The PA-X Peace Agreement Database and Dataset has recently been updated. The sixth data release is now available to search and download, with 44 new agreements added, for a total of 1959 peace agreements between 1990 and mid-2022. The update also includes new additions to PA-X Gender and PA-X Local.
This blog by Dr Sanja Badanjak describes the trends emerging from the updated dataset for local negotiations, country- and conflict-level processes, and the stability of comprehensive agreements.
Changing characteristics of peace processes: New findings from the PA-X Database
Are comprehensive peace deals which result from formal, conflict-level peace processes a thing of the past? With so much of the literature on peace processes turning its gaze towards local peace negotiations practice, this is a reasonable question, and one to which the PA-X data can provide some answers. In this case, and unsurprisingly, the answers are both “yes” and “no.” Considering the most recent PA-X release, current up to June 2022 and covering all major armed conflicts and violent social crises, it seems more likely that we are seeing a change in the type of process that is prevalent, with more locally negotiated deals, and fewer and fewer agreements that are indicative of wide-ranging, formal, conflict-level, organised peace processes. While this suggests a change in how peace is negotiated and how peace processes are designed – it may be too soon to declare that the comprehensive deal itself is a thing of the past. When we consider only the conflict-level peace processes, the comprehensive agreement is still as present as it has been in the past and so the problem may in fact be that the conflict-level negotiations are simply not as common as they were in the 1990s. Additionally, preparatory, process-defining or pre-negotiation deals seem to be falling by the wayside, prompting questions about the ways in which peace processes are built and sustained. Judging from the data discussion below, thinking about this change in how peace processes are conducted may be a fruitful avenue of research in the coming years.
Exploring the data
Figures 1 and 2 provide the first step in this discussion, starting with the comparison of non-local and local peace deals. The first and clearest trend is that of the relative decline in the number of signed agreements that aim to resolve a larger conflict, be it regional, country-wide, or interstate (Figure 1). This is accompanied by a rise in the number of agreements that tackle a local offshoot of a wider conflict, or local consequences of a wider conflict (Figure 2). On their own, these two trends need not imply that peace processes have changed. The lower yearly counts of non-local agreements (Figure 1) may well be explained by the changing nature of conflict: for example, if the prevalence of localised armed conflict has increased over time, it is also likely that the peace processes would begin to match that change. Additionally, neither Figure 1 nor Figure 2 can be interpreted as supporting a claim that agreements that deal with substance of the conflict, and comprehensive deals in particular, are on the wane. These agreements are not common, but they are still present. To understand the drop in the number of non-local agreements, we should have a better look at this set of agreements on their own terms, separate from the rising number of local agreements in the PA-X data.
Figure 1: Non-local agreements, by year and stage. The bars that represent the yearly count of agreements are split by agreement stage: ceasefires, implementation agreements, other agreements, pre-negotiation and process agreements, renewals, and finally, at the bottom of each bar, substantive comprehensive agreements, and substantive partial agreements.
Figure 2: Local agreements, by year and stage. The bars that represent the yearly count of agreements are split by agreement stage: ceasefires, implementation agreements, other agreements, pre-negotiation and process agreements, renewals, and finally, at the bottom of each bar, substantive comprehensive agreements, and substantive partial agreements.
Thus, looking at the non-local agreements only, the comprehensive deal is indeed shown to be a rare event, with at most 8 recorded in a single year (Figure 4), and this has been the case for the entirety of the period covered by the PA-X data. These agreements tend to comprise less than a tenth of the total number of non-local formal agreements (Figure 3), and this is not unusual. These are complex, long agreements that often attempt to remake the entirety of a country’s political system, and many peace processes never result in an agreement of this type. Whether negotiations reach such a high threshold may be a poor indicator for the state and characteristics of negotiations even in the most conducive circumstances. However, and somewhat surprisingly, even with the high year-on-year variation in the number of comprehensive agreements, their prevalence, expressed as proportion of all agreements, is quite stable over time (Figure 4).
Figure 3: Proportion of comprehensive agreements in the total count of non-local agreements, by year, PA-X data, dotted line indicates the mean proportion for the 1990-2022 period.
Figure 4: Number of comprehensive non-local agreements, by year, PA-X data, dotted line indicates the mean number of such agreements per year, for the entire 1990-2022 period.
If the comprehensive deals are relatively stable, where is the drop in non-local agreements coming from? Figures 1 and 5 suggest that much of the drop may be in the pre-negotiation and process-defining agreements. Indeed, the proportion of such agreements in the total number of agreements (Figure 5) may be better at helping us understand how contemporary peace processes develop and change. The lowest yearly figures for this indicator are all found in the past ten years, and while pre-negotiation/process deals were making more than a third of all agreements in the 1990s, they are mostly falling below that threshold in the 21st century. The horizontal line in Figure 5 is the mean yearly proportion of pre-negotiation/process agreements, and the arrangement of yearly values around it is quite clear, with 2003 appearing as the threshold year, when pre-negotiation/process agreements began to make up a smaller portion of the total number of agreements.
Figure 5: Proportion of pre-negotiation agreements in the total count of non-local agreements, by year, PA-X data, dotted line indicates the mean proportion for the 1990-2022 period.
While this is one of many over-time trends that can be gleaned from the PA-X data, it may be the one that helps us understand how the more recent peace processes and negotiations differ from those we have seen 15, 20, or 25 years ago. In PA-X, this broad category covers a variety of agreements that all aim to set up the key parameters of the negotiations themselves. While many of these agreements are indeed focused on the pre-negotiation period, they can come at any time in a peace process, whenever the sides are trying to better define the process through which they plan to reach some form of substantive agreement. As agreements of this type drop, both in numbers and in proportion of all agreements, we should begin to focus more on the characteristics of peace processes rather than on the prospect of a major comprehensive deal among the warring sides and ask what drives this drop, and what its consequences are. Additionally, it is possible that the increase in local negotiation practices is related to this decline in national processes, be it from the frustration of the citizens and their organizations, or from the change in how third parties and mediators are perceiving the entry points for the path to wider settlement. Finally, the global view presented here is likely to be obscuring differences across regions, countries, and conflicts – any variation in how these trends present in particular settings would be important to uncover.
Considerations and further analysis
Some caveats should be noted and added to the list of questions on this research agenda. First, even in a collection with firm rules on inclusion such as PA-X, comparison over time may not be a perfect like-for-like. On the one hand, publishing and publicizing agreements has changed over time. Whereas agreements may have required time to become public, via publication in the official journals, reprints in the media or in academic books, we now see them published almost immediately, in online media, and via social media. Additionally, media, official journals, and academic publications would have been more likely to suffer from geographic bias, certainly prior to ~2010: as most media, policy work, and academic research are more likely to take place in cities, documents from rural and more distant areas might have been less likely to be published and publicized. This means that some agreements that might have been considered of local interest only would have only rarely become available for publication in a collection like PA-X in the 1990s but are easily available to us in the 2020s. We know that such smaller-scale negotiations and agreements are present in earlier conflicts (see, for instance, cases from the Bosnian conflict) but the extent to which the rise in their number is attributable to different publication options and media coverage as opposed to a change in the practice of peace-making is a matter that requires more research. This is relevant for the estimate of the local agreement numbers, though it should not affect any of the conclusions regarding the trends in non-local agreements.
The second caveat related to data collection relates to the inevitable delay between agreements being signed and their publication in the PA-X database. Some are only made public well after signing, some are not easily available online, and others need to go through a process of vetting and translation. The pace of additions to PA-X shows the existence of this type of time lag. While the PA-X data are current up to mid-2022, the newly added documents are as old as 2008. This potential source of bias may affect pre-negotiation and process agreements more than others, as they may be considered less salient than the substantive deals, and therefore less likely to see immediate publication.
Between these caveats and the quandaries posed by the data described above, the research questions on peace agreements and peace processes do not appear to be in short supply. The PA-X data point to simultaneous trends of thinning out of the peace processes at country- or conflict-level and increased reliance on local negotiations – but with relative stability in the number of comprehensive agreements, even when considering non-local agreements only. Understanding these awkwardly fitting trends will require a more comprehensive analysis of three layers of interconnected data: on armed conflict, peace negotiations, and the regional and international context in which conflict and negotiations are embedded.
Dr Sanja Badanjak is a Chancellor’s Fellow in Global Challenges at the University of Edinburgh School of Law, and is PeaceRep’s Data Manager for the PA-X Peace Agreements Database and Dataset.