Chile Vehicle Theft Recovery Market Report, Size, Share and Forecast 2026–2033

Overview: 101 vehicles a day, and a recovery rate measured on 39 per cent of them

Chile recorded 36,868 vehicle thefts in 2025 — 101.0 a day. The recovery rate on insured vehicles stood at almost 45 per cent at first year of claim across full-year 2025 — the second consecutive annual increase, and still far below 2017's 54.75 per cent and 2016's 61.34 per cent. Both halves of that statement are correct, and they describe two different populations: the thefts are national and include the uninsured majority, the recovery rate is measured only on insured vehicles — 2,211,320 units, or 31.8 per cent of a fleet of 6,951,716 vehicles.

That asymmetry is the subject of this report. A country can state its theft count with two independent series agreeing to within 1.07 per cent, and simultaneously have no measured recovery rate at all for 61.28 per cent of the vehicles it loses.

Three readers take something different from that gap. An insurer pricing theft cover is setting a premium against a recovery probability it can observe only on its own book — which is why the same Mitsubishi L200 is quoted at between CLP 60,774 and CLP 540,025 a month across Chilean carriers, a 788 per cent spread on an identical asset. A municipality buying detection is buying the only recovery instrument in the country with a published output and no published cost. And a plate-reader or telematics vendor is selling into a function with no budget line of its own, because recovery in Chile is a by-product of claims handling rather than a product anyone purchases.

Two movements say more about direction than the level does. The national count has been falling: the Observatorio del Crimen Organizado at Universidad Andrés Bello puts the peak at 46,242 thefts in 2022 against 36,868 in 2025 — down 20.27 per cent in three years. The insured count has been rising over the same window, from 13,733 in 2024 to 14,276 in 2025, because coverage is growing faster than theft is falling; the rate at which insured vehicles are stolen has now held at 0.65 per cent for three consecutive years. And the recovery rate has turned: at first year of claim, full-year 2025 came in at almost 45 per cent, the second consecutive annual increase, still 9.75 points below 2017 and 16.34 points below 2016.

That last movement is the most favourable change to this report's assumptions since it was written. The base case takes recovery from 45 per cent to 51 per cent by 2033; on the 2025 print that is no longer a hypothesis about recovery but an extrapolation of a two-year trend. It is still a forecast — two years is not a decade — and the risk chapter keeps it under a named falsifier rather than treating it as settled.

What this report sizes

The headline market is gross recovery value (GRV): the commercial value of stolen insured vehicles returned to their owners in a year. It is a derived construct and the formula is stated so it can be argued with:

GRV = insured parc × theft rate per insured vehicle × recovery rate × average insured vehicle value.

2025: 2,211,320 × 0.6455% × 45% × USD 19,000 = USD 122.1 million. Three of the four inputs are published in Chile. The fourth — average insured vehicle value — is not, and is the single soft input in the model; it is banded USD 14,000–24,000 in the sensitivity and its effect is shown explicitly.

USD million. The band is USD 175.3 million wide, 87.0 per cent of the base case — the honest width when four inputs move independently and one of them has no Chilean publication behind it.

The market in its smallest units, because the aggregate hides them

USD 122.1 million is USD 334,521 of insured vehicle value returned every day, against 101.0 vehicles taken. It is 17.60 vehicles coming back for every 101.0 stolen. Divided across the insured fleet it is USD 55.22 per insured vehicle per year — the entire recovery function, the ANPR network, the insurer claims units and the police units combined, returns about the price of a tank of fuel per covered vehicle per year. And divided across insured thefts it is USD 8,553 per insured theft, which is simply 45 per cent of the USD 19,000 average and is the recovery rate restated in money. Those are the same 122.1 million divided four ways, and they are stated here because a nine-figure national aggregate conceals how small this market is at the point where any of the three readers above would actually transact.

The construct is gross recovery value, and what it therefore excludes should be stated before anyone builds a business case on it. It excludes the cost of recovering anything — no published cost per recovery exists anywhere in the Chilean record, so no margin on this market can be computed from public inputs; it values every recovered vehicle at the full average even when the vehicle comes back damaged or is written off; it counts only vehicles found within the first year of claim, so a vehicle found in year two is recorded as never found; it says nothing about the uninsured 61.28 per cent of thefts; and it does not touch the illicit parts market those unrecovered vehicles feed. A net figure would be materially smaller, and no Chilean source publishes one. GRV is used because it is the only version buildable from published inputs — three of the four exact, the fourth banded and disclosed.

Key drivers: four levers, and which way each is now moving

Every path in the table above is the product of four inputs, and each is moving independently. Showing them apart is more honest than a single 6.47 per cent, because they can fail apart:

The decomposition is worth stating because it changes what to watch. Recovered vehicles go from 6,424 to 9,052 — up 40.91 per cent; the average value of a recovered vehicle goes from USD 19,000 to USD 22,261 — up 17.17 per cent; multiplied out that is 1.651 times, which is exactly the 122.1 to 201.5 move in the paths table. On the USD 79.4 million of added value, 62.83 per cent is more vehicles recovered and 37.18 per cent is more value per vehicle. Most of the base case is therefore volume, and most of that volume is coverage: this is a penetration story wearing a recovery story's clothes, and the two are argued separately here because they fail separately.

There is a fifth movement that is not in the model and pushes against all four. Imported parts and no minimum-stock obligation mean a damaged stolen vehicle reaches a write-off quickly, and a write-off closes the recovery file whether or not the vehicle is ever found. Parts economics therefore set a ceiling on the recovery rate that no camera network can lift, which is why the base case improves recovery by six points and not by sixteen.

And the cheapest available falsifier is the lever least connected to crime. Coverage grew 4.66 per cent in 2025 against insured theft growth of 3.95 per cent, which is the only reason the theft rate held flat while the count rose. A 21.43 per cent decline in average premium is not a durable competitive position, and if issuance stalls at the 2026 renewal the denominator stops growing and the base case loses its largest contributor with nothing having happened to crime at all.

What each lever is worth at 2033, one at a time

GRV is a product of four inputs, so holding three at base and moving the fourth traces the exposure directly. These are elasticities, not scenarios: they do not sum, and the USD 175.3 million band in the paths table is what happens when the four move together.

USD million. Two readings follow, and the second is the one to act on. The recovery lever — the assumption this chapter corrects — is the second-largest single exposure: USD 23.70 million at the terminal year, 11.76 per cent of the base case, so a two-year trend that reverses costs a ninth of the forecast with nothing else changing. And the largest downside is not crime at all. Coverage falling back to today's 31.81 per cent instead of rising to 38 removes USD 31.82 million, more than any recovery failure. The theft rate is the only lever whose upside exceeds every downside, and it is the one no participant in this market controls — which is why the panel above watches issuance, not policing.

Five findings

1. The recovery rate has fallen 26.64 per cent in nine years while the fleet grew. 61.34 per cent in 2016 to 45 per cent at full-year 2025, a trough reached in 2023–24 from which the rate has now risen for two consecutive years. At the 2016 rate, the same 14,276 thefts would have returned 8,757 vehicles instead of 6,424 — 2,333 more, worth USD 44.3 million.
2. The ANPR network carries 28.92 per cent of all insured recoveries. 1,858 of roughly 6,424 vehicles came back through a plate-reader system that the insurance association built and gives to municipalities free of charge, operating in more than 260 communes. It is the only recovery instrument in Chile with a published output — and no published cost.
3. The uninsured majority is unmeasured. 22,592 of 2025's thefts sat outside the insured universe, and outside the only statistic that says whether a vehicle comes back. The 45 per cent is not a national recovery rate; it is a rate for the third of the fleet that bought a policy.
4. Fifty-five thousand insured vehicles are still missing. Unrecovered over the last twelve years, worth USD 1.045 billion at the model's average value. Recovery is measured at first year of claim only, so the stock of never-recovered vehicles accumulates silently.
5. The count can rise while the rate holds — and that is a denominator effect. Insured thefts rose 3.95 per cent from 2024 to 2025, and the theft rate still held at 0.65 per cent, because the insured parc grew 4.66 per cent. Growth in insurance coverage is currently masking growth in crime.

Market segmentation

Five cuts, each reconciling to the same 36,868 thefts.

Cut two carries a correction. The trade press reported the Metropolitan Region as 65 per cent of national thefts; the regional table it published sums exactly to 31,052 and puts the region at 61.97 per cent. The component figures are used here, not the summary.

What this report is not

It is not a national recovery rate. It is not a claim that the ANPR network causes all recovery — the 1,858 are detections that ended in a recovery, and some would have happened without the camera. It is not a unit forecast of telematics subscriptions, for which no Chilean installed-base series exists. And it is not a model of the illicit market in stolen parts, which competes directly with the legitimate collision repair parts supply, is the destination of the unrecovered 55 per cent, and is one that no Chilean authority measures in currency.

Table des matières

1. Overview: the count, the rate, and the third of the fleet that is measured
2. Competitive landscape: the recovery function has no buyer
3. Value chain: seven stages and the four-hour window
4. Investment: what a recovery is worth, and who pays for it
5. Technology: plates, pings and the cloning problem
6. Market risk: five risks, five falsifiers
7. Regulatory: article 443, bill 17352-07, and the missing mandate

Paysage concurrentiel

The recovery function has no buyer

Chile does not have a vehicle recovery industry in the ordinary sense. It has a recovery function, assembled out of four actors who are not trading with each other: an insurance association that funds the detection technology and gives it away, 260-plus municipalities that operate it, a police division that holds the stolen-vehicle register, and a vehicle owner who receives the benefit without ever paying for the instrument that produced it.

That structure has one consequence a normal market does not: there is no price for a recovery. Nobody buys one. The instrument is donated, the operator is salaried, the enforcer is budgeted, and the beneficiary is paid out in kind. Any valuation of this market therefore has to be built from the value returned, not from a transaction — which is what the Overview does.

The one scaled instrument, and its numbers

The Asociación de Aseguradores built a plate-recognition system with a Chilean developer and began installing it under public-private agreements with municipalities in 2021. It is given to councils at no charge and operated with the Asociación Chilena de Municipalidades, Vespucio Norte and Carabineros.

• Coverage: more than 260 communes.
• Recoveries in 2025: 1,858 vehicles with an encargo por robo — 28.92 per cent of all insured recoveries in the year.
• Cumulative since 2021: more than 5,200 vehicles, of which 2025 alone accounts for 35.73 per cent.
• Throughput: each mobile unit processes more than 150 plates an hour.
• Concentration of output: Maipú 119, Quilicura 95, Pudahuel 89 recovered in 2025 — three communes in Santiago accounting for 16.31 per cent of the network’s national output.

That last line is the competitive fact that matters. The network’s productivity is not evenly distributed across the 260 communes; it is concentrated where patrol density, plate volume and theft density coincide. Coverage is national. Yield is not.

The telematics layer, which is a different business

Above the plate network sits a commercial telematics sector that sells monitoring, not recovery. It is where the revenue in this market actually transacts.

GPS Chile is the clearest domestic benchmark: founded in 2001, more than 20 years operating, and by its own count more than 40,000 vehicles monitored across mining, construction, logistics, operational leasing, agricultural machinery, cash transport, passenger transport, fuel and waste. It is the largest single documented installed base in the country and the only one that puts a number in public.

Wisetrack Corp runs the other end of the market: mining, heavy machinery, energy, large carriers, forestry, refrigerated distribution, dangerous cargo, cash transport. Its product set is explicitly anti-crime — a kit marketed for protecting cargo, vehicles and drivers — plus fatigue and distraction detection, which is safety telematics rather than recovery telematics.

Multinationals (Geotab, Webfleet and others) sit above both, selling fleet platforms where tracking is a feature rather than the product.

The structural point: the telematics sector and the recovery function overlap only at the moment of a theft. A monitoring subscription sold on fuel efficiency becomes a recovery instrument once, if ever. That is why the market has a large installed base and a small number of recoveries attributed to it — and why no participant publishes a recovery rate by vendor.

Enforcement capacity is the binding constraint on the other side

Carabineros reports its own numbers through the Glosa N°15 mechanism attached to its budget. During the anti-carjacking plan run in the Metropolitan Region the force recorded 2,086 detentions, more than 31,000 vehicle controls, 640 workshop inspections and more than 600,000 plate reads. The last figure is the one to hold onto: a single police operation generated 600,000 plate reads — against the 5,200 recoveries the ANPR network has produced since 2021. Reads are abundant; recoveries are not. Detection is not the scarce input. Interception is.

Structure verdict

Moderately consolidated at the technology layer, monopsonistic at the funding layer, fragmented at the operating layer. One association funds detection nationally; a few vendors supply hardware and platforms; 260 municipalities operate; one police division holds the register. No participant has an incentive to publish a cost per recovery, which is why none does.

The moat is not the camera. Cameras are commodity. The moat is the encargo por robo register and the plate database bound to it — a national list of wanted vehicles that a private entrant cannot assemble, and that a municipality cannot use without the association’s data.

The funder is far more concentrated than the operator

The actor paying for detection is a four-company oligopoly, and that is the structural fact the table above hides. In 2025 the industry issued 4,182,519 voluntary vehicle policies, a record, and four insurers wrote 77.9 per cent of them: Bci Seguros 1,498,372 (35.8 per cent), HDI 913,944 (21.9 per cent), Suramericana 467,042 (11.2 per cent) and Zurich Chile 373,997 (9.0 per cent).

Two consequences follow. A programme funded out of member premium can be decided by four balance sheets, which makes the plate network faster to move than a statutory system and considerably more fragile than one, because it has no floor under it. And a four-firm funder facing more than 1,800 models and 72 brands from 28 countries — the association’s own count — cannot price theft risk at model level by hand. The eightfold premium spread documented in the Investment chapter is a consequence of that arithmetic, not an inefficiency anybody chose.

Below the funder the structure inverts: 260-plus municipalities, no contract value, no exclusivity, no published output per commune. Concentration sits at the top of this market and fragmentation at the bottom, and the benefit lands on an owner who pays for none of it. That asymmetry is why the adjacent commercial layers — the fleet management platforms and the fuel card and fleet payment businesses — hold a customer relationship this market has never had: they bill the vehicle owner directly for a service the owner consciously buys, which is precisely what no participant in vehicle recovery can do.

Chaîne de valeur

Seven stages, and a four-hour window

A stolen vehicle’s chance of coming back is decided long before anyone looks for it. The chain below runs from the offence to the disposition, and the loss almost always occurs at stage two.

1. Theft. Increasingly violent and increasingly organised. OCRIT at Universidad Andrés Bello identifies at least twelve transnational criminal organisations operating in Chile and names motor-vehicle theft as one of their financing lines. The mode has shifted from stealing parked cars for parts to taking moving vehicles at gunpoint, then dismantling or exporting them.
2. Report and encargo. The owner files a denuncia; Carabineros enters an encargo por robo into the SEBV register. Nothing downstream can happen until this entry exists — the plate reader has nothing to match against.
3. Detection. A mobile reader matches a plate against the register, or a telematics unit reports a position, or a patrol identifies the vehicle visually.
4. Interception. A municipal patrol or Carabineros stops the vehicle. This is where detection converts to recovery or does not.
5. Custody. The vehicle goes to a corralón. Storage, peritaje and administrative release follow, and the owner bears a delay that has nothing to do with whether the car was found.
6. Identification. Is this the vehicle on the encargo, or a clone carrying its plates? VIN verification decides whether a recovery is a return or a second loss.
7. Return or total loss. Repair and return, or write-off and salvage — with the salvage value flowing back against the claim.

The clock is legally load-bearing

Chilean policies condition theft cover on immediate notification, and insurers enforce it. In November 2025 the Corte de Apelaciones de Santiago upheld an order against Renta Nacional Compañía de Seguros Generales for 189 UF — about CLP 7.49 million — after the insurer refused a theft claim on the ground that the police report had been filed more than fifteen hours after the event. The court held that the insured had reported as soon as he could establish the vehicle was gone, and that the declaration was made the same day.

Two things follow. A fifteen-hour delay was enough for an insurer to deny a claim outright, so the reporting step is not administrative — it is the point at which the loss becomes either a recovery attempt or a dispute. And the case shows the failure mode this report’s central number hides: in a market where 45 per cent of insured thefts are recovered, some of the 55 per cent that are not were never properly in the system.

Stage six is where the count quietly double-counts

Plate cloning is the reason a plate-matching network alone cannot finish the job. A cloned vehicle carries a legitimate plate from an identical model; the reader flags it, the stop happens, and the VIN says it is not the stolen car. Chile’s insurance association points to a specific enabler it says it helped close: the ability to incorporate a company in a single day, which let paper companies launder ownership of stolen and cloned vehicles. A legal change followed. The report treats that as the association’s account of its own lobbying, not as an independently verified causal claim.

Where the value actually leaks

The cumulative stock is the number to hold: roughly 55,000 insured vehicles stolen over the last twelve years remain unlocated, worth USD 1.045 billion at this report’s average-value assumption. Recovery is reported at first year of claim only. A vehicle found in year two is recorded as a failure in the year it was taken and never reappears in the statistic — so the 46 per cent is a first-year rate being read as a total rate, and the stock of never-recovered vehicles accumulates without appearing in any headline.

The geography of the leak points the wrong way

62.0 per cent of encargos are registered in the Metropolitan Region and 9.10 per cent in Valparaíso; Aysén and Magallanes together account for 28 cases in ten months. The ANPR network’s output is similarly concentrated. But export routes leave the country, and the association reports recoveries made abroad. The detection footprint is densest where the thefts are, not where the vehicles go — which is a structural limit on how far patrol-mounted readers can push the recovery rate, and the reason the base case does not assume the 2016 rate returns.

Stage seven is decided by parts, not by damage

Whether a recovered vehicle is repaired or written off is the last place this chain leaks value, and in Chile it is a supply question rather than a damage question. The association notes that most replacement parts must be imported and that no legal obligation exists to hold minimum stock in the country, and insurers name parts availability, cost and replenishment time among the sector’s structural problems. A recovered vehicle waiting on an imported panel sits in a corralón for weeks, and at some point the storage and the delay cost more than the write-off.

That feeds directly back into the recovery statistic this report is built on. A write-off closes a file that a later recovery might have reopened, so the total-loss decision moves with the import cycle as well as with damage — which is a mechanism, not a hypothesis, and the risk chapter carries it as one. The repair-side economics of the destination market sit in the automotive collision repair series; the freight chain that carries the parts sits in the transport and logistics series.

The claims operation this function lives inside

The insured system handled more than 467,000 vehicle claims in 2025 — about 40,000 a month — and theft was 3.06 per cent of them. The recovery function is therefore a three-per-cent activity inside a claims operation thirty times its size, and it is judged by a metric, first-year recovery, that nothing else in that operation uses.

Service friction is measured separately and barely at all: Sernac received 2,172 vehicle-insurance complaints in the second half of 2025, roughly 1 per cent of the cases the industry handled in the period, and 50.5 per cent were resolved in the customer’s favour. That is the association’s own account of its service record, and it is the nearest thing in the public record to an outside check on the notification-and-denial behaviour the November 2025 appellate ruling exposed at stage two.

Activité d'investissement

What a recovery is worth, and who pays for it

There is no disclosed transaction price anywhere in this market. No insurer publishes what a recovered vehicle is worth to it, no municipality publishes what the reader costs to run, and no vendor publishes a price per recovery. What follows is built from the two things that are public: the number of recoveries and the value of the vehicles.

The fourth line is the one that changes how a reader should think about this market. The unrecovered stock is eight and a half times the annual value returned. This is not a business with a growth problem; it is a business with a backlog.

The cost side is the disclosure gap

The network that produced 1,858 recoveries and USD 35.30 million of returned vehicle value is the only recovery instrument in Chile with a published output. Its cost is not published — not the hardware, not the operating agreements with 260 municipalities, not the data platform, not the staff. That means the single most useful number in this market, cost per recovery, cannot be computed from the public record.

A participant who published it would move this market. Until someone does, the only defensible statement is the ratio a buyer can actually act on: at USD 19,000 per vehicle, the network’s break-even cost per recovery is high enough that almost any plausible operating budget clears it — and the reason the programme survives is not that its return is uncertain, but that nobody has to prove it.

The denominator is doing the work in the headline

Insured thefts rose 3.95 per cent from 2024 (13,733) to 2025 (14,276). The theft rate did not move: 0.65 per cent in both years. It held because the insured parc grew 4.66 per cent, from roughly 2.11 million vehicles to 2,211,320.

That is a warning about the series everyone quotes. A rising count with a flat rate is what coverage growth looks like; a flat count with a rising rate is what a deteriorating market looks like; and Chile is currently producing the first. Anyone reading the annual count as a trend in crime will mis-time this market by a year or two.

Price is not yet risk-based, and that is the investable anomaly

ComparaOnline priced the five most-stolen models across Chilean insurers, on 2015 vehicles, private use, CLP 5 UF deductible. The monthly premium for the same vehicle varies enormously between companies:

An eightfold spread on an identical vehicle is not a pricing disagreement about risk; it is evidence that the market has no shared model of theft risk per model-year. Whoever builds one — using the association’s own stolen-model ranking, which already shows the top ten models carrying about 30 per cent of insured thefts — prices the book better than the market currently does. That is the clearest commercial opening this report identifies, and it sits upstream of recovery: better-priced theft cover is what funds the recovery instrument.

Where capital is actually going

No disclosed M&A. No transaction in Chilean vehicle recovery or telematics was disclosed in the period reviewed, and the sector’s capital comes from premium loading rather than equity. The investment story is therefore not consolidation; it is instrumentation — readers, telematics units, and the data layer that binds a plate to an encargo.

Three things are worth underwriting, in order of evidence:

1. Fixed detection where mobile detection has saturated. The network is patrol-mounted and its yield is concentrated in three Santiago communes (16.31 per cent of national output). Fixed gantries on the exit corridors solve the problem the value chain identifies — that detection is dense where thefts happen and thin where vehicles leave.
2. Cross-border repatriation. The association already reports recoveries abroad. Nobody has published a unit economics for it.
3. Model-level risk pricing. The eightfold premium spread is the anomaly a data business can close.

Sensitivity, stated plainly

Average insured vehicle value is the model’s one unpublished input, so its effect is shown rather than buried: a USD 5,000 move on a USD 19,000 base is 26.32 per cent of the entire 2025 market. A reader who believes the average stolen Chilean vehicle is worth USD 14,000 should read USD 90.0 million where the report says 122.1, and a reader who believes it is USD 24,000 should read USD 154.3 million. The direction of the forecast is unaffected; the level is not.

The pool that funds recovery, and four numbers that do not tie

Every recovery in this market is paid for out of vehicle premium, so the size and direction of that pool matters more than any single vendor’s prospects. In 2025 the industry wrote 4,182,519 voluntary vehicle policies, a record, 10.79 per cent above the 3,775,047 of 2024, for UF 33 million of premium — 26 per cent of the general-insurance premium. At the same time the average premium was falling: Diario Financiero’s 14 May 2026 insurance supplement, citing the association, puts it at about UF 11 against UF 14 in 2024 — a 21.43 per cent decline. A record number of policies bought at a falling average price is the shape of a market competing on distribution rather than on risk, and it is the commercial background to every model-level pricing gap in this chapter.

Four published figures describe that one segment, and they do not reconcile:

The 1.89 ratio is the problem. A single-vehicle annual policy cannot produce 1.89 policies per insured vehicle, so either the policy count carries renewals, part-year certificates and endorsements inside the year, or the parc is a point-in-time stock being compared with a full-year flow. Neither explanation is published, and the report does not invent one. What it does say is that the report uses the parc, because the parc is the model’s denominator, and that every affordability argument about the recovery instrument depends on which unit is used: at UF 14.92 per insured vehicle the funding pool per unit is nearly twice what it is at UF 7.89 per policy.

The penetration figure carries a smaller version of the same problem. The association states 32.3 per cent of the parc for 2025; its own two published numbers, 2,211,320 over 6,951,716, divide to 31.81 per cent. Half a point changes nothing in this model, but a series whose published percentage and published ratio disagree is a series a reader should handle rather than quote.

Value by region, in money rather than counts

The Overview segments thefts. This is the same split carried into currency, using the report’s average insured value and recovery rate:

State the assumption plainly: this applies one average vehicle value and one recovery rate to every region. Both are almost certainly false in practice — Antofagasta’s mix is mining-linked pickups and the Hilux carries the lowest recovery rate in the market at 31.4 per cent — so the table is a count-weighted allocation of value, not an independent regional measurement. It is published because Chile has no regional recovery series at all, and a count-weighted allocation is the only regional value figure that can be built from evidence.

The base case, year by year

Compounded at 6.47 per cent from 122.06. The 2033 print is 201.6 against the published 201.5 — rounding of the published CAGR, not a second forecast.

The Hilux number: recovery is an asset characteristic

The association publishes one model-level recovery rate, and it belongs to the highest-volume model in the country: the Toyota Hilux at 31.4 per cent over the twelve months to mid-2025 — 13.6 points below the market and the lowest in its own table. Toyota was the most-stolen brand in the same window at 13.75 per cent of insured thefts, with the Hilux, RAV4 and Land Cruiser together above 2,250 units. A pickup worth more dismantled than returned is the clearest evidence in this market that recovery probability is an asset characteristic rather than a system constant — and it is exactly the input a model-level risk price would need.

Two more facts from the same series. 81 per cent of thefts happen with the vehicle parked, which is why plate reading works at all: the vehicle has to move eventually, and the four-hour window at stage four is where that movement is caught or lost. And the named destinations are dismantling, resale under a lien (venta con prenda) and movement abroad toward Bolivia and Venezuela — three outcomes with three different recovery probabilities that no Chilean series separates.

The data layer, not the camera, is what would close this. Building a theft-risk price per model-year across 1,800 models and 72 brands is an underwriting problem before it is a hardware one, which is the subject of the AI in insurance series.

Technologie et innovation

Two different questions, routinely confused

Chilean recovery runs on two technologies that answer different questions and are frequently discussed as if they were one.

• A plate reader answers “is this number wanted?” It knows nothing about the vehicle if the plate is false, and it cannot find a vehicle that is not driving past it.
• A telematics unit answers “where is this vehicle?” It knows nothing about whether the plate has been swapped, and it can only answer for vehicles that carry it.

The ANPR network covers the first question nationally and the second not at all. The telematics sector covers the second for its subscribers and the first only through integration it does not control. No Chilean participant offers both at national scale.

The plate layer, measured

The system is mobile, mounted on municipal patrol vehicles, and it works while the unit drives: more than 150 plates an hour per unit, matched in seconds against the encargo por robo register, with an alert routed to the operator. Six units were fitted to Santiago’s municipal security vehicles in a single 2023 deployment, and the first hit came within thirty minutes of the first patrol.

Three design consequences follow, and each is a ceiling:

1. Coverage is a function of driving, not of geography. A commune that stops patrolling stops detecting. The association’s own commentary on its 2025 ranking emphasises “territorial coverage and operational continuity” — which is an acknowledgement that the network’s output is an operating variable, not an installed asset.
2. Yield concentrates. Maipú at 119, Quilicura at 95 and Pudahuel at 89 recovered vehicles in 2025 are not three times better equipped than their neighbours; they are where plate volume, theft density and patrol hours intersect.
3. Cloning is a hard stop. A cloned vehicle carrying a legitimate plate from an identical model will be flagged and stopped, and the VIN will say it is not the stolen car. The technology that catches clones is the same technology that generates false positives on them, and the resolution is a physical VIN inspection that no camera performs.

The telematics layer, and why its installed base does not translate into recoveries

The largest documented installed base in Chile is GPS Chile, with more than 40,000 vehicles monitored after 20-plus years and over 1,500 clients across mining, construction, logistics, operational leasing, agricultural machinery, cash transport, passenger transport, fuel distribution and waste. Wisetrack Corp serves the heavy end — mining, energy, large carriers, forestry, dangerous cargo, cash transport — and has extended into fatigue and distraction detection, which is driver-safety telematics rather than recovery telematics.

That base is real and it is not a recovery base. A monitoring subscription sold on fuel efficiency, utilisation and preventive maintenance reports position continuously; whether that position is acted on in a theft depends on a protocol between the vendor, the client and Carabineros that nobody publishes. The consequence for this report is blunt: there is no Chilean series of recoveries attributable to telematics, so the 28.92 per cent attributed to the plate network is the only channel share that can be computed at all.

The connectivity substrate is not the constraint

Chile closed 2025 with 9,698,786 5G connections and 22.7 million mobile-internet accesses (SUBTEL), on a population of roughly 20 million. Connectivity does not limit what a telematics unit can do in this country. Cost, installation and the absence of a reason to fit one to a privately owned car are the limits.

What is missing, stated as technology rather than policy

• No national fixed detection network. Detection is patrol-mounted, so it follows patrol routes. The value chain’s central inefficiency — detection density following theft density rather than exit routes — is a technology-deployment fact, not a funding one.
• No VIN-level verification at point of resale. The cloning problem is solved at the transaction, not at the roadside, and Chile’s resale process is not instrumented to solve it.
• No OEM data-sharing arrangement. Connected vehicles built in the last decade know where they are. No agreement in the public record obliges a manufacturer to surrender that position to an insurer or a police force at the moment of a theft — which is why aftermarket installation remains the only route to a telematics recovery.
• No recovery analytics layer. The register knows a vehicle is wanted; the reader knows it passed; nobody publishes how long the median recovery takes, from which detection type, in which commune. The absence of that dataset is why this report’s channel segmentation stops at one channel.

The technology verdict

Growth-stage, moderate innovation, low patent activity. The core technologies are mature and commodity-priced; the value sits in the integration and the register, not the hardware. The next real step change is not a better camera — it is binding telematics position to the encargo register in real time, which would let a vehicle that never passes a patrol car still be found.

The asset, not the camera — where the connected vehicle already answers

Every vehicle built in the last decade with a factory telematics module already knows where it is. The constraint is that nobody in Chile has a right to that position at the moment of a theft: no published arrangement obliges a manufacturer to surrender it to an insurer or a police force. The technical substrate is ordinary — automotive embedded systems are standard content in a new vehicle rather than a premium option — and the commercial substrate is absent, which is why aftermarket installation remains the only route to a telematics recovery in this country.

Two structural facts explain why no vendor bridges the gap. The parc spans more than 1,800 models and 72 brands from 28 countries — the association’s own count — so the aftermarket is the only common denominator across it. And a plate is the only identifier that all 6,951,716 vehicles already carry, which is precisely why the plate network reached national scale and the telematics layer did not. A vendor that bound a position to the encargo register in real time would find vehicles that never pass a patrol car, which is the one thing patrol-mounted reading structurally cannot do.

There is a second-order technology risk in the same facts. A parc with that many models and that many imported parts is a parc in which repair economics diverge sharply by model, and the write-off decision — the one that closes a recovery file — is taken on parts availability rather than on damage. That is a data problem before it is a hardware one, and no Chilean series measures it.

Risque de marché

Five risks, each with the number that would settle it

This is a market whose forecast rests on four inputs, three of which are published and one of which is not. The risks below are ordered by how much of the 2033 base case each one can move, and each carries a falsifier — a specific observation that would tell a reader the risk has crystallised rather than an adjective describing it.

Risk 1 — Insurance penetration does not move, again

The insured share of the fleet was 32.3 per cent at end-2025 on a parc of 6,951,716 vehicles — 31.81 per cent if the association’s own two published numbers are divided — and the ten-year average is also 32 per cent. That is the single most important fact in the forecast, because the base case has penetration reaching 38 per cent by 2033 on the strength of one year’s 0.9-point gain. If penetration instead behaves as it has for a decade, the base case collapses to the low case: USD 131.9 million in 2033 and a 0.98 per cent CAGR, not 201.5 million at 6.47 per cent.

Falsifier: two consecutive years with the insured share at or above 34 per cent. One year is noise; two is a trend. The series is published annually by the association and is not revised.

Risk 2 — The recovery rate is partly a settlement convention

Recovery is measured at first year of claim. An insurer that settles a theft claim faster, or writes a vehicle off earlier, converts a possible late recovery into a closed file — and the rate moves without a single vehicle changing hands. The 26.64 per cent decline since 2016 is consistent with a genuine deterioration (the association points to dismantling and export) and with a change in settlement practice. This report cannot separate the two, and says so rather than picking the more flattering reading.

Falsifier: the recovery rate and the total-loss ratio on theft claims moving in the same direction across two consecutive annual reports. If write-offs rise while recoveries fall, part of the decline is bookkeeping.

Risk 3 — The uninsured majority enters the measured universe

61.28 per cent of stolen vehicles generate no claim and therefore no recovery statistic. If Chile extends compulsory motor cover to the vehicle — its current mandatory policy is a personal-accident cover attached to the circulation permit, not vehicle cover — then the denominator of this market roughly doubles in a single year.

That is a growth event and a distortion at the same time. The market would grow by arithmetic, not by performance, and every historical comparison in this report would break.

Falsifier: a compulsory-vehicle-insurance bill reaching its second constitutional trámite in either chamber. First-trámite motions on this subject are frequent and almost none advance.

Risk 4 — Substitution from the vehicle itself

Every connected vehicle built in the last decade knows its own position. The moment a Chilean insurer offers a premium discount tied to a factory-fitted system, the aftermarket install becomes optional and the telematics layer restructures around manufacturers rather than local vendors. The association’s plate network is less exposed — it finds vehicles regardless of what they carry — but the commercial layer above it is fully exposed.

Falsifier: a Chilean insurer advertising a theft premium differentiated by factory telematics rather than by model. None appears in the record reviewed; the current differentiation is entirely by model and year.

Risk 5 — The asset moves faster than detection

OCRIT identifies at least twelve transnational criminal organisations active in Chile and names vehicle theft as a financing line; the association notes vehicles leaving the country or being exchanged for drugs. Detection is patrol-mounted and concentrated where thefts occur — 61.97 per cent of encargos are in the Metropolitan Region — while the exit routes are not.

If export accelerates, recoveries fall even if detection improves, because the vehicle is no longer in the country to be found. The base case’s recovery improvement from 45 to 51 per cent assumes detection gains outrun export. That is an assumption, not an observation.

Falsifier: the association reporting cross-border recoveries as a rising share of total recoveries. It already reports recovering vehicles abroad; it does not publish the share.

The definitional risk that sits under all five

Three Chilean series describe vehicle theft and none describes the same population. A market this dependent on the theft count is therefore exposed to a counting change as much as to a crime change:

The third series is not a fifth of the problem. It is a different problem: a count concentrated 84.4 per cent in the Metropolitan Region, against 62.0 per cent for all encargos, is the signature of a violent-carjacking count rather than an all-theft count. Any reader who divides one by the other will manufacture a trend that does not exist.

The rate has turned, and that moves which risk is live

Risk 2 treats part of the recovery rate as possibly a settlement convention rather than a physical outcome. The 2025 print narrows the question without closing it. Full-year 2025 came in at almost 45 per cent at first year of claim — the second consecutive annual increase — so the 2016–2023 deterioration has stopped and partly reversed, while the level remains 9.75 points below 2017 and 16.34 below 2016. What is unexplained is no longer the direction. It is the gap.

The mechanism the association itself supplies is parts. With most replacement parts imported and no legal obligation to hold minimum stock in the country, the repair-or-write-off decision moves with the import cycle as much as with damage — and a write-off closes a file that a later recovery would have reopened. That is a testable mechanism rather than an adjective, so it gets its own falsifier.

Additional falsifier for Risk 2: a sustained fall in imported-parts cost or replenishment time — the constraint the association states itself — without a corresponding gain in the first-year recovery rate. If parts get cheaper and the rate does not move, the write-off story is not doing the work and the decline was physical.

And the risk the turn creates rather than removes: a rate that has risen for two years inside a forecast that assumes it keeps rising is now dependent on the continuation. If 2026 prints flat at 45 per cent, the low case’s 45 per cent recovery assumption stops being a floor and becomes the central case, and the base case loses the recovery contribution documented in the Investment chapter.

Paysage réglementaire

Article 443, bill 17352-07, and a mandate that does not exist

Chile regulates vehicle theft heavily and vehicle recovery barely at all. The criminal law is specific, the police register is national, and there is no instrument that obliges anyone to fit, fund or operate a recovery device.

The criminal layer

Theft of a motor vehicle is addressed in the second paragraph of article 443 of the Código Penal, which carries a heavier penalty than theft of property from a vehicle or from a public place — the distinction the code draws between taking the car and taking what is inside it. Receiving or handling a stolen vehicle sits in the receptación provisions.

In January 2025 a cross-party motion, bill 17352-07, proposed two changes aimed squarely at this market: postponing the point at which a substituted sentence can be served, and extending the flagrancia window for robbery and receiving of motor vehicles — the period during which a suspect can be treated as caught in the act. It was filed by deputies Alessandri, Leiva and Longton, entered its first constitutional trámite in the Chamber of Deputies, and carried no urgency.

The motion’s own background is the third counting series this report had to reconcile: 7,758 denuncias in 2024, against 10,413 in 2023 and 12,524 in 2022, and 6,100 in 2020 and 6,294 in 2021. It also documents the geographic shift — Valparaíso roughly doubling from 120 cases in 2020 to 277 in 2024, Biobío from 65 to 193 with a peak of 312 in 2023. Those are the numbers a legislator works from, and they are roughly a fifth of the encargo count in the same period. Chile’s lawmakers and Chile’s insurers are legislating and pricing against different definitions of the same crime, and nothing in the record reconciles them.

The register, and why it is the real infrastructure

The operational keystone is the encargo por robo: an entry in Carabineros’ SEBV register that a vehicle is wanted. Every downstream technology consumes it. A plate reader with no register matches nothing; a telematics position with no register is a location, not a recovery.

Carabineros reports its recovery and search activity through Glosa N°15, the transparency mechanism attached to its annual budget. In 2025 the institution executed CLP 1,647,455,185,496, of which roughly 77 per cent was personnel and about 15 per cent goods and services — a ratio that tells you the force’s constraint is people and hours, not equipment. The same report documents the search-and-locate function for persons and vehicles within the SEBV.

The insurance layer

Chilean theft cover is conditioned on immediate notification, and the condition is enforced. The November 2025 Corte de Apelaciones ruling against Renta Nacional — 189 UF, about CLP 7.49 million, for a claim denied on a fifteen-hour reporting delay — establishes that the reporting step is a legal threshold with money attached, not an administrative formality. It also means the population of “unrecovered” vehicles includes some that were never properly entered.

Pricing is regulated at the conduct level, not the risk level. Nothing in the framework requires an insurer to differentiate premium by theft exposure, which is why the identical Mitsubishi L200 carries a monthly premium between CLP 60,774 and CLP 540,025 across companies — a 788 per cent spread.

The corporate layer, and the loophole the industry says it closed

The association identifies one legal change it says it drove: closing the use of same-day company incorporation to launder ownership of stolen and cloned vehicles. “Before, there were paper companies and they cloned the cars,” its vehicle-theft prevention manager told La Tercera, describing a change driven from the insurance world. This report records it as the association’s account of its own lobbying. It is not an independently verified causal finding, and the recovery-rate line does not turn on it.

What does not exist

• No mandatory recovery device. No Chilean instrument in the record reviewed requires a private vehicle to carry a tracking unit, an immobiliser or a plate-readable identity.
• No national fixed detection mandate. The ANPR network is voluntary, insurer-funded and municipal. Its 260-commune footprint is an accumulation of agreements, not a statutory network.
• No compulsory vehicle cover. Chile’s mandatory policy attached to the circulation permit is a personal-accident cover; it does not insure the vehicle, which is why 68 per cent of the fleet has no theft exposure to transfer and therefore no claim to recover.
• No VIN verification at resale. The cloning problem is a transaction problem and the transaction is not instrumented.
• No published data-retention rule for plate reads. How long plate data is held, who may query it and when it is deleted does not appear in the record reviewed — a gap that becomes material the moment the network scales.

The regulatory verdict

Moderate complexity, compliance-driven, with the compliance sitting on the insurer rather than the technology. Nobody certifies a plate reader. The approvals that matter are municipal (permission to mount and operate), contractual (the association’s data terms) and prudential (the insurer’s claims conduct). The absence of a mandate is simultaneously this market’s weakness — no floor under demand — and its opportunity: a single compulsory instrument, from VIN verification at resale to mandatory cover, would reprice the entire chain in one legislative cycle.