Commodity Price Shocks and Civil War: A New Approach

Current literature theorizes a relationship between economic shocks and internal conflict, yet has produced contradictory findings. I propose a new theoretical and empirical approach for studying commodity price shocks and find that the effects of price shocks on civil war are heterogeneous and substantial.

Revise and resubmit, Economics Letters.

Download the latest version here.



Learning during Insurgency [with Andrew Shaver]

We introduce a comprehensive declassified dataset on insurgent violence in Afghanistan and challenge recent research claiming the escalation of modern insurgencies follows a predictable pattern. Prominent research claims the rate of conflict escalation and learning by insurgents adheres to a power law dynamic, a pattern observed in various unrelated physical and social contexts. Governments affected by insurgent violence might use these patterns to anticipate and potentially thwart some attacks. While this line of research is intriguing, we demonstrate these findings are an artifact of biased conflict microdata. We find instead that dynamic learning during insurgency, a concept quantified as the relative rate of learning, varies throughout the course of insurgency. Patterns of learning by insurgents and counterinsurgents in conflict environments are complex and reflect strategic imperatives. The only predictability we observe in substate conflict is that dynamic learning by insurgents never attains a steady state: at times, insurgents learn more quickly than their opponents; at others, they learn more slowly.

Download the latest version here.

  • Iraq Times Series Decomposition - IED clear rate
    Iraq Times Series Decomposition - IED clear rate
  • Afghanistan Times Series Decomposition - IED clear rate
    Afghanistan Times Series Decomposition - IED clear rate

Terrorism, Ideology and Target Selection

How do ideological motivations influence terrorists’ choice of targets? While the ideological determinants of target selection have been the topic of anecdotal conjecture, no rigorous empirical work has been done to answer this question. To address this gap in the literature, I present a theoretical and empirical model. Using novel data on Western European terrorism from 1965 to 2005 and a multinomial logistic extension of statistical backwards induction, I find that ideology is the only consistent predictor of target selection under strategic constraints. To my knowledge, this study presents the first rigorous evidence that ideology matters even within strategic environments. These results are particularly robust for nationalist-separatist and sectarian groups, even when excluding domestic terrorism in the United Kingdom and Spain. I also find strong evidence that government responses are affected by recent tactical success and institutional path dependence.

Please email for the latest draft.



Places to Hide: Terrain, Ethnicity, and Civil War [with David Carter and Andrew Shaver]

We argue that areas with both relatively rough terrain and significant variation in terrain ruggedness are ripe for the survival of minority groups, even in the face of state repression. Historically, rugged terrain has facilitated the formation of distinct ethnic identities through its influence on early human settlement patterns. What’s more, areas with significant variation in terrain ruggedness were ideal for the survival of ethnic groups facing external threats. The Quechua in Peru are an important example, as the Cuzco province in Peru contains highly rugged mountainous terrain, but also the fertile Sacred Valley, which has been the breadbasket of the region for centuries. The terrain of Cuzco allowed the Quechua to produce a food surplus while also serving as sanctuary from Spanish incursions. In the short run, many of the same geographic features that gave rise to early ethnic diversity allow targeted ethnic groups to survive and mobilize against the state. Simply put, bad geography has played a critical role in shaping ethnic diversity and given restive groups places to hide.

The survival of ethnic groups nestled in areas with highly rugged terrain has a previously missed effect on the risk of intrastate violence. We argue that there is an important missing connection between terrain ruggedness and ethnicity in the literature on political violence. Although the civil war literature has established that mountainous terrain is associated with civil war onset, it does not empirically say much about the linkages between rough terrain and civil war. We argue that rugged terrain has both direct and indirect effects on a state’s propensity for civil conflict. To understand how rough terrain influences civil war, it is important to also take into account its immediate, or direct, effect on the government’s ability to project power, as well as its historical, or indirect, effects on the emergence and persistence of potential challengers.


Combining new province-level data on terrain ruggedness, ethnic groups and civil war, we investigate these direct and indirect effects of terrain ruggedness on conflict propensity. We show that a significant proportion of the effects of geography are indirectly transmitted through several ethnic mechanisms. These findings suggest new avenues of research investigating how the complex relationship between ethnic and geographic factors influences the likelihood of intrastate violence.

Download the latest version here.




Market Distortion and Revolutionary Resolve [with Richard Jordan].

Rebel groups often generate income from participating in illegal activities. These organizations rely on criminal behavior to enable a revolutionary project of conquest. Yet such activity can confuse patrons and rivals, undermining support from sympathetic communities and muddying claims of self-determination. How then do rebels differentiate themselves from criminal cartels and signal revolutionary resolve?

To answer this question, we first develop a formal model. Criminal groups attempt to maximize uncertainty over their political ambitions, as this increases the likelihood that citizens and government leaders mistake them for revolutionaries and allow substantial policy concessions. By contrast, rebels seek to minimize uncertainty over their political aspirations so that citizens and leaders take those intentions seriously. We assign rebels and criminal organizations to a continuous type space describing their preferences over allocating resources to public and private goods.

We find that rebel groups whose types are otherwise similar to those of criminal organizations will forego potential economic gains from illegal activity to distinguish themselves from cartels. Consequently, most rebel groups tend to behave similarly, even when their underlying preferences are quite diverse. In turn, this concentration of rebel behavior increases the costs to criminal organizations of pretense, confronting them with a choice: they must either bear large costs to become indistinguishable from rebel groups, or they must forfeit any attempt at serious deception. This decision causes criminal groups to cluster as well, so that the observed behavior of rebel and criminal organizations follows a two-peaked distribution, even though the [underlying preferences] of these groups are far more heterogeneous.

The applications of this model are immediate. Rebel groups will tend to adopt practices that enterprises otherwise uninterested in conquest and political reform would not. These practices will often produce observably suboptimal economic institutions in order to send a strong signal to domestic audiences. For instance, rebel groups will deliberately undermine the profitability of their illegal practices to separate themselves from common criminals.

We examine these costly signals in the context of Colombia, where the leading insurgent organizations, the Revolutionary Armed Forces of Colombia (FARC) and National Liberation Army (ELN), openly and actively participate in the illegal coca trade. While an important revenue source for both groups, engagement in drug cultivation and export has obfuscated their historical identities and political aspirations. We present novel qualitative evidence that the FARC and ELN have and continue to engage in otherwise unprofitable distortion of the coca leaf market to signal their resolve to revolution. Put simply, rebels sacrifice potential profits to reassure civilians and political leaders of their revolutionary commitments. We supplement these qualitative accounts with archival evidence suggesting that drug cartels, prominent in an earlier period of the coca trade, sent costly signals of political ambition to obfuscate their identity as criminal enterprises.  Leadership of these illicit organizations, however, recognized the degree to which their political overtures and overt spending on public goods would induce concessions from government officials and bolster support from sympathetic communities.

This paper makes substantial contributions to our understanding of the political economy of rebellion. In particular, we highlight how rebels and criminal cartels sacrifice financial gain to secure potential concessions from state leaders and support from civilians.  Importantly, we demonstrate that there are limits to how much criminal organizations will invest in political and public spending commitments. Similarly, rebel organizations recognize that unfettered participation in criminal activities can drive away potential supporters and jeopardize their revolutionary ambitions.

Corporate Production of State Capacity: Evidence from a Natural Experiment [with Faisal Ahmed]

Under what conditions do multinational corporations produce state capacity? Recent research in economics and political science has focused on the means by which foreign firms undermine economic and political development.  This scholarship, largely drawing on evidence from countries rich in natural resources, has highlighted the role foreign corporate institutions play in enhancing the stability of authoritarian regimes through the generation of political and economic rents.  Increasing the value of state capture may further undercut state capacity by increasing the value of rent seeking behavior for rebel groups.  Yet previous studies ignore how multinational corporations produce state capacity by investing foreign capital in local police and national military forces.  This gap is due, in part, to the difficulty of identifying when and where firms enter into formal security contracts with state forces.  What’s more, even if such information were readily available, the intensity with which foreign companies invest in local state capacity is endogenous to the value of their historical investments in immobile resources.

We overcome both of these challenges with a natural experiment, embedded in Colombia’s ongoing civil conflict.  Using detailed geographic information on the historical boundaries of British Petroleum and Occidental Petroleum’s mineral leases, productivity measures of their top producing wells, and plausibly random shocks to the value of oil exports, we investigate how state rents and foreign corporate investments developed over the last half century.  We then introduce novel evidence acquired from the actual cash-for-security contracts these for foreign firms signed with Colombia’s national police and military.  These contracts, released to Human Rights Watch as part of an ongoing investigation, detail the extent to which multinational corporations have enhanced local state capacity. In addition to funding the reallocation of troop and officer units to oil-producing municipalities, these multinational firms set up temporary hospitals, purchased ammunition and other counterinsurgency supplies, and financed the creation of civilian informant networks.  These corporations further enhanced the human capital of local national forces by bringing in counterinsurgency experts to train troops and officers how to employ newly developed anti-guerrilla tactics.

To investigate the causal effect of corporate production of state capacity on insurgent activity, we draw on an incredibly rich set of economic and conflict microdata.  Our instrumental variables approach leverages plausibly exogenous variation in the international market price for oil, in addition to random variation in climatic conditions favorable for the production of coca, a dominant income source for Colombia’s primary rebel groups.  We also take advantage of the fact that the timing of these corporate-military security arrangements coincided with the otherwise unrelated collapse of the cartel system and termination of the aerial smuggling routes used to traffic coca.  We combine these natural experiments with our instrumental variables design and demonstrate that foreign firm involvement in local security arrangements had a profound causal impact on the character and intensity of rebel violence in mineral rich areas. In the period prior to foreign investment in state capacity, rebels engaged in frontal assaults on state forces to capture oil rents.  The intensity of this fighting was heightened by external shocks to the value of oil exports.  After enhancing state capacity in oil-rich municipalities, however, foreign firms were able to deter contestation over natural resources, even when those resources experienced a price boom.  Importantly, we find that these results obtain even after adjusting for windfalls in rebel capacity (during periods of high coca yield, and high prices).

This paper challenges the view that multinational firms undermine state capacity by consolidating political power, economic rents, and enhancing the prize of state capture.  Although some foreign firms may undercut political and economic development, others produce state capacity by investing in local police and military forces.  We use novel data on cash-for-security contracts and conflict microdata to demonstrate the causal impact of foreign-backed counterinsurgent capacity on rebel violence. This paper also illustrates the limitations of these security externalities.  In particular, we focus on violent contestation over emeralds and other precious gemstones in the mining municipalities of Colombia.  Our empirical analysis shows that even during a period of substantial increases in state capacity, rebels continued to engage military units forcefully to gain control over the extraction and export of emeralds.  Corporate production of state capacity, while beneficial in targeted areas, is not a panacea for state weakness.



Climate Shocks, Price Dynamics, and Human Conflict [with Patrick Signoret]

How do climate shocks impact human conflict? Growing evidence demonstrates that increases in ambient temperature and anomalous variation in rainfall significantly increase the risk of subnational conflict. These patterns are commonly attributed to two mechanisms: direct, heat-induced aggression, and indirect, economic disruption. This article addresses how various forms of violence—from spontaneous, individual aggression to organized, insurgent operations—respond to similar climate shocks. We also investigate how weather conditions impact local prices of staple goods, and use an instrumental variables approach to disentangle the direct, physiological and indirect, economic mechanisms underlying the climate-conflict relationship. We evaluate these questions in the context of Indonesia using high-resolution climatic, consumer price, and agricultural data, and exploit the most comprehensive microdata on violent crime, communal violence, and insurgent operations currently available for academic research.

We focus on three findings. First, interpersonal violence increases significantly as ambient temperatures rise, while the frequency of insurgent operations exhibits the opposite relationship. A standard deviation increase in ambient temperature, for example, increases the frequency of battery and assault by 15%. The same temperature shock decreases the intensity of insurgent fighting by 24%. Second, climatic conditions in agricultural sub-districts drive subnational variation in consumer prices. Third, climate-induced increases in consumer prices drive up non-economic and economic violent crime, but only in sub-districts that rely on production elsewhere for rice, soybeans, and other essential commodities. Importantly, these forms of interpersonal violence remain unchanged or decline in rice and soybean producing sub-districts as farmgate prices rise. These findings are robust to accounting for numerous potential sources of bias, including measurement error due to highly variable microclimates, exposure to subnational market fragmentation, and the introduction of surface water irrigation technologies. Our results clarify how climate shocks influence the occurrence of human conflict, challenge prominent findings in recent scholarship, and yield implications for policy-makers crafting interventions to reduce violence associated with contemporary and projected climate shocks.

Download this paper here.

— Supported by The Asia Foundation.


Wildfires and Human Conflict. 

Each year, wildfires destroy millions of hectares of vegetation, displace countless humans, further endanger at-risk species, generate uncontrollable respiratory hazards, and quicken the pace of global warming. However, our understanding of the impact of wildfires on human conflict is limited. Advances in remote sensing enable planetary-scale tracking of fire hazards and new microdata on a variety of human conflict permit novel assessment of the links between wildfires and violence. Here we quantify the impact of climate-induced wildfires on violence in Indonesia. We study data on the location, scale, and crops destroyed by 63,540 wildfires as well as 38,332 precisely georeferenced acts of interpersonal conflict. Our results suggest a robust link between the spread of wildfires and increased human conflict. The relationship is substantial and persists across crop and timber producing regions, and criminal and political violence. These results provide the first evidence that wildfires enhance human conflict, with important policy implications.






Connecting the Red Corridor: Infrastructure Development in Conflict Zones [with Oliver Vanden Eynde, Jamie Hansen-Lewis and Jacob N. Shapiro]

We introduce a unique, integrated dataset on Maoist activity, three flagship pro- grammes for rural infrastructure development (PMGSY, RGGVY, and USOF), and a dedicated programme targeted at India’s Left Wing Extremism (LWE) regions. Our data reveals that Maoist affected villages were not targeted differentially for the flagship programmes, but did attract more Integrated Action Plan (IAP) projects. The relationship between Maoist activity (at the village or district level) and programme performance appears to be complex. Regression results are partially in line with a large body of qualitative evidence on the importance of disruption. Nevertheless, we note some cases in which programme completion is faster (USOF) or appears to be cheaper (PMGSY and IAP) in Maoist affected areas.