ConflictMeter’s latest movement table contains one result that demands explanation. El Salvador has fallen from 23rd of 80 countries in the authorised 2025 Vintage baseline to 71st of 81 current Global FULL countries. That is a movement of 48 ranking places away from rank 1—the largest fall in the current comparison set.
This is not merely a quirk of the ranking universe changing by one country. El Salvador’s headline ConflictMeter score also fell by 27.59 points, to a current reading of 33.52. Its present profile is uneven: Pressure 52.35, Activation 18.61, Economic pressure 0.00 and Narrative fragmentation 33.52. The striking feature is the very low current Activation reading relative to Pressure.
That invites an obvious research question: does the dramatic transformation of public security under President Nayib Bukele help explain the fall?
The answer is plausible, but not simple. ConflictMeter is not an event-coding system and it does not assign points to a named political leader or policy. It measures how closely current conditions align with a set of conflict-associated conditions. A movement therefore needs to be investigated through the structure of the score, source timing and external evidence—not reverse-engineered into a single political explanation.
El Salvador really has undergone an extraordinary security change
For years, El Salvador was synonymous with exceptionally high gang violence. In 2015 the country recorded 6,656 homicides. By 2024, official figures reported 114 homicides, or about 1.9 per 100,000 people. Reuters later reported that the government said the homicide rate fell again to 1.3 per 100,000 in 2025. These official figures describe a profound change in everyday security conditions. Associated Press reported the 2024 figures, while Reuters described the continuing 2026 anti-gang campaign.
The change followed a state of exception introduced in March 2022 after a sudden surge in killings. The government combined emergency powers, mass arrests, military-backed policing and the construction of the CECOT mega-prison. By April 2026, Reuters reported that more than 91,000 people had been detained under the crackdown.
For a model concerned partly with mobilisation and the conditions under which organised internal conflict can escalate, it is reasonable to investigate whether the destruction or incapacitation of organised gang structures, the collapse in homicide and the restoration of state control over previously contested neighbourhoods could reduce some activation-type signals.
But “more order” is not the same thing as “lower political risk”
There is a crucial counterweight. The same security programme has involved the prolonged suspension of constitutional protections and has been criticised extensively by human-rights organisations. Human Rights Watch says gang violence has markedly declined while also documenting mass arbitrary detention, enforced disappearance, torture allegations, due-process violations and a major concentration of executive power. Its 2026 country report says more than 90,000 people had been arrested under the state of emergency.
Amnesty International’s July 2026 investigation similarly argues that the emergency regime has become a sustained security system involving mass detention and serious rights violations. Amnesty’s report focuses particularly on due process, detention conditions and institutional change.
Those facts matter because a civil-conflict conditions index is not a liberal-democracy score and it is not a simple crime index. A government can suppress one pathway to organised violence while simultaneously creating different forms of institutional stress. The direction of the net headline score therefore cannot be read as a verdict on the government, its democratic quality or the desirability of its policies.
What the current ConflictMeter anatomy suggests
The most useful clue is the composition of El Salvador’s current reading.
- Global: 33.52
- Pressure: 52.35
- Activation: 18.61
- Base: 31.21
- Economic pressure: 0.00
- Narrative fragmentation: 33.52
The country is therefore not uniformly low across the model. Structural Pressure remains above 50, while Activation is much lower. That matters. One possible interpretation is that some underlying structural conditions remain present, but the current environment contains far less evidence of active mobilisation than the 2025 Vintage baseline did.
That interpretation would be consistent with a state that has severely weakened the organisational capacity of violent non-state actors. It is also consistent with a broader lesson about ConflictMeter: the same underlying pressure can produce very different headline conditions depending on whether activation is high or low.
The score movement is too large to dismiss as rank noise
Ranking changes can be misleading when the comparison universe changes. Here, however, the underlying score moved sharply as well. El Salvador’s current score is 27.59 points below its 2025 Vintage reading. That makes the movement analytically interesting even after allowing for the ranking universe changing from 80 to 81 countries.
The country page also reports LOW current revision sensitivity and MEDIUM historical revision exposure. That does not eliminate source effects, but it reduces the case for treating the entire fall as a data-revision artefact.
A hypothesis worth testing, not a conclusion to declare
The strongest defensible hypothesis is therefore narrower than “Bukele fixed El Salvador, so the score fell.” The evidence supports asking whether the collapse in gang violence and the dismantling of organised criminal control materially reduced the activation side of the model. The current profile is compatible with that possibility.
But the same period also brought extraordinary imprisonment, weakened due-process protections and a concentration of political power. Those developments could affect other dimensions in ways that a single headline movement obscures.
The research task is therefore to decompose the change further: which underlying source families moved, how much of the 27.59-point fall is attributable to Activation versus structural components, and which changes survive a Vintage-versus-Reconstructed comparison?
That is exactly the kind of question the movement table is supposed to generate.
ConflictMeter note: A lower ConflictMeter score is not a statement that a country is “good”, democratic or safe in every sense. The index measures alignment with a defined set of civil-conflict conditions. It is not a probability forecast and rank movement is not a desirability judgment.
Explore the data
See the El Salvador country profile, the movement table, and the ConflictMeter methodology.