Manufacturing convergence in the Southern Cone: New evidence for the industrialization period

ABSTRACT

The objective of this paper is, on the basis of new evidence, to contribute to the analysis of the performance of the manufacturing industries in Brazil, Chile, and Uruguay during the state-led industrialization period and in comparison with a developed country. Specifically, this paper estimates the productivity gap between Brazil, Chile, and Uruguay relative to the United States in order to reveal whether convergence took place at the industry level. The results identify changes within the industrial sector in the three Latin American countries. In short, manufacturing in Brazil achieved substantial changes, which were reflected in favourable structural change and manufacturing convergence. Moreover, manufacturing convergence accelerated in Brazil in the 1960s, when the development model based on industrialization deepened. Structural transformation was weak in Uruguay and mild in Chile, and the ability to reduce technological gaps was limited to industries based on natural resources with medium and high levels of industrial protection. The latter must also be linked to the different pace of industrialization in these two countries, especially in Uruguay, where the industrializing impulse was exhausted very early on.

KEYWORDS:

• Industrialization
• productivity
• comparative perspective

Acknowledgements

I thank the editor Alfonso Herranz and two anonymous referees for their useful comments.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Notes

1 I approach the concept of productivity considering only the quantity of labour (Ark 1993). Therefore, I focus on labour productivity, defined as the value added per worker. This means leaving aside differences in other factors’ productivity and also in the quality of labour (such as those associated to the educational level). Besides, the impact of changes in the number of hours worked per employee cannot be explored here either, due to a lack of data, since the statistics available for the period do not provide that information. Therefore, the reader must keep in mind that I always refer to labour productivity measured as value added per worker.

2 Although Argentina was a natural candidate to be included in this work, it is excluded due to the lack of the necessary data to construct a benchmark using the industry of origin approach. More precisely, the absence of physical quantities and output values at the product level is determinant for its exclusion. Frankema and Visker (2011) compared industrial labour productivity in Argentina and Australia, for the period 1907–73, but they used the official exchange rate to calculate the benchmark between the two countries, due to unavailable conversion rates based on purchasing power parities (the industry of origin approach).

3 Moreover, few countries offer a detailed industrial census comparable to the American one, which has made the US the subject of several studies on cross-country productivity comparisons. In further research, if data become available, it would be interesting to compare the countries under study with other developed and developing countries, such as the Asian newly industrialized economies.

4 The countries are: Tanzania, Zambia, India, China, Indonesia, Egypt, Hungary, Poland, East Germany, Morocco, Czechoslovakia, Portugal, Taiwan, the USSR, Korea, Mexico, Brazil, Spain, Australia, the UK, Finland, Sweden, West Germany, France, Japan, Canada, Belgium, Netherlands, and the US. See the compiled benchmarks in: https://www.rug.nl/ggdc/docs/icopsumtab2.pdf

5 The assumptions behind the products matching are quite strong: products must have comparable qualities and similar market structures, and prices must only reflect different currency values.

6 The matching procedure between Uruguay and US is explained in detail in Lara (2012); for the Brazil–US benchmark see Ark and Maddison (1994).

7 The number of products matched in Chile is only representative at the major branch level; this means that it is not possible to present results at the industry or branch level as in the case of Brazil and Uruguay.

8 For Brazil I used the UVRs estimated originally by Ark and Maddison (1994).

9 Lara (2012) presents the list of products matched between Uruguay and the US for 1988.

10 This database is compiled by the Division of Productive Development and Entrepreneurship (ECLAC) and is called Programa de Análisis de la Dinámica Industrial (PADI). The database provides information about value added and employees for Latin American countries since 1970.

11 I thank the authors for providing me the complete database.

12 The series of output and employment based on the industrial surveys begin in 1966. Colistete (2007) filled the gap before 1966 using the 1949 Brazilian industrial census and applying linear interpolation.

13 In a future step, it could be interesting to explore new sources in order to produce more benchmarks whenever possible. However, data limitations are a very binding constraint. In particularly, as far as I have explored in the case of Chile, the required data are not available in other industrial surveys or industrial censuses, and therefore it is not possible to construct a benchmark for a more recent period than 1939.

14 This coverage ratio for Uruguay is similar to other cases in which a developing country is compared to the United States for a similar period (Ark and Maddison 1988; Pilat 1991; Ark 1993).

15 The high relative labour productivity level in chemicals between 1945 and 1960 should be viewed cautiously and studied in depth using other secondary sources.

16 Textiles and chemicals showed higher relative labour productivity in Uruguay than in US during the 1940s and 1950s. This outstanding result, in the case of the textile industry, may be attributed to the important increases of labour productivity during the import substitution period in Uruguay. Camou (2010) explained that this sector included new products and new techniques, favoured by state protectionism, and widened its export profile in the 1940s. On the other side, the fact that chemicals includes petroleum may lead to distortionary results. Correcting for this distortion would require a broader analysis with additional sources on this sector.

17 Although the sample size is small in the comparison between Uruguay and US, the ADF test can still be applied (Stock and Watson 2012), and Zivot and Andrews present different tests in order to deal especially with small samples (Zivot and Andrews 1992).

18 After the 1980s, though, Lavopa (2015) found that Brazil lost ground compared to the American level.

19 This can be clearly observed in the most recent data of the Maddison Project Database.