1. INTRODUCTION
In the recent decades, the view of the way in which the first globalisation (1880-1930) impacted the Latin American economies has changed considerably. In contrast with the rather negative portrayals provided by structuralism and dependency theory, the idea that global integration opened up opportunities from which countries could profit according to their own resource endowment and domestic situation (Bulmer-Thomas Reference Bernstein1994), or even the notion of «commodity lottery» (Díaz-Alejandro Reference Cárdenas1988), led to more nuanced approaches to this crucial experience. In the more sophisticated interpretations of Latin American performance during the export era there is place for winners and losers, with internal factors playing a role as important as the one granted to external forces.
However, even in the more optimistic versions of current interpretations, Mexico has been portrayed more as an example of failure than of success on many counts. This literature grants that Mexico’s export sector showed larger diversification than most countries’ within the region and that its exports grew at a remarkable pace for over 40 years. Nevertheless, it also underlines that in Mexico the export sector remained small, exports per head were modest and, above all, industrial growth was disappointingFootnote 1 . This picture is confirmed in the more recent—revised—edition (2014) of Victor Bulmer-Thomas’ canonical work, which provides more detailed information on sources and methodology in the appendices. According to this account, by 1913 Mexico’s industry would have contributed with barely 12 per cent to GDP, far below Argentina, Chile and even a small, rural country like Guatemala—and its position would have even worsened by 1929Footnote 2 . Was Mexico less industrialised than Guatemala, for real?
This perception is at odds with a whole array of research that sustains the idea that Mexico started to industrialise well before ISI policies took hold in Latin America (Cárdenas Reference Bulmer-Thomas1987; Haber Reference Díaz-Alejandro1989; Cerutti Reference Bulmer-Thomas1992). This literature sheds light upon progress made in a variety of industrial branches, from textiles to steel mills, and upon the fact that import substitution had practically been completed in Mexico for non-durable consumer goods by 1929. In fact, a consensus has emerged among these scholars that not only Mexico had made significant steps towards industrialisation before the Great Depression, but that this phenomenon took place as an endogenous outcome of export-led growth (Haber Reference Fernández Hurtado2006; Salvucci Reference López Mateos2006; Kuntz Ficker 2007)Footnote 3 . How to reconcile this view with the apparently sluggish performance summarised above? The answer is rather simple. Conventional national accounts data contain a mistake as a result of which industrial output before 1929—and its share in the nation’s GDP—is underestimated. This error consists in the misleading identification of industry with manufacture, ignoring the contribution to Mexican industrial production made by the metallurgical sectorFootnote 4 . By incorporating the value added from metallurgy to the net output of manufactures the share of industry in GDP grows accordingly, placing Mexico among the most industrialised countries of Latin America by the end of the export era.
Section 1 of this contribution traces the origins of the available data on the sectorial distribution of GDP and shows how we know that metallurgy has been excluded from industrial net output. The second one discusses the reasons why this may have happened and why it is a mistake. In section 3 we estimate the share of the so-called mining output that should be accounted as the value added by metallurgy and incorporated to the industrial sector, and present a new estimate of sectorial GDP that better portrays the actual state of development of the Mexican economy in the early decades of the 20th century. The paper concludes by discussing some implications of these results for the interpretation of Mexico’s performance during the first export era and for the comparative analysis of Latin American countries.
2. THE AVAILABLE ESTIMATES OF MEXICO’S GDP BY SECTORS, 1895-1945
Let us start by recognising the fragility of Mexico’s GDP estimates for the period before 1950 (Riguzzi Reference Kuntz Ficker2009, p. 350). Figures are scant and there is almost no explanation of the way in which they were constructed. For lack of better data, scholars may find it necessary to use them as rough indicators of Mexico’s macroeconomic performance over time. Hence the importance of rectifying the available estimates to the extent of our knowledge for them to provide a truer picture of the structure and characteristics of the Mexican economy at that time.
The organised account of Mexico’s national product started only in 1946 with the publication of the first time series of net national income by sectors from 1929 to 1945 (Sáenz Reference Kuntz Ficker1946)Footnote 5 . Later on, Mexico’s adherence to the Agreements of the International Monetary Fund led to the establishment of a specialised agency within the Bank of Mexico aimed at generating a system of national accounts, which gave place to the publication of national income estimates for the period 1939-1945 (INEGI Reference González Reyna2003, p. 27). It was only in 1960 that an officially promoted publication (Beltrán et al. 1960) aimed at providing long-term series of some macroeconomic indicators, with the purpose of giving «an overview of the accomplishments achieved by the country» thanks to the Mexican revolution (López Mateos 1960, p. xiii). In a chapter titled «National product» a retrospective GDP series starting in 1895 was presented for the first timeFootnote 6 . As its author explained
The estimates of gross national product, in total and by activities, from 1895 to 1910 and from 1921 to 1938 … were made … extrapolating by means of volume indices of production (Pérez López 1960, p. 574).
This rather fragile estimate (what we will call the «Beltrán» series) became one of the two standard sources for Mexico’s GDP accounts for the years before 1939, in total as well as by sectors. It is precisely this series (for the years 1895-1938, in constant 1950 prices) the one that appears in the first editions of INEGI’s broadly used volumes of historical statistics, and the basis for ECLA’s compilation of macroeconomic indicators for Latin American countriesFootnote 7 . It is also the one employed by Bulmer-Thomas in his classic book and by those who rely on itFootnote 8 .
A somewhat different series stems from Banco de Mexico (BM) and was built under the direction of Leopoldo SolísFootnote 9 . Although in this case there is no explanation as to its origin or method of construction, Reynolds states that this was actually a revision of the former (Beltrán) seriesFootnote 10 . In any event, these are the data employed by Solís and, later on, by Enrique Cárdenas on their studies of Mexico’s economic development in the 20th century (Solís 1970, pp. 79-81; Cárdenas Reference Bulmer-Thomas1987, pp. 194-195)Footnote 11 .
Generally speaking, those who have used any of these series do not undertake further analysis of the quality and features of the data, the possible biases to be found, or the way in which they were originally builtFootnote 12 . Neither is there usually a comparison between the two sources. For our purposes, Figure 1 presents the differences between them as to the share of mining/oil and manufacturing in total GDP.
SHARE OF MINING/OIL AND MANUFACTURING SECTORS IN MEXICO’S GDP ACCORDING TO DIFFERENT SOURCES
As may be observed, for this period of time INEGI follows Beltrán, while Solís follows BM. For the sake of simplicity, we may then speak of INEGI’s vs. BM’s figures to comment on these differences. First, the share of the mining/oil sector in total GDP is larger for INEGI than for BM in 1895, 1910 and 1929, and smaller in 1921. This is not the result of a larger share of manufacturing in INEGI’s estimates (as compared with those of BM), though, as its participation according to INEGI is also larger in all four benchmark years than that registered by BM. In the case of manufacturing, the difference goes from 1 to 1.6 percentage points, and is of little importance in terms of the issue at stake, which is underestimating the share of industry in total GDP.
Let us start from the basic: how do we know that there is such an underestimation? In the sources where GDP series are provided there is generally no explicit account of the methodology or of the composition of each sector. However, there is plenty of indirect evidence that points to that conclusion.
For a long time it was customary in dealing with the mining-metallurgical sector to call it just «mining»Footnote 13 . Aiming perhaps at avoiding confusion, the rest of the industrial sector was usually called «manufacturing», or even «manufacturing industry», a name that excluded metallurgy, a distinctive form of industrial activityFootnote 14 . In fact, early accounts of industrial development explicitly excluded metallurgy, as the Memoria of the first industrial census (1930), states
The Industrial Census embraced mainly manufacturing industries, but some non-manufacturing industries were included, as electricity plants … On the other hand, there were industries not included in this Census, as mining and oil …Footnote 15 .
Inasmuch as the census became the source to studies on industrial growth, the latter excluded metallurgy from what was considered «industry». For one early example, in the same volume coauthored by Beltrán et al. (1960), a chapter deals with the «capitalisation of national industry, according to census data», in which metallurgy is utterly absent (Fernández Hurtado Reference Cerutti1960, p. 616). Almost three decades afterwards, Enrique Cárdenas, in his pioneering work on Mexico’s industrialisation before 1929, relied on the same census to estimate industrial value added by branch, and thus left out metallurgy from his account of the industrial sectorFootnote 16 . More recently, Bulmer-Thomas provides us with data on the «structure of manufacturing output» in which the source for Mexico is the very same industrial census, thus excluding metallurgyFootnote 17 .
On the other hand, there is evidence that even when mining was considered part of industries in the original sources, later on it ended up being reallocated within the primary sector. It all started because although it was implicitly accepted that there was a secondary branch embedded within the «mining» sector, it was hard to separate it from the total «mining» output, as reported from earlier sources. It is for this reason that INEGI chose to include the extractive subsector among «industries» rather than the «primary sector» (Table 1).
GDP BY ACTIVITIES ACCORDING TO INEGI
Source: INEGI (Reference Flores Clair1985, I, table 9.2).
As mentioned before, ECLA relied on Beltrán for the years before 1939,—that is to say, the same source as INEGI. However, the criterion by which ECLA decided to «disambiguate» this activity was by renaming the «mining» (or extractive) sector as «mining and quarrying,» thus leaving to the reader the decision as to whether this was a primary or a secondary activity (Comisión Económica para América Latina, Reference Salvucci1978, pp. 148-151). Hence while both sources used the same original figures, the results in terms of the contribution to GDP by sectors were quite different, as may be observed in Table 2.
GDP BY SECTORS ACCORDING TO INEGI AND ECLA (PERCENTAGE UPON TOTAL GDP)
Note: percentage shares calculated from absolute values provided by the sources. Na: not available.
Sources: INEGI (Reference Flores Clair1985, I, p. 334); CEPAL (Reference Salvucci1978, pp. 148-149).
Maybe because «mining and quarrying» did not sound much like an industrial activity, users of ECLA’s data decided that it was part of the primary sector, attributing to industry only what in this source appears as «manufacturing industry.» This is at least what Bulmer-Thomas did, and it is for this reason that in his indicators of Latin American industrial development by countries Mexico appears below Argentina, Chile, and Guatemala (and barely above El Salvador) by 1913 and also below Uruguay and Brazil by 1928 (Table 3)Footnote 18 .
MANUFACTURING OUTPUT IN SELECTED LATIN AMERICAN COUNTRIES, ACCORDING TO BULMER-THOMAS (SHARE OF GDP)
Source: Bulmer-Thomas (2014, p. 145 and 206). Na: not available.
1 1920 for Brazil, Guatemala and El Salvador; 1910 for Mexico.
2 1929 for Chile; 1930 for Uruguay.
On the other hand, INEGI’s classification is correct in considering that there was value added in the so-called «extractive» branch of industry (see Table 1). However, it is also inaccurate as it ignores that part of it was composed by the raw material, namely, ores as extracted from the soil, and thus belonged to the primary sector. As a result, industrial output ends up being overestimated if one uses INEGI’s summary figures by sector (originally in absolute values) to calculate percentage shares (Table 2). A more precise account of industrial net output would include value added by metallurgy but put aside inputs, comprising, of course, ores.
3. WHY WAS METALLURGY EXCLUDED FROM THE INDUSTRIAL SECTOR?
There are several reasons why the metallurgical value added may have been excluded from industrial output in early estimates of Mexico’s GDP by sectors. Three of them come to mind.
First, because it was not considered an industry. Before 1929 Mexican metallurgy did not usually produce finished articles but rather performed the basic process of smelting and refining ore minerals in order to obtain concentrates or metal bullion. This might be the reason why it was considered part of the extraction process, a primary activity in itself. However, this would be a rather misleading way to look at the difference between extraction and processing, between ores and basic metal products, one that would neglect the dramatic change that represented the introduction of smelting and refining plants into Mexico’s industrial landscape.
The metallurgical industry reached considerable proportions between its establishment in the early 1890s and the eve of the Mexican revolution (1910). By then three plants owned by ASARCO provided 1.3 million tons in annual smelting capacity (Mexican Yearbook 1911, p. 210); another one, owned by the Guggenheims and considered at the time «the larger custom smelter in North America», added another 500 thousand tons. All together, the more than four dozen plants operating in Mexico had a total smelting capacity of 6.5 million tons per yearFootnote 19 . With respect to silver-lead blast furnaces, 45 out of 124 that existed in North America were located in Mexico, for an «annual charge capacity [that] was slightly over 40 per cent of the United States annual total». Even though the Mexican revolution may have halted progress during the 1910s, technological advancement continued during the 1920s as selective flotation was introduced, making of Mexico the second world producer of lead and one of the world’s largest producers of zinc (Bernstein 1964, pp. 38-40, 139)Footnote 20 .
In fact, by the end of the export era most of Mexico’s «mineral» production and exports had some degree of industrial processing. According to an official source, in 1927 Mexico’s total output of mining products (by value) included 5 per cent of raw minerals, 20 per cent of concentrates and 75 per cent of metallurgical products, the latter composed in >90 per cent by gold, silver and lead refined bars, lead mixed bars and copper bars (Anuario 1929, p. 2). As the United Nations’ Standard Industrial Classification stated in 1949, these products were to be considered part of the «Non-ferrous metal basic industries»Footnote 21 .
The second possible reason is that the metallurgical industry was (largely) foreign owned. As is well known, there is a difference between GDP and gross national product (GNP) that allows distinguishing between what is produced within the country’s boundaries by nationals and foreigners (GDP), and what is produced only by its own citizens, within or outside the nation’s boundaries (GNP). The most commonly used indicator is the former, as it allows to measure a country’s economic performance, while the latter places the accent upon the performance of its citizens, inside their nation or abroad. Early estimates of Mexico’s total output came under the name of «Producto Nacional Bruto», that is to say, GNP (Fernández Hurtado Reference Cerutti1960; Pérez López 1960). More than a conscious decision, it seems to be an inaccurate way to label data that aimed at representing Mexico’s GDP, as there was no attempt at estimating the contribution of Mexico’s citizens abroad. However, it cannot be ignored that these first approaches took place amid an atmosphere of combative nationalism that could have (unconsciously?) led to pay less attention to those activities that were not under direct control of national entrepreneurs and the revolutionary State. In any event, there is no reason whatsoever to dispense with metallurgical net output just because it was mainly foreign owned. In fact, there were some manufacturing industries in which foreign investment was present, although to a lesser extent, and they were still included.
The third reason may be because most of the so-called «mining» product was aimed at foreign markets. Before dealing with it, let us first make a slight amendment to that assessment. At the beginning of the export era practically all of the mining output (other than some specie for domestic use) was export oriented. Later on, after the emergence of the metallurgical industry, a large part of the raw material extracted and all of the concentrates produced remained in Mexico for further processing, while most «metallurgical products» (92 per cent in 1927) were sold abroad (Anuario 1929, p. 2). In any case, this would not be a valid reason to ignore the contribution of this branch to industry in the computation of GDP.
As obvious as it seems, some confusion emerges as long as the literature on the first export era tends to identify the export sector with the production of primary products and the domestic sector with the place where industry was expected to appear. For one example, while talking about the emergence of a manufacturing sector in Latin America Bulmer-Thomas refers to «domestic industry geared to the home market»Footnote 22 . This does not mean, of course, that he or anybody else considers explicitly that any industrial activity oriented to the external markets should not be included in that country’s industrial GDP. It suffices to say that most of Argentina’s industrial output by 1929 was composed of articles that were to a large extent for export, like frozen meat and wheat flour, as the same author recalls (Bulmer-Thomas 2014, p. 83, 198). But if «straw hats from Ecuador and Colombia» are to be considered industrial products for export (Bulmer-Thomas 2014, p. 83), why should not be the products of the Mexican metallurgical industry?
4. HOW TO ESTIMATE THE METALLURGICAL SHARE OF MINING OUTPUT?
There is some difficulty at separating metallurgical value added from total «mining» product. Most historical series related to this sector consist of aggregate figures of what in the best of cases is called «mining-metallurgical output»Footnote 23 . On occasions, export series distinguish between raw material and metal exports, but only for a few products, or for a broader sample of products but just for a few yearsFootnote 24 . There are only two sources that explicitly differentiate mining output from metallurgical output. The first one comprises the period 1897-1907. According to it, a steady 53 per cent of the total mining-metallurgical output (by value) was composed by metallurgical products within those years (Flores Clair Reference Coatsworth1985, p. 142). There is reason to believe that industrial value added is underestimated in this account, particularly as we move on in time, as it is unlikely that its share remained stagnant while the smelting industry was growingFootnote 25 . The second source has already been mentioned and provides a detailed account of Mexico’s «main mineral and metallurgical products» for the years 1926 and 1927 (Anuario 1929). According to it, the value added by the metallurgical industry (including «metallurgical products» but not concentrates) reached 75 per cent of total mining-metallurgical output in 1927Footnote 26 .
If the metallurgical share of «mining-metallurgical» output started at 50 per cent in 1895 and ended up at 75 per cent in 1927, in order to provide an estimate for 1910 the only big assumption that we need to make here is that by then it found itself at some point between the two given figures. As the industry experienced sustained growth before the Mexican revolution it would not be daring to suggest that its share reached 65 per cent of total «mining-metallurgical» output in 1910. Knowing that these are not exact figures but reasonable estimates, we have applied these shares to the data provided by INEGI to recalculate the percentage distribution of GDP by sector. Table 4 presents the results.
GDP BY SECTORS, ORIGINAL AND CORRECTED FIGURES (PERCENTAGE SHARES CALCULATED FROM CONSTANT VALUES, 1950 PRICES)
Sources: Original data in INEGI (Reference Flores Clair1985, I, pp. 313-319). Revised shares for metallurgy estimated from Flores Clair (Reference Coatsworth1985, p. 142) and Anuario (1929, p. 2). For 1910, see text.
1 Includes agriculture, cattle raising, forest, and fishing.
The revised shares in Table 4 give a more precise view of the extent to which structural change had progressed in Mexico within this period. Between 1895 and 1929 the agricultural sector (broadly taken) diminished its size from 33 to 21 per cent of GDP and the extractive stage of mining remained stable, while at the same time industry, in its manufacturing and metallurgy components, grew from 12 to 20 per cent of GDPFootnote 27 .
We may now use these new shares to reassess Mexico’s place among the Latin American countries in terms of industrial development, as presented in Table 5.
INDUSTRIAL OUTPUT IN SELECTED LATIN AMERICAN COUNTRIES, WITH REVISED FIGURES FOR MEXICO (SHARE OF GDP)
This table provides a small difference but an entirely different picture. In the first third of the 20th century, Mexico ranked first and compared closely with Argentina in terms of industrial development, followed at some distance by Uruguay, Chile, Guatemala and BrazilFootnote 28 . Moreover, while Chile seems to have experienced a setback between 1913 and 1928 and Brazil a rather modest progress, Mexico and Argentina appeared as the only countries in which the share of industry in GDP was consistently growing during these yearsFootnote 29 .
5. CONCLUDING REMARKS
It is not easy to tell when Mexico’s GDP data by sectors started to include metallurgical products as a part of industrial output. In the 1930s activity in the mining-metallurgical sector declined and, as soon as the economy recovered from the Great Depression, manufactures started to outgrow metallurgy. The combination of these events makes it hard to identify at what point metallurgical value added was taken out from «mining» and incorporated to industrial net output. Starting in 1940 official publications usually considered these activities as part of industrial production, but did not necessarily separate the extractive phase from the industrial value added (Secretaría de la Economía Nacional 1942, p. 546ff). In the late 1940s Mexico received advice on this matter by international organisations and adopted United Nations’ Standard Industrial Classification as soon as it was published (United Nations Statistical Office 1949). Thus it is certain that at least after that year national accounts distinguished correctly between mining and industrial output—the latter including metallurgy.
Mexico’s GDP shares by sector presented in Table 4 come as no surprise for those familiar with the way in which the Mexican economy evolved during the export era. This country enjoyed the most diversified export sector within the Latin American region. It embraced products from agriculture, forest, cattle raising, mining and oil activities. Some of them, even in agriculture, experienced some degree of processing before heading for export, and the mining sector generated a large-scale, technologically advanced metallurgical branch. The outcome was an export basket with a significant industrial component, a not so common feature within the Latin American context. In fact, as we have shown here, 75 per cent of Mexico’s mineral exports was made of refined metals, the product of basic metallurgy, which contributed with 7 per cent of GDP in 1929. Although the export sector grew at a fast rate between 1870 and 1929, having started at a very low level its size was moderate by the end of the period. Perhaps for this reason, export expansion did not hinder—and most probably promoted—the development of an industrial sector oriented to the domestic market.
By the end of the export era, Mexico held one of the biggest and more diversified industrial plants in Latin America. Domestic industry supplied almost 80 per cent of the apparent consumption of non-durable consumer goods and 35 per cent of intermediate goods. Its textile industry was larger than that of Argentina and only comparable with Brazil’s; its iron and steel industry had no parallel in Latin America. Import substitution industrialisation had made considerable progress before the 1929 Great Depression hit the Mexican economy forcing its shift towards inward-looking growth. That explains why industrialisation advanced so fast amid ISI policies during the late 1930s and in the 1940s.
It is still true that Mexico’s industrial development fell short with respect to some Latin American countries in per capita terms (Bulmer-Thomas 2014, p. 145). The reason is to be found in the large share of the Mexican population that lived in the countryside until well into the 20th century. It is also accurate that some of Mexico’s industries were in the beginning unprofitable and thus in need of tariff protection (Bulmer-Thomas 2014, p. 147). But those are different matters. In terms of the composition of GDP by sectors Mexico compared favourably with the largest Latin American countries (table 5). Only by incorporating this feature into the analysis we can make sense of the performance of the Mexican economy and formulate fruitful comparisons with other Latin American economies about industrialisation amid export-led growth.
