5 quick insights into PA-X data

Dr Sanja Badanjak offers five quick insights from data on 30 years of peace processes from around the world, drawn from the PA-X Peace Agreements Database.

On July 1st 2020, PSRP published the fourth release of the PA-X Peace Agreements Database and Dataset. Since its very first iteration as the Transitional Justice Peace Agreement Database in 2015, the PA-X team have been continuously expanding the main database. Collecting and organizing peace agreements is now so routine for the PA-X team that we go about our daily work without even thinking about what this work does – it transforms peace agreement documents into data publicly available to a variety of audiences.

What does the PA-X do?

PA-X is a database and archive of all peace agreements signed since 1990, covering all stages of negotiations and all types of conflicts.

The data are available to search, so that anyone with an interest in what peace agreements say can visit the PA-X website, search by country, region, date, provisions, type of agreement, or words of interest. Every search result provides multiple ways to view, export, and interpret the data, including as a list of results, full document downloads, spreadsheet and csv format, and timeline.

With 30 years of data and over 1800 peace agreements, PA-X search tools and visualisations give a vital overview of the mass of data that the database contains. Not only do they make the data approachable, but they help to tell the story of some of the trends we see in peace agreements and peace processes. Here are some quick insights from our newest release.

What do the data show? Five quick insights.


1. Since 1990, there have been peace agreements signed on every continent, with the exception of Antarctica.

The map in Figure 1 shows the agreements aiming to resolve internal conflict in each country, since 1990. Some countries, shown in darker shades on the map, have seen more negotiated agreements of all types than others: for example, Colombia, Bosnia and Herzegovina, Sudan, South Sudan, Syria, and Philippines have had some of the most active peace processes. This should not be equated with more peace, as the number of agreements has more to do with the length and (in)tractability of conflict than anything else. Agreements and negotiations are likely to fail, and peace processes never proceed in a linear way towards conflict resolution. This visualization of peace processes shows how convoluted the processes can get.

Figure 1 (Click here for full size.)

2. There is no regularity in the number of agreements we see over time…

…with the exception of the relative decline of the interstate agreements that aim to resolve interstate conflict. As internal conflicts are the dominant type of conflict in the 21st century so far, so are agreements aiming to resolve internal conflict dominant in our data. Unlike internal, and often sub-national conflicts, interstate wars tend to be well covered in the media. Similarly, when agreements are reached in these conflicts, they tend to be publicized. Nevertheless, the PA-X team’s tracking of media and reports found very few instances of interstate agreements that aim to resolve interstate conflict in the past several years, as Figure 2 demonstrates.

Figure 2 (Click here for full size.)

3. Local peace agreements are growing in popularity.

One trend that is clear from the PA-X data is the growing number of agreements that can be characterized as ‘local’ agreements in the context of internal conflict (IntraLocal label in Figure 3). In the description of our PA-X Local sub-database, we define these as written agreements ‘that deal in some way with local issues, involve local actors, and deal with forms of local/communal violent conflict’. However, it is unclear whether this is an artifact of the data collection process or a newly developing feature of conflicts and negotiations. Improved availability of information and online resources makes it more likely that a locally relevant agreement will now be reported and published. Working on PA-X has revealed that one is more likely to find local agreements in historical cases that have been well-documented, such as those of Bosnia and Herzegovina and Croatia, where the PA-X team were able to access ICTY archives and numerous newspaper reports. Nevertheless, there is also increased international mediation efforts in local conflicts and areas of local fallout of broader conflicts, such as in cases where the HD Centre for Humanitarian dialogue has been active. This may mean that external and internal mediation efforts have been turning more towards local negotiations over the past years.

Figure 3 (Click here for full size.)

4. Regional agreements reflect long-term periods of political change.

Regional distribution of agreements over time results from long-term processes of political change, specific to regions. The trends, if any, are endogenous to the regions. For example, agreements from Europe are present in great numbers in the early 1990s and taper off almost completely afterwards. The initial surge in the 1990s stemmed from the collapse of the USSR and Yugoslavia, with conflicts (and agreements) persisting where there was no initial resolution of conflicts or where borders were later contested. In the 2000s, most internal conflicts are fought in Africa and Asia, hence they comprise most of the PA-X collection in that period. Figure 4 shows this regional distribution of PA-X agreements.

Figure 4 (Click here for full size.)

5. There is no recipe for a peace agreement.

There are some topics that are more likely than others: for instance, issues of security are tackled more often than regulation of the banking system is, which is only to be expected. However, there is no standard of what needs to be included nor the level of detail that needs to be addressed. Topics that are included and the level of detail are contingent on time and context and vary greatly. This variety is well-demonstrated by looking at how long agreements are (Figure 5). The statistics are stark: the median length is just two pages, meaning that half of the PA-X collection is less than two pages long. The length at the 75th percentile is just five pages. The mean length is 6.2 pages, driven by a small number of long agreements. There are only 35 agreements with more than 50 pages, and the longest one among them is the 2016 Final Agreement to End the Armed Conflict and Build a Stable and Lasting Peace in Colombia, at 323 pages. The shortest agreement in the PA-X collection is just 329 characters long, the Customs stamps (Technical Dialogue) Agreement from the Kosovo-Serbia normalisation process.

Figure 5 (Click here for full size.)

These quick findings are just scratching the surface of what the PA-X data have to offer. If there is one thread that connects these disparate bits of information on peace agreements and peace processes, it is one of restraint in interpretation. These data show only individual facets of multidimensional phenomena that appear in the context of peace negotiations and peace processes. Better insights require approaching the subject matter from multiple angles and multiple perspectives. Good resources for further exploration of the data include:

Explore PA-X now.

See instructions on how to cite PA-X data. You can also view the PA-X codebook, detailing all coded categories.