Can Arms and Ammunition Flows Data Inform Conflict Early Warning and Early Response?

By Joshua Bata and Erica Mumford

Data & Policy Blog
Data & Policy Blog

--

Arms flows to the military junta in Myanmar enable documented attacks on civilians and atrocities against the Rohingya ethnic minority. Myanmar is not an isolated case in this regard. Conventional arms and ammunition continue to be the primary tool for violence in armed conflicts. Flows of conventional arms into unstable regions have thus been treated as important early warning signs in the build-up to armed conflict. If properly identified in a timely manner by conflict early warning and early response (EWER) systems, arms and ammunition flows can alert the international community to the risks of armed conflict escalation, outbreak, or relapse.

The topic of monitoring arms flows in EWER systems is long standing. However, informing modern EWER systems with a comprehensive scope of arms and ammunition flows data remains challenging in practice. This commentary represents an opening salvo for a platform that UNIDIR will provide for practitioners and partners from the arms control and EWER communities in 2024 to enable the cross-fertilization of ideas and ways forward in integrating arms and ammunition flows data into EWER systems.

Can arms and ammunition flows data help to predict conflict?

Monitoring uncontrolled arms and ammunition flows to identify conflict risks can be compared to the monitoring of fault lines to determine the probability of an earthquake. Instead of creating deadly tremors, weapons flows into a particular country or region can increase the risks associated with armed conflict onset, intensity, and duration. Just as the occurrence of an earthquake is difficult to predict, so is identifying if, or when arms and ammunition flows might fuel full scale armed conflict. Monitoring arms and ammunition flows, then, serves similar functions to seismographs that detect movement in the earth’s crust as early warning alerts.

Arms and ammunition flows as a variable to predict the type, likelihood, intensity, duration, or emergence of armed conflict is not a novel idea. The monitoring of licit and illicit international arms transfers has been used to identify trends and patterns, estimate causal effects, make inferences, test theories, or explain outcomes related to conflict. Calculations of ammunition needs and use during times of war are also notably different when compared to times of peace, and therefore ammunition acquisitions and flows could give a more accurate real-time picture of what is to come.

Trends and challenges in using arms flows data for EWER

Despite the existence of various sources of information and data on arms and ammunition flows (see figure 1), global EWER systems like The Early Warning Project, Violence Early Warning System, CrisisWatch, Conflict Forecast, and ACLED’s Conflict Alert System have not systematically included such information to inform their analysis or predictions. A selection of sources gathering and processing arms flows data, as identified through a wider mapping conducted by UNIDIR, is summarized in the table below.[i] The findings of UNIDIR’s mapping can help us to learn why this gap in conflict EWER systems exists, and what the international community can learn.

Figure 1: Indicative sources of arms and ammunition flows information[ii]

Accessibility

The use of arms flows data in EWER systems is influenced by the accessibility of datasets. Most of the mapped data repositories in figure 1 make data publicly available for licit transfers, whereas those providing more comprehensive illicit transfer data are available only for States or subscription based. The frequency of releasing data and the time lag between collection, reporting, and sharing data for these arms flow data sources varies, which also influences their utility for EWER systems. The majority of the analyzed data sources release their reports or data either annually or on an ad hoc basis.

ACLED and the UN Comtrade release data most frequently, but there are differences in the scope of their coverage. On one hand, UN Comtrade contains data on ammunition and small arms flows, but it does not contain information on major conventional arms in a form that can help with early warning. On the other hand, the ACLED database only captures a small part of arms flows, primarily those reported in the media. In terms of coverage, monitoring ammunition and its price are even more rare, as data on ammunition feature lower levels of transparency when compared to that for major arms flows, and even for flows of small arms.

The most reliable and comprehensive arms flows data heavily depends on information made available by States. Arms control and transparency instruments like the UN Register of Conventional Arms and the Arms Trade Treaty have helped to create reporting and transparency norms which facilitate information sharing. Yet, the frequency and format in which most States submit their reports are not readily operable for EWER systems. At the same time, some official government sources such as national reports to parliament or other national monitoring bodies contain more information and can be made available more frequently than those submitted under international instruments.

The information that States publish as part of their annual reporting process and ad hoc investigations by governments often contain more information than those submitted to the UN or other international and regional instruments. Systematizing data collection from these reports or investigations into databases that contains all types of flows and conventional weapons may be useful for EWER systems.

Open-source data gathering is another vital attribute for EWER systems, as transparent methods of collecting and sharing information are essential to distinguish them from intelligence-gathering. Open-source data sources include not only official national reports and documents, but also defence company announcements and industry publications as well as specialized arms flow-focused social media. However, the systematic and consistent use of advanced data collection techniques, including web scanning, automated or semi-automated event coding, and open-source intelligence (OSINT) tools in creating “big data” on weapons flows remain under-utilized.

Mainstream media under-reporting, mis- or dis-information on arms flows, and the lack of resources from academic or research institutions to tap into more advanced digital technologies and tools are hurdles to creating and maintaining a global arms flows database based on open-source platforms. Bellingcat and Calibre Obscura show the potential of crowdsourcing without big budgets. However, transforming these initiatives into a more global coverage and readily usable data format for EWER systems are also challenging, as it would require expert judgment and considerations on establishing a high evidentiary bar before publishing. Such a pursuit implies time-consuming and meticulous work to validate arms flows data and minimize the risks that could result from publishing mis- or dis-information.

Common tools for gathering arms flow information “in the field” include interviews with authorities or key informants, eyewitness reports, illicit market information, documentation of weapons (identification and tracing) and investigations, and public opinion surveys. Because these methods are labor-intensive, highly contextual, geographically constrained, and may be highly political, these tools are more readily applicable for regional, national, and local or community EWER systems, as opposed to global EWER systems.[iii]

There are critical trade-offs in using different information sources on arms and ammunition flows for EWER systems. Official State reports take time to be collected and published, but they usually represent verified data when taken at its face value. Company or defense industry reports are usually released on an ad-hoc basis, but these are useful for data triangulation. In contrast, information from social media may be easily available and quicker to gather, but it can require considerable verification efforts.

Scattered Databases

The integration of data covering both licit and illicit weapons flows can inform a comprehensive understanding of early warning risks throughout the full life-cycle of arms, from their production to end-use. Nonetheless, the current landscape of data repositories is fragmented in its coverage of licit and illicit flows, owing to challenges related to lack of specialized knowledge for identification and experience to make judgment calls on which reports are reliable versus those which are based on unfounded rumors, among others highlighted above.

There is currently no publicly available database which covers both licit and illicit global flows of all types of conventional weapons. The potential for a global ammunition flows database is also lacking in available data sources. A large portion of arms flows information and data repositories cover licit, or legal, weapons flows, while critical data sources on illicit flows remain limited or inaccessible for public use. Databases such as the United Nations Office on Drugs and Crime (UNODC) Monitoring Illicit Firearms Initiative or INTERPOL’s iArms include data on illicit arms and diversion, however the latter requires subscription or authorization for access. The Stockholm International Peace Research Institute (SIPRI) Arms Transfers Database remains the most utilized data source on contracts and deliveries of major conventional weapons, but it only includes information on some types of light weapons and missiles.

Data concerning fragile and conflict-affected countries

The level of geographic coverage of mapped information sources revealed a predominance of data at the country-level, focusing on international transfers between two States and international movements of weapons. Around half of the mapped data repositories aspired to have global coverage, with regional, domestic and contextual or targeted coverage being less common. Yet data availability on arms flows at the global level remains aggregated, lacking temporal and spatial granularity. Most available sources do not provide disaggregated data on end-users, reported diversion attempts, illicit flows, or seized arms; the type of data that would be most useful for EWER systems. Is an international database containing such data feasible?

Analysts have shown the possibility of collecting granular, open-source data that may be useful for EWER systems, as in the case before Russia invaded Ukraine. Months before the attack, analysts noted the build-up of arms along Ukraine’s border suggesting an imminent war. This represented a traditional early warning signal of an imminent conflict — looking at the movement of arms and troops, but with the aid of modern data collection tools. Even though some doubted the possibility of the invasion despite the clear presentation of the military build-up, it showed that OSINT tools, including Google maps and satellite imagery by the Conflict Intelligence Team and Middlebury Institute of International Studies, can be used to sound the alarm before an invasion. However, here too, lies a similar hurdle. Such information is easily accessible, but usually not verified. The challenge is to harness the benefits of arms and ammunition flow data collected from these tools for EWER systems, and at the same time minimize the risks that can undermine alerts and derail response due to questionable sources.

What can arms and ammunition flows data contribute to modern EWER systems?

Despite the challenges of collecting and analyzing arms and ammunition flows data to predict armed conflict outbreak and intensity, developments in conflict prediction suggest that such information — when combined with other conflict-relevant data — can nonetheless provide more robust early warning alerts. The Global Urban Analytical for Resilient Governance is an EWER project at the Alan Turing Institute which combines geography, politics, culture, economics, and arms flows data to predict peace or conflict. In overlaying routes between towns and cities with the various layers of data, the EWER project reported 94% accuracy in predicting conflict onset in a city or region, and 82% accuracy in predicting “peace breaking out” in a conflict zone 12 months in advance. Yet still, proponents of this project underscored issues in collecting weapons trade data due to the timing of data releases, and the resource and labor-intensive nature of the data collection.

Modern conflict EWER systems have significantly benefited from ever larger volumes and increasingly precise data. Sophisticated models for conflict monitoring, data fusion capabilities, and the use of artificial intelligence and emerging technologies have all contributed to “real time” data generation for early warning. Within this landscape, how can we better utilize the increasingly available, yet scattered arms flows data for armed conflict EWER systems in the 21st Century?

The current state of practice suggests further improvement can be made in the accurate analysis and quality of arms flows data, and how to make them useable and actionable in EWER. Collaboration among experts and enhancing the political will to monitor weapons flows will be critical to improve the accessibility, completeness, and granularity of arms flows data. Prospects for such an endeavor are supported by the fact that it will always be preferable to monitor arms flows and ring the bell for preventive action, then to wait and count fallen bodies.

Footnotes

[i] UNIDIR published a list of information sources on licit and illicit arms and ammunition transfers, including seizures and diversion for the Arms-Related Risk Analysis Toolkit in 2021. See the list in pp. 102–104, https://unidir.org/publication/arms-related-risk-analysis-toolkit. This commentary assessed 21 information sources where arms and ammunition flow data can be found.

[ii] The list in the figure is neither comprehensive nor an endorsement, but rather an illustrative compilation of active sources (as of 31 March 2023) of information on arms and ammunition flows. The figure shows that there is no one-stop-shop to comprehensively and consistently gathers all types of arms and ammunition flows.

[iii] Local early warning systems are also considered civilian-focused and civilian-led systems. They have also been referred to as third generation early warning systems characterized by their focus on micro-level conflict indicators, local ownership, and the ability to harness community-based methods. https://doi.org/10.5334/sta.857

**

About the Authors:

Joshua Angelo E. Bata is an Associate Researcher in the Conventional Arms and Ammunition Programme of the United Nations Institute for Disarmament Research (UNIDIR). His current research focuses on ammunition profiling that helps authorities reduce armed violence by surfacing the dynamics of the illicit supply of ammunition. Joshua has worked on non-state armed groups’ use of uncrewed systems such as drones and supported testing the Arms-Related Risk Analysis Toolkit under the Preventing Armed Conflict and Armed Violence workstream.

Erica Mumford is a Program Manager at the Konrad-Adenauer-Stiftung (KAS) New York Office, where she oversees dialogue programs at the United Nations. Previously, Erica was Associate Researcher in the Conventional Arms and Ammunition Programme of the United Nations Institute for Disarmament Research (UNIDIR) in Geneva, where she led the research project on integrating conventional arms control and conflict prevention.

***

Publisher’s Note:

Readers may be interested in contributing to the special collection that Data & Policy Editors are organising on Data for Peace: How Can Novel Data Sources and Technology Enhance Peace? For more information see the Call for Papers (deadline for submissions October 2, 2023).

****

This is the blog for Data & Policy (cambridge.org/dap), a peer-reviewed open access journal exploring the interface of data science and governance. Read on for five ways to contribute to Data & Policy.

--

--

Data & Policy Blog
Data & Policy Blog

Blog for Data & Policy, an open access journal at CUP (cambridge.org/dap). Eds: Zeynep Engin (Turing), Jon Crowcroft (Cambridge) and Stefaan Verhulst (GovLab)