Bulmer-Thomas V., Coatsworth J., Cortes-Conde R. The Cambridge Economic History of Latin America: Volume 2, The Long Twentieth Century

Подождите немного. Документ загружается.

Cambridge Histories Online © Cambridge University Press, 2008

P1: GDZ

0521812909c07 CB929-Bulmer 0 521 81290 9 October 6, 2005 12:28

282 Richard Salvucci

Obviously, some countries, especially those with similar export baskets,

display some correlation in the net barter terms of trade. Argentina and

Uruguay (r = .53), Brazil and Colombia (r = .57), and Mexico and Peru

(r = .53) exhibit this sort of behavior, although for time series, the cor-

relations do not seem to be extremely high. But many more of the pairs

of correlations are low or even negative, pointing instead to a diversity of

experiences. Moreover, examining the degree to which the terms of trade

cointegrate or trend together suggests little common behavior, even where

Brazil and Colombia, Argentina and Uruguay, or Peru and Mexico are con-

cerned. In other words, the national experiences of the export economies

remain, at some level, irreducibly distinct, a point Daniel D

ıaz Fuentes

emphasized in an account of the operation of the gold standard in Latin

America in the interwar period, or which, for purposes of comparison,

Angus Deaton had made for export economies in Africa. At best, it seems,

pairs of countries display common behavior, but that is all.

Nevertheless, there are patterns in the data that call for some comment,

particularly because they apparently represent the reversal of trends that

characterized the half-century between 1820 and 1870. Although it is not

true that there was a uniform deterioration in the net barter terms of trade

across Latin America, there was some tendency for a decline to occur in

many of the major economies if, by decline, we mean a lower average in the

terminal decade than in the initial one (see Table 7.6). Thus, in this sense,

Argentina, Brazil, Mexico, Peru, and Venezuela experienced some deteriora-

tion in the terms of trade, and in Mexico, the deterioration was pronounced

because the terms of trade had fallen by 60 percent over the period. But

Colombia, Chile, Cuba, and Uruguay all experienced improving terms of

trade, and for Cuba, the reversal was all the more dramatic considering the

sharp fall in the relative price of Cuban exports earlier in the nineteenth cen-

tury. Moreover, if we look instead at the time trend ﬁt to each series, we see

a slightly different result. We may regard Argentina, Cuba, and Colombia

as essentially trendless because their positive coefﬁcients do not differ sig-

niﬁcantly from zero. Chile and Uruguay each present, to a greater or lesser

extent, a positive trend. Brazil, Mexico, Peru, and Venezuela all present a

negative trend and, in the case of Mexico and Peru, the deterioration ranges

from 1.6 to 2.4 percent per year. Thus, the “commodity lottery” favored

some countries, worked decidedly against the interests of still others, or

was irrelevant to a few more. A common pattern, much less a general-

ized, secular deterioration in the terms of trade, is difﬁcult to ﬁnd. Indeed,

there seems to be little apparent connection between the growth of output

Cambridge Histories Online © Cambridge University Press, 2008

P1: GDZ

0521812909c07 CB929-Bulmer 0 521 81290 9 October 6, 2005 12:28

Export-Led Industrialization 283

Table 7.6. Net barter terms of trade. By decades, 1870–1930, and time trend in

percent per year

Brazil Mexico Colombia Cuba Chile Argentina Uruguay Peru Venezuela

1870–79 116 177 102 85 65 127

1880–89 123 140 129 88 70 98 83 136

1890–99 128 119 151 94 64 82 102 108 171

1900–09 100 100 100 100 100 100 100 100 100

1910–19 91 94 107 105 110 101 119 84 97

1920–29 93 73 120 92 91 77 61

Country Trend (percent per year) R

Argentina −0.19 −0.003

Brazil

∗

−0.76 −0.193

Chile

∗

1.15 0.330

Colombia 0.08 0.004

Cuba 0.07 −0.011

Mexico

∗

−1.61 0.809

Peru

∗

−2.43 0.663

Venezuela

∗

−1.10 0.236

Uruguay

∗

1.45 0.785

∗

Signiﬁcant at the 99 percent level. R

is adjusted for degrees of freedom.

Notes and Sources: All series have been rebased to 1900–09.Inafew instances, it has been necessary

to interpolate a missing year or two to complete the series. Most of the data employed has been

published. I include here only the principal sources. For Argentina, I am indebted to Alan Taylor.

ForBrazil, see Reinaldo Gonc¸alves and Amir Coelho Barros, “Tend

encias dos termos de troca:

a tese de Prebischeaeconomia brasiliera – 1850/1879,” Pesquisa e planejamento econ

omico, 12, 1

(1982), 109–31; for Chile, Thomas B. Birnberg and Stephen A. Resnick, Colonial Development. An

Econometric Study (New Haven, CT, 1975), 267; for Colombia, Jos

e Antonio Ocampo, Colombia

ylaeconom

ıa mundial, 95–97; for Cuba, Birmberg and Resnick, Colonial Development, 273, and

Linda K. Salvucci and Richard J. Salvucci, “Cuba and the Latin American Terms of Trade: Old

Theories, New Evidence,” Journal of Interdisciplinary History, 31, 2 (2000), 197–222; for Mexico,

Edward Beatty, “The Impact of Foreign Trade on the Mexican Economy: Terms of Trade and

the Rise of Industry, 1880–1923,” Journal of Latin American Studies, 32 (2000), 399–433; for Peru,

Seminario and Beltr

an, Crecimiento econ

omico en el Per

u, 301–302; for Uruguay, Luis Bertola, “An

Overview of the Economic History of Uruguay since the 1870s,” EH.Net Encyclopedia, edited

by Robert Whaples, July 6, 2004, http://eh.net/encyclopedia/?article=Bertola.Uruguay.ﬁnal; for

Venezuela, Baptista, Bases cuantitativas, 311, 313.

Cambridge Histories Online © Cambridge University Press, 2008

P1: GDZ

0521812909c07 CB929-Bulmer 0 521 81290 9 October 6, 2005 12:28

284 Richard Salvucci

and changes in the terms of trade given that both improvements and dete-

riorations are observed, whereas GDP increases.

The difﬁculty, to repeat, is that such questions can be handled convinc-

ingly only at a disaggregated and, here speciﬁcally, national level. Consider,

for instance, Mexico and Peru. Here, as Edward Beatty explains, the dete-

riorating terms of trade reﬂected not an overall decline in the demand for

Mexican exports but rather the fate of silver, of which Mexico was a rel-

atively large exporter and producer. Although it is true that demand for

silver was relatively inelastic, a reﬂection of the increasing demonetization

of the metal internationally after 1870,Mexican production and export

actually increased, which, in turn, depressed its market price. The same

held true for Peru, at least in the early years of the series, because Peru-

vian production and export of silver expanded as well. Silver comprised

a quarter of Peruvian exports by value in the mid-1890s, and as late as

1910, still represented 10 percent of Peruvian exports. Later, Beatty pro-

poses, the deterioration in the Mexican terms of trade was driven by the

rise in the price of its import basket, especially during World War I and its

aftermath. According to Thorp and Bertram, the same held true for Peru,

whose import prices after 1915 rose dramatically, much more so than, say,

Brazil’s and Chile’s, whose terms of trade did not decline by proportion-

ally as much. But Beatty emphasizes that the purchasing power of Mexican

exports, the so-called income terms of trade (export revenues deﬂated by an

import price index), strongly increased over the period, especially when the

barter terms of trade were falling, at an overall rate (1880–1923)of4.8 per-

cent per year. The rate of increase of the purchasing power of Peruvian

exports (calculated from data compiled from Seminario and Beltr

an) (1896–

1929)isanalmost identical 4.6 percent. Thus, for both Mexico and Peru,

the deterioration in the net barter terms of trade was offset by an increase

in the purchasing power of exports. And, in both Mexico and Peru, the

export base diversiﬁed, with Mexico producing and exporting henequen,

gold, cattle hides, and nonferrous metals. Peru, argue Thorp and Bertram,

had an unusually diverse export base for Latin America, with sugar, copper,

cotton, rubber, and petroleum successively driving export performance.

Whatever the vicissitudes of international demand, both Mexico and Peru

experienced sufﬁciently rapid productivity increases in the export sec-

tor to maintain, if not increase, their share of international commodity

markets.

The situation for Venezuela was somewhat similar in that its net barter

terms of trade deteriorated, but its income terms of trade nevertheless rose.

Cambridge Histories Online © Cambridge University Press, 2008

P1: GDZ

0521812909c07 CB929-Bulmer 0 521 81290 9 October 6, 2005 12:28

Export-Led Industrialization 285

The overall change in the income terms of trade between 1870 and 1920,

using data supplied by Asdr

ubal Baptista, was positive, at somewhat less

than 1 percent per year. But this is deceptive. The purchasing power of

Venezuelan exports is more accurately described by a nonlinear trend in

which the income terms of trade rose at about 4.4 percent per year between

1870 and 1895 and then fell at somewhat less than 1 percent per year between

1895 and 1920. The explanation for this pattern appears to be tied to changes

in the price of coffee, which accounted for a major share of exports before

petroleum took hold in the 1920s, as well as generally sharp import price

increases in the latter part of the period. Still, simply describing the pattern

of the Venezuelan terms of trade as deteriorating would be quite misleading.

Like Mexico and Peru, the purchasing power of its exports rose.

ForBrazil, Leff indicates that both the net barter and income terms of

trade rose after 1870, but his analysis is limited to the years down to World

WarI.Ifwelook instead at the more comprehensive analysis by Reinaldo

Gonc¸alves and Amir Coelho Barros, they ﬁnd that the net barter terms of

trade fell between 1870 and 1939 at a yearly rate of 0.98 percent and, indeed,

between 1870 and 1913 as well, albeit at a very low (0.09 percent) rate. This

series, in turn, has been used by the Instituto Brasileiro de Geograﬁa e

Estat

ıstica to calculate the income terms of trade. Between 1870 and 1930,

the purchasing power of Brazil’s exports rose at a rate of 2.1 percent per year

and, indeed, did so predictably except during the years between 1914 and

1922.

Thus, even where the net barter terms of trade declined – which is to

say, even where relative price trends did not favor Latin America – the

purchasing power of exports rose between 2.1 and 4.6 percent per year,

doubling roughly every ten to thirty years. This ﬁnding is entirely consistent

with the argument made by Enzo Grilli and Maw Cheng Yang in their

highly inﬂuential study of the terms of trade of the developed countries

between 1900 and 1986: the fall that occurred in the net barter terms of

trade of the primary commodity exporters after World War II may have

been more than compensated by an improvement in their income terms of

trade. Their conclusion is worth quoting:

In the presence of a strong improvement in the purchasing power of commodities

and in the income terms of trade of non-oil-exporting developing countries, the

negative welfare signiﬁcance of falling relative prices of non-fuel primary commodi-

ties should not be overstated . . . The presumption . . . is that the negative effects of

declining real export prices may have been at least in part mitigated by productivity

growth.

Cambridge Histories Online © Cambridge University Press, 2008

P1: GDZ

0521812909c07 CB929-Bulmer 0 521 81290 9 October 6, 2005 12:28

286 Richard Salvucci

In other words, in Brazil, Peru, Mexico, and Venezuela, it was productivity

growth in agriculture, mining, and extractive industries that drove down

prices, not a lack of demand.

This analysis could also be extended to countries whose terms of trade

were positive or essentially trendless. For example, from 1902 through 1930,

Cuba’s income terms of trade grew at a rate of 3.4 percent per year. When

we consider that other research has concluded that the purchasing power

of Cuban exports grew throughout the nineteenth century as well, we may

reasonably conclude that, whatever happened to the relative price of Cuban

exports, their purchasing power grew over the century from 1830 through

1930.Similarly, the Chilean income terms of trade grew at a very rapid

rate between 1888 and 1930,around 6.1 percent yearly. More precisely, they

grew at this rate until 1919, when a clear structural break marking the end

of World War I occurred, at which point the trend rate of growth fell by

65 percent! The explanation for this break was the impact of the end of

the war (and the invention of synthetic substitutes by German scientists)

on nitrate production in Chile, which accounted for a quarter of GDP. Jos

Antonio Ocampo also argues that the purchasing power of Colombian

exports rose dramatically between 1870 and 1898.

Although it is true that an argument concerning the deterioration in the

net barter terms of trade appears less than persuasive, it would be premature

to conclude that the terms of trade were therefore not an issue. Although

the literature is not conclusive, and perhaps cannot be, there is some indi-

cation that volatility in the terms of trade is associated with slower growth

and structural change. This makes intuitive sense because a string of good

years (or bad) in the export sector sends an unambiguous signal to domes-

tic markets. The problem, among others, is what counts for evidence of

volatility, or what would be considered an unacceptably high (or desirably

low) level of price instability. One conventional measure is to divide the

standard error of the regression (a measure of the difference between ﬁtted

and observed measures of the terms of trade) by their mean value. This

asks, in essence, whether a time trend is a useful predictor: do they grow

(or decline) steadily or not? As a rule of thumb, the standard error of the

regression should not be more than 10 to 15 percent of the mean of the

dependent variable. As the data in Table 7.7 show, the trend regressions

are all reasonably good models, with the relevant statistic ranging from

2 percent (Uruguay) to 6 percent (Venezuela) and a median value of 4.1.

By this standard, Cuba, Mexico, Peru, and Uruguay were relatively less

volatile in their net barter terms of trade than were Argentina, Colombia,

Chile, Brazil, and Venezuela. One suspects that there is a connection

Cambridge Histories Online © Cambridge University Press, 2008

P1: GDZ

0521812909c07 CB929-Bulmer 0 521 81290 9 October 6, 2005 12:28

Export-Led Industrialization 287

Table 7.7. Stability of the terms of trade

Percent

Argentina 4.3

Brazil 5.6

Chile 4.6

Colombia 4.3

Cuba 3.9

Mexico 2.7

Peru 3.6

Venezuela 6.0

Uruguay 2.0

Median 4.1

Note: Deﬁned at thestandard error of the regres-

sion of the terms of trade divided by the mean

of the dependent variable.

Source: Based on author’s data in Table 7.6.

Higher numbers mean greater instability.

between relatively higher volatility in the terms of trade (susceptibility to

external shocks, in other words) and the difﬁculties that Argentina, Brazil,

and Chile had with adherence to the gold standard. In other words, it

was the intrinsic and growing unpredictability of the external environment

rather than a transfer of resources from the periphery to the center that

made satisfactory macroeconomic performance difﬁcult for Latin America

between 1870 and 1930.Toput it simply, a ﬁxed exchange rate (the gold

standard, in this case) will magnify the effect that ﬂuctuations in the terms

of trade have on domestic output. Flexible exchange rates insulate a coun-

try’s economy from changes in the terms of trade. A ﬁxed exchange rate

and deteriorating terms of trade, such as Mexico experienced after moving

to gold in 1905, was unfortunate and may well have contributed to the

outbreak of revolutionary violence in 1910.

STRUCTURAL TRANSFORMATION:

THE ORIGINS OF INDUSTRIALIZATION

As Edward Beatty has argued, historians’ views of the link between export

growth and industrialization have undergone a substantial change in recent

decades. For one thing, as Beatty points out, by 1930,asubstantial amount

Cambridge Histories Online © Cambridge University Press, 2008

P1: GDZ

0521812909c07 CB929-Bulmer 0 521 81290 9 October 6, 2005 12:28

288 Richard Salvucci

of industrialization was already underway. What happened beginning with

the Great Depression in Mexico, as Enrique C

ardenas has suggested, for

instance, or after World War II elsewhere in Latin America, was not a

dramatic new departure but a continuation of changes that had begun

well before. In some cases, and Mexico has been well studied by Stephen

Haber, there can be no doubt that the true origins of modern industri-

alization can be traced back to the 1880s, if not before. Also, as Beatty

has noted, the export economies provided a foundation of physical infras-

tructure (e.g., rail networks and port facilities) or state capacity that was

to be crucial in activating and directing later efforts at import-substitution

industrialization. Still, the degree of industrialization that occurred between

1870 and 1930 should not be exaggerated. The share of agricultural out-

put in national product remained substantial, especially when measured

by employment. The data in Table 7.8 suggest this quite convincingly,

even when due allowance is made for uncertainties in their provenance,

calculation, and interpretation. From the standpoint of economic devel-

opment, most of Latin America remained poor and sparsely populated

as late as 1900.Neither condition is consistent with a high degree of

domestic industrialization, and in the absence of restrictions on imports

or deliberate attempts to foster industrialization, the result of an export

boom was more likely an expansion of the market for imported manufac-

tures rather than an expansion of domestic production. This was especially

true of resource-rich economies that had a strong comparative advantage

in foodstuffs, ﬁbers, minerals, and other raw materials. There, the prob-

lem of Dutch disease or the chronic overvaluation of the exchange rate

driven by the rapid generation of foreign exchange made industrializa-

tion especially difﬁcult. This was an old problem for the Latin American

export economies reaching well back into the silver booms of the colo-

nial period in Mexico and there had been other episodes in Brazil, Chile,

Costa Rica, Cuba, and Peru as well in the intervening years. By and large,

the degree of structural transformation that had occurred in Latin Amer-

ica by 1900 was about typical for the relative level of per capita income

in the countries for which we have data. Argentina, perhaps, is the one

example of a relatively wealthy country that had less industrialization than

one might expect – notice, in particular, the contrast with Uruguay, where

Luis B

ertola ﬁnds that the products of the modern meat industry, cattle-

breeding industry, the wheat ﬂour complex, and the production of wine

led to a particularly rapid increase in the share of manufactures in GDP in

the 1920s.

Cambridge Histories Online © Cambridge University Press, 2008

P1: GDZ

0521812909c07 CB929-Bulmer 0 521 81290 9 October 6, 2005 12:28

Export-Led Industrialization 289

Table 7.8. Sectoral composition of output, 1870–1930 (percentages rounded)

Primary Secondary Tertiary

Argentina

1900 49 22 29

1935 33 30 37

Brazil

1910 36 14 50

1920 32 17 51

1930 31 17 52

Chile

1930 34 16 50

Mexico

1900 29 12 59

1910 29 12 60

1920 35 15 50

1930 28 15 53

Peru

1900 40 15 45

1910 39 18 43

1920 40 17 43

1930 43 15 42

Uruguay

1870 47 36 18

1910 44 37 19

1930 40 44 16

Venezuela

1920 35 8 57

1930 37 14 49

Notes: Argentina is measured over 1875–1900 and 1900–35. Chile is 1907–30.For Mexico,

1920 is 1921.Uruguay is 1872 (1870) and 1912 (1910).

Primary includes agriculture and ranching in Argentina and Uruguay, agriculture and min-

ing in Mexico, and agriculture and petroleum in Venezuela. Secondary includes industry

and construction in Argentina and Venezuela, and manufacturing and construction in

Mexico and Venezuela. Tertiary includes services variously deﬁned: for example, trans-

portation, commerce, and government in Argentina; distribution and services in Peru; and

in Mexico, commerce, transportation, communications, electricity, and services.

Sources: Argentina: Ana Mar

ıa Cerro, “La conducta c

ıclica de la actividad econ

omica de

Argentina en el per

ıodo 1820–1970.” Brazil: Marcelo de P. Abreu and Afonso S. Bevilacqua,

“Brazil as an Export Economy, 1880–1930,” in Enrique C

ardenas et al., An Economic History

of Latin America, 1, 33. Chile: Markos J. Mamalakis, The Growth and Structure of the Chilean

Economy, 15.Mexico: calculated from La econom

ıa mexicana en cifras, 1990, 11th ed. (Mexico

City, 1990), 121–2.Peru: Bruno Seminario and Arlette Beltr

an, Crecimiento econ

omico en

el Per

u: 1896–1995, 285–6.Venezuela: calculated from Asdr

ubal Baptista, Bases cuantitativas

de la econom

ıa venezolana, 59.Uruguay: Luis B

ertola et al., El PBI de Uruguay, 1870–1936,

30.

P1: GDZ

0521812909c07 CB929-Bulmer 0 521 81290 9 October 6, 2005 12:28

290 Richard Salvucci

There was, nevertheless, more than a casual association between export

booms and industrialization. In Colombia, for instance, the cotton textile

industry got underway in Antioquia and was a product of the region’s cof-

fee boom. Santiago Montenegro emphasizes that the capital that ﬁnanced

cotton textiles in the early twentieth century came from a variety of sources

in Colombia, including merchants’ trading proﬁts. In Chile, Cariola and

Sunkel suggest that the expansion of nitrate exports contributed to the for-

mation of a national market, especially in consumer goods such as footwear

and bottled beer, and to the building of a transportation network. Of course,

these examples pale in importance to developments that occurred in Brazil

and Mexico. Here, industrialization seems to have been more extensive

and subject to a host of institutional, ﬁnancial, political, and economic

inﬂuences that conditioned its subsequent evolution in important ways.

In Brazil, the rapid growth of coffee production and export at the end

of the nineteenth and during the ﬁrst quarter of the twentieth century

coincided with the consolidation of republicanism and the beginning of

industrialization. Brazilian scholars and historians agree that income gen-

erated by coffee exports was a proximate source of industrialization. As

Marcelo de Paiva Abreu has observed, Brazil’s large share of international

coffee production implied considerable control over pricing and rendered

growers less resistant to protective tariffs that promoted domestic industri-

alization even though they raised costs. The contrast with Argentina was

striking, for here the country was a price-taker and hence wedded more

closely to free trade and laissez faire with, as we have noticed, a corre-

spondingly reduced share of industrialization outside of meat processing

and packing.

In Mexico, both Stephen Haber and Edward Beatty have demonstrated

that the output of light industry and, in particular, cotton textiles, beer, and

cigarettes began to increase substantially in the 1890s. Haber, in particular,

has emphasized the importance of developing capital markets in Mexico

for mobilizing ﬁnance, something that, in his view, was more important in

Mexico than in Brazil, where the basic infrastructure for internal exchange

was not yet in place. Beatty shows that the D

ıaz government relied on a

regime of high tariffs (especially high effective protection, which increased

value added, or national product). Beatty also ﬁnds that Mexico engaged in

a kind of industrial policy as well as in patent reform, all with an eye toward

encouraging manufacturing. Nevertheless, as Haber has argued, small mar-

kets and protection tended to produce high costs and inefﬁciencies. These

were disguised by the sharp depreciation of silver to which we have already

P1: GDZ

0521812909c07 CB929-Bulmer 0 521 81290 9 October 6, 2005 12:28

Export-Led Industrialization 291

alluded because the fall in silver raised the relative prices of tradables, such

as manufactures, and increased their production. Other tradables, such as

mineral resources, would have beneﬁted as well, which again suggests the

complementarity between primary export expansion and industrialization

in the later nineteenth century.

An indirect way of judging the progress of industrialization is by mea-

suring the extent of import substitution. Domestic production of tex-

tiles would displace imports to an extent, whereas industrialization would

increase the share of capital goods imported to produce consumer goods.

In Brazil, the relation was very clear. Data assembled by Richard Graham

show that the share of textiles in British imports fell from around 60 per-

cent in 1870 to about 36 percent in 1910.Atthe same time, the share of

capital goods in imports rose from 26 to 42 percent, a pattern consistent

with the use of imported machinery to produce textiles that formerly had

been imported.

In Mexico, Beatty has demonstrated a similar phenomenon. Coarse cloth

imports fell by 4.3 percent per year between 1895 and 1908, tobacco products

at the rate of 5.5 percent, and beer at 4 percent. All were objects of import

substitution. At the same time, imports of producer goods rose at nearly 11

percent per year.

Finally, we might contrast the situation in Mexico or Brazil with

Argentina, whose industrialization was limited. As Carlos D

ıaz Alejandro

observed, the Argentine government provided limited support for indus-

trialization before 1930, perhaps fearful of limiting the market for its own

agricultural exports. Nevertheless, D

ıaz Alejandro emphasizes that much

pre-1930 manufacturing growth can be attributed to export expansion:

meat packing and dairy processing, milling, wool washing, and so forth.

Between 15 and 20 percent of the gross value of manufacturing before 1930

was exported.

CONCLUSIONS

When London caught a cold, it was once said, Buenos Aires contracted

pneumonia – the consequence of tight links between Britain and Argentina,

and of the ability of the gold standard to transmit macroeconomic shocks. In

recent years, Argentine economic historians have judged pneumonia prefer-

able to systemic collapse. Our conclusion might follow macroeconomic

historian Bradford deLong, who argues that the choice of an exchange rate