Key Determinants for a Resilient Development
Resilient development makes sure that, through inclusive systems-building and capacity development, individuals and communities have what they need to be better prepared to cope and recover from crises[1]. Resilience is, hence, vital to move forward into sustainable and inclusive development.
One way to achieve resilience is by adapting to climate change and its future effects. Strong adaptation skills lead to a greater resilience level. These two terms are intrinsically linked to each other — not only adaptation leads to resilience, but resilience is a property needed for having the capacity to adapt.
It is essential to ask how to enable climate change adaptation so that it can happen timely. There are many ways to do it and conditions that need to be in place beforehand. As I explained in a previous article (paste the link), having a plan to adapt to the effects of climate change is just not enough.
“Resilience is vital to move forward into sustainable and inclusive development.
A considerable number of widely researched determinants help countries in this process, such as evidence-based planning, monitoring, strong leadership, collaboration and learning from others. However, they do not set an order of priority and do not guide countries on what determinants should be considered first[2]. An important reason for this absence of comprehensive guidance might be lack of data or centralized macro-data, something prevalent in this field.
In this article, however, I try to demonstrate the importance of some variables to achieve adaptation — not only for climate-related hazards and disasters but for all kind of them.
Data sources and variables for the statistical models
To know the key determinants that are more likely to push adaptation further, the University of Notre Dame’s Global Adaptation Initiative (ND-GAIN) was used. The reason behind this election is data availability (for countries and years) and the relevance of such indicators for adaptation. More concretely, to simplify the assessment, the indicators that were selected were:
- Economic readiness: Captures the ability of a country’s business environment to accept investment that could be applied to adaptation that reduces vulnerability (reduces sensitivity and improves adaptive capacity).
- Social readiness: Captures the factors such as social inequality, ICT infrastructure, education and innovation that enhance investment mobility and promote adaptation actions.
- Governance readiness: Captures the institutional factors that enhance the application of investment for adaptation.
The indicators from this index are compared against three independent variables:
- Number of deaths
- Houses destroyed
- Houses damaged
From the UNDRR database[3] from any disaster[4]. The reasons behind the selection of these two independent variables are data availability and the direct effect of climate change adaptation.
If a country has strong adaptation skills, it should see its disaster-related losses and damages reduced (including the number of deaths and number of houses damaged and destroyed), as well as its vulnerability to the effects of climate change. Since not only natural hazards are included, we are talking here about a resilient development.
The relation between vulnerability and readiness has been already demonstrated by Sarkodie and Strezov (2019). They studied the relationship between vulnerability (which positively affects disaster losses — more vulnerability, more losses) and the readiness indicators above, using the same index for the data.
While measuring vulnerability is more generic (the propensity or predisposition of human societies to be negatively impacted by climate-related hazards), in my assessment, I attempt to be more specific by analyzing more concrete damages and losses, which cause more public attention, in contrary to general vulnerability.
Not all countries have the same exposure levels. Exposure is defined as the presence of people, livelihood, environmental services and resources, infrastructure, or economic, social, or cultural assets in places that could be adversely affected by physical events and are subject to potential future harm, loss, or damage. Hence, exposure is also included in the model — given the influence it can have on total damages and losses. This data is extracted from the same index mentioned earlier.
Another independent variable, which is essential for the assessment is GDP per capita. Usually, the least developed countries are the ones that suffer more damages from the climate change effects. Therefore, the poverty level may influence the independent variables (number of deaths and houses damaged and destroyed). This data is extracted from the World Bank.
Statistical models
In order to measure if economic, social and governance readiness, exposure and GDP per capita have any influence on the direct consequences of climate change, three different models were studied:
Model 1:
Model 2:
Model 3:
Yit is either number of deaths, houses damaged, or houses destroyed.
t-x shows that the data is lagged (moved back x amount of years). In Model 1, all the data is regressed in the same year. In Model 2, the dependent variables (those on the left) are moved two years back, and in Model 3, five years back. The only variable that is not moved five years back is GDP/capita. It doesn’t make sense for politicians (the ones that at the end of the day take the decisions in all aspects, including climate change adaptation initiatives) to think about actions five years in advance since the change of mandate usually occurs every four years. Hence, a GDP/capita of five years ago may not be relevant.
Relationships between indicators and climate change adaptation: Panel data analysis
After running multiple tests and models for panel data analysis, from 9 models I had in total, only three show global statistical significance, have good results and passes additional statistical tests. In other words, the relationship between the independent and dependent variables is caused by something different than chance and is aligned to research.
The results are shown below:
1. Model 1: When Y is houses destroyed
2. Model 2: When Y is houses destroyed
3. Model 3: When Y is deaths
What do these models tell us?
From these results, the following conclusions can be drawn:
1. GDP per capita is a crucial component for high resilience, which is aligned to the fact that the wealthiest countries are the least vulnerable to adverse shocks. GDP per capita doesn’t, however, affect the number of people that die in a disaster in the models.
2. Past governance readiness from five years ago (institutional factors that enhance the application of investment for adaptation) plays a key role in reducing the number of people that dies in a disaster. This time-lagged effect may be explained by the fact that adaptation projects take place to be implemented.
3. Exposure, constant through time for all the countries in the dataset, positively affects the number of deaths due to a disaster.
4. As long as we move the variables back in time, the adjusted R squared of the model increases, meaning that the time component is essential. This might be because investment takes time to take place, and projects take time to be completed. Furthermore, the adjusted R square is not too big most of the times, which suggests that more variables have a role to play in the number of deaths, houses damaged and destroyed after a disaster. More relevant data could not be added due to unavailability, such as the corruption perception index.
Conclusions
An enormous challenge when creating this model was the lack of available open datasets, making the present model incomplete. Although there is this significant constraint, this assessment gives us essential insights into what affects resilient development. Despite one cannot change exposure to disasters, since it is primarily a geographical component, the change must come from governance. If public institutions facilitate investment for adaptation, each country will better adapt to future adverse shocks. This facilitation must come as soon as possible since the existence of facilities doesn’t imply that each country can complete all the projects in the same year. They take some time to complete, which explains why governance readiness from the previous five years is vital to reduce losses.
“Despite one cannot change exposure to disasters, since it is primarily a geographical component, the change must come from governance.
On the other hand, it is also important to have enough resources available. It is hence not a surprise that the poorest countries are generally the most affected by disasters. On this point, the wealthiest countries must keep assisting those more vulnerable to better adapt to the imminent consequences of climate change.
Technical notes
Given data gaps, the countries that are included in this assessment were: Costa Rica, Cambodia, Ecuador, Panama and Sri Lanka for the years: 2000–2019. I did not pick further past years in order to make this analysis more relevant to the present. Older data may not be relevant for the statistical model due to lower impacts from climate change.
I used pooled OLS for the regression analysis to assess the relations since both null hypotheses for random and fixed effects were rejected.
The number of observations from my dataset, and the fact that the countries are from different parts of the world, make it relevant to extrapolate the conclusions globally.
Aknowledgements
I would like to acknowledge Santiago Lema Burgos for his inputs and editorial work.
References
[1] Regional United Nations Development Group (R-UNDG) Eastern and Southern Africa (ESA) & Western and Central Africa (WCA), Strategic Framework to Support Resilient Development in Africa. Available at: https://www.preventionweb.net/files/57759_undgframeworkforresilientdevelopmen.pdf
[2] Some examples include https://www.acq.osd.mil/eie/downloads/CCARprint_wForward_e.pdf and https://www.theccc.org.uk/publication/progress-in-preparing-for-climate-change-2019-progress-report-to-parliament/.
[3] UNDRR DesInventar Sendai. Available at: https://www.desinventar.net/DesInventar/main.jsp
[4] Not only natural hazards, since there was not enough data for the statistical models to work.
