To determine the intensity of a conflict, the GCRI uses a scale from 0-10 which allows a set of useful gradations.
Intensity level 0 is equivalent to a situation where no conflict is present in the country at all.
Existing models of conflict risk assessment not only differ in their definition of conflict, but also in their understanding of a conflict’s profile.
In short, they mostly put emphasis on only one type of conflict and do not address the potentially distinct causes that favour the outbreak of e.g. conflicts over national power as well as conflicts on a subnational level in one model.
The GCRI offers a comprehensive methodology that distinguishes three dimensions of conflict every country faces:
The risk of confrontation with other states, the risk of internal conflicts over government control, and the risk of internal conflict over issues apart from government power itself, such as secession or resources.
As a composite index, it provides policy makers with a sound basis for qualitative risk assessment, supports coordinated actions focused on anticipating, mitigating, and preparing for violent conflict, and promotes decision-making on preventive action.
To the interested public, it offers a transparent tool that goes beyond aggregated national risk assessments and allows for a quick and scientifically sound overview of countries at risk. It thereby helps to convey information about:
1. What are the underlying factors that may lead to conflict?
2. Which countries show the highest risk for violent internal or interstate conflict?
3. Which countries or regions are more prone to a certain kind of conflict?
4. How does a country’s risk for those conflicts change over time?
Building on the results of earlier research projects, such as the JRC’s INFORM index and the Drivers of Vulnerability Monitor (HCSS, 2012), JRC reviewed the outcome of more than 200 articles on quantitative conflict research to determine a set of variables that are highly correlated with the onset of interstate, national power, and subnational conflicts.
In addition, further indicators were included based on a series of interviews with country and conflict experts and practitioners at the EEAS. We identified a set of main concepts for every pillar that, in theory, are relevant to determine the risk stemming from a state’s political institutions and performance, the social fabric, its security situation, geographical or structural aspects, and its economic situation.
The structural conditions thereby increase the risk for violent conflict in every country, for example if the neighbour plunges into civil war or democratic institutions break down.
These structural conditions work in both directions, for example a longer period of domestic peace should decrease the risk for intrastate conflict.
The result are the five risk areas and the attached variables as shown in the graph below.
The risk modelling approach in the GCRI rests upon the combination of two statistical models.
The first is a logistic regression, the second a linear regression. Using historical data as input, these analyses produce weights (or coefficients) for each variable.
Weighting a country’s most recent data with these coefficients, a probability for the outbreak of war is produced by the logistic regression, and a probable conflict intensity by the linear regression.
Combining these two values, each country is assigned a final GCRI risk score.
This risk score depends on whether a country’s probability score exceeds a certain threshold or not.
If it does, then the country is assigned the value produced by applying the linear regression model; otherwise, it is assigned a value of zero.
This way, the final GCRI score combines both the probability of a conflict occurring, and the likely intensity of the conflict if it does occur.