Report: Understanding and Forecasting Political Instability and Genocide for Early Warning
Charles R. Butcher, Benjamin E. Goldsmith, Dimitri Semenovich, Arcot Sowmya
Australian Government – AusAID, Asia-Pacific Centre for the Responsibility to Protect, The University of Sydney, The University of New South Wales
September 2012
Genocide is not an inevitable feature of the modern world. Nor, when the killing has started, is the process inexorable. Genocides can be prevented, or, at least, stopped when they begin. The fact that genocide continues to occur, and continues to attract a range of international responses from the ignorant to the anemic and, occasionally, the forceful, no doubt reflects the reluctance of concerned major powers to become militarily involved in foreign conflicts where traditional national interests are not at stake. Information, however, also has a role to play. Accurate and reliable forecasts of genocide can act as a ‘force multiplier’ by increasing the efficacy of prevention and intervention strategies, and, where these fail, improving the chances of successful prosecution to deter other leaders from committing these crimes in the future. And while no forecasting model can be a substitute for political will, adequate forewarning and monitoring should alleviate some of the uncertainty associated with deployments in foreign lands and close the window for states to obfuscate and avoid real opportunities (some would say obligations) to prevent genocide.
In this report, we discuss the design, results, and usefulness of a quantitative model to forecast genocide. Adopting a widely used definition, we take genocide to mean the ‘promotion, execution, and/or implied consent of sustained policies by governing elites or their agents – or in the case of civil war, either of the contending authorities – that result in the deaths of a substantial portion of a communal group or politicized non-communal group’.1 This definition includes the targeting of groups because of their ethnic or communal identity as well as the targeting of groups based upon political beliefs, or ‘politicide’. Further discussion of this and related definitions, and of issues surrounding the use of the terms ‘genocide’ and ‘politicide’, can be found in the Appendix to this report. For brevity, we use the term ‘genocide’ to refer to events of both genocide and politicide in the remainder of this report. Such events include the 20th century’s most notorious cases of mass-killing – for example Pol Pot’s murderous reign in Cambodia during the mid-1970s and the killing of at least half a million Tutsi and moderate Hutu in Rwanda in 19942 – along with some less well known cases – such as Burundi from 1965 to 1973 and Sri Lanka from 1989 to1990. Our model identifies the Central African Republic (CAR), the Democratic Republic of the Congo (DRC), and Chad as the top three states at risk of the onset of a new genocide in the period to 2015.
We begin this report by detailing how an ‘early warning’ system capable of identifying countries at the highest risk of genocide might enhance prevention, intervention and prosecution efforts. This is followed by a brief overview of global and regional trends in the occurrence of genocide  after World War II. We then discuss the design of our forecasting model and how the forecasts for 2011-2015 should be interpreted. The register of fifteen ‘at-risk’ states for the years 2011-2015 is then presented. Predictors that place these states at high risk, along with some historical and contemporary examples, are then discussed. We conclude by reflecting upon some future directions for forecasting events of massive human rights violations. In addition to discussing definitional issues, the appendix provides a list of the predictors used in the models and data sources. (…)
Read the full report.