Conflict forecasting and its limits

Abstract

Research on international conflict has mostly focused on explaining events such as the onset or termination of wars, rather than on trying to predict them. Recently, however, forecasts of political phenomena have received growing attention. Predictions of violent events, in particular, have been increasingly accurate using various methods ranging from expert knowledge to quantitative methods and formal modeling. Yet, we know little about the limits of these approaches, even though information about these limits has critical implications for both future research and policy-making. In particular, are our predictive inaccuracies due to limitations of our models, data, or assumptions, in which case improvements should occur incrementally. Or are there aspects of conflicts that will always remain fundamentally unpredictable? After reviewing some of the current approaches to forecasting conflict, I suggest avenues of research that could disentangle the causes of our current predictive failures.

Supporting rebels and hosting refugees: Explaining the variation in refugee flows in civil conflicts

“…Finally, in addition to the in-sample analysis, I also perform out-of-sample cross-validation procedures, which can be an effective indicator of the model's success (Chadefaux, 2014(Chadefaux, , 2017bWard, Greenhill & Bakke, 2010). This procedure reinforces the causal claim and helps to overcome the overfitting problem (Beck, King & Zeng, 2000;Chadefaux, 2017a). Results support my argument, and including rebel support in the model furthers our understanding of why some countries host more refugees than others.…”

Risk Forecasting in the Light of Big Data

“…Essentially, conflict forecasting approaches have assumed three different forms: (1) individual experts who summarize available information and form a judgement, (2) collections of such experts who are brought together to form a consensus view, and (3) data-driven computer models that use patterns of past actions to predict future physical behavior. Recent progress in computational methods, and in particular in text analysis, has enabled automation of the data-drive approach significantly, thus allowing far more comprehensive sources of real-time information to be analyzed, thereby facilitating the transition from structural to short-term registration of tensions and other conflict characteristics [47].…”

“…The constantly rising availability of these types of documents offers both opportunities and challenges for risk forecasting of political instability [4,47,50]. While these changes in the global news media industry, as well as worldwide internet accessibility, provide significant potential for novel applications, the inherent process of generating event data, which is not static but rather very dynamic, also creates a number of technical challenges and requires frequent, if not continuous, validation activities from users.…”

“…In addition to news, information from social media is also attracting increasing attention, the advantage of which is both the considerable amount of data and the very nature of turning every user into a potential reporter, thus potentially extending the range of events covered considerably and facilitating real-time coverage of a wide range of events [47,51]. However, their use poses considerable challenges, as social media diffuses information and mobilizes people, and most content, such as Twitter content, is either not relevant for the purposes of conflict prediction or has been deliberately implanted with false information for manipulation [47,52].…”

Forecasting Social Conflicts in Africa Using an Epidemic Type Aftershock Sequence Model

“…The analysis and prediction of social conflicts is an active and important area of research. Yet, the current state of affairs is that it is not easy to find a forecasting methodology that can deliver accurate predictions, see for example [1][2][3][4].…”