Mexico's Security Strategy Is a Failure, Critics Say, Despite 46% Homicide Drop Claim
The government says homicides dropped 46%. The bodies say otherwise. Since Sheinbaum took office, an estimated 35,000 people have been killed.
President Claudia Sheinbaum and Security Secretary Omar García Harfuch have spent months touting a statistic they say proves their security strategy is working: a 46% average decline in daily homicides since September 2024, translating to roughly 39 fewer murders per day. The number comes from internal security data and has been repeated at the presidential morning press conference, the mañanera, as proof that the administration's approach is delivering results.
Critics say the number is a fiction. Not because it's fabricated, but because it measures the wrong thing, or measures it in a way that obscures more than it reveals. The argument, advanced by civil society organizations, victims' families, independent analysts, and some state-level officials, is that the federal government has reclassified crimes, adjusted methodologies, and selectively presented data to produce a narrative of progress that doesn't match the reality on the ground.
The numbers are complicated. Mexico's security data comes from multiple sources: the Executive Secretariat of the National Public Security System, which compiles reports from state and municipal police forces; the INEGI statistics institute, which conducts its own surveys; and the federal government's internal tracking. These sources frequently disagree. The SESNSP, which has historically been more transparent than the federal government's preferred metrics, has produced figures that contradict the official narrative.
What isn't complicated is the human cost. Since Sheinbaum took office in October 2024, an estimated 35,000 people have been killed in homicides, according to civil society tallies. That figure, if accurate, would make her administration one of the most violent in Mexico's modern history, even as the government points to declining rates. The contradiction is mathematically possible: if the baseline was extraordinarily high, a significant decline can still leave the absolute numbers catastrophic.
García Harfuch, a former Mexico City police chief who survived a 2020 cartel ambush that killed several of his bodyguards, has become the public face of the security strategy. His appearances at the mañanera are lengthy, data-heavy, and aggressively defensive. Critics accuse him of using the platform to build a political profile for a potential 2030 presidential run, a claim the government denies. His critics point to a pattern of presenting favorable data while dismissing contradictory evidence from state-level authorities.
The structural problem is deeper than any single statistic. Mexico's security apparatus is fragmented across federal, state, and municipal levels, with overlapping jurisdictions, incompatible databases, and political incentives that reward optimism over accuracy. Governors have strong reasons to underreport violence in their states. Federal officials have strong reasons to aggregate data in ways that produce favorable trends. The result is a data landscape where almost any number can be defended and almost no number can be fully trusted.
For the families of the 35,000 dead, the debate over methodology is an obscenity. They don't need a statistic to tell them what happened. They need the killings to stop. The government says it's working. The data says it's complicated. The bodies say something else entirely.
The comparison to previous administrations is instructive. Under AMLO, the government adopted a similar strategy of presenting favorable data while dismissing contradictory evidence. The result was a documented increase in violence that the official numbers failed to capture. Sheinbaum's team has refined the approach, using more sophisticated data presentation and a more polished spokesperson in García Harfuch. The underlying dynamic, however, remains the same: the government needs the numbers to work, and when they don't, the methodology changes.
Independent analysts have noted that the 46% figure relies on a specific comparison period that flatters the current administration. September 2024 was one of the most violent months in recent memory, driven by the Sinaloa Cartel internal power struggle following the arrest of Ismael Zambada. Starting the clock at that peak produces a dramatic-looking decline. Starting it six months earlier, or using a rolling annual average, produces a much less impressive picture.
The international dimension complicates matters further. The DEA's declaration this week that the Sinaloa Cartel and CJNG are its top priority directly contradicts the narrative that Mexico's security situation is improving under its current leadership. If the two most powerful criminal organizations in the country are still powerful enough to be the DEA's number one target, the claim that homicides are falling rapidly requires more explanation than the government has provided.
None of this means the security strategy has failed entirely. Some states have seen genuine improvements. Some municipal police forces have been strengthened. Some programs are working. But the gap between the government's message and the lived experience of millions of Mexicans living in violence-prone areas is wide enough to drive a truck through, and it's getting wider.