Forecasting in International Relations

by

Nils W. Metternich, Kristian Skrede Gleditsch, Christoph Dworschak

• LAST REVIEWED: 07 December 2020
• LAST MODIFIED: 12 January 2021
• DOI: 10.1093/obo/9780199743292-0179

Introduction

Forecasting has always been a central aspiration in the study of international relations. Early efforts to develop a “science of politics” typically emphasized the ability to anticipate future events as a key advantage. The behavioral revolution likewise gave strong prominence to the positivist equivalence between explaining phenomena and the ability to predict. However, despite many programmatic statements, predictions have largely remained aspirations until relatively recently. In this bibliography, we will provide an overview of the existing literature on forecasting, and on applications to conflict in particular. This overview will be organized around differences in specific approaches to forecasting, largely categorized by methodology. However, forecasts can differ along a number of dimensions that cut across specific methods. First, there are differences in the scope of forecasts. Some predictions are highly case-specific and predict outcomes for a single unit and particular issue area. Other models have global scope and seek to assess the risks of some general phenomenon across a wide range of units. Between these two extremes there are various more delineations, focusing on a confined sets such as regions or specific groups of actors. Second, forecasts differ considerably in their time horizons, ranging from daily to periods as long as 100 years. Third, forecasts differ in the scope of what analysts try to predict. Many forecasts are limited to predicting the value of the main response variable of interest, based entirely on what is known about predictors ex ante. Other forecasts entail more ambitious efforts to predict larger systems of variables, including the relevant predictors. Finally, one can distinguish between model- and data-driven approaches to forecasting. Purely model-driven forecasts are based on simplified theoretical or mathematical models and a priori assumptions, sometimes devoid of any empirical inputs at all. At the other extreme, there are purely data-driven models, possibly without any assumptions about the structure or functional form of the relationships at work. In between these two extremes, there is a wide range of forecasting approaches, combining a priori modeling assumptions with parameters that are calibrated by empirical data. This article concludes with an overview of the interests in forecasts among policy and applied communities, including some of the barriers that have prevented effective communication of research results, pointing to future directions of forecasting conflict. Nils W. Metternich acknowledges support from the Economic and Social Research Council (ES/L011506/1). Kristian Skrede Gleditsch acknowledges support from the European Research Council (313373).

General Overviews

Forecasting in international relations has received new attention because of advances in statistical modeling, data collection, and computational performance. In addition, government funding (see O’Brien 2010) has fostered a huge number of new projects. Schneider, et al. 2011 provides an overview of these efforts in the introduction of an issue that also features a helpful categorization of approaches (see Brandt, et al. 2011). Schrodt, et al. 2013 provides a review of advances in the field. A 2017 Journal of Peace Research special issue on forecasting in peace research collects articles that apply and discuss state-of-the-art approaches to conflict forecasting (see Hegre, et al. 2017, cited under Statistical and Stochastic Models). D’Orazio 2020 and Bara 2020 offer the latest summaries of advances in the field of conflict prediction specifically. The edited volume Choucri and Robinson 1978 shows that interest in forecasting has a long record in the study of international relations, and Feder 1995 highlights an important interest in forecasting applications in the intelligence community.

• Bara, Corinne. “Forecasting Civil War and Political Violence.” In The Politics and Science of Prevision: Governing and Probing the Future. Edited by Andreas Wenger, Uursula Jasper, and Myriam D. Cavelty. London: Routledge, 2020.

  This chapter aims to introduce readers in an accessible and nontechnical manner to the forecasting framework, with specific examples from conflict research.

• Brandt, Patrick T., John R. Freeman, and Philip A. Schrodt. “Real-Time, Time-Series Forecasting of Political Conflict.” Conflict Management and Peace Science 28.1 (2011): 41–64.

  DOI: 10.1177/0738894210388125

  Part of an issue on forecasting in international relations (see also Schneider, et al. 2011), the authors propose a particular forecasting framework as well as a good overview of existing prediction approaches.

• Cederman, Lars-Erik, and Nils B. Weidmann. “Predicting Armed Conflict: Time to Adjust Our Expectations?” Science 355.6324 (2017): 474–476.

  This article summarizes the promises and challenges of predicting political violence, reviewing contemporary contributions to the field.

• Choucri, Nazli, and Thomas W. Robinson. Forecasting in International Relations: Theory, Methods, Problems, Prospects. San Francisco: W. H. Freeman, 1978.

  An edited volume that collects the state-of-the-art forecasting approaches to international conflict prediction at the time of publication, and also alludes to the limitations of forecasting.

• D’Orazio, Vito. “Conflict Forecasting and Prediction.” In Oxford Research Encyclopedias: International Studies. New York: Oxford University Press, 2020.

  This piece offers an overview of the components of empirical models for conflict prediction, and reviews studies applying these methods across the different areas in peace and conflict research.

• Feder, Stanley. “Factions and POLICON: New Ways to Analyze Politics.” In Inside CIA’s Private World: Declassified Articles from the Agency’s Internal Journal, 1955–1992. Edited by H. Bradford Westerfield, 274–292. New Haven, CT: Yale University Press, 1995.

  This chapter reviews the use of prediction models in the US intelligence community, based on declassified information. Some of the models described are based on Bueno de Mesquita’s work.

• O’Brien, Sean P. “Crisis Early Warning and Decision Support: Contemporary Approaches and Thoughts on Future Research.” International Studies Review 12.1 (2010): 87–104.

  DOI: 10.1111/j.1468-2486.2009.00914.x

  This article focuses on introducing the DARPA-funded Integrated Crisis Early Warning System (ICEWS) program, but also covers a lot of issues that arise in current forecasting approaches.

• Schneider, Gerald, Nils Petter Gleditsch, and Sabine Carey. “Forecasting in International Relations: One Quest, Three Approaches.” Conflict Management and Peace Science 28.1 (2011): 5–14.

  DOI: 10.1177/0738894210388079

  This is an introduction to an issue on forecasting in international relations, but at the same time it highlights current and competing approaches to prediction.

• Schrodt, Philip A., James Yonamine, and Benjamin E. Bagozzi. “Data-Based Computational Approaches to Forecasting Political Violence.” In Handbook of Computational Approaches to Counterterrorism. Edited by V. S. Subrahmanian, 129–162. New York: Springer, 2013.

  DOI: 10.1007/978-1-4614-5311-6_7

  Even though this is published in a terrorism handbook, it focuses on conflict studies more generally.

• Ward, Michael D., Brian D. Greenhill, and Kristin M. Bakke. “The Perils of Policy by P-Value: Predicting Civil Conflicts.” Journal of Peace Research 47.4 (2010): 363–375.

  DOI: 10.1177/0022343309356491

  This article highlights the benefits of out-of-sample predictions, rather than judging statistical findings only on the significance of individual variables. It provides a template for current forecasting approaches.