How Brazil built a world-beating aircraft manufacturer

A guest post by Pedro Franco.

Although Donald Trump is hard at work deindustrializing America with tariffs, reindustrializing the country is still a worthy goal, and I’m hoping we can get back to it soon.

A lot of Americans seem to have a despairing attitude about their country’s ability to build anything at all, or to dominate new industries. There’s definitely an inferiority complex with respect to China in particular. There’s also a lot of skepticism toward industrial policy, with a lot of libertarians claiming that it always fails, and MAGA people turning up their nose at it in favor of tariffs.

And yet when I look at the aircraft manufacturing industry, I see that China’s most valuable entrant is at number 12 on the list, while the 7th spot is held by Embraer, a company from Brazil. You usually don’t think of Brazil as a manufacturing powerhouse, and yet it has one of the world’s top jet making companies — ahead of any manufacturer from East Asia.

How the heck did that happen? I was discussing this with Pedro Franco de Campos Pinto, a Brazilian economics lecturer at Musashi University in Tokyo, who’s very interested in the history of his country’s industrial policy. He agreed to write a post for me about how Brazil was able to beat the odds and build a world-class aircraft company. His post is full of potentially valuable lessons for the U.S., as we think about how our country can reindustrialize in the face of Chinese competition.

---

Latin America (LA) remains something of an understudied area in economics. There is a remarkable number of different experiences and policies LA countries have gone through in the 20^th century; e.g., decades long mass immigration, concerted efforts to industrialize via import substitution, hyperinflation… Aside from that, understanding LA’s current state, composed mostly of countries caught in the middle-income trap, could be very important for rising economies in South and Southeast Asia.

Noah’s ongoing interest in Industrial Policy (IP) led me to mention my home country of Brazil’s experiences: the contrasting examples of the Zona Franca of Manaus (ZFM), a special economic zone, and Embraer, a once state-owned,¹ now public company. The latter can be considered an exemplary success of IP, but one achieved despite a near collapse in the 90s and not quite the way expected by its originators, whereas the former is arguably one of the most misguided and costly "successes" in IP history and serves as a warning for countries attempting IP without due care. Let’s start with the more negative example of the ZFM and end with the more positive example how Embraer achieved great heights (pun intended) later.

One small caveat, as the military dictatorship in Brazil (1964-1984) played an outsized role in both policies. Both personally and as an economist, I abhor the dictatorship and what it did to Brazil, but I’ve tried to keep my own personal feelings towards that institution aside.

The ZFM and how it came to be…

As the name suggests, the ZFM is a special economic zone with primary effects on a single city, Manaus, where manufacturing firms have lower taxes and some simplified bureaucracy. Although tentatively initiated in the 50s by a democratic government as a part of a larger effort to “develop” the Amazon, it was the military dictatorship’s 1967 law created the ZFM as we know it and endures until now.² The overarching goal was simple: to develop the Amazon region by industrializing it.

Why Manaus? The city of Manaus is located right in the heart of the Amazon, in the Amazonas state, on the shores of the Amazon River. This is quite important for what follows, because the Amazon River is large enough to handle fairly large container ships.³ Unfortunately, the Amazon basin does not connect Manaus to other major industrial cities directly, meaning that most shipping must navigate the full length of the Amazon River first and then either continue via the ocean or transfer to trucks in Belém, Pará. Manaus itself had experienced a substantial boom and bust due to the rubber boom in the late 19^th century/early 20^th century; despite the bust, the city had continued to grow and by the 1950s it was among the top 20 largest cities in Brazil and by far the biggest city “inside” the rainforest itself.

Map by ClausHansen via Wikimedia Commons

And why this preoccupation with developing the Amazon? The dictatorship did care about economic growth in general, but that’s likely to be of secondary concern in this case. Instead, the likely biggest reason for these efforts remains one of the silliest in economic history: paranoia that some other country could claim portions of the Brazilian Amazon without a larger population presence there, famously summarized during a 1966 speech by the then dictator Castelo Branco which, loosely translated, was “use it or lose it”.⁴ To be clear, the Amazon is not a very hospitable place; it was and remains sparsely populated, and it is largely devoid of significant natural resources aside from things like lumber, some agricultural potential and its inherent biological diversity. The notion that any of Brazil’s much smaller Amazon neighbours or, even more ridiculously, the US, is interested in “taking” the Amazon is self-evidently not very credible, but somehow remains(!) popular in some Brazilian circles, at least until recently.

Returning to the ZFM, in a very real sense, these tax incentives worked. If the goal was to populate Manaus, then the simple fact is that it went from a small city of around 100kin the 1950s to 2m nowadays, growing 4x faster than the total Brazilian average. It is striking that there is a large industrial complex in a metropolis quite literally surrounded by the Amazon forest and River.

Photo by Portal da Copa via Wikimedia Commons

How much is this attributable to the ZFM? To reference the most comprehensive data-driven study (which I will come back to many times) I could find, Holland et al., the fact that similar states and cities that did not benefit from the ZFM and did not grow nearly as much is fairly conclusive in its own right.

More direct evidence comes, again via Holland et al., by looking at Manaus’s GDP in terms of industrial activity, which went from somewhere between 40% just before the ZFM passed to a peak above 60% in the 1980s, before returning to around 40% again in the mid-90s, around the time of the liberalization and end of the Brazilian hyperinflation. Although the number varies depending on economic activity, around 100k workers are currently directly employed in the ZFM and there’s a good case to be made that, per Enrico Moretti’s research, there are substantial local multipliers resulting from the industries located in the ZFM. It’s very hard not to conclude that the ZFM accomplished the goal of getting people to live in Manaus due to the factories located there.

The consequences of misguided IP

At this point, you’re waiting for the other shoe to drop and, indeed, I could and will list a bunch of things showing why this isn’t worth the cost. But let’s take one step before this and ask ourselves a fundamental question: does the ZFM make any sense?

Why are manufacturing industries located where they are? E.g., why did Detroit have the automobile industry, why does China have its manufacturing hubs on the coast, etc.? A full answer would take much more than a simple column to answer, but clustering effects (i.e., being close to other companies and workers is helpful), the presence of a well-educated workforce and low transportation costs are important parts of the answer and on all three counts, the ZFM makes no sense whatsoever. Fundamentally, you don’t want to locate your manufacturing away from industrial clusters, with a poorly-educated workforce and far away from where your products are going to end up; that’s a recipe for wildly inefficient production.

In terms of clustering, despite the deindustrialization of the 90s and 00s, Brazil still has quite a few manufacturing clusters in the South-Eastern and Southern states. Not surprisingly, these were also the biggest and most industrialized states before the ZFM. The creation of the ZFM resulted in deagglomeration of some industries (electronic goods and motorbikes are two clear examples) as manufacturing plants moved to Manaus. As we know now, this almost certainly made it harder for these and related industries to take advantage of clustering effects, which are typically considered very important in manufacturing.

In terms of educated workforces, I have to be a bit speculative here. Holland et al. claims there’s no data before the 1980s that would give us an idea of how well-educated workers were in Manaus vis-à-vis other areas and I’m inclined to believe them.⁵ Still, there was a large disparity of GDP per capita in the 1950s and 60s between the Amazon state and more developed states: São Paulo and Rio de Janeiro had something like 7x more GDP per capita.

Source: Wikipedia

It seems very improbable that education in the Amazon state was even close to that of other states, especially as access to education at the time was very low in all of Brazil, with only around 30% of under 20s attending school. The quality of education was also quite low. I.e., Manaus likely had subpar educational outcomes at a time when Brazil was doing a very poor job of educating its people. It is true that the state of Amazonas and city of Manaus now have significantly better educated populace compared to similar, less developed states/cities (one of Holland et. al’s best arguments in favour of the ZFM), but Manaus is not a paragon of education in Brazil and it does not have many good universities either.⁶

Finally, in terms of transportation costs. As you might expect, a significant portion of the manufacturing done in the ZFM is sent to other Brazilian states, maybe around 30%; that is, a big portion of this manufacturing ends up being sold in other, more populous and richer states.⁷ The ZFM also imports a lot of input goods, as discussed below. Hence, the baffling issue of having a large manufacturing base in a location that adds at the very least 12-14 days of shipping⁸ to your logistics, increasing costs significantly, both in terms of time and money.

So why do firms do it? Why manufacture in Manaus at all? Because the tax incentives are not just large, but they have become de facto permanent for all practical purposes. The original project envisioned the tax incentives to finish in the 1990s but this is Brazil we’re talking about: whenever a certain privilege is granted by government to a group, that group will fight tooth and nail to keep it. In this case, after multiple extensions, the current deadline for ending the tax incentives is 2073. And this is not a decision from the dictatorship either; these extensions took place after Brazil’s return to democracy, showing how this IP resulted in a natural and powerful lobby group.

How much does this cost the government? Coming back to Holland et al, even in their most generous analyses,⁹ conclude that each job created by the ZFM costs the government (in 2018) around R$25k, very close to the R$33k of Brazil’s GDP per capita at that date; I would be willing to guess that it’s likely that the actual cost of each job is higher than Brazil’s GDP per capita, which is quite astounding. Another way of looking at things is that the amount of tax foresworn corresponds to around 0.4% of GDP, an incredible amount for a single city that has 0.9% of Brazil’s population.

There’s one last, important topic to consider here. Has this at least resulted in technological development of the industries of Manaus? This is an area I’m less confident about, but what evidence I can find isn’t great. Manufacturing in Manaus was initially largely last-step assembly stage, where components of goods where mostly imported from elsewhere. This is, of course, a common step when developing local industries, but because of all the factors listed above, including the lack of larger clusters and better universities, the logistics costs, etc., as best as I can tell, there’s been some, but overall small development of local suppliers or new technologies. For one thing, a World Bank report stated that the ZFM is “losing competitiveness and is finding ever harder to attract new companies”, which is not exactly a great sign.

So, let’s turn to a case study instead. Honda, which has over 75% of the motorbike market in Brazil, boasts of its manufacturing of motorbikes in the ZFM. Although it does the final assembly in the ZFM to take advantage of the tax incentives, 72% of suppliers are located in São Paulo state (2600km away as the crow flies, at least 9 days by boat, one way), 20% in Manaus proper, meaning there are claims that there are more manufacturing jobs in São Paulo state for motorbikes assembled in Manaus. I don’t need to point out the insanity of this logistical arrangement, stemming from all the factors listed above.

Photo by Jose Serra via Wikimedia Commons

Another way to see this is that, according to Holland et. al, the ZFM imports around $7.4b of inputs from other countries (and $4.5b from other states) and exports a piddling $0.48b back into the world.¹⁰ This is not the sign of a competitive, cutting-edge industry, but one which seems largely stuck at the assembly-stage of manufacturing, over 50 years after its creation. Indeed, given that import tariffs for most manufactured goods are high in Brazil, I’ve heard many argue that this is a very coddled industry.

I could add here a discussion about the environmental impacts, but this analysis has already become too long; suffice to say, the little evidence there is shows little impact either way. Not too surprising, since most of the cutting down of the Amazon stems from demand for cattle raising in other states. But this also means there’s little reason to claim that the ZFM is a positive thing for the Amazon forest, and, indeed, I suspect that on net it’s had a negative effect.

Finally, and to return to the bigger picture, even the strongest defences towards the ZFM end up being rather weak; claims about how the cost of taxes isn’t too high or the educational system may be a bit better than other places in Brazil simply isn’t good enough to justify the ZFM.¹¹ And perhaps no better argument against the ZFM is the fact that manufacturing in Manaus would likely wither without the incredibly generous tax incentives that it receives.

Let’s summarize. More than 50 years after its creation, the ZFM has achieved the goal of populating the Amazon or, at least, of growing the city of Manaus. It may have had some positive effects, but overall, it remains a hugely expensive “success”, both in terms of foregone taxes and the creation of completely illogical logistics and the declustering of industries.

Pretty shocking stuff. And not surprising: the goal of this IP was never to create a world-class industry and it shows. Good IP should be about targeting industries where you temporarily give certain advantages, e.g., as tariffs and subsidies, and maybe offer longer term support in other ways (such as infrastructure and government assistance in diverse ways for technological development) but with the goal of exposing these industries to market competition, sooner than later. Protection should only last until the sector is ready. And the IP should be done in places that might be able to develop themselves fully, whether by some natural advantage or continued support in other ways (e.g., education). As we can see, the ZFM pretty much does the opposite of all of this in nearly every respect.

In Brazil itself, the ZFM is a perpetual object of discussion. Critics have attacked the ZFM since close to its inception, but the built-in lobbying from firms has proved quite strong. Yet it’s such a hugely costly blunder for Brazil as a whole, one that actively hinders the development of better manufacturing capacity. Not to say that the ZFM should be ended immediately, given that around 100k jobs directly depend on its existence, but some kind of phasing out of the ZFM and/or change in the policy could potentially have big, positive effects in Brazil as a whole in the long-run.

Embraer’s teammates: ITA and DCTA

So! After looking at the costly “success” that ZFM is for Brazil, it follows logically to ask ourselves, can a country like Brazil get IP right? There is at least one striking example where that happened, for aircraft manufacturer Embraer, which is widely lauded as a stunning example of manufacturing prowess in Brazil. The example of Embraer is a good illustration of the ups and downs that even successful IP can have and how a focus on some key aspects, research support, exports, among others, seems to have played a role in its current success.

To understand Embraer, we need to first begin with the foundation of the Aeronautics Technological Institute and the Department of Aerospace Science and Technology. The first of these is a public but military-run university, much better known by its Portuguese acronym, ITA; the second, also better known by its acronym DCTA, is an aeronautics research centre and is also military-run. Both were founded by a rather interesting figure, Casimiro Montenegro Filho.

As a brief aside, although not well known even in Brazil, Casimiro was a bit of an aviation pioneer, having founded and flown the first airmail flights in Brazil in the 1930s and moving from the army to the air force at a relatively high rank¹² when the latter was founded in the 1940s, which allotted him the influence needed for what comes next. He visited the US in the mid-40s and took a great interest in MIT and how effective it was as a technological institute. Despite Brazil being a much poorer country at the time,¹³ he wanted to try to create a similar institution in Brazil and even convinced a MIT professor, Richard Habert Smith, to be the first rector of the DCTA. After extensive lobbying of successive governments, he succeeded in spearheading the creation of the CTA (as it was known at the time) and assisted in the creation of ITA.¹⁴

We need to focus on both ITA and DCTA right from the start as their joint role in Embraer’s success should not be underestimated. Both were created with the goal of achieving a high degree of excellence in aeronautics engineering and seem to have largely succeeded in this respect. Why they were successful, unlike many other Brazilian attempts to create high quality educational institutions, is an interesting conundrum, one that is useful to speculate a bit on.

Part of the reason may be as military-run institutions, I suspect that neither has had funding issues comparable to other Brazilian higher education and research institutes,¹⁵ although I wasn’t able to find any direct evidence of this. They also seem to have largely kept to their original goals throughout their history, of developing engineering talent and knowhow for use in Brazil, both military and civilian. The city they are located in, São José dos Campos, was deliberately chosen to be within driving distance of both São Paulo and Rio de Janeiro, the megacities of Brazil, allowing ITA, in particular, to more easily attract students and engineers from both cities.

Both were successful in attracting talented engineers from abroad in their early years and have managed to sustain and develop links abroad. ITA itself has a fearsome reputation, no doubt helping to attract good students; its entrance examinations are famous in Brazil for probably being the hardest in the nation and covering topics well outside of the Brazilian high school system. This reputation, in particular, is important, because although Brazil’s primary and secondary education systems are quite poor, Brazil is large enough that there always will be outstanding students every year which will be attracted to elite institutions like ITA.

Regardless of how they’ve managed to achieve this, both institutions were and remain fundamental in creating and sustaining Embraer, by providing qualified personnel, expertise and projects throughout the years. The closeness of the three institutions is quite literal, all three within a 10-minute drive from each other. Indeed, it’s a fairly safe bet that the existence of all three is why São José dos Campos is such a successful city in Brazil; they likely became anchors that helped attract relatively high-tech manufacturing and technological development in the area. Directly and indirectly, it’s a safe bet that they helped transform the small and sleepy town of around 50,000 in 1950 into a city of 700,000 (a growth of 14x compared to 4x for Brazil as a whole), with a GDP per capita amongst the highest in Brazil, 50% above average.

The rise of Embraer

Let’s now return to Embraer itself. It was founded by a former student at ITA in 1969/70, as a result of a successful, simple aeroplane prototype developed at the DCTA, with the goal of commercialising the prototype. Brazil had had some success with commercial aeroplane manufacturers before and during WWII, producing at its peak over 250 planes per year, but this was largely driven by military demand and these companies quickly collapsed after the war, with only a couple of smaller producers operating in Brazil before Embraer’s success eclipsed them.

Embraer was founded as a majority owned state company, with its founders coming from both ITA and DCTA. The company understood early on that it would need to focus on a niche product and said initial product would be the EMB-110, a small commuter plane for both military and civilian use that could operate well in areas with poor infrastructure, as was the case in Brazil. Embraer’s focus in the first 20 years or so was in similar types of aircraft (i.e., small airplanes) and in that respect, it worked: in its first peak in 1989, Embraer was selling nearly USD1b (in 2000 dollars) with half being exports and it had around 2500 employees. Given its focus on a relatively niche product, this was clearly a successful company. Note, however, that I called this the first peak… we’ll get back to this.

Source: Vertesy & Szirmai (2010)

How did Embraer achieve this success? We’re already explored one hugely important factor, the general and high-quality expertise provided by ITA and DCTA, but several other factors also played a big role, many of them quite typical in IP. Firstly, as a state-owned company, the government provided the substantial initial investment required, not a trivial issue given the relatively underdeveloped capital markets of Brazil in the 60s, and continued to fund projects in the first 2 decades or so. Secondly, the military was an important source of orders throughout its early history, providing it with a steady flow of orders. Thirdly, the company, jointly or separately from ITA and DCTA, kept close links with foreign aeroplane manufacturers, having created partnerships as early as the 70s.¹⁶ Fourthly and in a similar vein, relatively early on Embraer was already exporting substantial amounts of its planes and this focus on exports remained throughout its history.

This, in particular, is an important point. As I discussed briefly above, Brazil had implemented an import substitution program that focused on industrialization as early as the 1930s, but things really took off starting in the 1950s and such programs lasted until around the late 1980s, early 1990s. In a way, similar to the case of the ZFM, these programs were successful in getting Brazil to both grow and industrialise rapidly; indeed, Brazil’s growth from 1950 to 1980 was around 7% a year and was likely only behind Japan amongst major economies as the fastest growing, allowing Brazil to reach middle-income status sometime in the 1970s.

But as the name suggested, this was an IP to create industries focused on selling to local markets, not export markets. This was achieved through various means, but one key focus was to incentivize FDI by having high import tariffs and tax incentives, as well as investing substantially in infrastructure. Having exports not be a big part of this IP was likely a big mistake and a major factor leading to the lost decade of the 1980s. I believe that a substantial part of Brazil’s growth woes ever since stem from this too.¹⁷ The reason why is, in short, without the discipline of markets (such as exports), manufacturing firms in Brazil simply didn’t continue to invest and push to become as efficient as they were after their initial FDI. Why should they? Protected from competition and with a big internal market, there was no need to do so. One way to see this is how TFP growth in Brazil has been mostly very low since around 1980.

Embraer, happily, never had this issue and it always seemed to be at least trying to keep up with the aeronautical engineering frontiers. Given how important exports were and remain for it, it has pretty much always faced market competition to discipline it and has largely been able to keep pace with other aeronautical companies in similar niches.

Finally, one surprising factor that likely contributed significantly to Embraer’s growth is the success of the agriculture sector in Brazil, in two ways.¹⁸ Firstly, there was the enormous expansion of the agricultural frontier of Brazil that took place from the 1970s and onwards, as you can see below; a lot of this expansion took place in the first decades of Embraer and, particularly in the regions of the 1980s onwards, these were areas that had exceedingly poor infrastructure, incentivizing the use of commuter planes. Secondly, in areas outside of the frontier, there was a push to mechanise farms, which included the use of agriculture planes, a type of plane that Embraer also makes and has always had significant sales with. I should add that in this vein and as others have suggested, the sheer size of Brazil always meant it was going to have significant demand for planes to connect the country, although I should also note that the major Brazilian airlines of the time were mostly flying American planes.

Source: Junior et al. (2023)

The fall and rise again of Embraer

Let’s return to the issue of the first peak, as things turned for the worse in the early 1990s. This was a very tough period for Brazil, as it endured the pinnacle of hyperinflation and had low/negative GDP growth. World GPD growth fell dramatically around that time and the US in particular, a major market for Embraer’s planes, went into recession. Finally, Embraer made a big bet to create a new plane with an Argentine aircraft manufacturer, FMA, that unlike other foreign partnerships, was unsuccessful and the resulting plane was too costly for the market. Overall Embraer sales would collapse to around 25% the level of 1989 and the firm was in a dire situation, as the government itself was incapable of providing significant funds at the time.

With a more liberal, in the classic sense, government in power and despite protests, only one path forward seemed viable and Embraer was privatised in 1994. This was a significant decision in allowing Embraer to flourish, as it injected capital into a company desperately needing some and, by most accounts, the increased freedom allowed Embraer to seek out more international partnerships which greatly reduced R&D costs and transformed it from a vertically integrated company into one that outsources a lot of its components, allowing for greater efficiency. Embraer also benefited from significant export subsidies at that time, probably the only way the government could still help out.¹⁹

Embraer also took a big bet around that time that paid off exceedingly well: it decided to start manufacturing regional jets in addition to its previous focuses.²⁰ Again, this was a smart bet on a niche product, as the US market, the biggest market for regional planes, was dominated by inefficient turboprop planes in the early 90s and fuel-efficient jets were primed to take over. Indeed, the Canadian manufacturer Bombardier was first to take the plunge and was very successful in the beginning.

This leads to a long story of the competition between Embraer vs Bombardier. To keep things brief, despite or perhaps because of Bombardier’s initial head start in the area, Embraer was able to quickly catch up, leading to a competitive duopoly that lasted for around 10 years. However, thanks to its less expensive aeroplanes²¹ and a huge strategic blunder by Bombardier of trying to manufacture larger planes, Embraer was able to take the reins definitely by the mid-2000s.²²

Source: https://ars.els-cdn.com/content/image/1-s2.0-S0048733316301457-gr3_lrg.jpg

And this is where things more or less lie to this day. Embraer is now the biggest regional jet manufacturer in the world and is by far the most technologically advanced manufacturer in Brazil. Indeed, with Boeing’s current woes, there’s been a lot of discussion of whether Embraer might take the plunge itself and try to enter the larger jet market. Personally, I doubt that it will or should, given what happened to Bombardier and how the company perpetually chooses to focus on more niche products, but who knows?

Let’s get back to economics. The “moral of the story” is that Embraer is a good example of a very successful IP that, in many ways, followed the classic path as (I understand it) recommended by the literature for IP. That is, Embraer had strong initial government support/intervention in different ways, but a lot of this was eventually reduced; it always had a strong focus on exports; there was always a focus on keeping up with the technological frontier via foreign partnerships and via the important indirect, government support for technology and skilled labour via ITA and DCTA; it chose niche products where it could compete from the very start. It also benefited from being privatised and being given more freedom, meaning it was successful in a way that was not at all envisioned when founded as a state-owned company. Compared to other cases of IP in Brazil, Embraer’s privatization and its strong focus on exports from the start likely played key roles in its current success.

The broader picture of IP in LA

To end things, let me address something at Noah’s request. One of my sources for this column, the Odd-Lots podcast, had an episode about Embraer and although they get most things about Embraer itself right, I think it’s worth pointing out their broader picture of Brazil does not conform to my understanding at all! That is, unless there’s been a big change in economic consensus that I’m unaware of, their claims about import substitution policies are very much not what mainstream Brazilian economists believe in. In brief, the success of these policies in the 50s to 70s cannot be used to justify the subsequent lost decade of the 80s (more like 15 years in reality), nor the tepid growth (in GDP and TFP terms) since then. This is not meant as a broad stroke against all IP, but specifically against how these policies were pursued in LA.

Furthermore, the podcast raises questions about whether the liberalization of the 80s and 90s was “premature” and why there’s been a lack of clear IP since then. The latter, together with the more general question of how Brazil can become developed, is worthy of a serious discussion (and perhaps another column!), but I’d seriously question how “premature” this liberalization was. Taking the 50s as a starting point until the opening up of the 80s/90s (which reduced tariffs and allowed previously restricted imports, such as cars), that gives us 30/40 years(!) for an industry to be ready to compete with imports.²³ To cite one example the podcast highlights, the fact that native Brazilian automotive companies such as Gurgel failed despite decades of protection (27 years of existence, in Gurgel’s case), is indicative of why import substitution is considered, in the long-run, a failure.

To end this rather long digression, there’s a couple of points the podcast claims, in particular, that I believe are plain wrong, no matter how you feel about import substitution in general. For example, in its defence of import substitution, it states there were “ambiguous failures which as in [sic] like nuclear power and computers in Brazil”. As far as I understand things, the attempt made during the 1980s to create a computer chip industry in Brazil by raising import tariffs to extraordinary high levels despite, among many, many other things, lack of engineering talent or a significant pipeline to develop it, remains one of most criticized episodes of IP that I am aware of.²⁴

The podcast also states that firms created by this policy, such as Embraer, Petrobras and Vale, “took decades and loads of failures before any of these companies became profitable.” To focus on just Embraer, it’s true that in some ways, one can argue that Embraer only truly became “successful” after it entered the regional jet business (i.e., over 25 years after its foundation, 45 years if one focuses on foundation of DCTA), which would suggest that IP can take a very long time before it works… but that’s arguably not the case. Embraer saw significant success (and profits!) before its crisis of the early 1990s and most of the factors that would lead it becoming the even bigger success that it is today, i.e., ITA and DCTA, the foreign partnerships, the focus on exports, etc., were already present then. Many cases of IP do take a long time to happen, but you should be able to see things working well before a quarter-century has passed!

And as a truly very last point about both the ZFM and Embraer, note that there’s a strong irony at play here. Both policies were (essentially) military projects, but with differing goals and methods, and both can be said to have accomplished said goals, but with vastly different consequences for Brazil as a whole.

---

1. Majority state owned, to be clear.

2. Holland, Márcio, et al. "Zona Franca de Manaus: impactos, efetividade e oportunidades." FGV, available at: http://site. suframa. gov. br. Acesso em 11 (2019). https://eesp.fgv.br/sites/eesp.fgv.br/files/estudos_fgv_zonafranca_manaus_abril_2019v2.pdf

3. The current record being seems to be Post-Panamax sized ship of 5500 TEU (https://www.revistaportuaria.com.br/noticia/19966).

4. Integrar para não entregar, i.e., integrate the Amazon region with rest of Brazil or it will be handed (lit delivered) to others.

5. The lack of more and better data is likely the biggest reason why LA is understudied in economics.

6. https://ruf.folha.uol.com.br/2023/ranking-de-universidades/principal/ is a fairly comprehensive ranking and only 2 of the top 100 Brazilian universities are in Amazonas, compared to 16 in the state of São Paulo.

7. Holand et al. presents data that allows us to get a very rough estimate that maybe around $25b of $80b is sent to other states, but I wish I had been able to find clearer data on this.

8. https://www.transamazonas.com.br/travessia-de-balsa-belem-manaus-transporte-fluvial-de-carga-carro-moto-caminhao-picape/ estimates this as the time to transport merchandise to and from Belem, the big port city at the edge of the Amazon. From there, merchandise can be transported by trucks or boats. Given the inefficiencies involved, in practice, this is very much an underestimate.

9. Using generous local job multipliers and taking into account tax intakes, i.e., using net taxes.

10. There are some studies claiming that some goods are re-exported from other states and should be counted as exports from the ZFM, but this just highlights the insane logistics of the ZFM.

11. If you can read Portuguese and are willing to go over Holland et. al, you’ll see that the very best arguments they can make still paint a fairly damning picture of the ZFM, if only by virtue of how weak the defense is.

12. Either Major or Lt Colonel, it’s not quite clear.

13. Having around 10% of GDP per capita of the US at that time.

14. And other military commanders of the time, to be clear, this was not a solo effort. The exact nature of his contribution for ITA itself is unclear, but there’s several sources attributing to him some merit in getting ITA to function.

15. To be clear, like virtually all Brazilian institutions, they’ve faced significant budget cuts/swings in the past. This is me guessing, but I’d imagine that they probably have not had the dramatic swings that other research institutions have faced and they’ve probably been able to budget themselves better; both consequences of being military-run, which helps protect their funding both during and after the dictatorship. Money from Embraer projects itself has probably helped significantly too.

16. Two important early partners consisted were Piper and Aermacchi, both relatively successful small aeroplane manufacturers.

17. But certainly not the sole explanation. The litany of problems that Brazil’s economy has are myriad, to list just a few: poor educational systems, poor infrastructure, myriad difficulties stopping companies from scaling up, a truly insane tax system, heavily government directed investments, incompetent and sometimes corrupt bureaucracies, excessive regulations, slow and capricious law systems, etc.

18. As another aside, note that, to my best of my knowledge, the agriculture sector in Brazil had a much lighter touch compared to the industrial sector in that time period, allowing for significantly more “exposure to market forces”; the implications are quite obvious. Not to say that there were not important contributions from the government, the state-run research institute Embrapa has played an outsized role in both expanding the frontier and increasing Brazilian agriculture productivity, for example, but overall, agriculture faced less direct and indirect government intervention.

19. Many other aircraft manufacturers were also receiving subsidies at that time, such as Bombardier.

20. Regional planes are smaller planes, usually under 100 passengers, that are used more or less exclusively for short-haul flights, particularly connecting lower traffic airports to bigger ones.

21. Bombardier’s business planes were adapted to become regional jets, whereas Embraer’s were designed as such. This is in addition to the efficiencies achieved post-privatization as explained.

22. Bombardier decided to compete directly with Boeing and Airbus in manufacturing a narrow-body jet, one step above regional jets. Cost overruns and manufacturing delays ended up costing Bombardier so much that it was forced to sell off many of its assets, including its own regional jet line. Bombardier now only manufactures business jets.

23. I should point out the opening was not as complete as is sometimes portrayed, as tariffs remained (and remain) relatively high overall: https://portalibre.fgv.br/sites/default/files/2021-03/artigo-centro-de-estudos-do-setor-externo_agosto2010.pdf

24. The nuclear power program, although slightly more successful, is also probably a misguided effort, given both the potential and actual success of renewable energy in Brazil.