Variants of violence: How classifying conflicts helps us solve them

Editor's note: This article is part of series covering CEPR's Reducing Conflict and Improving Performance in the Economy (ReCIPE) programme. Dominic Rohner is the Research Director, Oliver Vanden Eynde is the Head of Engagement, and Emma Verhille is the Research Officer at ReCIPE.

Scholars studying war face a core dilemma: to explore broadly or delve deeply. Carrying out big picture studies covering every country in the world allows us to understand general patterns, but this comes at the cost of missing nuance, institutional details, and major trade-offs. We are, for example, able to understand the likely overall (‘on average’) implications of climate change (Hsiang et al. 2013) or mineral price shocks (Berman et al. 2017). This helps build a basic understanding of the problem but often obscures important nuances. It is, for example, striking that climate change has much more severe effects in the face of settler-herder competition (Eberle et al. 2025, McGuirk and Nunn 2025, Bloem et al. 2025) and the impact of natural resource exploitation strongly depends on institutional factors (Mehlum et al. 2006, Fetzer and Kyburz 2024).

Why study conflict at the country level?

Though a big-picture overview provides a valuable starting point, micro-level evidence on specific countries is ultimately necessary. Having a controlled institutional setting with less unobserved variation helps us to cleanly identify causal effects, and more easily understand major mechanisms and channels of transmission. Still, the challenge of generalisability remains: does a given finding for, say, Somalia also apply to Kenya? In some cases, it may; in others, it may not.

Carrying out a given study design for 200 countries separately is impracticable for several reasons: 

1. The resources and workload required would be overwhelming.
2. Often countries are selected depending on the availability of sources of ‘exogenous’ variation. Such sources of variation, allowing for, say, an instrumental variable approach, may be available for Somalia but not Kenya.
3. The economics profession does not give enough credit for replication studies in new settings. Put differently, a tenure-hungry scholar has very low incentives to re-do a study that has already been conducted in 20 countries, as this will typically result in lower academic impact.

One way to meaningfully boost the external validity of within-country studies is to think about comparable settings. In the past, there have been numerous attempts at classifying country-conflict profiles (Kalyvas and Balcells 2010), but these are often done manually, which can make such classifications vulnerable to cognitive biases. In Rohner, Vanden Eynde and Verhille (2025), we develop a data-driven approach for classifying types of conflict across countries. In order to be useful, a classification cannot have too few categories, otherwise it lacks nuance; for example, if we were to just distinguish between poor versus rich countries, too much information would be lost. On the other hand, if we have too many categories, we are unable to synthesise; for instance, if every country was in a separate category, the purpose of classification would be obviously defeated.

Classifying types of conflict across countries

In our particular case, we classify countries binarily along three dimensions, resulting in eight total categories—striking a balance in the trade-off between nuance and traceability. While numerous distinct dimensions could be considered, we focus on three that have garnered particular attention in the existing evidence base (Rohner 2024a, b).

The first key dimension is economic, which captures the state of the economy in line with the theoretical underpinning that productive legal activities increase the opportunity cost of appropriation and fighting. There is solid empirical evidence showing that negative income shocks tend to fuel fighting (Miguel et al. 2003, Vanden Eynde 2018). We classify countries into two income categories: poor (low and lower-middle income countries) and rich (upper-middle and high-income countries), using the World Bank country income classification, with thresholds defined by GNI per capita in US$.

The second dimension we consider is political/institutional. More inclusive politics, such as those in democratic states, allow individuals to express their views more peacefully. The state is typically more accountable, lowering grievances, commitment problems, and asymmetric information frictions. Recent empirical evidence shows the scope for inclusive politics to reduce conflict (Cederman et al. 2010, Marcucci et al. 2023). In our research, we classify countries into two categories, democratic and non-democratic, based on contestation and participation criteria (Boix et al. 2013), meaning that political leaders are chosen through free and fair elections and according to a threshold value of suffrage.

Finally, the third dimension is security/state capacity. Greater administrative and military capacity makes a state a less likely target for would-be warlords. Higher fiscal and legal capacity leads to a series of virtuous cycles, boosting not only the economy but also prospects for peace (Besley and Persson 2011). The role of security guarantees for fruitful public policies has also been demonstrated empirically (Berman et al. 2011, 2013). In our research, we operationalise this by classifying countries into two categories again: secure and non-secure, based on countries’ security apparatus score in the Fragile States Index (The Fund for Peace 2023). Our security indicator encompasses military capacities, police practices, perceived citizen trust in security institutions, organised crimes and homicides, security threats to the state (such as bombings, rebel movements, terrorism), or irregular security forces that serve the interests of political leaders.

Below we display an overview figure depicting the eight categories along the three dimensions of income, democracy, and security at baseline (Rohner et al. 2025). We see stark differences, with richer, more democratic states experiencing less conflict. The dominant conflict types also vary: warlordism and rebellion are more predominant in weaker, poorer states; state repression is more frequent in non-democracies; and rich countries suffer mostly from terrorism. We also show in detail which countries belong to which categories, highlighting the potential for external validity of country-level results.

Figure 1: Division by type of violence for different categories of countries between 2007 and 2022

Source: Rohner et al. (2025). Notes: The colour-coding represents the overall conflict likelihood and pie charts represent the dominant conflict types. How to read the numbers: 70.59% of the 34 poor non-democratic and non-secure countries at war experience majority state-based events. The darker the red, the more violence a type of country experiences. Poor includes low and lower-middle income countries, rich includes upper-middle and high-income countries, based on the World Bank country income classification (thresholds defined according to GNI per capita in US$ (Atlas method). Non-secure countries are defined using the Security Apparatus component from the Fragility States Index; countries with a score exceeding 6 are classified as non-secure. Democratic countries are determined according to contestation and participation criteria, by the Boix-Rosato Dichotomous Coding of Democracy.

Implications for studying conflict and peace

In terms of main takeaways, it is crucial to think about the external validity of any empirical findings. One can think of policy interventions as helping countries to move in specific dimensions towards categories that are more peaceful. However, there are important complementarities between dimensions (Rohner and Thoenig 2021). For example, in poor, undemocratic, and unsecure countries, strengthening security capacity alone often fails to reduce conflict, as it merely shifts violence from warlordism and rebellion to state repression. However, if the country was to simultaneously state capacity and inclusiveness, the benefits are disproportionally greater. This highlights the need for a systematic approach to designing policy interventions that reduce conflict risk. Successful recipes for peace ought to tackle several major challenges at once. Classifying country and conflict types boosts the external validity of specific studies, allowing us to better understand how particular policy dimensions interact. Beyond the micro- and macro-level, more work at the meso-level is necessary.

References

Berman, E, JN Shapiro, and JH Felter (2011), “Can hearts and minds be bought? The economics of counterinsurgency in Iraq,” Journal of Political Economy 119(4): 766–819.

Berman, E, JH Felter, JN Shapiro, and E Troland (2013), “Modest, secure, and informed: Successful development in conflict zones,” American Economic Review 103(3): 512–517.

Berman, N, M Couttenier, D Rohner, and M Thoenig (2017), “This mine is mine! How minerals fuel conflicts in Africa,” American Economic Review 107(6): 1564–1610.

Besley, T, and T Persson (2011), “Pillars of prosperity: The political economics of development clusters,” Princeton University Press.

Bloem, JR, A Damon, DC Francis, and H Mitchell (2025), “Herder-related violence, labor allocation, and the gendered response of agricultural households,” Journal of Development Economics 176.

Boix, C, M Miller, and S Rosato (2013), “A complete data set of political regimes, 1800–2007,” Comparative Political Studies 46(12): 1523–1554.

Cederman, L-E, A Wimmer, and B Min (2010), “Why do ethnic groups rebel? New data and analysis,” World Politics 62(1): 87–119.

Eberle, UJ, D Rohner, and M Thoenig (2025), “Heat and hate: Climate security and farmer-herder conflicts in Africa,” Review of Economics and Statistics: 1–47.

Fetzer, T, and S Kyburz (2024), “Cohesive institutions and political violence,” Review of Economics and Statistics 106(1): 133–150.

Hsiang, SM, M Burke, and E Miguel (2013), “Quantifying the influence of climate on human conflict,” Science 341(6151): 1235367.

Kalyvas, SN, and L Balcells (2010), “International system and technologies of rebellion: How the end of the Cold War shaped internal conflict,” American Political Science Review 104(3): 415–429.

Marcucci, A, D Rohner, and A Saia (2023), “Ballot or bullet: The impact of the UK’s Representation of the People Act on peace and prosperity,” Economic Journal 133(652): 1510–1536.

McGuirk, EF, and N Nunn (2025), “Transhumant pastoralism, climate change, and conflict in Africa,” Review of Economic Studies 92(1): 404–441.

Mehlum, H, K Moene, and R Torvik (2006), “Institutions and the resource curse,” Economic Journal 116(508): 1–20.

Miguel, E, S Satyanath, and E Sergenti (2004), “Economic shocks and civil conflict: An instrumental variables approach,” Journal of Political Economy 112(4): 725–753.

Rohner, D, and M Thoenig (2021), “The elusive peace dividend of development policy: From war traps to macro complementarities,” Annual Review of Economics 13(1): 111–131.

Rohner, D, O Vanden Eynde, and E Verhille (2025), “Variants of violence: Classifying conflict types and policies for peace,” Economic Policy, eiaf009.

Rohner, D (2024a), “Mediation, military, and money: The promises and pitfalls of outside interventions to end armed conflicts,” Journal of Economic Literature 62(1): 155–195.

Rohner, D (2024b), “The peace formula,” Cambridge University Press.

The Fund for Peace (FFP) (2023), “Fragile states index.”

Vanden Eynde, O (2018), “Targets of violence: Evidence from India’s Naxalite conflict,” Economic Journal 128(609): 887–916.