Public Release: 

A way of predicting civil war

New Scientist

THE dictator General Pervez Musharraf leads a country on the brink. As the war in Afghanistan intensifies, he is torn between appeasing the US and pacifying Pakistani rioters baying for his resignation. The danger his country faces is obvious, but researchers in the US have devised a system they claim could predict much earlier any countries approaching civil war.

Their "conflict barometer" gives a week-by-week measure of the scale of unrest. Although the barometer's forecasting prowess remains to be proved, its developers say it could have presaged the slide of Algeria and Sri Lanka into civil war. They also believe it could help the US and Britain decide how long to fight Operation Enduring Freedom.

Raw material for the barometer arrives daily in the form of several thousand Reuters news stories. A sentence-analysing program called a parser classifies events into roughly 200 categories. This task had to be automated, says the program's developer, Doug Bond of Harvard University, since humans cannot keep pace with reports.

From the category counts, Bond and his colleague Craig Jenkins, a political sociologist at Ohio State University, calculate the proportions of events involving civil protests, repressive government actions and outbreaks of violence.

They feed these three factors into an equation to give a nation's "conflict carrying capacity" or CCC. A score of 100 signifies stability, zero equals chaos. Before 11 September, the US typically scored around 98, Britain in the 90s, and countries such as Sri Lanka, 60 or below.

In a recent paper in The Journal of Conflict Resolution, Jenkins and Bond traced a decade of CCC scores for seven countries. They found that the scale provided warnings six to nine months ahead of civil war in Algeria and Sri Lanka, and Peru's march back to relative stability. "A close observer might have done as well, but I think it helps to have the indicator out in front of you," says Jenkins.

Independently, John Freeman at the University of Minnesota has shown that the CCC is also a strong predictor of currency collapse. The Swiss Peace Foundation, for example, uses CCC scores to help identify countries whose currencies are at risk.

Jenkins and Bond have found that nations fail when violence, civil strife and government repression mount and stay high for many months-keeping the CCC below 85 or so. "It looks like the gestation period of a crisis is six to nine months," says Jenkins. That makes him think that both the intensity and duration of fighting in Afghanistan must be weighed in assessing the risk of destabilising neighbouring states.

Gary King, professor of government at Harvard, points out that even though only two or three nations implode each year, the consequences for the rest of the world are momentous. "The only sponsors of international terrorism are failed states." He values the CCC's ability to monitor countries in real time. "This is a tremendous help," he says. "It provides knowledge we couldn't get otherwise."

While Jenkins emphasises that the CCC's forecasting ability needs to be evaluated and honed through further studies, he thinks it could already help governments spot crumbling regimes. He is particularly worried about Pakistan and Tajikistan, which, according to the index, are teetering on the edge of instability.


Author: Robert Adler More at: The Journal of Conflict Resolution (vol 45, p 3)

New Scientist issue: 27 October 2001


Disclaimer: AAAS and EurekAlert! are not responsible for the accuracy of news releases posted to EurekAlert! by contributing institutions or for the use of any information through the EurekAlert system.