Editor’s note: This column is based on CEPR Discussion Paper No. 21648 “Refineries, pipelines and petroleum fields – the impact of oil on conflict around the world”.

During the 2026 Middle East oil shock, world oil markets have been volatile. Between February and April 2026, the average price of crude oil rose from $68 to $104 per barrel (World Bank 2026a). In April, the World Bank projected energy prices to surge by 24% this year, after attacks on energy infrastructure and shipping disruptions in the Strait of Hormuz triggered what it described as the largest oil supply shock on record (World Bank 2026b). For a locality sitting on oil reserves, that means that the gross market value of each barrel beneath the ground rose by more than half in only two months. The policy debate, on VoxEU and elsewhere, has understandably concentrated on the macroeconomic fallout: how the shock passes through to inflation, how central banks should respond, and why geopolitical oil price shocks hit harder than ordinary ones (Verduzco-Bustos and Zanetti 2026). Far less attention has been paid to a question that previous price surges have raised time and again: what does expensive oil do to violence in the oil-producing regions of low- and middle-income countries?

The honest answer from the academic literature has long been “we are not sure”. Blair et al. (2021) count more than 350 studies of the relationship between commodity prices and conflict, and in a meta-analysis of 46 natural experiments they find no systematic average effect. What does emerge is that commodities differ. Higher prices for labour-intensive agricultural goods tend to reduce violence by raising the opportunity cost of fighting (Miguel et al. 2004, Dal Bó and Dal Bó 2011, Dube and Vargas 2013). McGuirk and Burke (2022), for instance, predicted that the food-price surge following the Ukraine war would raise conflict in Africa’s food-consuming regions while reducing it in producing ones. Price increases for capital-intensive, spatially concentrated resources, by contrast, are more often linked to violence, as the value of controlling fixed rents rises. Berman et al. (2017a, 2017b) estimate that the mineral price boom of the 2000s may account for a substantial share of conflict across Africa. Yet even within the literature on oil, the evidence remains ambiguous (Cotet and Tsui 2013, Bazzi and Blattman 2014, Andersen et al. 2022).

In a recent paper (Dreher et al. 2026), we argue that part of this ambiguity reflects a mismatch between theory and measurement. Higher oil prices can generate violence through two distinct channels. First, they raise the value of controlling oil-bearing territory, giving armed actors stronger incentives to fight for the prize. Second, where local populations see few benefits from rising oil rents, price booms can sharpen distributional grievances and turn visible oil infrastructure into focal points for protest, sabotage, and other forms of unrest. Aggregate studies may find little or no effect of oil prices on conflict because they average across the wrong margins: different forms of violence, different types of oil assets, and different locations. 

Comparing oil and non-oil locations within the same country and month

We construct a monthly panel of 0.5×0.5 degree grid cells – roughly 55 km × 55 km at the equator – covering 131 low- and middle-income countries between 1989 and 2021, around 17.2 million cell-month observations in total. Georeferenced conflict events from the Uppsala Conflict Data Program are combined with detailed data on the locations of petroleum fields, extraction sites, refineries, and pipelines. Because world oil prices are set on global markets, no individual grid cell moves them. We therefore compare oil-bearing and non-oil cells within the same country and month as world prices fluctuate. This allows us to hold constant national politics, macroeconomic conditions, and month-specific shocks, while isolating differences between locations with and without oil. The design passes a battery of checks: there is no evidence of differential pre-trends between oil and non-oil locations, the results survive excluding the world’s ten largest oil producers; they hold for conflict onset, event counts, and alternative event datasets; and a placebo test using coconut oil prices yields an effect close to zero.

Conflict moves towards the prize

When oil prices rise, the probability of organised conflict increases significantly in oil-bearing cells and falls in the non-oil cells around them. A one standard deviation rise in the (log) oil price, corresponding to roughly a 55% price increase, raises the monthly probability of conflict in an oil cell by 0.054 percentage points. Relative to the baseline conflict probability of 0.69% in oil cells, that is an increase of about 7.7%. Cumulated over a large price swing, the effect is anything but marginal: comparing the lowest oil price in our sample period (December 1998) with its peak (July 2008), the predicted probability of conflict in oil cells is 0.51 percentage points – or 45.8% – higher at the peak (Figure 1 shows the predicted increases for each cell). At the same time, additional specifications show that the risk of conflict declines in first- and second-degree neighbouring cells.

Figure 1 Predicted increase in conflict risk in oil cells during the 1998–2008 oil price boom

Notes: The map shows, for each oil-bearing grid cell in the sample, the predicted decline in conflict probability had oil prices remained at their December 1998 level instead of rising to their July 2008 peak. Cells are grouped at cut-offs of 36% and 45%, the 25th and 75th percentiles of positive predicted effects: 938 cells show predicted declines of up to 36%, 1,874 cells between 36% and 45%, and 932 cells above 45%.

This spatial pattern explains why the literature has struggled to agree. Averaged across all cells in our sample, the predicted change in conflict probability from the same low-to-peak price swing is -0.02 percentage points – essentially zero – because the declines extend across far more neighbouring cells than there are oil-bearing ones. Weighting cells by their predicted conflict risk under the low-price scenario turns the average to +0.02 percentage points. At the level of aggregation used by most country studies, a large local effect nets out almost perfectly. Put simply: in the aggregate, little changes; locally, a lot does.

Not all violence moves the same way

The reallocation is driven almost entirely by state-based conflict – violence involving governments – and is strongest in petroleum fields, the locations of underlying resource wealth, rather than at operating extraction sites or downstream infrastructure. This pattern is consistent with territorial contestation. Petroleum fields mark where the oil wealth lies, whether or not it is currently being extracted, and when the world price rises, the value of controlling that territory rises with it. Operating extraction sites show no such response; if anything, state-based conflict there declines, arguably because production revenues strengthen governments’ capacity to protect these sites. 

Lower-intensity unrest follows a different logic. We find that higher oil prices are associated with more protests and violence against civilians around refineries and pipelines, rather than at petroleum fields. This pattern is consistent with a distributional-grievance channel: downstream infrastructure makes the processing and movement of oil rents especially visible, and can become a focal point for mobilisation where communities feel excluded from the gains. And unlike armed conflict, protests also tend to rise in neighbouring cells: civil unrest spreads into surrounding areas, whereas organised conflict is pulled out of them. Attacks on energy infrastructure increase across all facility types when prices rise. 

Who captures the rents matters

The response also depends on who owns the assets. Compiling ownership information at the facility level, we find that conflict responses to oil price shocks are significantly stronger where oil assets are domestically controlled, particularly by state-owned enterprises, and weaker, or even reversed, where foreign firms dominate. Attacks on multinational corporations do not increase. The effects are also amplified where politically excluded ethnic groups live near the oil, concentrated near international borders and far from national capitals, and dampened where institutions are stronger. Taken together, these patterns point to the political salience of oil rents. Where rents are domestically controlled, particularly by state-owned firms, higher prices appear more likely to intensify both struggles over control of valuable territory and grievances over how the resulting wealth is distributed.

What this means for 2026

Applied to the 2026 Middle East oil shock, our results suggest a sharp rise in conflict risk around oil-bearing locations. Based on the rise in average crude prices from $68.01 to $103.91 per barrel between February and April 2026, our estimates imply a 0.09 percentage-point increase in the monthly probability of conflict in the oil-bearing cells of low- and middle-income countries, about 6.7% relative to the February baseline, alongside predicted declines of 0.02 and 0.03 percentage points in first- and second-degree neighbouring cells.

Watching the wrong map

Conflict risk is still often monitored at the country level through country risk ratings, national conflict counts, and threshold-based definitions of civil war. Our results suggest that such measures can be misleading during oil price booms. They may conceal two very different responses: organised violence is reallocated toward petroleum fields as the value of controlling the underlying resource rises, while lower-intensity unrest escalates around refineries and pipelines, where rising oil wealth is especially visible. The policy implication is not simply to direct more security resources towards oil-producing countries. Effective prevention requires knowing which form of violence is likely to emerge, which type of oil asset is at stake, and where the risk is concentrated. Protecting petroleum fields addresses a different problem from preventing protests, attacks, and violence around downstream infrastructure. Policies that ignore these distinctions risk responding to the wrong threat in the wrong place. This matters especially when the sharp rise in oil prices is increasing both the value of oil-bearing territory and the political salience of oil rents. The 2026 oil shock will be felt not only in consumer prices. It will be felt on the map of violence in low- and middle-income countries, and aggregate statistics are the wrong place to look for it.

References

Andersen, J J, F M Nordvik and A Tesei (2022), “Oil price shocks and conflict escalation: Onshore versus offshore”, Journal of Conflict Resolution 66(2): 327–356.

Bazzi, S and C Blattman (2014), “Economic shocks and conflict: Evidence from commodity prices”, American Economic Journal: Macroeconomics 6(4): 1–38.

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

Berman, N, M Couttenier, D Rohner and M Thoenig (2017b), “Countering the mining curse”, VoxEU.org, 9 June. 

Blair, G, D Christensen and A Rudkin (2021), “Do commodity price shocks cause armed conflict? A meta-analysis of natural experiments”, American Political Science Review 115(2): 709–716.

Cotet, A M and K K Tsui (2013), “Oil and conflict: What does the cross country evidence really show?”, American Economic Journal: Macroeconomics 5(1): 49–80.

Dal Bó, E and P Dal Bó (2011), “Workers, warriors, and criminals: Social conflict in general equilibrium”, Journal of the European Economic Association 9(4): 646–677.

Dreher, A, J Pan and M Herrmann (2026), “Refineries, pipelines and petroleum fields – the impact of oil on conflict around the world”, CEPR Discussion Paper No. 21648.