Jump to ContentJump to Main Navigation
The Spread of Modern Industry to the Periphery since 1871$

Kevin Hjortshøj O'Rourke and Jeffrey Gale Williamson

Print publication date: 2017

Print ISBN-13: 9780198753643

Published to Oxford Scholarship Online: March 2017

DOI: 10.1093/acprof:oso/9780198753643.001.0001

Show Summary Details
Page of

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2022. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.date: 04 July 2022

Industrial Growth in South America

Industrial Growth in South America

Argentina, Brazil, Chile, and Colombia, 1890–2010

Chapter:
(p.318) 13 Industrial Growth in South America
Source:
The Spread of Modern Industry to the Periphery since 1871
Author(s):

Xavier Duran

Aldo Musacchio

Gerardo della Paolera

Publisher:
Oxford University Press
DOI:10.1093/acprof:oso/9780198753643.003.0013

Abstract and Keywords

This chapter uses new time series data to examine the industrialization in Argentina, Brazil, Chile, and Colombia since the early twentieth century. This allows the authors to uncover variation across countries and over time that the literature has overlooked. Rather than providing a single explanation of how specific shocks or policies shaped the industrialization of the region, their argument is that the timing of the industrial take-off was linked to initial conditions, while external shocks and macroeconomic and trade policy explain the variation in the rates of industrialization after the 1930s, and favourable terms of trade and liberalization explain de-industrialization after 1990.

Keywords:   South America, manufacturing, economic history, economic policy, trade policy, de-industrialization, liberalization, external shock, macroeconomic policy

13.1 Introduction

Between the late nineteenth century and the 1970s, the largest economies in South America had one of the most impressive rates of industrial catch-up in the world (Bénétrix, O’Rourke, and Williamson, 2015; Williamson, 2006).1 Yet, despite the fact that these South American countries are all commodity exporters and share a similar culture, religion, and colonial origin, there is wide variation in their rates of industrial growth. They industrialized not only at different points in time, but also at different speeds.

In this chapter we take a long-term view and examine the patterns of industrialization in Argentina, Brazil, Chile, and Colombia. Rather than trying to provide a single explanation of how specific shocks or policies shaped the industrialization of the region, we show that there is too much heterogeneity for a single theory to work. Thus, we provide a range of alternative explanations, arguing that differences in initial conditions explain differences in pre-1930 industrial growth, while external shocks and macroeconomic and trade policy seem to explain the variation in rates of industrialization from the 1930s onwards.

The traditional history of the industrialization of Latin America has unfortunately been studied by periods, and the data used has often been estimated differently across the periods, thus leaving us with no readily available long-term series. Therefore, writing a South American industrial history required us to put together long-term series of manufacturing value added (i.e., industrial GDP), manufacturing labour productivity, and the manufacturing labour force from 1900 to 2010 (see the Data Appendix). Our manufacturing series cover industrial establishments of all sizes, different technologies (artisan vs mechanized), and a wide range of products. Most of our series reflect true manufacturing value added, after the first industrial census in the country concerned. However, Colombian data before 1953 are estimates of the quantum of manufacturing production, and for (p.319) Brazil the early industrial GDP indices are proxies using data on raw materials and statistics on textile production (Haddad and Contador, 1975). In addition to the data on manufacturing value added, we gathered information on population, openness to trade, exchange rates, terms of trade, and other series.

This new database allows us, for the first time, to uncover variation across countries and over time that the existing literature had either ignored or studied in piecemeal fashion at the country level. Our main insight is that industrial growth in South America has not been homogeneous. For instance, Argentina and Brazil, the largest economies in the region, enjoyed rapid catch-up before the 1930s, while only Brazil and Colombia saw very rapid and sustained catch up on the US, the global leader, in the 1930s. The post-Second World War golden age of growth was experienced in the four countries studied, but only Brazil and Colombia saw faster industrial growth than the industrial leaders in Europe and Japan. During the 1980s and early 1990s industrial growth slowed in most of the region, barring Colombia, but by the first decade of the twenty-first century most of the economies started to speed up again, but not sufficiently fast to catch up on the industrial leaders.

With our new time series in hand, we cannot find evidence to support most of the traditional hypotheses that have been used to explain the industrialization of South America. There are four basic explanations in the literature. First, there is the ‘adverse shocks’ hypothesis (Tavares, 1972; Nations/ECLA, 1951; Prebisch, 1950; Furtado, 1959), which argues that as a consequence of adverse international shocks, such as wars, downturns, or commodity price shocks, the relative price of exports decreased and/or import scarcity rose, financing channels were interrupted or there was a scarcity of foreign exchange, and the terms of trade declined. This hypothesis is related to Dutch Disease forces in the sense that it posits that Latin America de-industrialized when there were commodity booms because of Dutch Disease effects, while the region re-industrialized when the terms of trade declined and the local price of manufactures rose. The second hypothesis has been characterized as the ‘endogenous industrialization view’, seeing industrialization as a product of export-led growth (Diaz-Alejandro, 1976; Dean, 1969). This thesis argues, in sharp contrast to the previous one, that South America industrialized when commodity exports thrived, because the export boom attracted foreign investment and, via an income effect, stoked the growth of domestic industrial demand. Thus, commodity booms led to increases in manufacturing value added, but also to improvements in industrial productivity (Lederman, 2005; Haber, 2006; Williamson, 2011a).

A third hypothesis sees industrial growth as the product of import-substituting industrialization (ISI), or explicit policies that included tariff protection, exchange controls, special preferences for firms importing capital goods for new industries, preferential import exchange rates for industrial raw materials, and an ample set of industrial policy tools that ranged from subsidies, targeted credit, pressure on foreign companies to open plants in the region, or the direct establishment of state-owned enterprises (Hirschman, 1968; Baer, 1972). Most of these policies were not undertaken simultaneously until after the Second World War, yet a modified version of the ISI model was applied before the 1930s (Versiani, 1980; Coatsworth and Williamson, 2004). Finally, there is the ‘stagnationist’ hypothesis. (p.320) It argues that even if protection did lead to increases in manufacturing value added, it did not provide the incentives for firms to improve productivity over time: even when manufacturing value added increased, labour productivity stagnated. Thus, the ‘stagnationist’ hypothesis sees ISI as a policy that only succeeded in the short to medium term (Colistete, 2009; Macario, 1964; Diaz Alejandro, 1970; Krueger, 1978; Haber, 2006; Bulmer-Thomas, 2003).

Because our data show that none of these hypotheses explains all cases during the whole century, we argue that the industrialization of Latin America cannot be understood by appealing to a single theory. The drivers of industrial growth in each country are not the consequence of a common shock, or the adoption of a single set of policies. In fact, the evidence suggests that the most important external shocks, such as the First World War, the Great Depression, the Second World War, and the 1980s debt crisis, had heterogeneous effects on industry in Argentina, Brazil, Chile, and Colombia. While some countries were damaged by some of these shocks, others thrived.

Our argument is that the timing of the industrial take-off was powerfully influenced by initial conditions rather than by policies or external shocks, but that the subsequent variation in industrial growth and de-industrialization experiences was due to the response of policy to changes in external conditions. Market size, literacy, and infrastructure development at the end of the nineteenth century all favoured industrial growth in Argentina and Chile, but not in Brazil and Colombia, which were slower at developing infrastructure and integrating their markets. Later on, market size, the spread of infrastructure, trade policy, and macroeconomic policy explain why industrial growth sped up in Brazil and Colombia after the 1930s, while it slowed in Chile and Argentina. Finally, we argue that favourable terms of trade and economic liberalization explain relative de-industrialization in South America between 1990 and 2012.

The heterogeneous nature of industrial growth in Latin America can be seen in Tables 13.1 and 13.2. In Table 13.1 we compare manufacturing value added growth rates in South America with those in the industrial leader, the United States, and three other key early industrializers (the United Kingdom, Germany, (p.321) and Japan). The wide dispersion in growth rates between our four Latin American countries, especially in the first three periods, can be seen immediately.

Table 13.1. Industrial GDP growth rates, South America vs global leaders (per cent, per annum, annual averages)

Leaders (average)

GER

UK

USA

JAP

ARG

BRZ

CHL

COL

1900–19

3.2

2.7

1.0

5.8

4.8

9.8

2.4

1920–30

2.6

1.0

2.9

3.9

6.5

2.6

1.5

3.1

1931–43

4.9

2.2

2.7

9.9

6.3

3.3

10.0

7.5

8.9

1944–72

5.6

4.7

4.3

3.1

10.2

5.3

8.7

5.4

6.9

1973–90

2.2

1.8

0.8

1.7

4.5

−0.6

4.0

1.8

3.9

1991–2009

0.8

0.03

−0.4

2.9

0.4

2.9

2.2

3.6

2.2

Notes: GER = Germany, JAP = Japan, ARG = Argentina, BRZ = Brazil, CHL = Chile, COL = Colombia.

Source: See Data Appendix. Developed country growth rates from Bénétrix, O’Rourke, and Williamson (2015).

Table 13.2. Convergence/divergence among South American nations and the developed country leaders

Growth in Latin America minus growth in leaders

Growth in Latin America minus US growth

Leaders (average)

ARG

BRZ

CHL

COL

ARG

BRZ

CHL

COL

1900–19

3.2

1.6

6.6

–0.8

–1.0

4.0

–3.4

1920–30

2.6

3.9

0.0

–1.1

0.5

2.6

–1.3

–2.4

–0.8

1931–43

4.9

–1.6

5.1

2.6

4.0

–6.6

0.1

–2.4

–1.0

1944–72

5.6

–0.3

3.1

–0.2

1.3

2.2

5.6

2.3

3.8

1973–90

2.2

–2.8

1.8

–0.4

1.7

–2.3

2.3

0.1

2.2

1991–2007

0.8

2.2

1.5

2.9

1.5

0.0

–0.7

0.7

–0.7

Source and notes: As in Table 13.1.

On the left-hand side of Table 13.2 we present these same South American growth rates, minus the average growth rate of the four leaders. Positive entries in the table imply industrial catching-up, while negative entries imply that the country in question’s industry is falling further behind the leaders. The right-hand side of the table presents similar data, this time subtracting the industrial growth rate of the United States alone, rather than of all four industrial leaders.

13.2 Pre-1920 Industrial Growth in South America

The period before 1920 has always been considered the Belle Époque of GDP and industrial growth in Latin America. However, the phrase ‘Belle Époque’ refers to Latin America’s performance in that epoch, relative to other periods, rather than to Latin America’s growth performance compared with that of the developed world. When industrial growth in South America is compared with that of the industrial leaders, only Brazil and Argentina have an outstanding performance. When the South American performance before 1920 is compared with the United States (on the right-hand side of Table 13.2), only Brazil fares favourably. Thus, the Belle Époque was indeed an important period of initial industrial take-off for Argentina and Brazil, yet was a period of rapid catch-up only for Brazil.

Argentina was the industrial front-runner in the late nineteenth and early twentieth centuries, despite the fact that it experienced a sustained commodity export boom. Its industrial growth rate before the First World War was between 6 and 11 per cent (but the war years lowered the 1900–19 average to 4.8 per cent, as shown in Table 13.1). This early industrialization was an endogenous, private sector-led process driven by the dynamism of the export economy. The industrial boom was tightly linked to the development of agriculture, which produced (p.322) forward and backward linkages, accelerating urbanization rates, and giving rise to a new consumer class that demanded manufactured goods. The initial industrial boom was dominated by the production of food, beverages, textiles, wool and leather, tobacco, and glass, with some important firms competing successfully with consumer goods imports. This was the beginning of what has been called an ‘easy’ import substitution process. During this first phase, the production of consumer goods represented 72 per cent of total manufacturing output. However, there were natural resource obstacles to the development of a competitive ‘heavy’ industry. These obstacles included the scarcity of coal, iron, and other minerals, which precluded the development of large-scale machinery and metallurgical firms. Firms thus relied heavily on imported intermediate and capital inputs (Rocchi, 2005; Diaz-Alejandro, 1970; Barbero and Rocchi, 2003).

Industrial growth in Brazil, Chile, and Colombia began later (much later in Colombia). In the 1910s Chile’s industrial output was growing at a modest annual rate of 3 per cent, while Brazilian industry was growing at 14 per cent. The rapid Brazilian industrial take-off seems to have started at the beginning of the twentieth century. Textile and other industrial firms enjoyed buoyant internal demand associated with coffee booms (and coffee valorization programmes2), while they also had easy access to foreign and local equity and debt finance between 1905 and 1914, helping firms finance machinery imports. The result was a rapid spurt in industrialization during the fifteen years before the First World War (Haber, 1991; Suzigan, 1986; Musacchio, 2009). Cano (1977) highlights the importance of the almost forced import substitution during the First World War, when the disruption to shipping and capital inflows led to a shortage of foreign exchange and imported goods. Domestic producers took up the slack by increasing capacity utilization (Suzigan, 1986).

In Chile, Palma (2000) argues that in spite of a virtual world monopoly of sodium nitrate exports, the Chilean economy avoided Dutch Disease forces in the 1890s by implementing an active policy of manufacturing protection that propelled the local industry.

We do not have data for Colombian industrial GDP growth in this initial period, but we know that there were two coffee booms, following the valorization programmes in Brazil (in 1906 and after the late 1920s). The coffee booms, combined with increased protection for manufacturing, created a long-lasting non-durable consumer products industry for the first time, as well as a few shorter-lived coffee machinery producers. Moreover, it was during the second coffee boom that (p.323) manufacturing was able to take advantage of the rise in market size due to investments in infrastructure (Echevarria, 1993).

In Argentina and Brazil, labour productivity growth was faster than that in the US during this period (Table 13.3). This supports the argument that Argentina and Brazil experienced endogenous industrialization with significant improvements in productivity (Haber, 2006).

Table 13.3. Average labour productivity growth rate: Argentina, Brazil, Chile, Colombia, and the United States

Argentina

Brazil

Chile

Colombia

US

1900–19

1.4

4.8

3.9

1.0

1920–30

5.4

6.1

0.4

–2.5

4.4

1931–43

–1.4

5.6

4.3

5.5

3.5

1944–72

3.5

4.0

3.8

3.7

2.1

1973–90

1.6

3.7

–0.5

2.7

2.6

1990–2009

1.3

2.6

3.5

1.1

3.9

Source: See Data Appendix.

13.2.1 Initial Conditions

What explains the early start of Argentina and the late start of Colombia? The heterogeneous timing of industrial take-off cannot be explained by any one of the industrialization hypotheses presented above. Instead, we argue that the explanation lies in the heterogeneity of their initial conditions, external conditions, policy, and outcomes. That is, heterogeneous initial conditions and the speed of technology adoption (related to absorptive capacity and market demand) explain the timing of industrial take-off better than any of the theories presented above. The rest of this section thus focuses on how initial conditions such as the stock of human capital, market size, and transportation structures determined the timing of industrial take-off in South America.

By 1900, Argentina, Brazil, Chile, and Colombia had all consolidated their status as independent countries. Colombia still waged a civil war, but it would soon be finished, and there followed a long period of relative peace. The inflow of migrants and capital from Europe was at its peak, contributing to expanding domestic markets and integration into the world economy (Obstfeld and Taylor, 2004; Stone, 1999; O’Rourke and Williamson, 1999; Williamson, 2011b).

A casual inspection of the demographic and educational data in Table 13.4 shows that initial conditions (at the beginning of the twentieth century) differed greatly between our four countries. Argentina and Chile were closer to European levels of literacy, while Brazil and Colombia lagged behind. As Sokoloff and Engerman (2000) remind us, low literacy rates retarded economic growth and might have been a crucial obstacle that postponed the take-off of the industrialization process. The adoption of new technologies and the capacity to innovate is (p.324) (p.325) partly dependent upon a labour force (and management) that could learn new techniques.

Table 13.4. Population, urbanization, and illiteracy rates (per cent), and transportation infrastructure: Argentina, Brazil, Chile, and Colombia, 1900–2000

Population (millions)

Urbanization rate

Illiteracy rate

Railway (km per million pop.)

Road (thousand km per million pop.)

Population (millions)

Urbanization rate

Illiteracy rate

Railway (km per million pop.)

Road (thousand km per million pop.)

Argentina

Brazil

1900

4.7

41

40

3,572.8

n.a.

18.0

n.a.

65

851.6

n.a.

1920

8.9

54

32

3,981.7

n.a.

27.4

n.a.

65

1,041.3

n.a.

1930

11.9

57

25

3,192.5

17.8

33.6

n.a.

61

967.5

3.6

1940

14.2

60

18

2,913.6

28.7

41.1

31

56

833.1

5.1

1950

17.2

64

12

2,499.4

4.2

53.4

36

51

686.4

n.a.

1960

20.3

72

9

2,164.8

3.8

69.6

45

40

550.3

7.1

1970

24.0

79

7

1,665.3

8.4

95.7

56

32

332.8

11.9

1980

28.4

83

5.6

1,201.2

n.a.

123.0

66

25

241.1

n.a.

1990

33.0

87

4.3

1,082.2

n.a.

151.2

75

19

200.6

n.a.

2000

37.3

89

3.2

957.6

5.8

176.3

81

15

172.4

11.2

Chile

Colombia

1900

3.0

43

47

1,471.4

n.a.

4.0

n.a.

61

142.1

n.a.

1920

3.7

47

37

2,205.5

n.a.

6.2

n.a.

56

216.8

n.a.

1930

4.3

49

25

2,094.9

8.4

7.9

n.a.

48

329.7

3.0

1940

5.1

60

27

1,702.9

7.1

9.2

29

43

363.5

2.5

1950

6.1

61

21

1,396.0

n.a.

11.6

43

38

304.2

n.a.

1960

7.4

68

16

1,137.2

7.4

15.4

48

30

204.6

n.a.

1970

9.4

75

12

883.9

6.8

21.4

57

22

160.3

2.3

1980

11.1

81

8

568.1

n.a.

26.6

64

16

127.8

n.a.

1990

13.1

83

6

521.9

n.a.

33.0

70

11

64.1

n.a.

2000

15.2

85

4

324.8

5.3

39.8

75

8

53.1

2.7

Source: MOxLAD (2014).

Some have argued that the type of education imparted might also have been a deterrent to the adoption of sophisticated technologies. For example, Maloney (2002) shows that these countries failed to establish technical education during most of the nineteenth century. ‘By 1926, Australia had twenty-seven times more graduates of technical schools per capita than Argentina, perhaps the most educated country in Latin America’ (Maloney, 2002, p. 127). If this was the case, the genius to industrialize might have been embodied in European migrants who came with technical skills. In 1914, foreigners propelled industrial initiative in Argentina. More than 65 per cent of firms in Argentina were initiated and owned by first-generation immigrants (Diaz Alejandro, 1970).

Literacy and higher-order human capital may be good predictors of increases in per capita income levels, but they are insufficient to drive sustained manufacturing growth. Literacy and engineering rates alone would have led us to predict that Argentina and Chile, instead of Brazil, would be the industrial leaders by the mid-twentieth century (see Table 13.4 and Maloney, 2002).

Yet that did not happen. Brazil took the lead in terms of industrial sophistication and sheer size in the second half of the twentieth century. It started as a laggard: the initial industrialization of Brazil, between 1890 and 1930, did not have the kind of skill-biased technologies associated with the second industrial revolution (Goldin and Katz, 1996; Musacchio, Martínez Fritscher, and Viarengo, 2014). Brazil started rather with labour-intensive industry. Thus, the industrialization levels of Brazilian states in 1920 and 1940 are not correlated with educational levels. The Brazilian government did not introduce the kind of technical education necessary for the development of more sophisticated industries until the 1950s. The National Council for Research, a Brazilian version of the National Science Foundation, and the Technological Institute for Aeronautics (ITA) were launched in 1950, the National Institute of Applied Math (IMPA) in 1951, and the State University of Campinas—with a strong emphasis on engineering—was not founded until the early 1960s. The expansion of national laboratories and research centres continued in the 1970s and coincided with the dramatic changes in manufacturing revealed in Table 13.5 below, and with the rise of Brazil as a leader in innovation in areas such as agricultural research (e.g. with the creation of the Brazilian National Agricultural Research Company, known as Embrapa in 1973) and aeronautics.3

Table 13.5. Industrial value added by type of industry, Brazil and Colombia, 1940–95 (per cent)

c.1940

c.1955

c.1970

c.1980

c.1990

c. 1995

Brazil

Transformation of natural resources:

14

15

13

11

Consumer products (durables and non-durables)

59

41

34

28

Heavy industry and industrial inputs

17

24

27

30

Capital goods and high-technology products

5

15

20

24

Others

1

2

2

4

Colombia

Consumer products (non-durables)

n.a.

66.3

30

32

31

33

Consumer products (durables)

n.a.

4.6

14

15

16

18

Intermediate goods

n.a.

16

39

34

36

34

Capital goods

n.a.

12

9

10

9

8

Transportation equipment

n.a.

n.a.

7

9

8

7

Sources: Brazil, compiled from IBGE (1990, p. 386). Colombia from ECLA (1957, p. 274); Garay (1998, p. 463).

In any case, monocausal explanations of Latin American industrialization are always going to be insufficient to explain variation over time and across countries. Industrialization is a complex process that requires a multivariate explanation. For instance, one has to consider not just education, but urbanization levels, market potential, infrastructure, protectionist policies, and a variety of other important factors.

For instance, it is well known that urbanization is closely associated with increases in per capita income. Several studies even suggest that an increasingly (p.326) urban economy promotes industrial growth (due to the benefits of agglomeration effects). However, the increases in rents experienced by an increasingly productive natural resource sector can generate two types of urban employment: tradables, like consumer goods; and non-tradables, such as services. South America’s leading cities, Buenos Aires, Santiago, Sao Paulo, Medellín, and Rio de Janeiro were hybrids. Initially they relied more on the stimulus of internal demand based on exports, than on a supply-side push. Firms located in urban enclaves dealt first with processing and later with the substitution of imports. The Brazilian and Colombian urbanization processes came much later, but their industrial catch-up was extremely fast. After 1940, the acceleration of urbanization was probably caused by an industry-led boom attributable to the artificial change in relative prices favourable to manufacturing. This was certainly the case for Argentina and Chile, and in the Southeast of Brazil.

In Colombia urban concentration of the scale found in our other countries came much later. By 1912 there were two important urban centres, Bogotá and Medellín. The former was the political capital of Colombia, with 120,000 inhabitants, and focused on the production of foods. Medellín, with 70,000 inhabitants, had manufacturing focused on food, beverages, and textiles (Melo, 1987). Buenos Aires had 1.2 million inhabitants in 1910, Santiago 390,000 in 1900 and São Paulo 240,000 in 1900.

Moreover, the market potential for manufactured goods was not only determined by urbanization, but also by transportation costs (involved both in selling manufactured goods and in importing raw materials) (Krugman, 1993). Buenos Aires and Santiago, the key industrial hubs in Argentina and Chile, had the initial advantage of being close to the coast, and thus faced lower transportation costs for raw materials and enjoyed easier access to internal and external markets (by boat). In contrast, São Paulo, Bogota and Medellín had, initially, more difficult access to the coast. Until the early part of the twentieth century, the railway lines connecting the coast to the interior plateau of São Paulo were not fully developed. In Colombia, the location of the major urban centres of Bogotá and Medellín in the Andes probably made trade with the rest of the world, and within the country itself, harder than it was in any of the other three countries in our sample.

One way to gauge the transportation challenges of Colombia and Brazil is to look at their railway and road densities during the take-off period (Herranz-Loncán, 2014). Table 13.4 shows that Chile and Argentina had an initial advantage. Brazil, with its scattered population, was at a disadvantage and its transportation infrastructure did not compare well to those of Argentina and Chile, on a per capita basis, until after the Second World War. In Colombia, the poor rail and road infrastructure and the challenging topography complicated market integration until the second half of the twentieth century, when the large urban centres provided markets big enough for large-scale domestic industries to flourish. The road network in Colombia was small in relative terms even in the year 2000; comparable to the density of roads Brazil had had at the turn of the century.

Population size may be another factor determining the extent of specialization and diversification in manufacturing in these four countries. Murphy, Shleifer, and (p.327) Vishny (1989) described the benefits of size in models of ‘take-off’ or ‘big push’ industrialization, in which the take-off phase is characterized by a transition from a slow growth, constant returns to scale technology to endogenous growth and an increasing returns to scale technology. Size may also enhance industrial growth because it promotes product competition.

As we can see from Table 13.4, the Argentine economy had a size advantage when it came to its major urban centre (Buenos Aires). The country was more urbanized than any other country in the region and most of the population lived in the larger Buenos Aires area. That may explain Argentina’s initial take-off and initial leadership as the early industrializer in South America. Brazil, despite its larger population, was a much more rural country and before 1900 most of the population centres were scattered around the country, with extremely poor communication and road networks.

Rapid population growth and urbanization in Brazil and Colombia after 1940 help to explain those countries’ positive and sustained industrial performance and catch-up with Argentina in terms of manufacturing value added per capita. Population growth in Argentina and Chile did not keep pace with Brazil and Colombia.

13.3 Industrial Performance During the Inter-War Period

Industrial growth during the 1920s varied across South America. Argentina, Brazil, and Chile experienced growth spurts during the first half of the decade, but average growth rates then declined. For Argentina, the 1920s saw the most rapid industrial catch-up in the twentieth century. Industry grew at an average annual rate of 6.5 per cent, the result of export-led manufacturing demand and positive terms of trade. Incumbent firms in the ‘traditional’ sectors (food, beverages, tobacco, meatpacking houses, sugar mills, and tanning firms) experienced important capacity expansion and structural transformation, and new sectors developed, such as rubber products, chemicals, pharmaceuticals, machinery, and electrical equipment. The ‘front-runner’ was entering a second phase of industrialization. The share of the new sectors in manufacturing output increased from 13 to 21 per cent between 1920 and 1930. This process took place well before any implementation of an explicit state-led import-substitution strategy (Pineda, 2009). Argentina’s performance during the 1920s supports the endogenous industrialization hypothesis.

The decline in the Chilean nitrate industry in the 1920s was due to the improvement in the production of synthetic nitrates in the core countries. The decade started with a 50 per cent nominal devaluation and increased protection that produced a transitory spurt in manufacturing. However, as the real exchange rate appreciated in the long run, the rate of manufacturing growth diminished and even became negative in some years (Palma, 2000; Muñoz, 1968). Thus, the 1920s proved to be the weakest for manufacturing growth in Chile.

Brazil enjoyed rapid industrial growth in the 1920s, driven to a large extent by the rapid advance of coffee exports and national income. The decade saw favourable (p.328) terms of trade, a large importation of machinery for manufacturing, and a rapid expansion in coffee exports toward the end of the decade. According to Stein (1957) and Musacchio (2009), textiles continued to expand rapidly during this decade. At the end of the 1920s, the government instituted a coffee valorization programme to keep global prices high, while fixing the exchange rate to prevent its rapid appreciation. These measures not only provided an impetus for endogenous industrialization after the second half of the decade, but also allowed Brazil to maintain stable-to-favourable terms of trade during the Great Depression (Suzigan, 1986; Furtado, 1959).

The 1920s were years of unstable manufacturing expansion in Colombia. Industrial growth was fuelled by coffee booms, and large inflows of international capital after the Central Bank was set up in 1923. Incumbent firms expanded capacity and adopted modern management techniques. Foreign direct investment also helped to diversify the industrial sector by expanding oil manufactures. The development of the railroad network during the second half of the decade reduced the cost of transportation and its construction also increased the demand for manufactures. Qualitative evidence suggests that a mild process of endogenous industrialization was taking place in Colombia, aided by moderate protection.

Industrial output grew between 7.5 and 10 per cent per annum between 1931 and 1941 in Brazil, Chile, and Colombia, faster than in Europe and Japan, but not the US. In contrast, Argentina had industrial growth rates of only 3.3 per cent per annum, making it the only country of the four diverging from the industrial leaders.

Brazil and Colombia were not as hard hit by the depression thanks to the rapid recovery of their terms of trade, largely due to their coffee valorization programmes. In fact, Colombia experienced industrial growth of almost 9 per cent per annum during the 1930s and the Second World War, largely because of increasing coffee exports, a large depreciation in the exchange rate, protectionist policies, and an improved transportation infrastructure set up in the 1920s. This industrial growth spurt mostly involved consumer non-durable industries (Ocampo and Montenegro, 2007; Echevarría, 1993).

Chile probably suffered the most during the Great Depression: exports declined by 50 per cent between 1929 and 1932, imports by 83 per cent, and the terms of trade by more than 50 per cent. Chile needed to pursue another model and the policy reaction was immediate. The government engineered a devaluation of more than 300 per cent between 1932 and 1935, resulting in an increase of the real cost of imports of about 100 per cent. The result was an acceleration in average annual industrial output growth of 7.5 per cent (Muñoz, 1968). Manufacturing began an important structural transformation. In 1927 consumption goods still represented 83 per cent of manufacturing output, and Chile was unambiguously in its first phase of industrialization. However, the share of intermediate, heavy, and capital goods output in total manufacturing rose from 7.7 per cent in the early 1930s to 12.8 per cent by 1950, almost doubling. That is, the Great Depression accelerated the move towards import substitution of more sophisticated manufactures in Chile, and ISI policies played a big role in this. In 1938 the state development financial agency, CORFO (Corporacion de Fomento), was created to develop strategic plans (p.329) for agriculture, industry, and mining, and to develop domestic technology research (Ffrench-Davis et al., 2001).

The adverse shocks hypothesis, which sees negative shocks as opportunities for industrialization, finds some support during this period. Not only did industrial output grow fast in Brazil, Chile, and Colombia, but labour productivity did as well. The only exception was Argentina, which suffered a terrible blow to its terms of trade and had, overall, negative labour productivity growth in the decade.

Why was the Great Depression an obstacle for Argentina while it acted as a boost for the remaining three countries? The front-runner was already attempting its second industrial revolution by the 1920s, and needed access to foreign markets and a continued inflow of foreign capital to finance the structural change. These were unavailable during the 1930s, while the post-depression years delivered a fatal blow to temperate climate crop prices, and international capital flows to developing countries were almost nil. Thus, Argentina followed a very different path than did Brazil or Colombia, due more to bad luck than to bad policy.

13.4 The Second World War and Import-Substituting Industrialization

The golden era of import substitution industrialization was 1944–73, when governments in the region implemented explicit protectionist policies to promote the substitution of consumer goods, and to some extent intermediate goods in Brazil (Leff, 1968). Rapid industrial growth during this period has also been attributed to rapid urbanization; manufacturing surpassed agriculture as the most important employer (although Argentina underwent this process earlier in the century) (Hirschman, 1968; Baer, 1972; Baer, 2008).

While Colombia, Chile, and Brazil continued deepening their industrialization in the 1930s, expanding their textile sectors and beginning to develop other industries (Stein, 1957; Lederman, 2005), Diaz Alejandro (1970) argues that Argentina missed an opportunity to implement a targeted industrial policy that would have enabled that country to follow a smooth transition to a more advanced industrial structure. According to Diaz Alejandro, since Argentina never experienced another export boom (its terms of trade in the post-1945 period was much lower than in 1900–30), the endogenous industrial growth phase was over. Instead a new phase began, during which the country faced recurrent current account deficits and external disequilibria. According to Diaz Alejandro, after the 1930s Argentina experienced a long period of drifting away from its South American front-runner status (Taylor, 1998). Although Argentina’s terms of trade continued to decline, and it never experienced endogenous industrialization again, it did exhibit a relatively stable industrial growth rate after the Second World War, ranging between 4 and 6 per cent per annum. Brazil’s industrial growth rate ranged between 7 and 11 per cent, Chile’s was over 5 per cent, while Colombia maintained almost 7 per cent. The four countries were catching up on the world leader, the US, (p.330) but only Brazil and Colombia managed to keep converging on the UK, Germany, and Japan.

13.4.1 Protectionism and Industrialization

Students of ISI in Latin America stress that such policies were intended to promote industrialization in stages. That is, governments would sequentially promote new industries with increasingly high value added and ever greater technological complexity. Consumer goods and basic building materials industries would be promoted first because of their simple technology and their low capital requirements. Then governments would support more complex consumer goods industries, which required more sophisticated technologies and higher capital requirements. Finally, governments were to target consumer durables, steel, engineering, chemicals, and other heavy industries (e.g. Brazil and Argentina ventured into aerospace) (Baer, 1972; Love, 2005). This sequencing could include as a final link the development of a domestic capital goods sector, or a complex sector of industrial raw materials. In practice, the sequence did not work like this. Instead, some industries lobbied governments to avoid developing intermediate goods that could lead to expensive inputs (Baer, 1972).

The golden ISI years in Argentina lasted from the 1950s to the early 1970s. The Peron administration (1946–55) opted—by default—for an inward-looking industrialization model that accelerated under the leadership of Arturo Frondizi (1958–62), an advocate of the so-called ‘desarrollismo’ (‘developmentalist’ approach). Already in 1960, state-owned enterprises controlled basic sectors of the economy such as iron, steel and petroleum, energy generation, telecommunications, and transport. Multinational corporations were engaged in the production of vehicles, pharmaceuticals, petrochemicals, tobacco, agricultural equipment, and food processing (Katz and Kosacoff, 2001). Structural change in industry was significant between 1950 and 1970: the share of heavy industry in total manufacturing increased from 20.5 to 32.4 per cent, while the share of capital goods manufacturing jumped from 9.7 to 22 per cent. In contrast, the share of consumer goods production dropped by a third. By the mid-1960s, a domestic ISI model was in place; the question by then was whether the battery of protectionist and fiscal policies supporting it were sustainable.

Table 13.5 depicts the change in industrial structure in Brazil during the post-1945 period. We separate manufacturing from the extractive industries, and subdivide it into groups that broadly represent the stages predicted by the import-substituting industrialization hypothesis. Brazil made the largest leap in industrial sophistication during the post-Second World War period. Manufactured consumer goods (associated with the first stage of import substitution) fell from 60 per cent of the total in 1939 to less than 30 per cent by the 1970s. In contrast, the share of heavier industries producing metals, chemicals, plastics, and pharmaceuticals (associated with the second stage of import substitution) almost doubled between 1939 and 1980, going from 17 to 30 per cent of total manufacturing. The rapid increase in these Stage II industries during the 1960s and 1970s is associated (p.331) with the explicit development plans of the military government (1964–85) promoting heavy industries. Finally, industries associated with Stage III, such as mechanical industries, electrical and telecommunications equipment, and transportation equipment (automobiles and airplanes), gained momentum after the 1950s and reached 24 per cent of total manufactures by 1980. This was the heyday of industrial policy in Brazil, when the government provided the subsidized financing, infrastructure, and raw materials needed to develop some of the most sophisticated industries in the country. State-owned enterprises were key to developing the airplane, petrochemicals, electricity, and telecommunications sectors (Musacchio and Lazzarini, 2014).

By 1949 the local capital goods industry provided over 60 per cent of the domestic demand for industrial equipment. This development is especially impressive given that the nascent capital goods industry developed despite competition from foreign imports until at least the 1960s, during which time machinery imports benefited from preferential exchange rate treatment and duty-free importation. In fact, the development of the Brazilian capital goods industry was so impressive that the ‘domestic supply coefficient [for capital goods] was more than three times larger than in Argentina during the same years’ (Leff, 1968, p. 8).

In Colombia, ISI policies also involved the development of industrial banks and creative exchange rate devaluation mechanisms to reduce the chance of recurrent foreign exchange crises. Manufacturing underwent a significant transformation between 1955 and 1970 (see Table 13.5). The production of durables and intermediate goods increased rapidly, rising from 4.6 and 16 per cent of manufacturing value added to 14 and 39 per cent, respectively. These shares would then remain constant for most of the twentieth century.

An additional important fact about the post-Second World War period is that our ‘front-runner’, Argentina, lagged behind, at least relative to Brazil, Colombia, (p.332) and itself. Since there was no threat of external competition, Argentinian manufacturers focused on the domestic market and, according to Krugman (1993), experienced growth rates below their potential. But most importantly, ISI was an ‘incomplete model’ as the inadequate growth of industrial exports during this whole period was still an obstacle. Furthermore, attempts to develop large-scale, heavy industry required a continuous injection of public subsidies.

During the ISI golden age Chile was catching up on the US but lost the opportunity of catching up even faster, as did the UK, Germany, Japan, Brazil, and Colombia. Between 1950 and 1972, manufacturing grew at 5 per cent per annum in a context of very high monetary instability and stop–go macroeconomic policies. Chile did not implement a smooth ISI strategy, and the results were not stellar (Ffrench-Davis et al., 2001; Muñoz, 1968; Cortes Douglas, Butelmann, and Videla, 1981).

Having described the policy context and industrial performance in these four countries, we can now focus on the heated debates regarding the causes and limits of industrialization in Latin America during the ISI period. At least two important issues should be highlighted. First, the measurement of protection to domestic industrial producers is complex, and its correlation with industrial growth is debated. Second, there is still a debate as to what the right counterfactual is for Latin America in the post-Second World War period. Did Latin America industrialize rapidly because of ISI policies, or despite them? Could industrial growth rates have been higher if another set of policies had been implemented?

13.4.2 Overall Effect of ISI Policies

The contrasting performance of Brazil and Colombia on the one hand, and Argentina and Chile on the other, raises a puzzle. Was protection for domestic manufacturing behind the stellar performance of the former two countries? If it was, why did Chile and Argentina not achieve the same outcomes? In order to shed some light on this puzzle, Fig. 13.1 plots what we call real distorted import price indices for the four countries. These are the prices that local producers would have observed for competing imports. The index is the ratio of import prices (in domestic currency) to domestic prices. The prices of foreign goods in the domestic market are calculated by multiplying the import price index by the average tariff and the nominal exchange rate. Domestic prices are measured by the domestic price index for industrial goods (or the industrial GDP deflator). After a base year is selected, increases in the index above 100 imply that local producers faced greater protection from imports. The index is a useful but incomplete measure because it considers the effects of quantitative trade restrictions only via its effects on the domestic industrial price index.

Industrial Growth in South AmericaArgentina, Brazil, Chile, and Colombia, 1890–2010

Fig. 13.1. Real distorted import price indices for Argentina, Brazil, Chile, and Colombia, 1900–2012 (1939 = 100)

Note: The real distorted import prices index = ((1+ avg. tariff) × nominal exchange rate × import price index)/domestic industrial price index. We then index this series taking 1939 as the base year (1939 = 100).

Source: See Data Appendix.

Fig. 13.1 shows the log scale of the real distorted import price index (1939 = 100). We focus on trends rather than on levels and interpret the evolution of the index for each country. In Brazil macro and trade policies generated increasing protection for domestic producers up to 1980. Protection in Colombia was roughly increasing (p.333) between 1950 and 1990. In Argentina it increased up to 1950, and declined thereafter. In Chile, the index declined between 1948 and 1972 and increased thereafter. It is surprising that the Pinochet government seems to have implemented a set of policies that, although they included substantial trade liberalization, overall tended to protect domestic industrial producers.

To understand the overall effects of ISI policies it is important to derive synthetic indices such as the ones plotted in Fig. 13.1. The different macro and trade policies implemented during the ISI period generated complex effects on the economy, and it is even possible that their overall effect did not involve protection from import competition. The point seems even more important to take into consideration when one considers the complex political economy influencing ISI policies (the impact of trade unions, shifts from left to right, dictatorships, etc.). However, it seems that when a country aligned macro and trade policies so as to deliver truly effective industrial protection, it also experienced higher industrial growth. Thus, there seems to be suggestive evidence that increasing protection was indeed associated with increasing industrial growth. The period from the 1940s to the 1970s was characterized by a catch-up of industrial output and productivity on the leaders. Even though not all of our countries experienced faster industrial output growth than the US, they all experienced labour productivity growth that was almost twice as fast as in the US.

(p.334) 13.4.3 The Post-Second World War Period as a Missed Opportunity

Did increasing protection lead to the highest possible industrial growth rates, or were there lost opportunities available that South American did not take, but that other regions at similar stages of development took? Fig. 13.1 suggests that more distortions were correlated with faster industrialization in Colombia and Brazil, but less so in Chile. Industrial catch-up on the US could be considered a clear indication of the success of ISI policies. But compared to other latecomers, how good were those ISI policies?

There are at least three reasons why the post-Second World War period could be considered a missed opportunity and/or a period of not so stellar industrial growth. First, using Asia as a counterfactual makes us question the importance of protectionist policies in promoting industrialization in Latin America. Second, the measures we have of industrial growth, the share of manufacturing in GDP, and labour productivity in manufacturing, are distorted by the fact that protectionism led to higher domestic prices for domestic producers. These rents were not necessarily a sign of progress for everyone in the economy. Third, protectionist policies also ended up hurting the capacity of South American countries to promote regional markets. Finally, the timing of the decline in manufacturing as a share of GDP (in the late 1970s) casts doubt on the hypothesis that ISI policies led to successful industrialization in the region.

In Northeast Asia the ISI policy mix included more incentives for domestic producers to export and compete in global markets, and more targeted tariff protection. For instance, in South Korea, even if tariff protection was spread around the same sectors as in Argentina or Colombia, tariffs were lower on average—for a shorter time—and were more targeted to promote specific industries or companies (Wade, 1990). The result of these policies was faster industrial growth than in the United States. In fact, manufacturing value added in South Korea grew at an average of 11.2 per cent in the post-1945 period, and continued at an accelerated pace in the 1980s (above 12 per cent per year) and in the post-1990 period (7.7 per cent per annum).

Comparisons with South Korea suggest that we need to improve our knowledge in at least two ways in order to better understand South America’s ISI experience. First, what is the right counterfactual for Latin America? More precisely, is industrial catch-up on the US enough, or should we develop a measure of potential catch-up to judge our countries’ performance during this period? Industrial catch-up on the world industrial leader may mask faster potential catch-up. After all, the industrial leaders were all on the brink of de-industrialization. Second, the East Asian ‘gang of four’ (South Korea, Taiwan, Hong Kong, and Singapore) were growing faster than Argentina and Chile. In addition, all four South American countries exhibited faster productivity growth in the 1930s than during the golden age of ISI (except for Argentina), which suggests that the potential for faster catch-up was there, but not achieved. If this conjecture is correct, our evidence may support the ISI stagnationist hypothesis.

(p.335) Beyond this counterfactual using Asia, we know that the price distortions depicted in Fig. 13.1 were high for Brazil and Colombia, and that high protection is correlated with faster industrial growth. However, we also know that the shares of industrial value added in GDP were larger during this period, but in part because there were large rents in the protected industries. That is, value added and productivity figures may be inflated by the presence of such rents. With distortions such as tariff protection came high internal prices that made value added seem higher than if prices had been set in international markets.

Moreover, the distortions introduced by governments in South America also affected their capacity to promote a regional market for manufactures. In contrast with East Asia, where some manufactured products had to be competitive in international markets, the prices of Latin American manufactured goods were uncompetitive on world markets; furthermore, they could not be exported to neighbouring countries either, because they also wanted to develop their own industries using a similar policy mix. Thus, Latin America missed important opportunities to spur interregional trade: low productivity and high trade barriers became major obstacles, especially for intra-regional manufacturing trade (Badia-Miró, Carreras-Marín, and Meissner, 2014). Interregional integration and industrialization did not progress, despite efforts such as the 1940 Pinedo Plan, the 1944 John Hopkins report on ‘Cooperación para la promoción del Intercambio in Argentina’, which proposed the creation of a regional free trade area, or the Andean Pact between Bolivia, Chile, Colombia, Ecuador, Peru, and Venezuela.

All in all, maybe with the exception of Brazil, until the 1960s South American manufacturing production was mostly to satisfy domestic demand. The manufacturing sector in the region was a net importer, requiring foreign exchange continuously, which produced recurrent balance of payments crises. But the exclusively inward-looking characteristic of industry started to change by 1970: the share of manufacturing in total exports increased, governments lowered average tariffs, and some quantitative restrictions were removed. In addition, the creation of the free trade areas of ALADI (1980), which all four countries joined, and MERCOSUR (1990), which changed the export profiles of Argentina, Brazil, and to a lesser extent Chile, also allowed the region to expand manufactured exports.

Finally, the timing of de-industrialization also casts doubt on the hypothesis that ISI policies led to successful industrialization in the region. In the 1970s, all four countries experienced a deceleration of their industrial growth rates from 4 to 2 per cent. This implies that de-industrialization, measured as a declining share of manufacturing in total GDP, started for most countries before their governments dismissed ISI policies and before the crisis of the 1980s. That is, de-industrialization happened not as a consequence of the demise of ISI, but during the ISI period. De-industrialization started in Chile in 1971 (while Salvador Allende was still in power, and before any major change to ISI policies), and in Argentina and Colombia by 1975, also before any important departure from ISI programmes. In contrast, Brazil did not engage in any major liberalization until the late 1980s and early 1990s.

(p.336) 13.5 The 1980S Debt Crisis and Its Impact in South America

The 1980s is again a period with mixed results. On the one hand, for Brazil and Argentina this is a decade of crisis. With the rapid rise in interest rates in the United States and the debt default of Mexico, capital markets were closed for these South American nations and their governments could not refinance their foreign debt. During this decade, industrial GDP growth rates decreased across the board, and despite rapid exchange rate depreciations, domestic industry suffered because of the contraction in domestic demand and the sudden stop in capital inflows (Frieden, 1991). Argentina and Brazil, in fact, ended up running hyperinflationary policies in the late 1980s, which forced their governments to open up and to establish fixed exchange rate regimes (in order to anchor prices). On the other hand, Chile and Colombia went through the crisis relatively unscathed, with moderate macroeconomic imbalances.

In all four countries, manufacturing output and productivity growth rates slowed substantially, and manufacturing as a share of GDP accelerated its decline. Simultaneously, the world leader, the US, and the close followers, the UK, Germany, and Japan, all experienced industrial output and productivity growth deceleration. Therefore, in relative terms Brazil and Colombia managed to catch-up, while Argentina and Chile increased its gap with these industrial leaders.

13.6 The 1990S and Beyond

During the 1990s, a period of rapid structural change, we observe a rebound from the dismal 1980s, but with extremely modest rates of growth. The industrial complex in Argentina and Brazil maintained average growth rates close to 2 per cent, while Chile’s industrial performance improved substantially, reaching average growth rates of 7 per cent. The success of Chile in the 1990s stems from the fact that its manufacturing sector gained international competitiveness, mostly in the so-called extractive industries, but also in some of the medium- and high-technology industries. According to our estimates, the Chilean manufacturing sector was the best performer of the 1990s and 2000s, with an average annual rate of growth of close to 4 per cent per year. This may be related to the fact that the Chilean terms of trade substantially improved after 1995, when the government ran a nominal exchange rate crawling peg to avoid sharp swings in the real exchange rate (Huelva and Núñez, 2010).

The 1990s were Colombia’s worst decade of the twentieth century in terms of industrial growth, which stood at just over 2 per cent average growth per annum. While trade and capital market reform advanced and the terms of trade improved, there was a cycle of exchange rate appreciation–depreciation–appreciation that ended up hurting manufacturing growth (Tovar, 1998; Ocampo and Montenegro, 2007).

Finally, during the first decade of the twenty-first century, Brazil, Chile, and Colombia experienced modest average manufacturing growth while Argentina’s (p.337) output experienced a sharp decline, followed by a sharp rebound after 2005. During this period Chile’s industrial growth rates were higher than in both the US and the other industrial leaders, while Argentina, Brazil, and Colombia did not lose too much ground on the US, and experienced industrial catch-up on the other industrial leaders (see Table 13.3).

13.7 Conclusion

This chapter has constructed new long-run series of industrial GDP growth and labour productivity growth, as well as a set of variables related to initial conditions, international trade, and macroeconomic policy. We have used these data to test four popular hypotheses: industrialization promoted endogenously by exports via an income effect in the domestic economy; industrialization occurred under adverse shocks that induced policies promoting industrialization; import-substituting industrialization induced rapid manufacturing productivity growth; and import-substituting industrialization promoted uncompetitive domestic firms.

Industrial catch-up on the global leaders (the US, the UK, Germany, and Japan) did take place. But catch-up was not experienced during the whole period, and its pace was uneven over time and across countries.

The initial conditions facing these countries in 1900 were varied, and had an important impact on their subsequent industrial development. Argentina was characterized by higher human capital, and urbanization and transportation advantages, compared to the other three countries, and had begun its industrial development earlier. Brazil’s size was not an advantage at this time: its large population was still poor, illiterate, sparsely located, and far from water transport. Strong industrial development only started at the turn of the twentieth century. Chile was relatively well endowed with human capital and low transport costs, but it had a small domestic market that was only moderately urbanized. It industrialized slowly. Like Brazil, Colombia’s population was relatively illiterate, poor, sparsely located, and far from water transport. It was the industrial latecomer of this group.

There was an important heterogeneity across countries and periods in terms of the causes of industrialization and the policies adopted. Brazil’s experience highlights the very different sources of industrial growth over time: strong endogenous industrialization (1900–30), an adverse shock and export boom (1930–44), ISI (1944–80), and weak endogenous industrialization (1991–2010). Chile’s failure to catch-up on the US and the other industrial followers in the early stages of the twentieth century highlights the fact that industrialization via exporting was not automatic, even if many initial conditions had been already achieved. Argentina’s failure to converge during the decade of fastest convergence in Latin America, the Great Depression, shows that even if most countries adopt similar policies, some are lucky and export and industrialize, while others do not. And Colombia’s impressive Great Depression and ISI industrial performance highlights the importance of combining protectionist policies with interventions to reduce the disadvantages of initial conditions.

(p.338) This chapter thus highlights the importance of considering the international context, internal policies, initial conditions, and the nature of the country’s export products, in understanding industrialization. The point is particularly important in the context of the literature on Latin America’s industrialization, as this has emphasized policies, while downplaying the importance of these other factors, as well as the heterogeneity of country experiences within the region.

Finally, the share of manufacturing in GDP followed an inverse U-shape in South America during this period. The relative importance of the industrial sector increased rapidly after the Great Depression, peaked in the 1970s, and then fell during a period of relative de-industrialization. We have seen that industrialization in most countries required protectionism. With a few exceptions, there was a large retrenchment from manufacturing once these economies started opening up to the world economy after the 1990s. We do not argue that the policies that promoted industrialization before 1980 should be tried again. On the contrary, our estimates show that the de-industrialization of South America started before the demise of ISI. Latin American policies before 1970 enjoyed only short-term success, and only in some industries did the region develop long-term comparative advantage.

Series on manufacturing value added for Argentina, Brazil, Chile, and Colombia have been produced by researchers and agencies at various times. The most frequently used sources include the World Bank’s World Development Indicators (WB) and the Economic Commission for Latin America (ECLA). A project initially based at Oxford University and now at the Universidad de la República, Montevideo, collected and collated substantial ECLA data and has made it easily accessible for free via an internet website named MOxLAD (MOxLAD, 2014). The data on this website are slightly different from what we collected directly from ECLA reports (ECLA, 1966). After careful analysis, we decided to use manufacturing value added series put together by local experts in each country, rather than the MOxLAD or ECLA data. These series are usually longer, behave similarly to the ECLA series, and incorporate substantial local knowledge. That is, they purge distortions associated with political manipulation of the data from the series.

We have made additional adjustments to these manufacturing value added series. For Argentina we use manufacturing GDP in constant 1960 local currency units (LCU), constructed by Orlando Ferreres for 1875–2012 (Ferreres, 2005). For Brazil, we use IPEA’s series of industrial value added in current LCU, deflated by the GDP deflator, for 1908–70. We then link this to the IPEA’s real industrial value added series in LCU (deflated using the industrial GDP deflator) for 1971–2012 (IPEA, 2014). For Chile, we use the Díaz, Lüders, and Wagner (DLW) manufacturing value added series (in 1996 constant LCU) for the period 1900–2004, and extrapolate this to 2005–12 using real manufacturing GDP growth rates from Banco Central de Chile (Díaz, Schwarzenberg, and Wagner, 2007). Finally, for Colombia, we use the real manufacturing GDP for 1925–2012 from Banco de la República (1998). The four series are converted into indices with 1960 as the base year (1960 = 100).

We calculate growth rates of manufacturing value added and manufacturing labour productivity in LCU. Since we do not have PPP exchange rates for the whole period, we (p.339) prefer to assume that policy-induced nominal exchange rate distortions are (eventually) translated into inflation, and are therefore accounted for in local currency series. Although this is not ideal, estimating century-long PPPs for the four countries is beyond the scope of this chapter. Furthermore, hyperinflation and exchange rate events in Argentina and Brazil suggest that this is probably the best way forward, as we know that the local expert series have taken hyperinflation, the adoption of new currencies, and exchange rate events into account. Thus, average long-term growth rates are comparable across countries and over time, although care was taken if important exchange rate or inflationary events took place precisely at the cut-off dates of our periodization.

References

Bibliography references:

Badia-Miró, M., Carreras-Marín, A., and Meissner, C. M. (2014). Geography, policy, or productivity? Regional trade in five South American countries, 1910–1950. National Bureau of Economic Research Working Paper No. 20790.

Baer, W. (1972). Import substitution and industrialization in Latin America: experiences and interpretations. Latin American Research Review 7, 95–122.

Baer, W. (2008). The Brazilian Economy: Growth and Development. Boulder, CO: Lynne Rienner.

Barbero, M. and Rocchi, F. (2003). Industry. In A New Economic History of Argentina (Eds, Della Paolera, G. and Taylor, A. M.). Cambridge: Cambridge University Press, 261–94.

Bates, R. H. (1998). Open-Economy Politics: The Political Economy of the World Coffee Trade. Princeton, NJ: Princeton University Press.

Bénétrix, A. S., O’Rourke, K., and Williamson, J. G. (2015). The spread of manufacturing to the poor periphery 1870–2007. Open Economies Review 26, 1–37.

Bulmer-Thomas, V. (2003). The Economic History of Latin America since Independence. Cambridge: Cambridge University Press.

Cano, W. (1977). Raízes da Concentração Industrial em São Paulo. Rio de Janeiro: Difel.

Coatsworth, J. H. and Williamson, J. G. (2004). Always protectionist? Latin American tariffs from independence to Great Depression. Journal of Latin American Studies 36, 205–32.

Colistete, R. P. (2009). Revisiting import-substituting industrialization in Brazil: productivity growth and technological learning in the post-war years. Paper presented at conference on Latin America, Globalization, and Economic History, 24 and 25 April.

Cortes Douglas, H., Butelmann, A., and Videla, P. (1981). Proteccionismo en Chile: una vision retrospectiva. Cuadernos de Economía 18, 141–94.

Dean, W. (1969). The Industrialization of São Paulo. Austin, TX: University of Texas Press.

Diaz-Alejandro, C. F. (1970). Essays on the Economic History of the Argentine Republic. New Haven, CT: Yale University Press.

Diaz-Alejandro, C. F. (1976). Foreign Trade Regimes and Economic Development: Colombia. New York: National Bureau of Economic Research.

Díaz, J., Schwarzenberg, R. L., and Wagner, G. (2007). Economía Chilena 1810–2000: producto total y sectorial: una nueva mirada. Pontificia Universidad Católica de Chile, Instituto de Economía, Documento de Trabajo 315.

Echavarría, J. J. (1993). External shocks and industrialization: Colombia, 1920–1950. D.Phil. dissertation, University of Oxford, 1993.

ECLA, United Nations (1966). The Process of Industrialization in Latin America. Santiago: United Nations.

(p.340) Ferreres, O. (2005). Dos Siglos De Economía Argentina: Historia argentina en cifras. Buenos Aires: El Ateneo: Fundación Norte y Sur.

Ffrench-Davis, R., Munoz, O., Benavente, J. M., and Crespi, G. (2001). The Industrialisation of Chile during Protectionism (1940–82). In An Economic History of Twentieth-Century Latin America, Vol. 3: Industrialization and the State in Latin America: The Postwar Years (Eds, Cardenas, E., Ocampo, J. A., and Thorp, R.). Houndmills, Basingstoke: Palgrave Macmillan, 114–53.

Frieden, J. A. (1991). Debt, Development, and Democracy: Modern Political Economy and Latin America, 1965–1985. Princeton, NJ: Princeton University Press.

Furtado, C. (1959). Formação Econômica Do Brasil. Rio de Janeiro: Ed. Fundo de Cultura Econômica.

Goldin, C. and Katz, L. F. (1996). Technology, skill, and the wage structure: insights from the past. American Economic Review 86, 252–7.

Haber, S. H. (1991). Industrial concentration and the capital markets: a comparative study of Brazil, Mexico, and the United States, 1830–1930. Journal of Economic History 51, 559–80.

Haber, S. H. (2006). The political economy of industrialization. In The Cambridge Economic History of Latin America, Vol. 2: The Long Twentieth Century (Eds, Bulmer-Thomas, V., Coatsworth, J., and Cortes-Conde, R.). Cambridge: Cambridge University Press, 537–84.

Haddad, C. L. and Contador, C. R. (1975). Produto, moeda e preços: Brasil 1861–1970. Revista Brasileira de Estatística 36, 407–70.

Herranz-Loncán, A. (2014). Transport technology and economic expansion: the growth contribution of railways in Latin America before 1914. Revista de Historia Económica/Journal of Iberian and Latin American Economic History 32, 13–45.

Hirschman, A. (1968). The political economy of import-substituting industrialization in Latin America. Quarterly Journal of Economics 81, 1–32.

Huelva, D. C. and Núñez, S. R. (2010). Cambio estructural de la industria manufacturera en Chile: 1979–2004. Revista de Economía Mundial 26, 27–51.

IBGE (1990). Estatísticas Históricas Do Brasil. Séries Econômicas, Demográficas E Sociais De 1550 a 1988. Rio de Janeiro: IBGE.

IPEA (2014). Ipeadata. Rio de Janeiro: IPEA. Available from www.ipeadata.gov.br/

Katz, J. and Kosacoff, B. (2001). Technological learning, institution building and the microeconomics of import substitution. In An Economic History of Twentieth-Century Latin America, Vol. 3: Industrialization and the State in Latin America: The Postwar Years (Eds, Cardenas, E., Ocampo, J. A., and Thorp, R.). Houndmills, Basingstoke: Palgrave Macmillan, 282–313.

Krueger, A. O. (1978). Liberalization Attempts and Consequences. Cambridge, MA: Ballinger and NBER.

Krugman, P. R. (1993). First nature, second nature, and metropolitan location. Journal of Regional Science 33, 129–44.

Lederman, D. (2005). The Political Economy of Protection: Theory and the Chilean Experience. Stanford, CA: Stanford University Press.

Leff, N. H. (1968). The Brazilian Capital Goods Industry, 1929–1964. Cambridge, MA: Harvard University Press.

Love, J. L. (2005). The rise and decline of economic structuralism in Latin America: new dimensions. Latin American Research Review 40, 100–25.

Macario, S. (1964). Protectionism and industrialization in Latin America. Economic Bulletin for Latin America 9, 61–101.

(p.341) Maloney, W. F. (2002). Missed opportunities: innovation and resource-based growth in Latin America. Economía 3, 111–67.

Melo, J. O. (1987). Las vicisitudes del modelo liberal (1850–1899). In Historia Económica De Colombia (Ed, Ocampo, J. A.). Bogotá: Siglo XXI editores, 135–89.

MOxLAD (2014). Montevideo–Oxford Latin American Economic History Database (Eds, University of Montevideo and Oxford University). Montevideo. Available at http://www.lac.ox.ac.uk/moxlad-database.

Muñoz, O. (1968). Crecimiento Industrial De Chile 1914–1965. Santiago, Chile: Universidad de Chile, Instituto de Economía.

Musacchio, A. (2009). Experiments in Financial Democracy: Corporate Governance and Financial Development in Brazil, 1882–1950. New York: Cambridge University Press.

Musacchio, A. and Lazzarini, S. G. 2014. Reinventing State Capitalism: Leviathan in Business, Brazil and Beyond. Cambridge, MA: Harvard University Press.

Musacchio, A., Martínez Fritscher, A., and Viarengo, M. (2014). Colonial institutions, trade shocks, and the diffusion of elementary education in Brazil, 1889–1930. Journal of Economic History 74, 730–66.

O’Rourke, K. H. and Williamson, J. G. (1999). Globalization and History: The Evolution of a Nineteenth-Century Atlantic Economy. Cambridge, MA: MIT Press.

Obstfeld, M. and Taylor, A. M. 2004. Global Capital Markets: Integration, Crisis, and Growth. Cambridge: Cambridge University Press.

Ocampo, J. A. and Montenegro, S. 2007. Crisis Mundial, Protección E Industrialización. Bogotá: Editorial Norma.

Palma, G. (2000). From an export-led to an import-substituting economy: Chile 1914–39. In An Economic History of Twentieth-Century Latin America, Vol. 1: The Export Age: The Latin American Economies in the Late Nineteenth and Early Twentieth Centuries (Eds, Cardenas, E., Ocampo, J. A., and Thorp, R.). Houndmills, Basingstoke: Palgrave Macmillan, 217–64.

Pineda, Y. (2009). Industrial Development in a Frontier Economy: The Industrialization of Argentina, 1890–1930. Stanford, CA: University Press.

Prebisch, R. (1950). The Economic Development of Latin America and Its Principal Problems. Lake Success, NY: ECLA and UN Department of Economic Affairs.

Rocchi, F. (2005). Chimneys in the Desert: Industrialization in Argentina during the Export Boom Years, 1870–1930. Stanford, CA: University Press.

Sokoloff, K. L. and Engerman, S. L. (2000). History lessons: institutions, factors endowments, and paths of development in the New World. Journal of Economic Perspectives 14, 217–32.

Stein, S. J. (1957). The Brazilian Cotton Manufacture: Textile Enterprise in an Underdeveloped Area, 1850–1950. Cambridge, MA: Harvard University Press.

Stone, I. (1999). The Global Export of Capital from Great Britain, 1865–1914: A Statistical Survey. Houndmills, Basingstoke: Macmillan.

Suzigan, W. (1986). Indústria Brasileira: Origem e Desenvolvimento. São Paulo: Brasiliense.

Tavares, M. C. (1972). O processo de substituição de importações como modelo de desenvolvimento na América Latina. In Da Substituição de Importações ao Capitalismo Financeiro (Ed., Tavares, M. С.). Rio de Janeiro: Zahar, 47–150.

Taylor, A. M. (1998). On the costs of inward-looking development: price distortions, growth, and divergence in Latin America. Journal of Economic History 58, 1–28.

Tovar, J. (1998). La Industria Manufacturera, 1967–1995. In Colombia: Estructura Industrial e Internacionalización 1967–1996 (Ed., Garay, L. J.). Bogotá: Ricardo Alonso, 453–530.

(p.342) United Nations/ECLA (1951). Economic Survey of Latin America, 1949. New York: United Nations.

Versiani, F. R. (1980). Industrial investment in an export economy: the Brazilian experience before 1914. Journal of Development Economics 7, 307–29.

Wade, R. (1990). Governing the Market: Economic Theory and the Role of Government in East Asian Industrialization. Princeton, NJ: Princeton University Press.

Williamson, J. G. (2006). Globalization and the Poor Periphery before 1950. Cambridge, MA: MIT Press.

Williamson, J. G. (2011a). Industrial catching up in the poor periphery 1870–1975. National Bureau of Economic Research Working Paper Series No. 16809.

Williamson, J. G. (2011b). Trade and Poverty: When the Third World Fell Behind. Cambridge, MA: MIT Press.

Notes:

(1) Maria de la Paz Ferro and Daniel Habermacher provided superb research assistance for this work.

(2) The coffee valorization programmes in Brazil, and later on in Colombia, were government programmes restricting the supply of coffee by stockpiling it to keep international prices stable. The first programme operated from 1906 to 1914, and subsequent programmes were tried in the 1920s, until Brazil began burning large amounts of coffee in 1931. By World War II Colombia and Brazil had organized a cartel under the Inter-American Coffee Agreement and by 1962 the cartel was global, with quotas for two dozen countries. These programmes stabilized the terms of trade and made coffee the most important source of foreign exchange for Brazil and Colombia throughout most of the twentieth century. For a detailed description of the programmes see Bates (1998).

(3) To develop a native aeronautics industry, ITA worked closely with the Brazilian air force and the state-owned airplane manufacturer Embraer, which adopted key technologies from foreign firms.