Side by side views of buildings before and after the 2017 Puebla earthquake.

Why Building Codes Matter, and Why They’re Hard to Capture

Francisco’s study showed that 91% of the buildings that collapsed in Mexico City were built before 1985, when modern seismic provisions were introduced. Code era is one of the strongest signals of risk: older standards allowed systems like flat slabs or weak first stories that underperformed in earthquakes.

But translating code era into underwriting intelligence is difficult. In the U.S. for instance, building codes aren’t adopted nationally, but are enforced at the county or even municipal level. That means two buildings on the same street, built in the same year, may have been designed to different standards depending on local adoption. Traditional models rarely capture that nuance, while ResiQuant does—giving underwriters a clearer picture of not just when a property was built, but the standards it was built to meet.

Soil, Structure, and the Geography of Risk

Mexico City is well-known for its seismic microzonation that was created after the learning from the 1985 Michoacan earthquake. Built atop an ancient lakebed surrounded by firmer soils, the city experiences different ground motion depending on location.

In 1985, the Mw8.0 Michoacán earthquake mainly affected tall buildings (six stories and higher) concentrated on the lakebed zone, where long-period shaking was amplified. By contrast, the 2017 Puebla earthquak—an intraslab event—produced high-frequency shaking that disproportionately damaged low- to mid-rise buildings (1–5 stories) in the transition zone. The same city, the same peril, but with two very different damage patterns.

For insurers, the message is that peril is not monolithic. Soil–structure resonance can determine which buildings are exposed, and those interactions often operate at the neighborhood or even block level. That nuance rarely surfaces in portfolio models — but it’s decisive when losses hit.

(a) EW acceleration time-histories recorded by different stations across Mexico City. (b) Pseudo-acceleration response spectra at the same stations and direction across the city. Source: Francisco A. Galvis, Eduardo Miranda, Pablo Heresi, Héctor Dávalos, and Jorge Ruiz-García, “Overview of collapsed buildings in Mexico City after the 19 September 2017 (Mw7.1) earthquake,” Earthquake Spectra (2020).

The Recurring Lesson: Known Vulnerabilities, Overlooked Risks

Nothing about these failures was new. Flat slabs, soft stories, and torsional irregularities in corner buildings have all been documented repeatedly in earthquakes from Mexico to California to Turkey. And yet, they remain embedded in commercial properties worldwide.

The challenge for underwriters isn’t a lack of knowledge about what types of vulnerabilities exist. The problem is not knowing which specific buildings carry those risks. Without full visibility into a structure’s design, known weaknesses remain invisible in the underwriting process.

From Field Reconnaissance to Scalable Underwriting Intelligence

This reconnaissance study shows that detailed inspections of collapsed buildings, forensic tracing of failure modes, and cross-checking against code history and soil conditions are the kind of granular intelligence that can provide insights about which structures have a higher likelihood of failure.

But underwriters can’t send PhDs to walk every block or evaluate every building in their submission pile. That kind of reconnaissance, historically, only happens after a disaster.

ResiQuant was built to change that. By embedding structural-engineering expertise into AI models, ResiQuant performs virtual inspections that surface most of these known vulnerabilities with engineering-grade analysis before binding.

Explore the detailed analysis and methodology in the full paper.