The theory

The argument was not stupid, and it is worth stating in its strongest form before taking it apart. Brazil in the 1970s was running an import-substitution development strategy, and the country had a balance-of-payments problem. Computers were obviously going to matter. If Brazil simply bought them from IBM, Burroughs, HP, and Olivetti, it would import that technology forever and never build the capability to make it. But if domestic firms were shielded from those multinationals for a defined period, they could learn, mature, and emerge able to compete on equal terms.

The word to hold onto is defined. The protection came with an expiry date written into it from the beginning.

The machinery arrived in stages. CAPRE, the coordinating commission created in 1972, first began controlling computer imports. In 1979 it was replaced by the SEI (Secretaria Especial de Informática), reporting directly to the Presidency — a signal of how strategic this was considered. Then came the law itself: Lei nº 7.232, de 29 de outubro de 1984, the Política Nacional de Informática, signed under João Baptista Figueiredo, the last president of the military government. Its Article 1 created the CONIN (national informatics council), provided for the SEI, created informatics export districts, authorized the CTI technology centre, and instituted a national plan and a special fund.

Two political facts about that law are genuinely remarkable. It passed unanimously. And it did so on an improbable coalition: nationalist military officers, the domestic business class that wanted the market, and the left-wing opposition that wanted the jobs and disliked the multinationals. Almost nobody in Brazilian public life was against it. Its term was eight years — set to expire in October 1992.

The practice

What the policy actually produced, for anyone living inside it, was a domestic computer industry that mostly cloned foreign designs and sold them late and expensive.

The consumer end of it is the part a whole generation remembers. Brazilian machines of the 1980s were largely clones of the Sinclair ZX81 and ZX Spectrum, the Apple II, the MSX standard, and the IBM PC — the Microdigital TK line, the CP-series machines, and their many rivals. They worked. People learned real programming on them. They were also, by the standards of what existed abroad at the same moment, obsolete on arrival and priced as though they were not.

The sharp edge of the law was the doctrine of the similar nacional: if a domestically-made product was deemed functionally equivalent, the imported one could be refused. In principle this protected the local maker. In practice it meant a Brazilian buyer could be denied a better foreign product because a worse domestic one nominally did the same job — and it is precisely this clause that would trigger the international incident described below.

So a parallel economy formed. Equipment came across the border, especially from Paraguay, and the people running the most current hardware in Brazil were often the ones breaking the law to do it. A policy meant to build domestic capability had made smuggling into a technical strategy.

There was also simply too much industry for the market. Dozens of manufacturers appeared to serve demand that could not support them, an over-proliferation that later commentators from inside the sector identified as the structural error, distinct from the protectionism itself.

The trade war

The United States objected, and the documentary record here is unusually clear because it sits in presidential archives.

In September 1985, Reagan directed the US Trade Representative to open an investigation into Brazil's informatics policies under Section 301 of the Trade Act of 1974 — one of the first such cases the administration self-initiated. In October 1986 the policies were formally determined to be unreasonable and a burden on US commerce. A US Commerce estimate put American losses at several hundred million dollars a year. Parts of the investigation were suspended at the end of 1986 after Brazil made commitments about applying its law more flexibly.

Then in November 1987 Reagan announced sanctions: tariffs to offset roughly $105 million in lost sales, and a prohibition on importing Brazilian informatics products covered by the reserve. The stated trigger is the similar nacional doctrine in action — Brazil had refused to let an American software company license its product on the grounds that a domestic company made something functionally equivalent. Brazil's software law, Lei nº 7.646/1987, came out of the same pressure.

This is the part of the story usually left out of both the nationalist and the free-market retellings: the reserve was not simply an internal economic experiment. It was a foreign-policy problem, and by the late 1980s it was costing Brazil leverage on everything else.

The ending, which was on schedule

The Collor government moved to dismantle it early, sending Congress a bill in June 1990 to revoke Lei 7.232/84 outright. Congress amended it into something else, and the result was Lei nº 8.248, de 23 de outubro de 1991 — which redefined what counted as a "national company" so foreign capital could participate, and replaced blunt protection with fiscal incentives tied to local R&D investment. That incentive structure, repeatedly amended, is still the backbone of Brazilian technology industrial policy today.

But note the detail that gets lost: the reserve itself expired on time, in October 1992, exactly eight years after the law that created it. The sunset was honored. What changed early was everything around it.

The market opened, imports arrived, and much of the protected industry did not survive contact. The trade shows tell the story from the other side — at Anhembi became Latin America's largest technology event precisely in the years the reserve was crumbling and the market was opening.

What it actually left behind

The honest verdict has two halves, and picking only one is how this subject usually gets argued badly.

The hardware bet failed. Brazil did not produce a globally competitive computer manufacturer. The firms that survived did so by leaving: Itautec, Scopus, and Cobra ended up in banking automation rather than computers, and Itautec, the last one still making computers, exited that business in 2013. The theory that eight sheltered years would produce companies able to face the world on equal terms did not come true.

The human capital bet worked, sideways. The reserve trained a large generation of engineers who had to build, adapt, reverse-engineer, and maintain systems locally, because there was no option to phone a vendor abroad. That capability did not evaporate when the companies did. It is a direct ancestor of Brazil's genuinely world-class banking automation, of a serious domestic software and services industry, and of engineering firms that competed internationally on their merits — Cyclades, which took Brazilian console-server engineering to Silicon Valley, and Datacom, still building networking equipment in Rio Grande do Sul, both have roots in this period. The industry that grew was not the one the law was aiming at.

And there is a quieter legacy that is harder to put in a balance sheet. An entire cohort of Brazilian technologists learned computing in the 1980s on locally-made clones, because those were the machines that existed. Whatever the policy cost — and it cost a great deal, in money and in lost years — it put computers in the country early enough for that to happen. People who would go on to build Brazil's internet started on hardware the reserve created.

Both things are true at once: it was an expensive failure at what it said it was for, and it seeded the people who built what came next. Industrial policy is rarely graded on the outcome it promised.