Annex A. Methodological annex
Existing measures of fragility
The academic and policy world has been grappling for some time with methods and measures for best conceptualising states of fragility. That fragility is a multidimensional concept is now well accepted in both spheres, although there is still debate over defining those dimensions.
One conceptualisation of a multidimensional framework for fragility proposes a disaggregation based on a conception of statehood as the antithesis to fragility. Statehood in turn is defined by effectiveness and legitimacy (Goldstone et al., 2000; Marshall and Cole, 2014; Rice and Patrick, 2008); a three-dimensional conception of state authority, state legitimacy and state capacity; or some variation on this theme (Carment, Samy and Landry, 2013; Call, 2010; Grävingholt, Ziaja and Kreibaum, 2015). Empirical assessments of fragility tend to try to measure these dimensions of statehood based on domains such as political performance, economic performance and social performance. The George Mason University State Fragility Index, for example, has indicators of effectiveness and legitimacy across four domains: security, political, economic and social. The scores from the individual domains are combined to provide overall effectiveness and legitimacy scores, which are in turn combined to give an overall state fragility score. Data are available for 167 contexts as of 2014, with a historical time series going back until 1995. Because the index is aggregated based on concepts of effectiveness and legitimacy, it is difficult to analyse variations in the four domains. This potentially hides some significant differences between developed and developing countries, for example, or within the group of developed countries, which may face differing challenges.
The Country Indicators for Foreign Policy (CIFP) fragility index assesses state performance in terms of authority, capacity and legitimacy in six dimensions: governance, economics, security and crime, human development, demographics, and environment. It uses 75 indicators. Data have been made available for 2010, 2011 and 2012, but it is unclear whether the index has been calculated in subsequent years. Although separate scores for authority, capacity and legitimacy are calculated, it is unclear which indicators contribute to each of the three dimensions.
Moving away from a definition of fragility associated with statehood, the Fragile States Index measures fragility as a combination of pressures and capacity to respond to those pressures in three dimensions: social, political and economic. The 3 primary dimensions are further deconstructed to 12 sub-dimensions, which are defined by 88 indicators. The methodology involved in generating the dimension scores is highly sophisticated and based on triangulating three primary sources of data, using the Conflict Assessment System Tool analysis platform. Data are available for 178 contexts and the index has been calculated on an annual basis since 2005. While novel in its conceptualisation of fragility as a combination of pressure and capacity to respond, the index itself does not distinguish between these two sets of drivers of fragility, conflating indicators for both into the same score. It is thus difficult to establish whether fragility in one dimension is more a function of high pressures or low capacity to respond. This in turn has implications for policy making and donor assistance.
The Index of State Weakness in the Developing World is a fragility assessment based on four domains: security, political, economic and social. The 4 domains are measured with 20 individual indicators, and data are available for 141 developing countries, defined as those with a gross national income per capita below USD 11 115 and with a population above 100 000. The index was designed to help policy makers identify countries at risk of violence or conflict including terrorism (Rice and Patrick, 2008). The elegance of this index lies in its simplicity. However, it does not advance thinking on conceptualising fragility beyond existing indicators, and was only measured in 2008. Furthermore, it focuses only on developing countries, so a holistic view of fragility around the world is not possible.
A number of other indices attempt to capture concepts related in some degree to fragility, or at least to some component of fragility. The Global Peace Index (GPI) of the Institute for Economics and Peace (IEP), for example, measures internal and external peace based on political and security domain indicators. The GPI covers 163 contexts from 2008 onwards. The Worldwide Governance Indicators project (WGI) calculates indicators for 215 economies from 1996 to 2014 on 6 dimensions of governance, including political stability and the absence of violence (Kaufmann and Kraay, 2015). The Country Policy and Institutional Assessment (CPIA) measures 4 domains of state capacity – economic management, structural policies, policies for social inclusion and equity, and public sector management and institutions – for 95 economies from 2005 to 2014 (World Bank, 2015). None of these indices or indicators alone comprehensively addresses fragility.
The OECD’s move towards a risk-based approach to fragility is a significant departure from these previous attempts to measure fragility. Although there are extant measures of “risk”, these measures are not directly coupled to the concept of fragility as defined by the OECD. For example, the Global Conflict Risk Index (GCRI), the Peace and Conflict Instability Ledger, the Index for Risk Management (INFORM), and the World Risk Index are all empirical measures of various risks.
The GCRI measures the risk of conflict and conflict intensity with respect to five risk domains: political, social cohesion and public security, conflict prevalence, geography and environment, and economy. It uses 22 indicators to measure the risk of conflict in the near future as well as the intensity of ongoing conflict for 137 contexts, going back to 1989 (De Groeve, Vernaccini and Hachemer, 2014). Although the index uses the same domains as the OECD fragility framework used in this report, it is one step removed from a measure of fragility in that it looks at risk but not explicitly at coping capacity.
The Peace and Conflict Instability Ledger is a ranking of 163 contexts based on their estimated risk of experiencing major political instability or armed conflict. It is calculated biannually based on five indicators representing four risk domains: political, economic, social and security. The results are presented as a risk ratio, which is interpreted as the relative risk of instability in a country or context, compared to the average estimated likelihood of instability for 28 of the member countries of the OECD. The estimated likelihood of instability for these average OECD countries between 2010 and 2012, for example, was 0.008. For this same period, Afghanistan, which was ranked at the top of the instability ranking, was 36 times more likely to experience conflict or political instability. Similarly to the GCRI, this index has been calculated to measure risk alone, rather than a holistic conception of fragility.
The INFORM index is an attempt to measure disaster risk as an interaction of hazard and exposure, vulnerability, and lack of coping capacity. These 3 dimensions of risk are measured by 21 indicators in the social, economic, political, security and environmental domains. The index has been calculated since 2012, and covers 194 contexts. Although the INFORM index is intended to measure risk, the methodology also includes measures of coping capacities.
The World Risk Index calculates the risk of experiencing an extreme natural disaster for 171 contexts. A country or context faces a high risk if it is highly exposed to natural hazards and if its society is highly vulnerable, i.e. risk is calculated as exposure multiplied by susceptibility. While exposure mainly relates to environmental events such as earthquakes and flooding, sustainability includes indicators of social and economic domains (United Nations University, 2014).
While the OECD fragility framework has some overlap in terms of the multidimensionality of fragility, specifically in five domains, the conceptualisation of fragility as being a combination of high risk and low coping capacity necessitates a methodology that, for each domain, selects indicators of both risk and coping capacity. This is a key distinguishing feature of the OECD fragility framework from both existing measures of fragility, and existing measures of risk.
Conceptualising fragility as risks and coping capacities
The OECD formally defines fragility as:
Fragility is a combination of exposure to risk and insufficient coping capacity of the state, system, and/or communities to manage, absorb or mitigate those risks. Fragility can lead to negative outcomes, including violence, the breakdown of institutions, displacement, humanitarian crises, or other emergencies.
The OECD fragility framework looks at current exposure to natural disasters or violence, for example, as well as the ability of a country or context to deal with future negative events. It is the combination of risks and coping capacities.
The understanding of fragility as a manifestation of risk is not new. The African Development Bank (AfDB) defines fragility as a “condition of elevated risk of institutional breakdown, societal collapse, or violent conflict” (AfDB, 2014). Similarly, the World Bank is currently adapting its approach to fragility to reflect multidimensional risks, in line with its 2014 World Development Report: Risk and Opportunities. The notion of risk also corresponds to previous work by the International Network on Conflict and Fragility stressing the vulnerability of fragile contexts to external shocks and their weak governance and response capacity (OECD, 2011). This reflects a broader effort by development actors to invest in risk management that flows from international policy agreements made in 2015, including the Sustainable Development Goals, the Sendai Framework for Disaster Risk Reduction, the Addis Ababa Action Agenda and the Paris Agreement. All of these stress the need to identify, reduce, manage and prepare for risk.
However, risk is a complex area of practice and theory with competing schools of thought (Box A.1). Risks in the OECD framework are interpreted as hazards, threats and vulnerabilities that can be both internally generated within a society or polity as well as externally driven threats, hazards or vulnerabilities from either other countries or from external environmental events. Fragility manifests in the presence of low or poor coping capacities to mitigate the threat from these risks. However, the OECD framework does not attempt to calculate risk by calculating the probability of an event occurring and the impact that it would have. Instead, it uses different indicators as proxies for risk and considers their levels in relation to the coping capacities of 171 contexts.
The concept of risk is shared across many sectors – economic, financial, business, political, health and so on. Risk has thus been defined in varying ways, although most definitions refer to likelihood and impact. The OECD 2013 High Level Risk Forum called risk “the potential damage caused by a single event or a series of events” and “a combination of two factors”. Those two factors are the “probability of the occurrence of a hazard”, with hazard defined as a “potentially harmful event”, and vulnerability, which is defined as “a measure of the exposure of human lives, health, activities, assets or the environment to the potential damage caused by such hazards occurring” (OECD, 2011). Similarly, in the humanitarian community, the risk of disaster is often referred to as a combination of hazard and/or exposure and vulnerability. For example, the IFRC defines disaster as an occurrence “when a hazard impacts on vulnerable people” (IFRC, 2016).
Other definitions also point to the potentially negative outcome and its impact. The World Economic Forum describes a global risk as an “an uncertain event or condition that, if it occurs, can cause significant negative impact for several countries or industries within the next 10 years” (WEF, 2016). Leading the social risk management work for the World Bank, Holzmann and Jørgensen defined risk as “uncertainty or unpredictability that results in welfare losses” (Holzmann and Jorgensen, 2000).
The World Development Report 2014 takes an action-oriented approach, arguing that taking on risk, “the possibility of loss” is necessary to pursue development opportunities or “the possibility of gain”. This requires a shift from ad hoc responses to proactive, systematic and integrated risk management (World Bank, 2014). Analysing avenues towards “reducing the risks of violence”, the World Development Report 2011 had already advocated a move “from sporadic early warning to continued risk assessment wherever weak institutional legitimacy, and internal or external stresses indicate a need for attention to prevention” (World Bank, 2011).
Risks can be categorised in different ways. The World Economic Forum distinguishes different categories of risk (economic, environmental, geopolitical, societal and technological) and grades their severity according to their likelihood and impact on a global scale. Holzmann and Jørgensen list natural, health, life cycle, social, economic, political and environmental risks and classify them as micro-, meso- and macro-level risks depending on the number of people they potentially affect (Holzmann and Jorgensen, 2000).
Regardless of the definition selected, the ability of any quantitative measure to estimate the probability of any event is highly contested and criticised (Taleb and Treverton, 2015). Prediction is hard (Berdal and Wennmann, 2010). In building the methodology for the OECD fragility framework, the possibility of developing new quantitative risk measures across five dimensions capable of being accurate, justifiable, sufficiently different from existing established measures, and strong enough to obtain agreement throughout the community was considered not only unlikely, but unobtainable. Instead, the approach developed selects risk indicators based on research that has discussed each one of the indicators as representing an increased likelihood or impact of a negative event. These are not used to build a purely quantitative measure of risk per se. Rather the intent is to develop a typology of fragility where each cluster is created quantitatively but analysed qualitatively. This innovative mixed-methodology, multidimensional approach builds on the strengths of qualitative and quantitative analysis to better understand fragility.
Sources: Berdal and Wennmann (2010); IFRC (2016); OECD (2013); Taleb and Treverton (2015); WEF (2015); World Bank (2000).
Indicator selection
The choice of indicators has been driven by a selection criteria in line with the OECD’s fragility concept of high risk and low coping capacity. In addition to the normal technical criteria for selecting good indicators, of particular importance was the selection of indicators based on their relationship to fragility: are they a cause of fragility or an outcome of fragility? Indicators that represent outcomes of fragile contexts do not offer clear guidance as to policies that can reduce fragility. For example, infant mortality is an indicator used in several fragility measures. However, infant mortality is arguably more of an outcome of a fragile health context than a cause of it. To decide on the selection of indicators, the following criteria was used:
-
Risk: do the indicators alter either likelihood or impact ex ante?
-
Coping capacity: what indicators would stop the risk cascading ex post if the risk occurred?
Using these criteria does not eliminate the challenge of separating some indicators into either a risk or a coping capacity category. For example, levels of armed personnel can be thought of as a coping capacity for dealing with insurgencies. It can also be thought of as contributing to the risk of violence. Methodological decisions have been made to account for this to produce the best approximation given such limitations.
Further, some coping capacity indicators have been used in more than one dimension. This introduces an unintended issue in aggregating the dimensions to provide the final 56 contexts used for analysing flows by fragility. Using indicators more than once in effect weights these indicators more than others. While statistical measures have been employed to minimise the effect of this, this method was still chosen as an alternative to dividing some of these indicators into one and only one dimension. For example, government effectiveness and rule of law are important not only for the security dimension, but also the environmental dimension. Forcing these into one or the other dimension arbitrarily creates an incomplete picture of the interconnectedness of coping capacities. Furthermore, while these are in effect weighted more highly as a result, this was believed to be justifiable given their cross-dimensional importance. Table A.1 lists the indicators of fragility used in the States of Fragility 2016 methodology.
The relationship between selected indicators and fragility
Economic dimension
Resource rent dependence: resource dependence leaves an economy open to shocks in the global system as oil and mineral prices fluctuate. Resource dependence has also been found to increase the propensity for violence through greed and grievance mechanisms (Collier and Hoeffler, 2004). The USAID Fragile States Indicators list views primary export dependence as a measure of economic effectiveness.
General government gross debt: poverty and economic decline put extra pressures on a state in terms of service delivery, and can cause or exacerbate frictions between those who “have” and those who “have not”. People in situations of vulnerable employment tend to be hardest hit in economic crises, and countries facing high levels of sovereign debt tend to be most exposed during times of economic crises. In fragile economies especially, economic grievances can often result in protest, violence and conflict.
Youth not in education, employment or education (NEET): when youth, especially young men, are not engaged in productive activity such as employment, education or training, they may pose a threat to social stability and conflict. Youth are more likely to be recruited as fighters and take up arms when their expected incomes from the formal labour market or agriculture are less than their expected incomes from fighting (Collier and Hoeffler, 2004). Moreover, low levels of secondary education, again particularly in males, is strongly correlated with the outbreak of civil war.
Aid dependency: aid dependency can increase risk of conflict – severe aid shocks (i.e. decreases in aid) have been found to alter the domestic balance of power and induce violence (Nielsen et al., 2011).
GDP growth rate: an economy growing strongly is less likely to see economic tensions leading to violence. Economies that go through periods of negative growth and growth shocks have an increased likelihood of conflict (Miguel, Satyanath and Sergenti, 2004). High rates of economic growth that result in increased economic inequality tend to exacerbate underlying tensions and may lead to an increased likelihood of conflict over the distribution of resources.
Unemployment rate: high rates of unemployment, particularly among males, may lower the opportunity cost of joining an armed rebellion that may offer enticing alternative income through looting and pillaging. High rates of unemployment may also breed or further grievances among ethnic groups or among opposition groups and government, further contributing to social and political discontent (Elbadawi and Sambanis, 2000).
Education: low levels of education can lead to dependence on low-skilled work, which tends to be the most vulnerable employment in an economy. High levels of education combined with low levels of economic opportunity, however, are also a dangerous mix, acting as a catalyst for violent conflict.
Regulatory quality: if governments have good regulatory quality – the ability to formulate and implement sound policies and regulations that permit and promote private sector development – economic shocks are more easily contained (Dorsch, Dunz and Maarek, 2015).
Remoteness: contexts located far from major world markets face a series of structural handicaps, such as high transportation costs and isolation that render them less able to respond to shocks in an effective way. They also have greater difficulty diversifying their economies, even with globalisation and the Internet. Remoteness increases the costs of acquiring necessary imports, thus creating vulnerability to price shocks on global markets as well as to domestic shocks (including, e.g. natural disasters). Remoteness is a structural obstacle to trade and growth and is particularly harmful in the case of many lower income small island developing states and landlocked developing countries.
Men and women in the labour force: labour force participation is an indirect measure of economic human capital, which in the long term is needed to ensure sufficient coping capacity in the economy.
Food security: food security is a fundamental indicator of a country’s resilience to both environmental and economic shocks. Lack of food insecurity has been identified as a link between social and economic tensions and the spillover into violence and conflict (Brinkman and Hendrix, 2011).
Environmental dimension
Natural disasters risk: the INFORM natural disasters index captures the risk of some commonly occurring natural disasters, which in turn are a measure of environmental risk. The CIFP also uses a disaster risk measure for environmental fragility.
Environmental health: environmental health measures the protection of human health from environmental harm. The component indicators measure air and water quality, pollution levels, and safe sanitation.
Prevalence of infectious diseases: research suggests that the prevalence of infectious diseases can increase the risk of violent conflict outbreak. Infectious diseases can lead to the emergence of ethnocentric cultural norms, which coupled with resource competition among ethnic groups can lead to an increased frequency of civil wars (Letendre, Fincher, and Thornhill, 2010).
Uprooted people: the presence of refugees and displaced populations can increase the risk of subsequent conflict in host and origin countries. A majority of refugees never directly engage in violence but refugee flows facilitate the transnational spread of arms, combatants and ideologies conducive to conflict, and also alter the ethnic composition of the state. They can also exacerbate economic competition (Salehyan, 2008). Some measure of refugee burden on a host setting or country is also used in GCRI, INFORM, CIFP indices.
Socio-economic vulnerability: vulnerable populations are less able to cope with hazardous environmental and economic shocks such as natural disasters or economic collapse. Socio-economic vulnerability can stem from various sources, such as inequalities and dependencies as well as fundamental levels of development or deprivation.
Core civil society index: social capital has been defined as “the set of rules, norms, obligations, reciprocity, and trust embedded in social relations, social structures, and society’s institutional arrangements which enables its members to achieve their individual and community objectives”. Social capital reduces the risk of social instability spilling over into social violence and conflict by allowing groups to overcome differences and resolve problems (Lederman, Loayza and Menendez, 2002).
Government effectiveness: perceptions of government effectiveness reflect a population’s underlying level of discontent or satisfaction with the political status quo. Low levels of perceived government effectiveness can lead to social discontent, political protest and ultimately political violence.
Food security: food security is considered a fundamental indicator of a country’s resilience to both environmental and economic shocks. At the same time, food insecurity has been identified as a link between social and economic tensions and the spillover into violence and conflict: thus, food security can be considered a buffer between tension and violence (Brinkman and Hendrix, 2011). The Fragile States Index and the CIFP use measures of food insecurity in their calculations.
Political dimension
Regime persistence: both entrenched democracies and entrenched autocracies can be considered politically stable in the sense that there is a low probability of regime breakdown. Transitions between regime types are a manifestation of political instability, and which provide opportunities for political violence (Hegre et al., 2001). The State Fragility Index uses regime durability as a measure of government effectiveness.
Political terror: state-sanctioned violence against its citizens is a manifestation of a collapse of state legitimacy, which research has identified as one critical measure of fragility. Furthermore, state repression often forces opposition groups towards other means of expressing dissent including violence (Regan and Norton, 2005). The GCRI and State Fragility Index use repression indicators in their calculations.
Perception of corruption: corruption can increase grievances and demands for political change that may trigger political violence and social unrest. Corruption can also fuel greed, which may provide motivations for opposition groups to try and capture the state through violent means, and for the state to use violent means to repress opposition (Le Billon, 2003). High levels of corruption increase the risk of political violence and instability. The Fragile States Index, CIFP and USAID Fragile States Indicators all use some measure of corruption in their index calculations.
Decentralised elections: a highly centralised state suffers bigger consequences when political and sectarian turmoil occur than those contexts that have managed to decentralise power. Many highly centralised states mask a suppression of sectarian tensions, which, when they do erupt, can result in violence and conflict, as seen in the Syrian Arab Republic (hereafter “Syria”) and Iraq. Centralisation also may increase the probability of a military coup leading to further political instability (Taleb and Treverton, 2015).
Restricted gender physical integrity value: in the absence of good global level data on gender-based violence (GBV), the restricted gender physical integrity value captures the extent to which GBV issues are considered in law and attitudes in any country. The index captures laws on domestic violence, rape and sexual harassment, as well as attitudes toward violence and prevalence of violence, prevalence of female genital mutilation, and reproductive autonomy. Many components of the index relate to the legal environment in a country and therefore reflect a component of political fragility. Restricted physical integrity has effects on women’s health outcomes, but spills over into economic and social outcomes as well.
Voice and accountability: a mechanism for channelling grievances and participating in the political process provides an outlet for pressures that may otherwise boil over into violence.
Judicial and legislative constraints on executive power: conflicts are more likely to erupt in political systems that suffer from a lack of rule of law and of checks and balances. By preventing action that oversteps legitimate boundaries of the state, checks and balances contain spillover effects from political instability (Grant and Keohane, 2005). The executive is less likely to be able to take control of the state, or to co-opt the military into performing actions which may lead to a cascading effect on violence.
Security dimension
Homicide rate: high homicide rates reflect a diminished capacity of government to perform its duties to protect people within its borders.
Level of violent criminal activity: violent criminal activity may undermine a state’s ability to exercise its monopoly on violence and increase risks to public security of persons and property (Tilly, 1985). Furthermore, organised crime undermines a state’s capacity and legitimacy by undermining its ability to provide public goods and services and making corruption the norm (van Dijk, 2007). When a state’s capacity and legitimacy are eroded, the potential for an outbreak of violent conflict either internally, or externally through transnational organised criminal activities, increases (Miraglia, Ochoa and Briscoe, 2012).
Deaths by non-state actors per capita: armed non-state actors undermine the state’s monopoly on the use of force and are drivers of security fragility (Schneckener, 2006).
Impact of terrorism: terrorism is intrinsically linked to the environment of safety and security. Terrorist attacks can cause already unstable situations to fall further into the precipice of violence.
Conflict risk: the GCRI is an index of the statistical risk of violent conflict in the next 1-4 years based on 25 quantitative indicators from open sources.1 The GCRI measures this with respect to five risk domains: political, social cohesion and public security, conflict prevalence, geography and environment, and economy. Twenty-two indicators are used to measure the risk of conflict in the near future as well as the intensity of ongoing conflict for 137 countries, going back to 1989. Although the index uses the same domains as the States of Fragility framework, the index is one step removed from a measure of fragility in that it looks at risk but not explicitly at coping capacity.
Battle-related deaths per capita: high levels of battle-related deaths indicate high security fragility and can contribute to further conflict and instability.
Restricted gender physical integrity value: in the absence of good global level data on GBV, the restricted gender physical integrity value captures the extent to which gender-based violence issues are considered in law and attitudes in any country. The index captures laws on domestic violence, rape and sexual harassment, as well as attitudes toward violence and prevalence of violence, prevalence of female genital mutilation, and reproductive autonomy. Many components of the index relate to the legal environment in a country and therefore reflect a component of political fragility. Restricted physical integrity has effects on women’s health outcomes, but spills over into economic and social outcomes as well.
Police officers per 100 000 population and armed security officers per 100 000 population: a state’s security apparatus ensures its monopoly over violence and control over territory, as well as public safety. With adequate police and security personnel, a state that is experiencing security instability will be able to respond quickly and in a way that can make further cascading effects – for example full-scale civil war – less likely to break out.
Rule of law: the rule of law provides a means of addressing grievances through means other than violence and conflict. In ethnically heterogeneous societies in particular, it has been found that a strong rule of law is associated with enduring peace (Easterly, 2001).
Control over territory: states that control their territory, however fragile the security situation has become, are resilient to total state collapse and failure (Rotberg, 2002).
Formal security alliances: countries that are members of formal security alliances are more resilient to conflicts spilling over from neighbouring countries, and tend to honour alliances including defence pacts, thereby reducing the effects of conflicts (Leeds, 2003).
Societal dimension
Gini coefficient: although the causal relationship between vertical inequality and conflict is debated (Stewart, 2010), high levels of income inequality can cause or exacerbate underlying social tensions as well as overall levels of poverty in the general population.
Gender inequality: research has found that countries characterised by gender inequality are more likely to be involved in interstate disputes and more likely to rely on violence to settle those disputes. It has also been found that high levels of gender inequality may lead to a greater propensity for intrastate conflict (Caprioli, 2005).
Horizontal inequality: horizontal inequalities within a society may affect social cohesion. Kaplan (2008) and others argue that state fragility is caused not only by weak institutions, but also by a lack of social cohesion that leads to the erosion of intergroup trust and an increased risk of conflict. Many post-conflict development programmes now have one component focused on rebuilding social cohesion to reduce the likelihood of relapse into conflict.
Uprooted people: the presence of refugees and displaced populations can increase the risk of subsequent conflict in host and origin countries. A majority of refugees never directly engage in violence but refugee flows facilitate the transnational spread of arms, combatants and ideologies conducive to conflict, and also alter the ethnic composition of the state. They can also exacerbate economic competition (Salehyan, 2008). Some measure of refugee burden on a host country is also used in the GCRI, INFORM, CIFP indices.
Urban growth rate (percentage): urbanisation and the speed of urbanisation have a positive relationship with crime rates (Muggah, 2014). Urbanisation has also been found to have a direct effect on levels of political protest, which heighten the risk of political conflict (Auvinen, 1997). The CIFP and USAID Fragile States indicators both use some measure of urbanisation.
Core civil society index: social capital has been defined as “the set of rules, norms, obligations, reciprocity, and trust embedded in social relations, social structures, and society’s institutional arrangements which enables its members to achieve their individual and community objectives”. Social capital reduces the risk of social instability spilling over into social violence and conflict by allowing groups to overcome differences and resolve problems (Lederman, Loayza and Menendez, 2002).
Access to justice: if citizens have mechanisms to resolve disputes in a peaceful manner using the legal system, grievances induced by social risk factors can be diverted from violent action and conflict, containing the effects of any realised risks.
Voice and accountability: a mechanism for channelling grievances and participating in the political process provides an outlet for pressures that may otherwise boil over into violence.
Indicator coverage and missing data
The 43 indicators selected do not cover all contexts and imputation techniques have been used to fill in data gaps. Lack of data is the primary reason why a context may not be included. The OECD fragility framework inclusion threshold is that at least 70% of data had to be available for a context for it to be included. In 2016, this yields a list of 171 contexts.
For some indicators it is possible to assume that contexts missing from the dataset have a certain value. For example, those missing from the battle deaths and deaths by non-state actors datasets can be assumed to have a value of 0. Where no reasonable assumption could be made, data are imputed using k-nearest neighbour (KNN) imputation that uses statistical inference to fill in missing data from the k most similar contexts. In the OECD fragility framework, this has been done using the 15 most similar contexts for each missing data point.
OECD approach compared to existing approaches to measuring fragility
Until 2015, the OECD had been using a “fragile states list”. However, the use of a list as the sole indicator of fragility has become increasingly incongruent with the growing recognition that fragility is a multidimensional issue (Grävingholt, Ziaja and Kreibaum, 2015). In addition, composite indices have a number of conceptual, methodological and practical drawbacks:
-
There is little agreement in the literature as to drivers of fragility, making any particular measure open to debate (Mack, 2010).
-
In general, composite indices require a judgement on whether indicators are good or bad. While there are methods of dealing with more complicated dynamics in composite indices, it has been noted that such approaches are seldom used (Grävingholt, Ziaja and Kreibaum, 2015).20 However, as understanding of fragility and resilience progresses so too does recognition that this is an oversimplification. For example, increasing resilience sometimes increases fragility.
-
The additive nature of composite indices implies that poor performance in one of the indicators selected can be offset by strong performance in another. While compensation between factors may be realistic in some cases, in general it is hard to justify.
-
Composite indices lists can be good for vertical comparisons of contexts, i.e. whether Country A is more fragile than Country B. However, they are limited in horizontal comparisons. For example, if two contexts score 0.5 in some measure of fragility, this does not provide any information about the composition of this score. This is problematic as it is recognised that certain combinations of weaknesses may pose larger threats than others though quantitatively they produce the same score.
-
Using a list conveys the impression that “fragile states” can be thought of as a homogenous group when in reality fragility is highly contextualised and can manifest in many varied and different ways (Gould and Pate, 2016).
-
Lists alone make it difficult to cluster contexts into sensible and cohesive groups, making them of limited use for the “operational task of crafting policies to counter state fragility” (Grävingholt, Ziaja and Kreibaum, 2015). For example, a fragility index could rank Haiti, Pakistan and Zimbabwe together masking the stark differences in their risk exposure and capacities.
-
Finally, and perhaps most troublesome, the OECD Fragile States List was developed as a tool for tracking support and aid flows. However, it was often misunderstood as a definitional tool and had a stigmatising effect. Use of such lists can result in negative outcomes for the contexts named as fragile, introducing the risk of donors entering into a damaging pro-cyclical aid flow cycle (Collier and Hoeffler, 2004).
To address these issues, and while recognising the usefulness of a list for donors, the OECD developed a multidimensional approach. Its aim is to provide a deeper understanding of fragility and be valuable for high-level analysis. By defining fragility as combinations of risks and coping capacities, the new framework is applicable globally. It is an approach where quantitative analysis provides the input for more qualitative descriptions of global fragility profiles.
The methodology of the fragility framework is based on a two-stage process that first examines countries in each of the five dimensions and then aggregates this information to obtain an aggregate picture of fragility. For each dimension, principal components analysis (PCA) is used to combine the risk and coping capacity indicators into two statistically derived components. Deriving two measures per dimension has distinct advantages over creating one composite index. First, using two measures allows for a greater understanding of the differences between contexts that would score equally when using a single measure. Second, using the first two principal components allows countries to be broadly grouped based on their similarities in all of the input variables. Third, each indicator is weighted by the amount of new information it brings to the data rather than on a set of normative judgements on their relative importance. With the components of each dimension calculated, countries are then clustered on similarities and classified descriptively. This mix of both quantitative and qualitative methods appropriately offers a more flexible approach to describing the diversity of fragility.
With contexts classified into groups within each dimension, the second part of the methodology aggregates this information to arrive at an overall picture of combinations of fragilities. To achieve this, the components of each dimension provide inputs to a second aggregate PCA. This is then used to produce a list of the 56 most fragile contexts. The broad approach is shown in Figure A.1.
The methodology is ambitious in its objectives, but is not without limitations. PCA allows a reduction of indicators into two core components that explain most of the variance in the original data. However, in doing so, invariably information is lost. The Stage 2 PCA exacerbates this loss of information. In short, the results of this approach are a summary of the initial indicators that is then interpreted in terms of fragility. However, this summary is both more informative and less arbitrary than any composite index based on the initial indicators.
Aside from the technical limitations, there are also certain practical limitations to what can be captured in any quantitative approach. For example, at the micro-level, fragility is not only a context-level risk, but can manifest in pockets of fragility at the subnational level. At the macro level, drivers and effects of fragility can be transnational, either through local spillovers or via interstate spillover. However, the unit of analysis for all datasets used for this report is at the country level, which limits the ability to capture such scale and dynamics. Going forward, it would be useful to define entry points for drawing on subnational data or linking up the global scan with regional and local exercises or concepts. Further, while data on governance are widely available, data on informal institutions are less so. While every effort has been made to include indicators of both, at this point the lack of quality data is an issue.
The differences between other types of composite indices and the OECD fragility framework are summarised in Table A.3.
Table A.3. Differences between traditional composite indices and the OECD mixed methods multidimensional approach.
Results
The OECD fragility framework uses PCA to reduce the number of indicators and create
two non-correlated components to analyse fragility. It is multidimensional, but visualising the results past three dimensions is challenging. In presenting the results it is therefore necessary to describe the visualisations used and how to interpret them.
A standard visualisation of the results of a PCA is a biplot. Biplots are a type of exploratory graph used in statistics, a generalisation of the simple two-variable scatterplot. A biplot allows information on both samples and variables of a data matrix to be displayed graphically. Biplots have been used to visualise the results of the 2016 OECD fragility framework.
Contexts are plotted on a scatterplot of the first two principal components. The x-axis represents the first principal component, which accounts for the largest proportion of variance in the data. The y-axis represents the second principal component, which accounts for the second largest proportion of variance in the data. Vectors are also plotted on the graphs that describe how much each indicator contributes to each principal component. The length of the vector represents the variance of the particular indicator it represents – the longer the vector, the higher the variance of the indicator observed in the data. The direction in which the vector is pointing represents the loading (or weighting) of the represented indicator on the first and second principal components. The angle between the vectors (the cosine of the angle between the vectors) represents the correlation between the indicators the vectors represent: the closer the angle is to 90 or 270 degrees, the smaller the correlation between the indicators. Angles of 0 or 180 degrees represent correlations of 1 and -1 respectively.
Figure A.2 is provided to assist in interpreting the following biplots.
Biplot interpretation:
-
because the arrows for Indicator 1 and Indicator 2 run more along the x-axis, the first principal component (x-axis) is most closely related to a combination of Indicator 1 and Indicator 2
-
the second principal component (y-axis) is most closely related to Indicator 3 because the vector for indicator 3 is closest to the y-axis
-
Indicator 3 contributes to both principal components
-
Countries A, B and C are similar because they are close to each other
-
Countries D, E and F are similar because they are close to each other
-
Country G is an outlier in its characteristics.
Biplots can be difficult to interpret for non-statisticians. To assist, countries have been clustered based on the first two principal components and qualitatively described. The approach uses hierarchical clustering, a statistical technique that groups contexts to maximise similarity among contexts within a group based on how similar they are in terms of their values on indicators (Wolfson, Madjd-Sadjada and James, 2004). As the number of contexts within any cluster is dependent on how many clusters are used, the boundaries of any cluster do not represent a real world property. As such, the OECD only uses these as an indicative aid to assist in the qualitative assessment of different types of fragility.
Stage 1: Dimensional fragility analysis
Economic dimension
The economic dimension in the fragility framework aims at capturing the vulnerability to risks stemming from the weaknesses in economic foundations and human capital including macroeconomic shocks, unequal growth, high youth unemployment, etc. Selected indicators in this dimension are presented in Table A.4.
The indicators from Table A.4 that contribute the greatest to the first two principal components are:
First principal component (x-axis): Long-term drivers of growth
-
Aid dependence and resource dependence, lack of economic interconnectedness, education, males in the labour force, regulatory quality, and food security.
Second principal component (y-axis): Labour market imbalances
-
Unemployment rate, NEET (youth not in employment, education and training), and females in the labour force.
Based on these two components, Figure A.3 and Table A.5 describe the economic fragilities of each cluster.
Environmental dimension
The environmental dimension in the OECD fragility framework aims at capturing the vulnerability to environmental, climactic and health risks to citizens’ lives and livelihoods. This includes exposure to natural disasters, pollution and disease epidemics. The environmental domain indicators are presented in Table A.6.
The indicators from Table A.6 that contribute the greatest to the first two principal components are:
First principal component (x-axis): Household level, community and state vulnerability
-
Socio-economic vulnerability, environmental health and food security.
Second principal component (y-axis): Natural and man-made environmental threats and hazards
-
Disaster risk.
Based on these two components, Figure A.4 and Table A.7 describe the environmental fragilities of each cluster.
Political dimension
The “political” dimension in the 2016 OECD fragility framework aims to capture the vulnerability to risks inherent in political processes, events or decisions; to its political inclusiveness (including elites) and transparency (corruption), and to its ability to accommodate change and avoid oppression. Selected indicators in this dimension are presented in Table A.8.
The indicators from Table A.8 that contribute the greatest to the first two principal components are:
First principal component (x-axis): Checks and balances present in political institutions and protection of human rights
-
Voice and accountability, judicial constraints on executive power, perception of corruption, legislative constraints on executive power and political terror.
Second principal component (y-axis): Political grievances
-
Regime persistence, decentralised elections.
Based on these two components, Figure A.5 and Table A.9 describe the profiles of each political fragility cluster.
Security dimension
The security dimension in the 2016 OECD fragility framework aims to capture the vulnerability of citizen security emanating from social and political violence. As such it includes indicators of citizen exposure to direct political and social violence. Selected indicators in this dimension are presented in Table A.10.
The indicators from Table A.10 that contribute the greatest to the first two principal components are:
First principal component (x-axis): Rule of law and state control of territory
-
Violent conflict risk, control over territory, level of violent criminal activity, rule of law and restricted gender physical integrity.
Second principal component (y-axis): Armed conflict, terrorism, organised crime and interpersonal violence
-
Homicide rate, formal alliances, battle-related deaths per capita and the impact of terrorism.
Based on these two components, Figure A.6 and Table A.11 describe the security fragilities of each cluster.
Societal dimension
The societal dimension in the OECD 2016 fragility framework aims at capturing the vulnerability to risks affecting societal cohesion that stem from both vertical and horizontal inequalities (inequality among culturally defined [or constructed] groups), social cleavages, etc. Selected indicators in this dimension are presented in Table A.12.
The indicators from Table A.12 that contribute the greatest to the first two principal components are:
First principal component (x-axis): Access to justice and accountability, and horizontal inequalities
-
Voice and accountability, access to justice, horizontal inequality, core civil society index.
Second principal component (y-axis): Vertical and gender inequalities
-
Gini coefficient, gender inequality and urbanisation.
Based on these two components, Figure A.7 and Table A.13 describe the societal fragilities of each cluster.
Overall principal component statistics
Table A.14 breaks down the contribution of each indicator within each domain to both the first and second principal components in terms of variance of the data explained, as well as how each indicator correlates with each of the first two principal components. It is clear that there is substantial variation in the extent to which indicators contribute to the reduction of each fragility dimension, and how well each indicator correlates with the principal components.
Stage 2: Aggregate fragility analysis
Once contexts are classified into groups within each dimension, the second stage of the methodology aggregates this information to arrive at an overall picture of combinations of fragilities. To achieve this, the components of each dimension provide inputs to a second aggregate PCA. That is, the ten principal components derived across all five dimensions are used as the indicators in the second-tier PCA. The results from this second-tier aggregate analysis produce the list of the 56 most fragile contexts, classified as “extremely fragile” and “fragile”.
The second-tier PCA produces ten fragility clusters of contexts, shown in Figure A.8, which are differentiated not only by the extent of fragility, but also in the dominant characteristics of that fragility. The first dimension of the PCA represents coping capacities and the second dimension represents the types of fragility. To arrive at the 56 fragile contexts, two arbitrary cut-offs have been made. In order for a context to be classified as “extremely fragile” it must score less than -2.5 on the first principal component of the aggregate PCA shown in Figure A.8. In order to be “fragile” a context must score between -1.2 and -2.5 on the first principal component.
References
AfDB (2014), Addressing Fragility and Building Resilience in Africa www.afdb.org/fileadmin/uploads/afdb/Documents/Policy-Documents/Addressing_Fragility_and_Building_Resilience_in_Africa-_The_AfDB_Group_Strategy_2014%E2%80%932019.pdf.
Auvinen, J. (1997), “Political conflict in less developed countries 1981-89”, Journal of Peace Research, Vol. 34/2, pp. 177–195, https://doi.org/10.1177/0022343397034002005.
Berdal, M. and A. Wennmann (eds.) (2010), Ending Wars, Consolidating Peace: Economic Perspectives, Routledge, London.
Brinkman, H. and C.S. Hendrix (2011), “Insecurity and violent conflict: Causes, consequences, and addressing the challenges”, World Food Programme, Occasional Paper, No. 24, July, pp. 1–28, http://documents.wfp.org/stellent/groups/public/documents/newsroom/wfp238358.pdf.
Call, C.T. (2010), “Beyond the ‘failed state’: Toward conceptual alternatives”, European Journal of International Relations, Vol. 17/2, pp. 303-326.
Caprioli, M. (2005), “Primed for violence: The role of gender inequality in predicting internal conflict”, International Studies Quarterly, Vol. 49/2, https://doi.org/10.1111/j.0020-8833.2005.00340.x.
Carment, D., Y. Samy and J. Landry (2013). “Transitioning fragile states: A sequencing approach”, The Fletcher Forum of World Affairs, Vol. 37/2, pp. 125-151.
Collier, P. and A. Hoeffler (2004), “Greed and grievance in civil war”, Oxford Economic Papers, Vol. 56/4, pp. 563–595, https://doi.org/10.1093/oep/gpf064.
De Groeve, T., P. Hachemer and L. Vernaccini (2014), Global Conflict Risk Index.
Dorsch, M.T., K. Dunz and P. Maarek (2015), “Macro shocks and costly political action in non-democracies”, Public Choice, Vol. 162/3-4, pp. 381–404, https://doi.org/10.1007/s11127-015-0239-x.
Easterly, W. (2001), “Can institutions resolve ethnic conflict?”, Economic Development and Cultural Change, Vol. 49/4, pp. 687–706, https://doi.org/10.1086/452521.
Elbadawi, E. and N. Sambanis (2000), “Why are there so many civil wars in Africa? Understanding and preventing violent conflict”, Journal of African Economies, Vol. 9/3, pp. 244–269, https://doi.org/10.1093/jae/9.3.244.
Goldstone, J. et al. (2000), “It’s all about state structure: New findings on revolutionary origins from global data”, Homo Oeconomicus, Vol. 21/3, pp. 429–455, http://www.homooeconomicus.org/lib/getfile.php?articleID=132&PHPSESSID=0f97b006dca504e8852fff437785fc97.
Goodwin, J. (2011), “Why we were surprised (again) by the Arab spring”, Swiss Political Science Review, Vol. 17/4, pp. 452–456, http://onlinelibrary.wiley.com/doi/10.1111/j.1662-6370.2011.02045.x/abstract;jsessionid=64D75D07956EA3FBE328C79A91B48B06.f04t03.
Gould, L.A. and M. Pate (2016), A Introduction: The State as a System 1, CRC Press.
Grant, R.W. and R.O. Keohane (2005). “Accountability and abuses of power in world politics”, The American Political Science Review, Vol. 99/1, pp. 29-43.
Grävingholt, J., S. Ziaja and M. Kreibaum (2015), “Disaggregating state fragility: A method to establish a multidimensional empirical typology”, Third World Quarterly, Vol. 36/7, pp. 1281-1298, https://doi.org/10.1080/01436597.2015.1038340.
Hegre, H. et al. (2001). “Toward a democratic civil peace? Democracy, political change, and civil war, 1816-1992”, The American Political Science Review, Vol. 95/1, pp. 33-48.
Holzmann, R. and S. Jørgensen (2000), “Social risk management: A new conceptual framework for social protection, and beyond”, Social Protection Discussion Papers, No. 21314, The World Bank, pp. 1–30, https://doi.org/10.1023/A:1011247814590.
IFRC (2016) “What is a disaster?”, International Federation of Red Cross and Red Crescent Societies, www.ifrc.org/en/what-we-do/disaster-management/about-disasters/what-is-a-disaster.
Institute for Economics and Peace (2015), Global Terrorism Index 2015: Measuring and Understanding the Impact of Terrorism, http://economicsandpeace.org/wp-content/uploads/2015/11/Global-Terrorism-Index-2015.pdf.
Justino, P. (2009), “Poverty and violent conflict: A micro-level perspective on the causes and duration of warfare”, Journal of Peace Research, Vol. 46/3, pp. 315–333/3, pp. 315-333, https://doi.org/10.1177/0022343309102655.
Kaplan, S. (2008), Fixing Fragile States: A New Paradigm for Development, Praeger.
Kaufmann, D. and A. Kraay (2015), “The Worldwide Governance Indicators (WGI) project 1996-2015”, Natural Resource Governance Institute (NRGI) and Brookings Institution, World Bank Development Research Group, http://info.worldbank.org/governance/wgi/index.aspx.
Le Billon, P. (2003), “Buying peace or fuelling war: The role of corruption in armed conflicts”, Journal of International Development, Vol. 15/4, pp. 413-426.
Lederman, D., N. Loayza and A.M. Menendez (2002), “Violent crime: Does social capital matter?”, Economic Development and Cultural Change, Vol. 50/3, pp. 509–539, https://doi.org/10.1086/342422.
Leeds, B.A. (2003), “Alliance reliability in times of war: Explaining state decisions to violate treaties”, International Organization, Vol. 57/4, pp. 801–827, https://doi.org/10.1017/S0020818303574057.
Letendre, K., C.L. Fincher and R. Thornhill (2010), “Does infectious disease cause global variation in the frequency of intrastate armed conflict and civil war?”, Biological Reviews, Vol. 85/3, pp. 669–683, https://doi.org/10.1111/j.1469-185X.2010.00133.x.
Mack, A. (ed.) (2010), Human Security Report 2009/2010: The Causes of Peace and the Shrinking Costs of War, Oxford University Press.
Marshall, M.G. (2008), “Fragility, instability, and the failure of states”, Council on Foreign Relations, October, www.cfr.org/conflict-prevention/fragility-instability-failure-states-assessing-sources-systemic-risk/p17638.
Marshall, M.G. and B.R. Cole (2014). “Global report 2014: Conflict, governance, and state fragility”, Center for Systemic Peace, Vienna.
Miguel, E., S. Satyanath and E. Sergenti (2004), “Economic shocks and civil conflict: An instrumental variables approach”, Journal of Political Economy, Vol. 112/4, pp. 725–753, https://doi.org/10.1086/421174.
Miraglia, P., R. Ochoa and I. Briscoe (2012), “Transnational organised crime and fragile states”, OECD Development Co-operation Working Papers, No. 5, OECD Publishing, Paris, https://doi.org/10.1787/5k49dfg88s40-en.
Muggah, R. (2014), “Deconstructing the fragile city: Exploring insecurity, violence and resilience”, Environment and Urbanization, Vol. 26(2011), pp. 1–14, https://doi.org/10.1177/0956247814533627.
Nielsen, R.A. et al. (2011), “Foreign aid shocks as a cause of violent armed conflict”, American Journal of Political Science, Vol. 55/2, pp. 219–232, https://doi.org/10.1111/j.1540-5907.2010.00492.x.
OECD (2013), Fragile States: Resource Flows and Trends, OECD Publishing, Paris, https://doi.org/10.1787/9789264190399-en.
OECD (2011), Managing Risks in Fragile and Transitional Contexts: The Price of Success?, OECD Publishing, Paris, https://doi.org/10.1787/9789264118744-en.
Regan, P.M. and D. Norton (2005), “Greed, grievance, and mobilization in civil wars”, Journal of Conflict Resolution, Vol. 49/3, pp. 319–336, https://doi.org/10.1177/0022002704273441.
Rice, S.E. and S. Patrick (2008), Index of State Weakness In the Developing World, The Brookings Institution, Washington, DC.
Rotberg, R.I. (2002), “The new nature of nation-state failure”, The Washington Quarterly, Vol. 25, pp. 83–96, https://doi.org/10.1162/01636600260046253.
Salehyan, I. (2008), “The externalities of civil strife: Refugees as a source of international conflict”, American Journal of Political Science, Vol. 52/4, pp. 787–801, https://doi.org/10.1111/j.1540-5907.2008.00343.x.
Schneckener, U. (2006), “Fragile statehood, armed non-state actors and security governance”, Private Actors and Security Governance, December, pp. 23–40.
Stewart, F. (2008), Horizontal Inequalities and Conflict: Understanding Group Violence in Multiethnic Societies, Palgrave Macmillan, Basingstoke, UK.
Taleb, N.N., and G.F. Treverton (2015), “The calm before the storm”, Foreign Affairs, February, Vol. 1–7, www.relooney.com/NS4053/00_NS4053_134.pdf.
Tilly, C. (1985), “War making and state making as organized crime”, P. Evans, D. Rueschemeyer and T. Skocpol (eds.), Bringing the State Back In, Cambridge University Press, Cambridge.
United Nations University (2014), World Risk Report 2014, Bonn.
van Dijk, J. (2007), “Mafia markers: Assessing organized crime and its impact upon societies”, Trends in Organized Crime, Vol. 10/4, pp. 39–56, https://doi.org/10.1007/s12117-007-9013-x.
Wolfson, M., Z. Madjd-Sadjada and P. James (2004), “Identifying national types: A cluster analysis of politics, economics, and conflict”, Journal of Peace Research, Vol. 41/5, pp. 607–623, https://doi.org/10.1177/0022343304045975.
World Bank (2015), Country Policy and Institutional Assessment, World Bank, Washington, DC, http://data.worldbank.org/data-catalog/CPIA.
World Bank (2014), World Development Report 2014: Risk and Opportunity - Managing Risk for Development, World Bank, Washington, DC.
World Bank (2011), World Development Report 2011, World Bank, Washington, DC, https://doi.org/10.1596/978-0-8213-8439-8.
WEF (2016), The Global Risks Report 2016, World Economic Forum, Geneva, www3.weforum.org/docs/GRR/WEF_GRR16.pdf (accessed 4 September 2016).
Note
← 1. While there are methods of dealing with more complicated dynamics in composite indices, it has been noted that such approaches are seldom used (Grävingholt, Ziaja and Kreibaum, 2015).