Maquila: one regime, several different businesses

≈ 3 min read · updated Jul 2026

Everyone quotes maquila's 1%, but that tax is the same for auto parts, garments, plastics, food and services, and each profile calls for a different location. The proposed study groups the profiles and scores them: which niche to enter and where to set up.

Maquila by profile · clustering the regime
The cluster map: which profiles coexist inside maquila
How to read it: on the left, each dot is an operation and each halo a group of similar operations · on the right, what defines each group, which one grows and who each one suits
A · Each dot is a maquila operation vertical: employment intensity · horizontal: source of value high low ← more imported inputs more local value added → Labor-intensive garment work Border auto parts Asunción services empty niche in Mexico and Brazil, not here Food with local inputs B · What defines it, who fits trait · who it suits · which grows Border auto parts growing ↑ imported inputs, big plants fits if: capital, few people Garments stable → high labor per unit exported fits if: labor is your edge Services (BPO) growing ↑ value added almost all local fits if: talent, no trucks Food (empty) open to enter in Mexico and Brazil, not here enter early: needs suppliers
Border auto parts Labor-intensive garment work Asunción services Empty niche (dotted) Halo: group formed by the algorithm
  • Clustering · k-means / hierarchical
  • Dimensionality reduction · 2 axes
  • MILP · installation score
In simple terms: each dot is an operation in the regime and the halos are the groups the algorithm forms by joining similar operations; the dotted outline marks a profile that exists in other countries and not yet here. Conceptual map of the method: the real groups, which one grows and the score that ranks where to enter come from running the analysis on the regime's data; the score comes from an integer optimization.
Installation score · example weights
The score table: how each location stands for setting up export manufacturing
How to read it: on the left, five factors with their weight and each location's position, more filled dots is a stronger position · on the right, the score those weights produce · Mexico via IMMEX, Honduras and Costa Rica via free zones
A · Five factors, position per location example weight · ordinal position 1-3 Paraguay Mexico Honduras Costa Rica Regime tax burden · 30%Relative labor cost · 20%Export logistics · 20%Industrial energy · 15%Framework stability · 15% waterway Waterway: river freight lowers the cost per ton; the extra transit days get planned for. B · Example composite score higher = stronger with these weights weights change with your product 01 Paraguay · maquila 1% 9.3 02 Honduras · free zone 8.5 03 Mexico · IMMEX 7.3 04 Costa Rica · free zone 6.5 scale 0-10 · position × weight Built with your numbers, the model returns the order for your product.
Filled dot: position on the factor, 1 to 3 Paraguay · highlighted column and row Mexico · Honduras · Costa Rica Waterway · cost-per-ton edge
  • Ordinal reading · no borrowed figures
  • Example weights · sum to 100%
  • MILP · score with your numbers
In plain terms: each location adds up by its position on each factor and that factor's weight; with these weights, which reward cost, Paraguay comes out first and the waterway brings the cost-per-ton edge in logistics. An illustrative table of the method: the position per factor is qualitative and ordinal, with no figures from other countries, and the only real number is the 1% tax (Law 1064/1997). The model we propose is built with the investor's numbers, their volume, input mix and destination, and weights, score and order come out of that. If your product rewards transit time or a trade already built, the weights change and so does the order; that sensitivity is part of the deliverable.

A company under maquila, Paraguay’s export assembly regime, produces for export and pays a single 1% tax on national value added (Law 1064/1997), with programs through the MIC and CNIME.

The 1% is the number everyone quotes, and it is the same for auto parts, garments or services. What changes the outcome is the niche and the location, and that part almost never gets discussed.

From map to score

The installation score has not been run yet; these are the inputs the study needs and what it returns.

The data: CNIME program records, central bank export series by profile, INE labor costs and freight rates by corridor.

The technique: clustering to build the profiles and mixed integer linear programming that picks niche and location subject to the investor’s capital, headcount and logistics.

The decision: which niche to enter and where to set up, with the cost of each discarded alternative in plain sight.

What the study will answer, concretely:

  • Which profile grows fastest and why.
  • Which niches are empty: profiles that exist in Mexico or Brazil and not here.
  • What drives location: why auto parts goes to the border and services to Asunción.
  • What each group lacks to scale: logistics, trained people, suppliers.

If your constraints change, the answer changes: the model we propose is built to recompute by scenario.

The exercise: garments here or in the region

A hypothetical exercise, declared as such: a foreign investor compares producing garments under maquila in Paraguay against the region’s reference schemes: Mexico’s IMMEX and the free zones of Honduras and Costa Rica. The comparison is qualitative and declared, and the only real number is the 1% tax already cited above.

The regional picture: Honduras and Mexico have the garment trade built and routes to market that work; Costa Rica shows where a free zone goes when it aims at higher-value niches. Paraguay comes to that table with the simplest rule, in force since 1997, hydroelectric energy at a low industrial tariff and a labor cost that stays competitive when INE data is set against each country’s official sources.

The waterway works in Paraguay’s favor: river freight moves the ton at low cost to overseas ports, and for cargo produced on a calendar, as garments are, that cost per ton weighs more than the extra transit days, which get planned for. The opportunity in view is building the textile trade’s skills, and that grows with every plant that comes in.

The score panel above orders that table: five factors with declared weights and an example composite.

Who this is for

  • The foreign investor choosing a profile and a location before entering.
  • The local manufacturer weighing a conversion to maquila.
  • Industry groups and policy aiming at the profiles with the strongest local pull.

What to ask for: the cluster breakdown, the niche benchmark and the installation score against your profile.

How to measure it: a dashboard built with public MIC data (profiles, employment and value added by group), updated every quarter.

This map is published to be argued with and to be used: if you know the regime from the inside and would group the profiles differently, write to us and we will go over it against the CNIME data. And if you are weighing setting up production under maquila, write to us with your case and we will put the project together for you: the structure, the numbers and the regime paperwork.

Get the cluster breakdown →

Initial diagnosis at no cost · we reply within 24 business hours