1

Dutch Disease in the tropics: commodity booms and export growth in Brazil

during the early nineteenth century

Christopher David Absell and Antonio Tena-Junguito (UC3M)1

First draft. August 2015.

Please do not cite.

Abstract: This paper explores the hypothesis that during the nineteenth century Brazil’s

export commodities were subjected to a perverse form of the Dutch Disease in which

the foreign exchange returns of the coffee boom served to appreciate the exchange rate,

effectively pricing the country’s other principal export commodities out of the

international market. We construct a composite index of domestic prices for the period

1821-1870 and elaborate nominal and real effective exchange rates, export weighted by

the geographic distribution of exports to 11 of Brazil’s principal trading partners. We

then review the empirical evidence of Dutch Disease in Brazil, positing that it

developed as a consequence of British West Indies slave abolition and the subsequent

rise to dominance of Brazilian coffee.

Introduction

The traditional narrative of Brazilian export growth during the nineteenth

century generally paints a pessimistic picture of the post-independence period. While

the coffee sector was expanding rapidly, it had not yet reached the growth rates that

would characterise the supposed late-nineteenth century ‘belle époque.’ The eighteenth

century gold rush had subsided, and Brazil’s other principal export sectors, namely

cotton and sugar, were gradually succumbing to creeping stagnation and decadence.

Insurmountable trade costs derived from hostile geography and poor transport

infrastructure, political and institutional instability, technological backwardness and a

scarcity of manpower (free or enslaved) and capital are standard explanations of the

Brazilian export performance experience of the post-independence period.2

The recent reconstruction of Brazil’s export series from independence to the end

of the first period of globalisation has put into question certain aspects of this traditional

narrative.3 It is evident that the period preceding independence was more dynamic then

the series based on official prices, and thus that previous interpretations of Brazilian

export growth based on the official series had understated the degree of this dynamism.

This finding complements a growing revisionist literature that emphasises the

1 Correspondence: christopherdavid.absell@clio.uc3m.es; antonio.tena@clio.uc3m.es. 2 Classic purveyors of the traditional narrative include Caio Prado Júnior, História Econômica do Brasil, 38a edição, (São Paulo: Editora Brasiliense, 1990). pp. 192-204; Celso Furtado, Formación económica del Brasil, (México D. F.: Fondo de Cultura Económica, 1962) pp. 113-123. On the decline of gold and the cotton and sugar industries in the eighteenth century, see Roberto Cockrane Simonsen, Historia económica do Brasil: (1500-1820): curso professado na escola livre de sociología e politica de São Paulo, (São Paulo: Companhia Editora Nacional, 1957); Werner Baer, The Brazilian Economy: Growth and Development, 5th Edition, (Connecticut: Praeger, 2001), pp. 16-18; Albert Fishlow, 'Brazilian Development in Long-Term Perspective,' The American Economic Review, 70: 2 (May, 1980), p. 102. 3 Christopher David Absell and Antonio Tena-Junguito, ‘Brazilian export growth and divergence in the

tropics during the nineteenth century,’ Universidad Carlos III de Madrid, Working papers in Economic

History 15-03, 2015.

2

dynamism of the post-independence Brazilian economy.4 The roots of the observed

export dynamism during this period are to be found in both the institutional changes

taking place in the slave labour markets of the Caribbean and the development of

Brazil’s factor endowment (specifically of land and labour). The abolition of slavery in

the British West Indies and subsequent divergence in the export performances of

Caribbean tropical agricultural producers was accompanied by the incredible expansion

of Brazil´s slave labour force and area of cultivable land, which remained largely

unrestrained by enforceable property rights institutions until mid-century.5 Amidst the

turmoil of Brazil´s principal competitors, domestic agricultural producers rapidly

responded to the country’s competitive advantage, yielding the export growth rates

reported by the corrected series.

Table 1: Descriptive statistics of Brazilian export performance by commodity.

1821-50 1850-70 1870-90 1890-13 1821-13 1850-13 1870-13

Growth (constant prices)

Coffee 9.72 2.98 2.03 4.34 5.09 3.08 3.19

Cotton 1.26 5.60 -6.52 4.90 1.38 1.46 -0.38

Sugar 4.57 0.13 0.28 -14.62 -2.06 -5.06 -7.53

Composition of exports

Coffee 34.63 47.72 58.53 59.88 48.87 55.50 59.01

Cotton 9.06 12.42 7.11 2.01 7.55 6.85 4.45

Sugar 31.24 16.30 11.75 2.13 16.52 9.73 6.66

World export share

Coffee 34.31 50.18 51.89 69.63 48.28 56.89 59.98

Sugar 10.33 7.39 4.02 1.86 6.89 5.16 3.42

Cotton 0.03 0.02 0.01 0.01 0.02 0.01 0.01 Source: Absell and Tena, ‘Brazilian export growth.’ Composition of exports is the arithmetic

average of annual data. The world export shares are arithmetic averages of quinquennial, annual, and

decennial data, respectively.

However this is only one part of the story of Brazil´s export performance during

the nineteenth century. While the slave emancipation shock and accompanying

expansion of the slave labour force and the area of cultivable land may help to explain

Brazil's initial export growth dynamism and the rapid expansion of its world export

4 Indicators of such dynamism include the level of wealth-holding, the demand for traditional modes of

transportation, and the activity in the market for informal loans. On each of these topics respectively, see

Zephyr Frank, 'Wealth Holding in Southeastern Brazil, 1815–60,' Hispanic American Historical Review

85: 2 (2005), pp. 242-246; Herbert S. Klein, 'The Supply of Mules to Central Brazil: The Sorocaba

Market, 1825-1880,' Agricultural History, 64: 4 (Fall 1990), pp. 1-25; Joseph Ryan, 'Credit where Credit

is Due: Lending and Borrowing in Rio de Janeiro, 1802--1900,' unpubl. PhD diss., UCLA, 2007. 5 On property rights see Lee J. Alston, Gary D. Libecap, and Bernardo Mueller, Titles, Conflict, and Land Use: The Development of Property Rights and Land Reform on the Brazilian Amazon Frontier, (Michigan: University of Michigan Press, 1999), p. 35; Marcelo de Paiva Abreu and Luiz Aranha Correa do Lago, 'A economia brasileira no Império, 1822-1889,' in Marcelo de Paiva Abreu (ed.), A Ordem do Progresso, Edição Atualizada: Dois Séculos De Política Econômica No Brasil, (Rio de Janeiro: Elsevier, 2014); Warren Dean, 'Latifundia and Land Policy in Nineteenth-Century Brazil,' The Hispanic American Historical Review, 51: 4 (Nov., 1971), pp. 606-625.

3

market share, these factors do not account for the loss of competitiveness in certain

export industries from 1850 onwards and the divergent export performance across

commodities. As can be seen in Table 1, the export growth rates of Brazil’s three

principal export commodities, coffee, cotton and sugar, diverged considerably across

the century. While the rise of coffee constituted an incredible commodity boom that

remained a fixture of the country’s export economy well into the twentieth century, both

sugar and cotton (with the fleeting exception of the American Civil War) gradually

faded from prominence. Such comparative export performance across commodities is

reflected in Brazil’s commodity composition of exports as well as the share of each

commodity in world exports.

With the exception of cotton, which was evincing signs of paralysis before

Brazilian coffee truly became a world player, the opportunities afforded by the

institutional turmoil in the Caribbean were available for the taking for all of Brazil’s

principal tropical export commodities. As has been demonstrated by estimates of the

number of slaves imported into Brazil and the number of sugar engenho and coffee

fazenda registrations, the rapid increase in the availability of factors of production

during the post-independence period was characteristic of both the north-eastern sugar

and south-eastern coffee producing regions. Yet after the closure of the slave trade in

1850 the sugar and cotton plantations of the north-east lost most of their slave

population to coffee fazendas in the south-east.6 This implies a loss of competitiveness

of those sectors concentrated in the north-east, a loss that is reflected in the results of

our constant market share analysis.7 So why did these sectors lose their initial

competitive advantage? Was it the trend of international prices, the shifting sands of

international demand, protectionist policy in the core, or the relative cost structure of

production that created this divergence? Or was there something more sinister at play?

Here we explore the roots of this divergent performance across commodities, paying

particular attention to a hypothesis that has been posited in the literature: that Brazil’s

other export commodities were subjected to a perverse form of the Dutch Disease in

which the foreign exchange returns of the coffee boom served to appreciate the

exchange rate, effectively pricing the other principal export commodities out of the

international market.

The paper is structured as follows. In the next section we briefly outline the

Dutch Disease hypothesis and the descriptive evidence from the literature. The

following section concerns the construction of our principal variables of interest, the

nominal (NEER) and real (REER) effective exchange rates. We then test the Dutch

Disease hypothesis by examining the relationship between these exchange rates and

export growth by commodity. The final section concludes.

Dutch Disease in Brazil: conjectures from the literature

Since the seminal work of W. Max Corden and J. Peter Neary on the

macroeconomic effects of booming sectors, the literature on the phenomenon of the

6 Herbert S. Klein, 'The Internal Slave Trade in Nineteenth-Century Brazil: A Study of Slave Importations into Rio de Janeiro in 1852', The Hispanic American Historical Review, 51: 4 (Nov., 1971), pp. 567-585. 7 Absell and Tena, ‘Brazilian export growth,’ pp. 21-23. While coffee, cacao and sugar showed a positive

competitiveness effect from1821-1850, this effect would disappear around mid-century, reappearing

again only in1890-1821 in the cases of coffee and cacao and, in the case of cotton, during the American

Civil War.

4

Dutch Disease or, in the words of Corden and Neary, ‘the coexistence within the traded

goods sector of progressing and declining, or booming and lagging, sub-sectors,’ has

expanded to constitute a booming sector in its own right.8 This literature generally

distinguishes between two principal effects of a commodity boom. The first is the

resource movement effect, in which the higher marginal product of the factors in the

booming sector draws resources out of non-booming sectors. Thus capital or labour may

be reallocated across sectors, generating growth in the booming sector and the

stagnation or decline of non-booming sectors. The second effect concerns the impact of

the boom on the demand for domestic non-tradeables, which increases as the result of

higher real income. Rising demand results in internal inflation, which causes an

appreciation of the nominal exchange rate thus pricing the non-booming sector out of

the international market. Moreover, it is evident that such booms may also generate an

excess demand or supply for money in the short run. A commodity boom will result in a

balance of payments surplus and the accumulation of international reserves, increasing

the monetary base and resulting in an excess supply of money. This will translate into

inflation which will serve to appreciate the real exchange rate.

In the case of Brazil, It has been posited that during the nineteenth century

Brazil's export sectors were subjected to a perverse form of Dutch disease, in which the

exchange rate, influenced heavily by a single export commodity, effectively priced

Brazil's other export commodities out of the world market. While the classic model of

Dutch disease occurs across various sectors of the economy and results in a process of

de-industrialisation,9 in Brazil this process took place within a single sector and across

agricultural commodities. As Nathaniel Leff noted,

In reflection of Brazil's stronger comparative advantage in coffee, the implicit sterling-milreis exchange

rate … was higher for coffee than for sugar or cotton … Both regions, however, had to face the same

foreign exchange rate. As coffee exports grew, they led to a higher exchange rate than would otherwise

have prevailed. This affected adversely sugar and cotton, which required a lower sterling-milreis rate in

order to export...10

Indeed, Leff’s hypothesis regarding the relationship between coffee revenues

and the nominal exchange rate has been confirmed for the second half of the nineteenth

century. Cardoso demonstrated that the nominal exchange rate was primarily influenced

by coffee export revenues. A percentage point increase in coffee export revenues was

associated with a -0.62 decline in the milréis-sterling nominal exchange rate; that is, an

appreciation.11 However, Cardoso did not explore the consequences of this relationship

for Brazil’s other export sectors. Other indicators of the effects of Dutch Disease in

Brazil during the nineteenth century may also be observed. The aforementioned

relocation of slave labour, principally from sugar to coffee plantations, can be

interpreted as a resource allocation effect, which, after slave abolition, continued to

draw on immigrant labour. Indications of the spending effect are less clear cut. Indeed,

8 W. Max Corden and J. Peter Neary, ‘Booming Sector and De-Industrialisation in a Small Open

Economy,’ The Economic Journal, 92:368 (Dec., 1982), p. 825. 9 For a survey of the economics of the Dutch disease, see Jeffrey G. Williamson, Trade and Poverty: When the Third World Fell Behind, (MIT Press: Massachusetts, 2011), Chapter 4. 10 Nathaniel H. Leff, 'Economic Development and Regional Inequality: Origins of the Brazilian Case,' The Quarterly Journal of Economics, 86: 2 (1972), pp. 256-257. Also on this point see Luis Catão, 'The Failure of Export-Led Growth in Brazil and Mexico, c. 1870-1930,' University of London, Institute of Latin American Studies Research Papers No. 31, 1991, pp. 14-15. 11 Eliana A. Cardoso, ‘Exchange Rates in Nineteenth-Century Brazil: An Econometric Model,’ Journal of Development Studies, 19: 2 (1983), p. 175.

5

it is evident that much of the return from coffee exports was absorbed by transport

costs, intermediaries (commisarios or ‘coffee factors’), and central or provincial

governments (by way of export taxation). The effect of the coffee boom on internal

demand was perhaps more discreet than theory suggests, the result of externalities in the

form of the construction of infrastructure or the development of financial markets which

effectively contributed to the expansion of the domestic market.12

Moreover, the Dutch Disease has been used as a causal explanation in other

commodity case studies of the period. Leonardo Monastério provided a similar

explanation for the decline of the jerked beef (charque) sector in Rio Grande de Sul,

observing a negative relation between the price of jerked beef and the real exchange rate

for the period 1837-1890, as well as a clear resource allocation effect in the form of the

exportation of slave labour from Rio Grande de Sul to Rio de Janeiro.13 Bradford

Barham and Oliver Coomes also observed evidence of Dutch Disease effects in the case

of the Amazonian rubber boom. The boom negatively affected other export

commodities of the region, such as rice and cacao, while the production of agricultural

commodities produced solely for domestic consumption, such as manioc, tobacco and

rum, increased, assumedly due to the spending effects of the boom.14

Despite the importance of the exchange rate to Brazil's export performance

during the nineteenth century, Leff's observation regarding this important mechanism

has, until now, gone untested.15 Here we take these observations seriously, and aim to

test the hypothesis that Brazil’s overall export growth was negatively affected by a form

of Dutch Disease in the tropical agricultural export sector. Before exploring this issue,

however, we outline the procedure for elaborating our principal independent variables

of interest: the nominal and real effective exchange rates.

A new effective exchange rate for Brazil during the nineteenth century

As mentioned above, the literature posits that the effective exchange rate acts as

the main transmission mechanism from the booming sector to the rest of the economy.

Thus we seek to construct a robust indicator of the effective exchange rate for Brazil.

We are also interested in comparisons between nominal and real effective exchange

rates in order to isolate price and competitiveness effects and to provide a comparison

with previous studies that have relied on an analysis of the nominal rate. Given that our

analysis will focus principally on the impact of the effective exchange rate on export

growth across agricultural commodities, it is pertinent that our index be trade-weighted

- specifically, export-weighted - to account for changes in the geographical distribution

of Brazil’s exports over time.

In order to calculate the trade-weighted REER for Brazil, we must first

overcome three major hurdles that have for a long time plagued researchers of

12 On the effects of the development of infrastructure on the domestic market, see William R. Summerhill, Order against progress: government, foreign investment and railroads in Brazil 1854-1913, (Stanford: Stanford University Press, 2003). 13 Leonardo M. Monastério, ‘FHC errou? A economia da escravidão no Brasil meridional,’ História e Economia Revista Interdisciplinar da Brazilian Business School, 1:1 (2005), p. 13-28. 14 Bradford L. Barham and Oliver T. Coomes, ‘Reinterpreting the Amazon Rubber Boom: Investment, the State, and Dutch Disease,’ Latin American Research Review, 29:2 (1994), pp. 88-91. 15 Leff did, however, include an examination of the determinants of the trend of sugar and cotton prices across the century, concluding that from 1874 to 1913 variations in the exchange rate were the most important determinant of these price trends. Leff , ‘Economic Development,’ p. 257.

6

nineteenth century Brazilian economic history. The first concerns the availability of

bilateral trade statistics. Here we strive to present the most representative REER

possible, and to do so we need bilateral trade data for Brazil´s principal trading partners

during the period in question. Instead of examining the official sources of Brazil´s

trading partners as has been done elsewhere,16 perhaps a limiting exercise especially

before 1850, we mine the official Brazilian publications for bilateral trade data. While

the accuracy of this data is extremely questionable,17 it gives an overall picture of the

geographical distribution of Brazil´s exports. Data is available for the periods 1836,

1841-1872, 1879-1891, and 1901-1913. The intervening periods have been interpolated.

Although the interpolation of bilateral trade data is empirically questionable, we are

principally interested in the long-run trend of the geographic distribution of Brazil’s

exports, and we assume no major changes during the periods of interpolation.

Moreover, the figure for 1836 has been assumed for the period from 1821-1836. We

hope to provide a more precise estimation of the geographical distribution of exports for

this earlier period in the second revision of this paper, although it is doubtful whether

the distribution of Brazil’s exports changed much during the first 15 years of

independence.

Table 2: Trading partner sample, Brazilian exports, 1836-1913

Country Period Percentage of trade

1836 1870 1913

Belgium 1835-1913 3.37 0.92 2.56

Chile 1830-1913 0.34 0.73 0.28

Denmark 1821-1913 3.09 0.15 0.23

France 1821-1913 4.84 9.67 12.28

Germany 1821-1913 8.23 5.38 14.07

Great Britain 1821-1913 24.95 43.41 12.69

Italy 1861-1913 -- 0.99 1.29

Portugal 1833-1913 6.87 5.39 0.50

Spain 1821-1913 1.92 1.75 0.54

Sweden and Norway 1830-1913 5.66 0.32 1.01

United States 1821-1913 17.56 28.61 32.42

TOTAL 76.82 98.33 78.87

Sources: 1836: J. J. Sturz, A Review, Financial, Statistical, & Commercial, of the Empire of Brazil and Its

Resources, (London: Wilson, 1837), p. 100; 1841-1849: Brazil, Collecção de mappas estatisticos do

commercio e navegação do Imperio do Brasil, (Rio de Janeiro: Typographia Nacional, various years);

1850-1870: Brazil, Relatorio do Ministerio da Fazenda, various years; 1871-73: Brazil, Estatistica do

commercio maritimo do Brazil, (Rio de Janeiro: Typhographia Nacional, various years); 1879-1891:

Brazil, Mappas estatisticos do commercio e navegação do Porto do Rio de Janeiro, (Rio de Janeiro:

Typographia Nacional, various years); 1901-1913: Brazil, Annuario estatistico do Brasil (1939/1940),

(Rio de Janeiro: Tip Departamento de Estatistica e Publicidade, 1939).

The extent of our trading partner sample is limited by the availability of

domestic price data. We need wholesale or (preferably) consumer price indices in order

to deflate the nominal exchange values for each country included in the sample. In some

cases (France, Spain), however, we were obliged to resort to the use of wholesale

indices due to the paucity of data available for the earlier years. For this first version of

the paper we were able to gather data for 11 of Brazil´s trading partners, as displayed in

Table 2. As can be seen, at worst, our sample covers around three-quarters of Brazil´s

16 Luis A. V. Catão and Solomos N. Solomou, ‘Effective exchange rates and the classical gold standard adjustment,’ The American Economic Review, 95: 4 (Sept. 2005), pp. 1259-1275. 17 Absell and Tena, ‘Brazilian export growth,’ p. 21, ft 72.

7

export trade and, at best, we are close to assembling a complete series.18 It is, to our

knowledge, the most comprehensive, and thus representative, sample of trading partners

used for the elaboration of the nineteenth century Brazilian effective exchange rate.19

Table 3: Commodity composition of wholesale price index, Brazil, 1821-1913.

INT

1821-1869: Lobo et al index (1919 weights): sugar, rice, cod fish, coffee, dried beef, wheat

flour, mandioc flour, beans, butter.

1870-1913: Catão index: cotton, rice, wheat, manioc, coffee, pinewood, beans, tobacco,

corn, herva mate, cocoa, sugar-cane, refined sugar, brandy, processed rice, cod fish, lard,

tar, kerosene, dried meat, beer, cement, leather and skins, manioc flour, wheat flour, wine,

matches, butter, grease, capital goods, pasta, textiles, processed coffee, cigarettes.

IMP

1821-1850: Federico-Tena: Irish beef, cotton textiles, wine.

1851-1869: British exports: guns, gunpowder, beer and ale, coal, coke and manufactured

fuel, cotton manufactures, hardwares and cutlery, leather, linen manufactures, iron, copper,

lead, linseed oil, woolen manufactures, cheese, brandy, tea, tobacco, wine.

EXP

1821-1969: Absell-Tena: raw cotton, cocoa, raw hides, rubber (from 1850 onwards).

Sources: Lobo et al, ‘Evolução dos preços’; Catão, ‘New wholesale price index’; Federico and Tena-

Junguito, ‘World Trade’; United Kingdom, Annual Statement, various years; Absell and Tena-Junguito,

‘Brazilian export growth.’

Here we are confronted by our second challenge: the elaboration of a reliable

domestic price index for Brazil. Luckily, we are blessed with the work of Luis Catão,

whose wholesale price index is sufficiently representative and realistically weighted to

be confidently used. The Catão index, however, only covers the period from 1870

onwards (the commodity composition of this index is shown in Table 3).20 The earlier

period, which covers the rise to dominance of the coffee sector and thus a vital period of

interest for our study of the Dutch Disease effect, is missing a comparatively robust

estimation of internal price tendencies. Existing price indices, while being important

contributions, are either not sufficiently representative, do not cover the entire period in

question, or are based on questionable assumptions.21 The most commonly used in

Brazilian historiography, elaborated by Eulalia Lobo and co-authors, while covering

18 The revised version of this paper will include other important trading partners such as Argentina and

Austria-Hungary. 19 Catão and Solomou´s NEER and REER for Brazil for the period 1870-1913 contained five trading

partners: Argentina, France, Germany, Great Britain, and the United States. See Catão and Solomou,

‘Effective exchange rates.’ 20 Luis A. V. Catão, ‘A new wholesale price index for Brazil during the period 1870-1913,’ Revista Brasileira de Economia, 46:4 (1992), 519-533. 21 As mentioned in the text, the first criticism is commonly aimed at Eulalia Maria Lahmeyer Lobo, Lucena Barbosa Madureira, Octavio Canavarros, Zakia Feres and Sonia Gonçalves, ‘Evolução dos preços e do padrão de vida no Rio de Janeiro, 1820-1930 - resultados preliminares,’ Revista Brasileira de Economia, 25:4 (1971), 235-265. The periodization problem is relevant to Oliver Ónody, A Inflação Brasileira

(1820-1958), (Rio de Janiero: s/n, 1960) and Raymond W. Goldsmith, Brasil 1850-1984:

Desenvolvimento Financeiro sob um século de inflação, (São Paulo: Harper & Row, 1986). The third

criticism is relevant for both Lobo et al, and C. R. Contador and C. Haddad, ‘Produto real, moeda e

preços: a experiència brasileira no periodo 1861-1970,’ Revista Brasileira de Economia, 36 (1975), pp.

407-440

8

over a century (1820-1930), contains an extremely limited commodity coverage

(minimum of 3, maximum of 9) and a questionable weighting system.22 In order to

obviate the problems of data scarcity and coverage, we have constructed a composite

index of three separate price indices that presents are more balanced (but still somewhat

problematic) view of price tendencies during this period. We combine the Lobo index

with a new import price index for this period as well as an export price index for a

number of selected commodities that were present in the domestic consumption basket

(according to Catão).

The issue of a reliable import price index is our third major hurdle. For much of

the period under question, available Brazilian sources do not list both the quantity and

value of principal imports. When available, the resulting unit value calculations differ

widely from their British equivalents. This is most likely due to the reliance of customs

officials upon an official price list of importable commodities that was modifiable only

by Act of Parliament.23 Thus, even with comprehensive archival work, it is unwise to

solely rely upon Brazilian sources for an accurate import price series. Here we follow

the work of Ronaldo Gonçalves, who calculated an import price index for the period

1850-1913 using the unit values and quantities of British exports to Brazil.24 We differ

somewhat in our treatment, however, for, unlike Gonçalves, we use imports by value

from the Brazilian sources to calculate the weighting system. While perhaps calculated

using inaccurate official price lists, the use of these values provides a snapshot of the

overall importance of each commodity in Brazil´s imports, rather than merely the

weight of such commodities in British exports. Furthermore, following Catão,25 to these

prices we have added the respective ad valorem tariff of each commodity in order to

provide a more balanced representation of the cost of these goods in the Brazilian

domestic market.26

Reliable import prices before 1850 for Brazil are extremely scarce. We thus use

the import price index constructed by Giovanni Federico and Antonio Tena-Junguito.27

While this index only contains prices for three commodities (the original contains a

22 Lobo et al., ‘Evolução dos preços.’ 23 Flávio Rabelo Versiani, ‘Industrialização e economia de exportação: a experiencia brasileira antes de 1914,’ Revista Brasileira de Economia, 34:4 (1980), p. 24. 24 Reinaldo Gonçalves, 'Índices de Comércio Exterior do Brasil,' Revista Brasileira de Estatística, 42: 168 (1981), pp. 331-362. 25 Catão, ‘New wholesale price index,’ p. 522. 26 The import price index is a Fisher price index using annual import values by commodity from United

Kingdom, Annual Statement of the Trade of the United Kingdom with Foreign Countries and British

possessions, various years. Data for ad-valorem tariffs comes from a variety of sources, including John

MacGregor, Commercial Statistics: A digest of the productive resources, commercial and financial

legislation &c. of all nations, Vol. IV.1 (London: Whitaker and Co., 1850), pp. 234-238; United

Kingdom, Return of rates of duty on principal articles of British and Colonial produce and manufactures

levied by tariffs of foreign countries, various years; Brazil, Tarifa das alfandegas do Imperio do

Brasil. (Rio de Janeiro: Typographia Nacional, various years); and Brazil, Tarifa das alfandegas. Rio de

Janeiro: (Typographia da Alfandega do Rio de Janeiro, various years). During certain periods tariffs were

presented in both specific and ad-valorem forms, assumedly to give Brazilian customs officials leeway in

the event of sudden changes in import prices, given that the revenue derived from the taxation of imports

was an important part of the both the Imperial and Republican government’s revenue structure. However,

we use only the ad-valorem rates, given that we possess no indication of when specific rates were applied

and lack a full series of these rates over the period in question. 27 Giovanni Federico and AntonioTena-Junguito, ‘World trade, 1800-1938: a new data-set,’ unpublished

mimeo, 2015, p. 138.

9

fourth, wheat flour, which has been removed due to its inclusion in the Lobo et al

index), two of these commodities (cotton textiles and wine) accounted for around 20

and 13 per cent of Rio’s imports by value in 1836 and 1849-50, respectively.28 So while

the index is by no means comprehensive, it does represent a number of the most

relevant goods for internal consumption missing from the Lobo et al index.

Additionally, to this mix we add a selection of export commodities not present in the

Lobo et al or Federico-Tena indices, namely raw cotton, cocoa, hides and rubber (after

1850 due to the scarcity of price data before this year). This data comes from Absell and

Tena-Junguito, and represent not the official values of exports given in Brazilian

sources but rather the corrected values based on international prices.29 From the new

indices we have removed the commodities that appear in the original Lobo et al index in

order to prevent double counting (butter and rice from the British series; coffee and

sugar from the corrected export series; see Table 3).

Figure 1: Distribution of indices in composite index, Brazil, 1821-1913.

In order to unite these three separate indices, we have constructed a composite

index using a simple weighting system. Given that each index is weighted separately

according to individual criteria, the composite index merely weighs each element by the

number of commodities contained in each index.30 This is essentially the equivalent of a

Paasche index with current period weights. As can be seen in Figure 1, the elaboration

of a Laspeyres index using fixed base period weights is problematic due to the periodic

changes in the distribution of the commodity composition of the index. A number of

caveats are in order here. The Lobo et al index, like the Catão wholesale price index,

28 Sturz, Review, p. 101; Brazil, Ministerio da Fazenda, Proposta e Relatorio do anno 1855, pp. 33-61. 29 Absell and Tena-Junguito, ‘Brazilian export growth,’ p. 38. A second revision of this paper will include

import prices of goods from Great Britain for the period 1827-1835, as well as data from Brazilian

sources for the 1840s. 30 In this sense we avoid the pitfalls of the construction of a consumption basket based on assumptions derived from scarce qualitative data. See the recent debate on this issue in Revista de Historia Económica / Journal of Iberian and Latin American Economic History, 33:1 (2015). However, this will bias the index in the sense that tradable goods will perhaps be excessively weighted. This is an issue we hope to address in future revisions of the paper.

0%

10%

20%

30%

40%

50%

60%

70%

80%

90%

100%

1821

1823

1825

1827

1829

1831

1833

1835

1837

1839

1841

1843

1845

1847

1849

1851

1853

1855

1857

1859

1861

1863

1865

1867

1869

INT IMP EXP

10

represents the prices of commodities in Rio de Janeiro. However, a case can be made

for taking these prices as being representative on a national level, given Rio´s

geographical primacy during the period in question.31 Additionally, one might argue

that the index gives excessive weight to tradable goods. Given qualitative observations

on the level of industrialization and (under-)development of the internal market,

especially prior to mid-century, however, it seems that many of the commodities in the

domestic consumption basket, particularly manufactured or refined versions of

domestically produced goods, were principally supplied by way of importation,

especially in those coastal cities that possessed greater market access. Finally, again

regarding the Lobo et al index, we have chosen to utilise a weighting system based on a

consumption basket observed in 1919, far from the period in question. The alternative

weighting system of 1856, however, is based solely on a single company (Companhia

de Luz Stearica) and thus its representativeness is extremely questionable.

Figures 2.1 and 2.2: Wholesale price level estimate and elements of composite index,

Brazil, 1821-1913

Figures 2.1 and 2.2 display the new wholesale price index for the period 1821-

1870, chained to the Catão series which continues until 1913, and the index´s

constitutive elements. As can be seen, the new index charts a course between the

31 Catão, ‘New wholesale price index,’ p.521.

0

50

100

150

200

250

1821

1825

1829

1833

1837

1841

1845

1849

1853

1857

1861

1865

1869

1873

1877

1881

1885

1889

1893

1897

1901

1905

1909

1913

Ind

ex (

18

70

=1

00

)

Fixed weights Average Current weights

0

50

100

150

200

250

1821

1828

1835

1842

1849

1856

1863

1870

1877

1884

1891

1898

1905

1912

Ind

ex (

18

70

=1

00

)

INT INT IMP EXP

11

gradual inflation of internal prices (INT) and the post-Napoleonic War deflation of

export prices (EXP). Import prices (IMP), on the other hand, follow a steady trend

before spiking up alongside export and internal prices during the War of the Triple

Alliance and the American Civil War in the 1860s. Thus the addition of the export and

import series serves to offset the inflationary bias of the Lobo et al series.

The final step in the calculation of the trade-weighted effective exchange rates is

to obviously collect the relative nominal exchange rates for Brazil and its principal

trading partners.32 In some cases (pre-unification Italy, Portugal) nominal exchange

rates were unavailable, further limiting our sample. As mentioned previously, we

calculate a trade-weighted nominal effective exchange rate (NEER), unadjusted for

inflation as well as a trade-weighted real effective exchange rate (REER).

Nominal and real effective exchange rates and the Dutch Disease in Brazil, 1821-

1870

Figure 3 compares the logs of our NEER and REER indices, defined in milréis

per foreign currency unit terms. A rise (fall) of the index signifies a depreciation

(appreciation) of the milréis relative to our sample of foreign currency units. The

general trend of the indices is comparable to Catão and Solomou’s findings for Brazil

during the period 1870-1913. However, our effective exchange rate indices span a

greater period of time and include a larger sample of trading partners. Given the

literature’s focus on the post-1870 ‘belle époque’ period, for this first version of the

paper we choose instead to explore the earlier, post-independence period until 1870.

Here we merely aim to provide a number of descriptive observations as to the

relationship between the exchange rates and export performance by commodity. Future

versions of the paper will include more rigorous econometric analysis.

Figure 3: Trade-weighted real effective exchange rate and nominal exchange rates,

milréis per foreign currency unit, Brazil, 1821-1913.

32 Data on nominal bilateral exchange rates come from a variety of sources including Markus A. Denzel, Handbook of World Exchange Rates, 1590-1914, (Surrey: Ashgate, 2010); A.B. Nunes, M.E. Mata, and N. Valério, ‘Portuguese economic growth, 1833-1985,’ Journal of European Economic History, 18 (1989), pp. 291-330; Juan Braun-Llona, Matías Braun-Llona,, Ignacio Briones, José Díaz, Rolf Lüders and Gert Wagner, Economía Chilena 1810-1995. Estadísticas Históricas; and the Global Financial Database.

3

3.5

4

4.5

5

5.5

6

182

1

182

4

182

7

183

0

183

3

183

6

183

9

184

2

184

5

184

8

185

1

185

4

185

7

186

0

186

3

186

6

186

9

187

2

187

5

187

8

188

1

188

4

188

7

189

0

189

3

189

6

189

9

190

2

190

5

190

8

191

1

lnREER lnNEER

12

To begin with, we consider the demand conditions during this period. Changes

in the demand for Brazil’s export commodities, or the tariff policies adopted by Brazil’s

principal trading partners, may serve to explain Brazil’s divergent export performance

across commodities. Indeed, demand for Brazil’s export commodities increased during

the nineteenth century alongside income growth in the core countries. Our constant

market share analysis showed that during the nineteenth century (1821-1913), ‘world’

demand for cotton and sugar was the driving force behind export growth, although

negative competitiveness effects were just as important and served to offset increased

demand. In the case of coffee, the competitiveness effect was more important,

especially in the periods 1821-1850 and 1890-1913, a conclusion also confirmed to a

certain extent by the REER trend.33 What were the determinants of this competitiveness

effect? Was it related to protectionist policy in the core? Did coffee enjoy greater

market access than sugar and cotton? In terms of protectionist policy, it is evident that

these commodities enjoyed (or suffered from) more or less the same levels of market

access during the post-independence period. For most of this period, Great Britain was

the principal market for Brazilian sugar.34 The commercial treaties forged between the

two countries after independence and the destruction of the Portuguese trade monopoly,

however, were somewhat asymmetrical in character. It has been argued that this

asymmetry was partly responsible for the poor performance of Brazilian exports during

the first half of the century.35 As Paulo Batista Jr. has shown, British colonial

preferences served to exclude non-colonial sugar and coffee producers from the British

market until the passage of the Sugar Act in 1846. Batista calculated an ad valorem

equivalent of 475 and 266 per cent on the eve of the Sugar Act for imports of Brazilian

sugar and coffee, respectively.36 With respect to coffee, given the British preference for

tea, the United States emerged to become Brazil’s most important client.37 As Catão

demonstrated, in the second half of the century the price and value of Brazilian coffee

largely followed the dictates of US national income.38 US duties for coffee declined

from five cents a pound to duty free status after 1832.39 Brazilian sugar did not achieve

similar levels of market access until the reciprocal treaty of 1891, although duties on

33 World demand is proxied by exports of 55 countries from Federico and Tena-Junguito, ‘World Trade.’ For the elaboration and results of the constant market share analysis, see Absell and Tena-Junguito, ‘Brazilian export growth,’ pp. 21-23 and appendix 3. 34 Importing in 1846-47 around 20 per cent of Brazil’s sugar alongside other important importers such as Austria and Portugal, in 1900 close to 25 per cent, although by the turn of the century the United States had replaced Great Britain as the principal destination for both sugar and coffee exports. Brazil, Collecção de mappas estatisticos do commercio e navegação do Imperio do Brasil, (Rio de Janeiro: Typographia Nacional, 1853), pp. 377-378; Brazil. Commercio exterior do Brazil (1900), Boletim do serviço de estatistica commercial da Republica dos Estados Unidos do Brazil, (Capital Federal: s.n., jun. 1900), p. 14. 35 Paulo Nogueira Batista Jr., ‘Política tarifária britânica e evolução das exportações brasileiras na primeira metade do século XIX,’ Revista Brasileira de Economia, 34:2 (1980), pp. 203-240. 36 A similar tendency can be observed for French colonial preferences, although France occupied a lesser share of Brazil’s exports during this period, 2 and 7.5 percent of Brazilian sugar and coffee exports, respectively. Brazil, Collecção de mappas, pp. 377-378. 37 In 1846-47 the US already imported around 42 per cent of Brazil’s coffee exports. This would increase to 75 per cent around the turn of the century. Brazil, Collecção de mappas, pp. 377-378; Brazil. Commercio exterior do Brazil, p. 14. 38 Catão, ‘Failure of export-led growth,’p. 9. 39 Topik, p.24.

13

cane sugar decreased from three cents per pound in 1816 to 0.75 cents in 1861.40

However, the US was a relatively unimportant market for Brazilian sugar during the

post-independence period, occupying only 3 per cent of Brazilian exports around mid-

century.41 Thus sugar’s restricted US market access should not have weighted as heavily

upon export performance as that of its restricted access to the British market. As far as

competition is concerned, however, sugar was at a distinct disadvantage. While the

aftermath of slave emancipation allowed Brazil to rapidly expand its export market

share of coffee, in the case of sugar a similar trend con be observed for Cuba.42 While

the determinants of the formation of Cuba’s comparative advantage is outside the scope

of this first version of the paper, it is evident that the country’s competitive advantage

allowed it to fill the gap in the market left by British West Indies producers. Moreover,

the cane sugar industry was stricken by a substitution effect which came in the form of

the development of European beet production throughout the century.43

In such a context, we now explore whether the Dutch Disease effects described

in the literature served to contribute to the competitive disadvantage of Brazil’s export

commodities. Firstly, we seek to confirm the influence of coffee export revenues on the

nominal effective exchange rate. As mentioned previously, Cardoso found strong

evidence of a negative correlation between the nominal exchange rate and coffee export

revenues, defined in terms of constant values. We attempt to reproduce these results for

the period from independence to mid-century. Cardoso defined a model that predicted

that the nominal exchange rate was a function of wages, monetary policy, coffee export

revenues and international prices. Using an instrumental variables approach, she

regressed the log of the nominal exchange rate (E) over the logs of food prices (the

Lobo et al index, W), monetary policy (a rough estimate of M1, H), British import

prices (P*), and the export quantities of coffee and rubber (X).44 We have followed this

methodology, with a few modifications. To begin with, instead of using export

quantities, we use the value of coffee exported in current prices from our corrected

series (lnV). This is important given that we are interested in the price effects of the

NEER. Furthermore, for best fit purposes we use the lag of monetary policy. For

international prices we use our new import price index. Additionally, we examine two

periods: one which includes the turbulent period of the 1860s and another which omits

it.

The results, as displayed in Table 4, confirm the direction and significance of

Cardoso’s results for the later period. The strength of the coefficient of V is slightly

smaller than later in the century, however, perhaps due to the fact that the coffee export

sector occupied a lesser share of Brazil’s export composition during the earlier period.

Furthermore, the impact of the 1860s only seems to marginally affect the size of the

coefficients. Still, it’s clear that coffee exports were associated with nominal exchange

rate appreciation in the post-independence period.

40 F. W Taussig, Some aspects of the tariff question: an examination of the development of American Industry under protection, (Cambridge: Harvard University Press, 1934), p. 100. 41 Brazil, Collecção de mappas, pp. 377. 42 Absell and Tena-Junguito, ‘Brazilian export growth,’ pp. 27-28. 43 Peter L. Eisenberg, The Sugar Industry in Pernambuco: Modernization without Change, 1840-1910, (California: University of California Press, 1974), p. 22. 44 Cardoso, ‘Exchange rates,’ p. 175.

14

Table 4: Determinants of the nominal exchange rate, 1821-1870.

LnE= α0 + α1 lnW + α2 lnH + α3 lnV + α4 LnP*

α0 α1 α2 α3 α4 R2 DW

1821-1860

5.47 0.26 0.38 -0.41 -0.58 0.75 1.10

(3.86) (2.57) (3.68) (-4.68) (-2.89)

1821-1870

4.95 0.28 0.33 -0.36 -0.46 0.79 1.10

(4.95) (2.99) (3.61) (-4.82) (-3.36)

Given this observation, how did the nominal exchange rate, heavily influenced

by the growth of the coffee sector, affect Brazil’s other export commodities? There are

two relevant questions to be asked here. The first, most obviously, is did the Dutch

Disease actually occur during this earlier period? The second is, if so, when did it

begin? Beginning with the second question, we posit that the process emerged from the

institutional turmoil of the post-independence period in the Caribbean and the

subsequent definition of Brazil’s competitive advantage in coffee. It is precisely during

this period that one observes a heightening of the accumulation of factors of production

in Brazil.45 Furthermore, as the differential between the NEER and REER demonstrates,

the considerable depreciation of 1826-1829 would have served to increase the

competitiveness of Brazil’s exports in trading partner markets. Although this

competitiveness declined during the decade of slave emancipation, perhaps due to the

impact of emancipation on the prices of tropical agricultural commodities, it would

return in 1840 and remain a characteristic of Brazil’s export performance until the

1860s. In fact, although the price hikes generated by the slave abolition shock served to

reduce competitiveness, Brazil remained competitive relative to its principal tropical

competitors. Figure 4 shows the prices of a select number of commodities of Brazil’s

principal tropical competitors from 1830-1850, encompassing the slave abolition shock

and its aftermath. While British West Indies producers, represented here by Jamaica and

British Guyana, experienced steep and enduring price increases after slave abolition, the

stubborn slave economies of Brazil and Cuba remained relatively unscathed.46 As

45 Herbert S. Klein, The Atlantic slave trade, (Cambridge: Cambridge University Press, 1999), pp. 210-211, appendix Table A.2, which gives the following figures for slave imports Brazil (in thousands of slaves): n1801-1810: 241.3, 1811-1820: 327.7, 1821-1830: 431.4, 1831-1840: 334.3, 1841-1850: 378.4, 1851-1860: 6.4. 46 The effect of slave abolition on the sugar plantations of the British and French colonies of the Caribbean is described in J.H. Galloway, The sugar cane industry: an historical geography from its origins to 1914, (Cambridge: Cambridge University Press, 1989), pp. 123-130. For an overview of the British West Indies see Stanley L. Engerman, 'Economic adjustments to emancipation in the United States and British West Indies,' The Journal of Interdisciplinary History, 13: 2 (Autumn 1982), p. 196. For Jamaica, Gisela Eisner, Jamaica 1830-1930, (Manchester: Manchester University Press, 1961). For British Guiana,

15

shown by the trend of the REER, these relative price changes translated into increased

competitiveness, driving export growth and resulting in a greater export market share.

Figure 4: Price of coffee, cotton and sugar, various countries, 1830-1850.

Assuming that this period provides the catalyst for the definition of Brazil’s

competitive advantage in coffee, we posit that one should observe signs of the Dutch

Disease after full emancipation occurred in the British West Indies in 1833. Figures 5.1

and 5.2 lend descriptive support to these hypotheses. The first shows Brazil’s NEER

(left hand axis) graphed alongside the world export share of four of Brazil’s products:

cacao, coffee, cotton, and sugar. In the cases of sugar and cotton we see a clear negative

trend after the great depreciation of 1826-1829. Beforehand, however, sugar showed a

slightly positive trend; it is not until the late 1830s that sugar’s share in world exports

began to decline. Similarly for cotton, although we only possess decennial estimations,

we see a one percentage point decline between 1830 and 1840 to a level that was

maintained during the rest of the period, with the brief interlude of the American Civil

War. The trend of cacao is ambiguous; even though, like cotton, it only occupied a

minor share of world exports, unlike cotton this share increased slightly over the period.

The second figure again shows the NEER (left hand axis) graphed alongside the shares

of each product in the commodity composition of Brazil’s exports. This Figure perhaps

gives a clearer indication of the conditions surrounding the development of the Dutch

Disease in Brazil. After the sharp depreciation of 1826-1829, both sugar and cotton

show signs of declining shares. Sugar, however, manages to maintain its share until the

later 1840s when it shows a clear decline. Again the case of cacao is difficult to include

in the narrative of Dutch Disease, as it more or less maintained its possession in the

composition of Brazil’s exports during the period under consideration.

Michael Moohr, 'The Economic Impact of Slave Emancipation in British Guiana, 1832-1852,' The Economic History Review, 25: 4 (Nov., 1972), pp. 588-607.

0

10

20

30

40

50

60

70

80

1830 1832 1834 1836 1838 1840 1842 1844 1846 1848 1850

GB

P p

er m

etri

c t

on

Coffee Jamaica to UK Coffee Brazil to US

Coffee Cuba to US Coffee Sauerbeck, Rio

Cotton British Guyana to UK Sugar Jamaica to UK

Sugar Cuba to US

16

Figure 5.1: NEER and world export market shares by commodity, 1821-1870

Figure 5.2: NEER and export composition shares by commodity, 1821-1870

Conclusions

Although the descriptive evidence points to the development of Dutch Disease

and its dissemination throughout the export sector, such evidence is not conclusive.

While it is evident that the shock of British West Indies slave emancipation led to the

0.00

0.10

0.20

0.30

0.40

0.50

0.60

0.70

0.00

20.00

40.00

60.00

80.00

100.00

120.00

182

1

182

3

182

5

182

7

182

9

183

1

183

3

183

5

183

7

183

9

184

1

184

3

184

5

184

7

184

9

185

1

185

3

185

5

185

7

185

9

186

1

186

3

186

5

186

7

186

9

Per

centa

ge

Ind

ex,

19

13

=1

00

NEER Sugar Cacao

Coffee Cotton Lineal (Sugar)

Lineal (Cacao) Lineal (Coffee) Lineal (Cotton)

0.00

0.10

0.20

0.30

0.40

0.50

0.60

0.70

0.00

20.00

40.00

60.00

80.00

100.00

120.00

182

1

182

3

182

5

182

7

182

9

183

1

183

3

183

5

183

7

183

9

184

1

184

3

184

5

184

7

184

9

185

1

185

3

185

5

185

7

185

9

186

1

186

3

186

5

186

7

186

9

Per

centa

ge

Ind

ex,

19

13

=1

00

NEER Sugar Cotton Cacao Coffee

17

development of Brazil’s comparative advantage in coffee, it is not certain whether the

resulting relationship between coffee exports and the trend of the effective exchange

rate were the principal determinant of the decline of Brazil’s other export sectors.

Certain aspects of the Dutch Disease argument don’t square up. What about the case of

cacao? Why did it not follow the decline of Brazil’s other dominant export sectors? Was

the decline of sugar really due to Dutch Disease effects, or was it simply because cane

producers couldn’t compete with more innovative tropical agricultural producers and

European beet production? We aim to address these questions in the second version of

this paper.