Seven Out of Ten. Lima Was Not Built by Architects.

Human Becoming

The Man Who Builds One Floor at a Time

His name is Jorge. He does not use his last name with strangers, and he would not use it here. He is fifty-three years old. He has been building his house in Villa El Salvador for twenty-six years. The first floor went up in 2000 — four rooms on a plot he purchased with no title, using concrete blocks he bought in monthly batches from a distributor in Pachacamac who extended credit to anyone who could prove they had a plot. He mixed the concrete himself. His brother-in-law, who had worked construction in Comas, showed him how to set the rebar. Neither of them is an engineer. Neither of them consulted an architect. The municipality did not issue a permit because Jorge did not apply for one, and the municipality did not ask because the municipality does not have enough inspectors to ask.

The second floor went up in 2008. It took eight years to save for the materials. This time Jorge hired two workers from his neighborhood — men who had built their own houses the same way, learning by doing, passing techniques laterally the way recipes pass through families. They used ladrillo pandereta for the walls: hollow clay bricks that are lighter than solid brick, cheaper by a third, and widely available at every ferreteria in the southern cone of Lima. Ladrillo pandereta is designed for partition walls. It is not designed to bear structural loads. Jorge used it for structural walls because everyone in his neighborhood uses it for structural walls, and because solid brick costs more than he can afford.

The third floor is half-finished. The columns are poured. The rebar extends upward from the roof of the second floor, exposed to air and rain, rusting at the tips. Jorge says the third floor will be for his daughter when she marries. He has been saying this for four years. The rebar has been exposed for four years. In earthquake engineering, exposed rebar that has corroded loses between 15 and 40 percent of its tensile strength depending on the degree of oxidation. Jorge does not know this number. What he knows is that the rebar is still there, and when he has money, he will pour the rest of the slab.

His house is not unusual. His house is the median case. According to the most comprehensive study ever conducted of Lima's self-built housing stock — a joint investigation by CISMID (the Japan-Peru Center for Seismic Research and Disaster Mitigation) and the National University of Engineering — seventy percent of Lima's residential buildings were constructed without the participation of an architect or engineer, without municipal permits, and without compliance with Peru's national building code.^[1] Seventy percent. In a metropolitan area of 11.1 million people. In a city that sits directly on the subduction zone where the Nazca Plate slides beneath the South American Plate at a rate of 61 millimeters per year.

Jorge's house is not an anomaly. Jorge's house is Lima.

Structural Read

The Economy Nobody Measures

The word is autoconstruccion. It translates literally as "self-construction," but the translation strips the term of its density. Autoconstruccion is not a DIY movement. It is not a lifestyle choice. It is the primary mechanism through which the majority of Lima's population has housed itself for the past six decades — a system of incremental, informal, owner-directed building that operates entirely outside the formal construction industry, the banking system, the regulatory apparatus, and the architectural profession.

The numbers are staggering when aggregated. According to estimates compiled by Peru's Chamber of Construction (CAPECO) and the Association of Developers and Builders (ADI Peru), autoconstruccion moves approximately 25 billion soles per year — roughly 6.7 billion US dollars at current exchange rates. That figure represents 4.1 percent of Peru's GDP.^[2] To put that in perspective: autoconstruccion generates more economic activity than Peru's entire fishing industry. It generates more than the country's textile sector. It is, by volume of capital deployed, one of Peru's largest industries. And it operates with almost no institutional oversight, no quality control, no insurance, and no engineering standards.

The mechanism works like this. A family acquires a plot — sometimes through formal purchase, more often through invasion (organized land occupations that have been a feature of Lima's urbanization since the 1950s) or through informal subdivision of agricultural land on the city's periphery. Title may come later, years or decades after occupation, through COFOPRI, the government's formalization agency. Or title may never come at all. Construction begins regardless.

The first phase is typically a single story: three or four rooms built with concrete block or fired brick, a corrugated metal or concrete slab roof, basic electrical wiring, and rudimentary plumbing. This phase is financed entirely from savings or informal credit. Banks do not lend to people without title. People without title build anyway.

The second phase comes years later — often a decade or more. CISMID's field surveys found an average interval of 16 years between the construction of the first floor and the beginning of the second.^[1] The second floor is built on top of the first using whatever structural system the first floor happens to have. If the original foundation was designed for a single story — and it almost always was — the second floor adds loads the foundation was never calculated to bear. But the family needs the space. A son marries. A daughter has children. An elderly parent moves in. The house grows upward because it cannot grow outward — the plot is already bounded on all sides by neighbors who built the same way.

Each additional floor changes the structural calculus dramatically. Carlos Zavala, the director of CISMID and Peru's foremost authority on seismic vulnerability in informal construction, has published research quantifying this effect. For a typical autoconstruccion house in Lima built with confined masonry and ladrillo pandereta walls, the probability of structural collapse in an 8.0 magnitude earthquake increases from approximately 0.07 for a single-story structure to approximately 0.3 when a second or third floor is added without proper engineering.^[3] That is not a marginal increase. That is a fourfold multiplication of collapse probability. And it affects not just the house in question but the houses on either side, which share party walls and whose structural systems are coupled through those walls.

The ladrillo pandereta problem deserves its own paragraph because it is the single most dangerous material choice in Lima's informal housing stock. Pandereta bricks are hollow clay units with thin walls and internal cavities. They were designed and manufactured for use as non-structural partition walls — the kind of wall that divides a living room from a bedroom but bears no vertical load and transfers no lateral force. In Lima's autoconstruccion economy, pandereta bricks are used as the primary structural walls in an estimated 40 to 60 percent of informal houses.^[3] They are used this way because they cost roughly 30 percent less than solid king kong bricks, because they are lighter and easier to lay, and because the hardware stores that serve the periphery stock them in greater quantity. The result is that hundreds of thousands of homes in Lima have load-bearing walls made of a material that shatters under lateral seismic loading — a material that was never intended to hold up a roof, much less two additional floors of concrete and masonry.

The economic logic is internally coherent even as the engineering logic is catastrophic. A family that earns 2,500 soles per month — roughly the median household income in Lima's peripheral districts — cannot afford an architect, whose minimum fee for a residential project runs between 8,000 and 15,000 soles. It cannot afford a structural engineer, whose foundation design alone would cost 3,000 to 5,000 soles. It cannot afford solid brick when hollow brick costs a third less. It cannot afford to build the entire house at once when the only available financing is its own monthly surplus. So it builds incrementally, informally, and structurally inadequately — not out of ignorance but out of rational economic calculation within constraints that the formal system has never addressed.

The municipality knows. The national government knows. CISMID has been publishing vulnerability assessments for two decades. The Peruvian Geological, Mining, and Metallurgical Institute (INGEMMET) has mapped the seismic hazard zones of Lima in detail sufficient to identify which neighborhoods will experience the most severe ground acceleration in the next major earthquake.^[4] The data exists. What does not exist is a mechanism to retrofit seven out of every ten houses in a city of eleven million people.

The Earthquake That Has Not Happened Yet

Lima's last catastrophic earthquake was in 1746. It was estimated at magnitude 8.6 to 9.0. It destroyed the city entirely and generated a tsunami that obliterated the port of Callao, killing an estimated 5,000 of Callao's 7,000 residents. The recurrence interval for major subduction earthquakes on the central Peru segment is estimated at 250 to 300 years.^[4] It has been 280 years.

The seismic gap is real and measurable. The Nazca Plate has been accumulating strain against the South American Plate for nearly three centuries without a major rupture on the Lima segment. CISMID, the United States Geological Survey, and multiple international seismological institutes have identified this segment as one of the highest-probability locations for a major earthquake globally.^[5] The question among seismologists is not whether a major earthquake will strike Lima. The question is when, and whether it will be an 8.0 or something larger.

The modeling for an 8.0 magnitude earthquake directly beneath Lima produces numbers that seismologists describe with clinical precision and city planners receive with visible dread. CISMID's scenario analysis, led by Zavala's team, estimates between 15,000 and 30,000 fatalities in an 8.0 event, with approximately 200,000 homes destroyed and an additional 350,000 rendered uninhabitable.^[3] The fatalities are concentrated in the informal housing stock — the autoconstruccion belt that wraps around the city's periphery in districts like Villa El Salvador, Villa Maria del Triunfo, San Juan de Lurigancho, Comas, and Carabayllo. These are the districts where pandereta walls carry three stories. These are the districts where foundations were poured for one floor and loaded with three. These are the districts where rebar rusts in exposed columns waiting for a fourth floor that may never come but whose corroded steel will fail regardless.

The 2007 Pisco earthquake — magnitude 8.0, centered 230 kilometers south of Lima — killed 519 people and destroyed 48,000 homes in a city of roughly 130,000.^[6] Lima is eighty-five times larger than Pisco. If the destruction rate scales even partially, the arithmetic is devastating. And the Pisco earthquake revealed a pattern that CISMID had predicted but had never observed in real conditions: the houses that collapsed first and most completely were the ones built with ladrillo pandereta as structural walls. The material failed exactly the way the laboratory tests said it would — the hollow cells crushed under lateral force, the thin clay webs shattered, and the walls disintegrated into rubble rather than deforming. Deformation is survivable. Disintegration is not.

The post-Pisco forensic analysis also confirmed a second CISMID prediction: multi-story informal houses collapsed at dramatically higher rates than single-story structures of similar construction quality.^[6] The added mass of upper floors amplified the seismic forces at the base. The connections between floors — often improvised, without proper anchorage between the slab and the walls below — failed. The upper floors pancaked onto the lower floors. In earthquake engineering, this is called "soft story collapse," and it is the mechanism that kills people in their own homes.

The Water Equation

The seismic vulnerability of Lima's informal housing stock cannot be understood in isolation from its water infrastructure, because the two systems interact in ways that compound the risk.

As of 2025, approximately 635,000 residents of metropolitan Lima lack access to piped water.^[7] These residents are concentrated in the same peripheral districts where autoconstruccion dominates — the hillside settlements of the southern cone, the northern cone, and the eastern expansion zones. They receive water from cistern trucks that climb unpaved roads to deliver water into rooftop tanks or ground-level containers. The water from these trucks costs between four and six times more per cubic meter than piped water from SEDAPAL, Lima's water utility.^[7] The poorest residents of Lima pay the most for the worst water.

The interaction with seismic risk is direct. Houses that rely on rooftop water tanks — typically concrete or polyethylene tanks holding 500 to 1,100 liters — carry an additional 500 to 1,100 kilograms of dead load on roofs that were not engineered for any specific load. In a seismic event, the water mass becomes a dynamic load that amplifies the oscillation of the structure. A 1,000-liter tank on the roof of a three-story pandereta house is not a water supply. It is a wrecking ball waiting for an earthquake to swing it.

After a major earthquake, the water system itself becomes a secondary crisis. SEDAPAL's distribution network runs through the same unstable soils that will amplify ground acceleration. The Rimac and Lurin river valleys, which supply Lima's reservoirs, are flanked by slopes that INGEMMET has classified as high landslide risk.^[4] The 2007 Pisco earthquake ruptured water mains throughout the affected area, leaving 80 percent of Pisco without water for over two weeks. In Lima, a comparable rupture pattern would leave millions without water in the immediate aftermath of the earthquake — the same millions whose houses have just collapsed or become uninhabitable. The compound catastrophe is not two separate events. It is one event with two faces.

SEDAPAL's own contingency planning acknowledges this. The utility's emergency preparedness documents, reviewed by the Comptroller General's office in 2024, identify 23 critical points in the distribution network where seismic damage would cascade into system-wide failure.^[8] The remediation plan for these 23 points has been in various stages of proposal since 2018. As of 2026, fewer than half have been addressed.

Pattern Confirmation

The Template That Does Not Scale

Lima is not the only city in Latin America built primarily through autoconstruccion. The pattern is continental. Mexico City's peripheral expansion follows the same logic — incremental, informal, family-directed construction on land of uncertain title, using whatever materials and labor the family can mobilize. Bogota's southern districts, Quito's hillside barrios, the favelas of Rio de Janeiro and Sao Paulo — all share the fundamental mechanism: populations excluded from the formal housing market build their own environments through a parallel construction economy that operates without regulation, insurance, or engineering standards.

But Lima occupies a unique position in this continental pattern because of the convergence of three factors that no other major Latin American autoconstruccion city shares simultaneously. First, extreme seismic exposure: Lima sits on one of the most active subduction zones on Earth, with a documented seismic gap approaching three centuries. Second, extreme material vulnerability: the prevalence of ladrillo pandereta as a structural material in Lima's informal housing stock appears to be higher than in comparable cities, where solid brick or concrete block dominates even in informal construction.^[3] Third, extreme water fragility: Lima is the second-largest desert city in the world after Cairo, entirely dependent on Andean river systems for its water supply, with 635,000 people already outside the piped network.

No other city in the hemisphere combines all three conditions at this scale. Mexico City has severe seismic risk but better water infrastructure and different construction materials in its informal sector. Bogota has extensive autoconstruccion but sits in a lower seismic hazard zone. Rio's favelas face landslide risk but not subduction-zone earthquake risk. Lima is the convergence point.

The Peruvian government's response has been incremental and insufficient, calibrated to political cycles rather than seismic ones. The "Mi Vivienda" social housing program, launched in 2002, has produced approximately 500,000 housing credits over two decades — a significant number, but one that addresses the formal housing deficit rather than the structural vulnerability of the existing informal stock.^[9] A family in Villa El Salvador with a three-story pandereta house does not need a new house. It needs a structural retrofit of the house it already has. The retrofit programs that exist — principally the "Bono de Proteccion de Viviendas Vulnerables a los Riesgos Sismicos," launched in 2022 — have reached fewer than 8,000 households in three years.^[9] At that rate, retrofitting the estimated 1.5 million structurally vulnerable homes in Lima would take 562 years. The seismic gap will not wait 562 years.

The engineering solutions exist. CISMID has developed and tested retrofit techniques specifically designed for Lima's autoconstruccion housing stock: fiber-reinforced polymer strips bonded to pandereta walls to increase their shear resistance, steel mesh jackets for under-reinforced columns, foundation underpinning for overloaded footings.^[3] The cost per house ranges from 5,000 to 15,000 soles depending on the severity of the deficiencies — a fraction of the cost of a new house, and a fraction of the cost of a collapse. The techniques work. They have been validated in laboratory testing and in limited field applications. What does not exist is a delivery mechanism that can reach seven out of ten houses in Lima.

The paradox is structural, not technical. Autoconstruccion produced the largest housing stock in Peru's history. It housed millions of people that the formal market refused to house. It created neighborhoods, commercial districts, and functional urban life where the planning profession saw only desert hillsides and river margins. The families who built these houses did not fail. The systems that were supposed to provide alternatives — affordable formal housing, accessible credit, universal water infrastructure, municipal inspection capacity — failed. Autoconstruccion filled the vacuum. And in filling it, it created a vulnerability so large that the engineering profession has no mechanism to address it at scale.

This is the signal. Not that Lima's houses are poorly built. That is a condition, not a signal. The signal is that the largest act of collective agency in Peru's modern history — the self-construction of an entire metropolis by its own residents — is simultaneously the largest concentration of seismic vulnerability in the Western Hemisphere. The same energy, determination, and resourcefulness that built Lima is the thing that made Lima fragile. And the earthquake does not care about the distinction.

Alternative Explanations

It is possible that the 70 percent figure overstates the true proportion of informal construction in Lima. CISMID's methodology relies on survey sampling and extrapolation, and the precise boundary between "informal" and "formal" construction is not always clear — some houses are built informally but later legalized; others receive partial professional input (a maestro de obra with decades of experience, for instance, whose practical knowledge may exceed that of a newly graduated engineer). Under a more generous definition that includes semi-formal construction with experienced but unlicensed builders, the figure might drop to 55 or 60 percent. This does not change the structural conclusion: the majority of Lima's housing stock was built without engineering calculation, and the majority of it uses materials and methods that are known to fail catastrophically in seismic events.

A second counterargument holds that Lima has experienced several moderate earthquakes in recent decades — magnitude 5.0 to 6.0 events — without catastrophic failure of the informal housing stock, suggesting that the vulnerability may be overstated. This argument misunderstands seismic dynamics. The energy released by an 8.0 earthquake is approximately 31,600 times greater than a 5.0 earthquake. Structures that survive a 5.0 with minor cracking can collapse entirely in an 8.0. The moderate earthquakes that Lima has experienced are not evidence of structural adequacy. They are evidence that the test has not yet arrived.

A third objection notes that some autoconstruccion builders have access to technical guidance through municipal "asistencia tecnica" programs and through informal knowledge networks, and that the construction quality is not uniformly poor. This is true and important. Not every informal house in Lima is a death trap. Some maestros de obra — master builders — have extensive practical knowledge and build structures that, while not code-compliant, are reasonably resistant. But CISMID's vulnerability assessments account for this variation: the 15,000-30,000 fatality estimate is a range, not a point, and the lower bound assumes that a significant proportion of informal houses perform better than the worst cases.

What is not known: The exact number of houses in Lima that use ladrillo pandereta as structural walls. Estimates range from 40 to 60 percent of the informal housing stock, but no comprehensive census of construction materials exists at the metropolitan level. The actual condition of rebar in exposed columns — the degree of corrosion, the remaining tensile capacity — has been studied in sample sets but not at population scale.

What is not confirmed: Whether the seismic gap on the Lima segment will produce an 8.0 earthquake or something different in character — a series of smaller events, a slow-slip event, or a rupture that extends to adjacent segments and produces something larger. The modeling assumes a single 8.0 event because that is the historical analog (1746), but seismic gaps do not always close in the same way they opened.

What would change the signal: If the Peruvian government scaled the retrofit program from 8,000 to 80,000 households per year, the vulnerability trajectory would begin to bend. If a material substitution policy made solid brick price-competitive with pandereta in the peripheral ferreteria market, new construction would improve. If SEDAPAL completed its 23 critical-point remediation before the earthquake rather than after, the compound catastrophe would be less catastrophic. None of these changes are technologically impossible. All of them are politically improbable at current rates.

Monitoring indicators: Track CISMID vulnerability assessment updates annually. Monitor CAPECO autoconstruccion expenditure data for volume changes. Track the Bono de Proteccion retrofit program's annual household count. Monitor SEDAPAL's critical-point remediation progress through Comptroller General audits. Watch for USGS and IGP (Peruvian Geophysical Institute) updates on strain accumulation on the Lima segment. Track INGEMMET landslide risk map revisions for Lima's watershed areas.