civil society, and the private sector ensure enforcement, compliance, and accountability. Civil society, including local NGOs and international humanitarian organizations, shapes post-disaster relief and resilience outcomes. Private sector actors participate through their interactions with the other stakeholders and through risk insurance mechanisms. Capacity building in environmental governance requires empowering the three sectors to address disaster displacement effectively.

The literature emphasized the importance of the environmental governance network as an inclusive, resilient, and effective system for disaster risk reduction. The network consists of local, national, regional, and international institutions and organizations, each with different roles and levels of agency throughout the disaster cycle. It is intricate and fluid, adapting its composition and actions based on the disaster type, phase of the cycle, and socioeconomic and political context. Theoretical models for successful environmental governance networks include decision-making power, expertise, resources, incentives for cooperation, flexibility, transparency, and accountability, among others^{33,34,35,36,37}. Horizontal and vertical coordination among stakeholders is a vital factor^38,39. Albris et al.⁴⁰ discussed guidelines for effective disaster governance, such as sharing knowledge, harmonizing capacities, institutionalizing coordination, engaging stakeholders, leveraging investments, and developing communication.

Recent evidence suggested that better disaster governance is achievable^34,41, but the complexities and challenges of a harmonized governance network remain. Successful disaster governance requires decentralization, with the central government traditionally being the primary, if not the only, actor. The disaster governance network must allow for relative autonomy at lower levels of government, such as local municipalities, and must facilitate the involvement of NGOs and the private sector. Despite recommendations from external experts, such as international organizations like UNDRR, the implementation of better governance often remains on paper⁴². The central government’s institutional tendency to maintain power and marginalize other stakeholders continues to present a significant challenge⁴³. This means that implementing good environmental governance requires overcoming challenges of decentralization and empowerment, such as bureaucracy, coordination issues, organizational ego, and competition for resources^43,44,45. If the central government lacks capacity, new regulations for better disaster governance can have counterproductive results. For example, the capture of risk reduction finance at the local level by authorities, along with a lack of trust and coordination with civil society and private sector investments, can increase disaster risk^41,44.

Good environmental governance relies on the foundational prerequisite of state capacity or government effectiveness^46,47. Zuo et al.⁴⁸ defined government effectiveness as a combination of the efficiency of government institutions, the quality of public and civil services, and the credibility of government policies. Among other significant factors such as economic openness, democracy, and corruption, government effectiveness is considered the most important factor in evaluating risk governance outcomes. When the government is effective, other stakeholders’ roles and actions can be facilitated, enforced, and regulated.

In addition to evaluating state capacity, it is necessary to assess the capacity of social and market actors in measuring the effectiveness of disaster risk governance. Successful disaster governance involves the assessment of the effectiveness of state, market, and social actors in disaster risk reduction, adaptation, and resilience building. The determinants of effectiveness in disaster risk reduction are similar to those of general capacity, as previously discussed. Overall, the theoretical framework of environmental governance consists of a network of state, market, and social actors, whose collaborative efforts take the form of comanagement, public-private partnerships, and private-social partnerships³⁰.

To gain a more comprehensive understanding of the factors influencing disaster displacement, we turn to the ND-GAIN governance, social, and economic readiness scores, which serve as important indicators of the effectiveness of state, social, and market actors in environmental governance. Specifically, the governance readiness score assesses the capacity of state actors to manage disaster risks, while the social readiness score evaluates the capacity of social actors, such as community-based organizations and civil society groups, to respond to and recover from disasters. The economic readiness score measures the capacity of market actors, including the private sector, to adapt to and mitigate the effects of disasters. Additionally, ND-GAIN produces a composite readiness score derived from the three components, allowing us to assess the overall readiness of the environmental governance network as a system. By examining the overall effectiveness of the system, as well as the specific roles played by each actor, we can identify opportunities for collaboration and improvement, ultimately leading to more effective disaster risk reduction strategies and reduced rates of displacement.

Data

This study draws data from two primary sources: the Global Internal Displacement Database (GIDD) and the Notre Dame Global Adaptation Initiative (ND-GAIN) database. Due to the limited availability of data in certain years and the exclusion of countries with multiple gaps in information, we constructed a panel dataset consisting of 92 countries from 2010 to 2020. The selection of these databases enabled us to perform a comprehensive analysis of the relationship between environmental governance and internal disaster displacement.

The Global Internal Displacement Database (GIDD) produced by the Internal Displacement Monitoring Centre is a leading source of annual estimates of internal displacement caused by conflict, violence, and disasters⁴⁹. The data is currently available for the period spanning from 2008 to 2023, providing the most comprehensive and reliable information at the country level. The GIDD categorizes disasters as either geophysical (e.g., earthquake, dry mass movement, and volcanic eruption) or weather-related (e.g., flood, storm, drought). These two types of disasters differ fundamentally in their origins, characteristics, and the methods used to predict and manage them. Geophysical events, typically caused by the movement of the Earth’s tectonic plates, are less frequent but can have large magnitudes. In contrast, weather-related events are more frequent, ranging from small-scale occurrences to large-scale catastrophes, and are directly and significantly impacted by climate change, necessitating comprehensive climate adaptation strategies. Therefore, this study will focus exclusively on weather-related disasters.

Moreover, to address the potential distortion of data trends caused by historical large-scale catastrophes with large displaced populations, we conducted an additional analysis excluding such large-scale weather-related events. The exclusion criterion was based on Munich Re’s⁵⁰ definition, stating that a catastrophe is considered ‘great’ if the number of homeless exceeds 200,000. Using population statistics from the World Bank and the total number of displaced individuals, both with and without the exclusion, we calculated the percentage of internal disaster displacement caused by weather-related disasters of all scales and only small scales for each country between 2010 and 2020.

The Environmental Change Initiative at the University of Notre Dame (ND-ECI) manages the Notre Dame Global Adaptation Initiative’s (ND-GAIN) Country Index, which summarizes a country’s vulnerability and readiness to enhance resilience to climate change and other global challenges from 1995 to 2020⁵¹. The readiness index used in this study evaluates a country’s ability to leverage investments toward adaptation actions and generates an overall readiness score based on three components: economic readiness, governance readiness, and social readiness. Specifically, economic readiness assesses the investment climate of a country using World Bank Doing Business (DB) indicators to determine how well it can attract adaptation investment. Governance readiness measures the stability and capability of government institutions based on the World Governance Indicators (WGI), including political stability and nonviolence, corruption control, regulatory quality, and the rule of law. Finally, social readiness examines the socioeconomic conditions that can enable adaptation efforts, such as social inequality, information and communication technology (ICT) infrastructure, education, and innovation.

Models and specifications

To investigate the relationship between environmental governance and internal disaster displacement, we will employ the Granger non-causality test⁵², which is an econometric method specifically designed to assess causal relationships in time series data. The test is particularly suitable for understanding whether changes in one variable precede or predict changes in another over time. Additionally, it accounts for potential lagged effects between variables, which is essential in analyzing dynamic systems where the impact of one variable on another may take time to unfold. The relationship between the two variables, using panel data with N cross-sectional units and T time series dimensions, can be expressed as:

$${y}_{i,t}=f({x}_{i,t-l})$$

(1)

for i = 1, …, N, l = 1, …L, and t = L + 1, …, T. L is the maximum number of lags used.

Moreover, given that Granger causality requires both time series variables to be stationary, a panel unit root test must be performed as a prerequisite. The general form of the test can be expressed as:

$$\Delta {y}_{i,t}={\rho }_{i}{y}_{i,t-1}+{u}_{i,t}$$

(2)

The null hypothesis of the test is that the panels contain unit roots (are non-stationary), while the alternative hypothesis is that at least one panel is stationary. Once we make sure that the variables do not contain a unit root process, we can proceed with the Granger non-causality test. We use a homogeneous approach to testing for Granger causality in heterogeneous panels, proposed by Juodis et al.⁵³. Following the linear dynamic panel data, equation (1) can be modeled:

$${y}_{i,t}={\alpha }_{0,i}+{\sum }_{l=1}^{L}{\alpha }_{l,i}{y}_{i,t-l}+{\sum }_{l=1}^{L}{\beta }_{l,i}{x}_{i,t-l}+{\epsilon }_{i,t}$$

(3)

The parameters α_0,i are the individual-specific effects, α_l,i are the heterogeneous autoregressive coefficients and ϵ_i,t are the errors. The heterogeneous feedback coefficients (also called Granger causality parameters) are denoted as β_l,i. Our goal is to determine whether these coefficients are significantly different from zero. Specifically, the null hypothesis is that the variable x_i,t does not Granger-cause the other variable y_i,t, or in other words, x_i,t is not helpful at predicting y_i,t. If we reject the null hypothesis, we can infer that x_i,t contains information that can improve the prediction of y_i,t.

In this study, we aim to investigate the link between internal displacement resulting from weather-related disasters and environmental governance, as measured by the ND-GAIN readiness index. Given the panel dataset consisting of 92 countries and spanning the period from 2010 to 2020, equation (3) can be rewritten as:

$$Readines{s}_{i,t}= \, {\alpha }_{0,i}+{\sum }_{l=1}^{L}{\alpha }_{l,i}Readines{s}_{i,t-l}\\ +{\sum }_{l=1}^{L}{\beta }_{l,i}Displacemen{t}_{i,t-l}+{\epsilon }_{i,t}$$

(4)

$$Displacemen{t}_{i,t}= \, {\alpha }_{0,i}+{\sum }_{l=1}^{L}{\alpha }_{l,i}Displacemen{t}_{i,t-l}\\ +{\sum }_{l=1}^{L}{\beta }_{l,i}Readines{s}_{i,t-l}+{\epsilon }_{i,t}$$

(5)

for i = 1, …, N (= 92), l = 1, …L (= 1, 2, or 3), and t = L + 1, …, T (= 11). We will allow a maximum of three time lags given the 11-year dataset, and the optimal lag length will be selected using the BIC criterion. The variable Readiness_i,t represents either a composite ND-GAIN readiness score or the individual component measuring economic, social, or governance readiness for each country. The variable Displacement_i,t presents the percentage of the population displaced as a result of disasters, including those of all scales and those of only small scales.

The null hypothesis for equation (4) or (5) can be expressed as:

$${H}_{0}:{\beta }_{l,i}=0,\,{{\mbox{for all}}}\,\,i\,\,{{\mbox{and}}}\,\,l$$

(6)

Against the alternative hypothesis:

$${H}_{1}:{\beta }_{l,i}\ne 0,\,{{\mbox{for some}}}\,\,i\,\,{{\mbox{and}}}\,\,l$$

(7)

We will first test the null hypothesis stated in equation (4) that disaster displacement has no Granger-causal effect on readiness scores. Subsequently, we will assess whether readiness scores have any Granger-causal effect on disaster displacement, as outlined in equation (5). We will conduct a series of Granger non-causality tests under three different scenarios: (1) displacement caused by weather-related disasters of all scales; (2) displacement caused by weather-related disasters of small scales; and (3) displacement caused by weather-related disasters of both all scales and small scales across three income levels. The Wald test statistic and its corresponding P-value will be reported to determine whether the null hypothesis can be rejected at different levels of significance.