Conclusions
We identified six measures of vulnerability to conflict (self-labeled and targeted in different ways) for detailed examination. These come from the Brookings Institution (Index of State Weakness in the Developing World), Carleton University (the Country Indicators for Foreign Policy Fragile States Index), the Fund for Peace (Failed States Index), the Center for Global Policy at George Mason University (the Goldstone and Marshall State Fragility Index), the Center for International Development and Conflict Management at the University of Maryland (the Conflict Instability Ledger measure), and the Economist Intelligence Unit (the Political Instability Index). As background for the analysis reported here, we also explored several other measures we found to be narrower in focus-not as sharply attentive to the concept of conflict vulnerability, but sometimes looking at governance, repression, violence, or other related or partial measures.57
To overcome the proliferation of terminology the measures use, but-more important to help us understand the underlying concept of vulnerability to conflict and how assorted concepts might relate to it, we mapped out a spectrum of vulnerability. We identified failing or failed states (three in particular), fragile ones (forty-eight in our analysis), those with some vulnerability (another seventy-two), and the residual set of those consistently and widely stable. We used an average analysis of consistently transformed and compared indices to rank all states and specify the spectrum. We also relied on a widely used division of drivers of vulnerability into four categories: social, economic, political, and security.
With such foundational tools, we used four approaches in the effort to understand the similarities and differences of the six indices. First, we looked at them qualitatively and on their own terms. his involved considering their self-definitions and the variables they include, as well as looking at some of the other key methodological approaches and despite being hindered by a considerable lack of transparency in how some indices were prepared. This approach to comparison clearly identified the CIL measure as different from others in terms of its close connection to the model and variables of the Political Instability Task Force, its therefore parsimonious use of input variables, and its leaning towards prediction of failure by a few states
Figure 10. Polity Measure, Executive Constraint, 1946-2008
as opposed to assessment of relative fragility across many. We did, however, conclude that this distinction was perhaps not very strong across projects. Similarly, the EIU measure also began to stand out because that project, too, indicates more relative attention to prediction of conflict, but even more clearly because of its somewhat less typical use of variables across the four categories, including more attention to fractionalization and governance variables; its use of other indices and expert judgment analyses in contrast to off-the-shelf data; and its stated connection to the PITF state failure model-despite its obviously more hybrid character.
With approach 2, we compared all measures across all countries they assess. Using correlations of each with all others, as well as the transformation of indices to standardized derivative forms and analysis of each against each other and relative to the average index values,we found that, in fact, the Carleton, George Mason, Fund for Peace, and Brookings measures were most closely related to each other and the average, more or less in that sequence (that is, Carleton and George Mason most characterizing the central tendency), and we characterized these four as most sharply focusing on understanding variations within the fragile-state set. Yet we also found that the Brookings index, to some degree like that of Maryland, more sharply distinguishes a small set of most fragile or failing states from others. he EIU measure, like that of Maryland, sets off a number of countries from the average pattern, but whereas Maryland tends to do so by emphasizing the most fragile or failing, the EIU measure does so for the less fragile, which began to suggest that it is tapping something quite different-something perhaps closer to political risk.
Approach 3 took us to a consideration of how the measures differ with respect to specific countries. Our hope was that we could identify some groupings of countries that specific measures were most likely to assess as being more or less vulnerable than other indices did, but such identification proved at best difficult. here was no difficulty in once again seeing the degree