The Sugar Cane Boom

The initial phase of colonization took place mainly along the Northeastern coast and led to a boom in the sugar production that caused Brazil to become the main world producer of sugar by 1700.Footnote ⁸ Sugar cane was cultivated mainly along the stretch of land from the current state of Rio Grande do Norte down to the intermediary latitudes of the state of Bahia (“Recôncavo Bahiano”) shown in Figure 1A.Footnote ⁹

Following Roberto C. Simonsen, we identify the sugar boom as the period from the beginning of the effective colonization of Brazil by Portugal until 1760. This includes the height of production and profits during the “century of sugar,” from 1570 to 1670. From 1650 on the value exported started to fall and by the mid-eighteenth century, it was only 60 percent of the historical maximum.Footnote ¹⁰

The sugar economy was a plantation system built on three essential elements: “latifundio” (a large estate with a single owner), monoculture, and slave labor. Sugar cane brought the large rural estate and the patriarchal and slavery-based society to Brazil and the rest of colonial America.Footnote ¹¹ Sugar mills displayed economies of scale; therefore, land was given to individuals who had enough funds to invest in the purchase of large numbers of African slaves.Footnote ¹² In 1729 Governor Luís Vahia Monteiro wrote that, “The most solid properties in Brazil are slaves, and a man's wealth is measured by having more or fewer (…), for there are lands enough, but only he who has slaves can be master of them.”

Following closely the idea of an extractive occupation discussed by Engerman and Sokoloff, the sugar cane society was built on extreme social inequality, and very small economic and political elites with concentrated powers.Footnote ¹³ A typical sugar mill employed between 60 and 100 slaves. A particularly large one could have more than 200. In the eighteenth century, slaves represented more than half of the population in the Northeastern captaincies, and between 65 percent and 70 percent in the plantation areas. At the other extreme of the social pyramid, the senhores de engenho (literally, sugar mill lords, usually translated as “planters”) constituted the local landed aristocracy. They were invariably white and held a wide range of social, economic, and political powers.Footnote ¹⁴

The planter did not fully own the land and could not transmit it to heirs, yet possession of sugar mills and plantations and the position of senhor de engenho gave enormous social prestige. The planters lived opulently and had a great deal of local power but had little access to royal power. Another important factor was Portugal's focus on setting rules to ensure a constant flow of rents. In the State of Pernambuco in the seventeenth century, about 80 percent of government revenue came from various taxes on sugar production and trade, which was the only activity allowed in some areas during certain periods.Footnote ¹⁵

The polarization between landlord and slave and the dominance of the plantation system targeted at exports to Europe constituted the foundations of the social, economic, and political structures associated with sugar production. The colonial rule exacerbated the concentration of resources and the extractive nature of the sugar activity. Given the tendency towards persistence of institutions discussed in both the empirical and theoretical literatures, the effects of initial conditions may be present still today. The power of local elites can persist through the overruling of state power and de jure institutions by economic power and de facto institutions ran by the elites themselves. Landed aristocracies throughout Latin America were often a symbol of the political landscape created by the sugar cane society.

The Gold Boom

The gold mining expansion in the central part of Brazil was extremely intense and concentrated in time. In 1695 explorers made the first significant discoveries of gold in Sabará and Caeté, in the current state of Minas Gerais. Gold was also discovered in areas in the current states of Mato Grosso and Goiás after 1720 and 1726, respectively. Brazilian production peaked in 1760 and soon after started to decline, losing most of its relevance already by the end of the eighteenth century.Footnote ¹⁶

From 1700 to 1770 Brazilian gold production was roughly the same as the entire production of the rest of the Americas between 1492 and 1850, and corresponded to 50 percent of the production of the rest of the world between the sixteenth and eighteenth centuries.Footnote ¹⁷

The news of the discoveries generated a gold rush that led to an unprecedented occupation of the central part of the country. A quarter of a century after the initial discoveries, the Center-South region, which was mostly uninhabited previously, accounted for 50 percent of the colonial population.Footnote ¹⁸ The collapse of the mining economy was similarly fast. The main urban center of the region fell from 20,000 inhabitants in 1740 to just 7,000 in 1804. By then, gold production had already declined to only 12 percent of its peak value.Footnote ¹⁹

A series of regulations related to exploration of precious metals, movement of goods and people, and taxation were put in place to guarantee to Portugal an amount of rent deemed adequate.Footnote ²⁰ To encourage exploration, individuals responsible for the discovery of new quarries had priority in the selection of a fraction of the relevant area. The remainder was allocated to other candidates through auctions and lotteries, with the size varying according to the number of slaves owned. After the initial allocation, miners were free to sell and trade their plots.Footnote ²¹

At least twelve distinct taxes were adopted at different moments in time, but two were the most common: the fifth (or quinto), which established that 20 percent of the gold produced belonged to the crown; and the capitation (or capitação), which charged a fixed fee per slave owned above age 12. To prevent tax evasion, individuals were not allowed to circulate gold in the form of powder. Processing of raw gold and its transformation into bars were used as a tool to enforce taxation, and therefore remained under tight control of the metropolis. For the most part gold bars could only circulate with the seal of the Portuguese Crown; that implied that the raw gold would have been melted in an official establishment (Casa de Fundição) and the requisite taxes would have been paid.

Nevertheless, despite great efforts from the metropolis, frauds were constant. As miners tried to evade taxes levied by the Portuguese Crown, the metropolis created more and increasingly oppressive devices to regulate and control production even further. The process evolved into an extremely antagonistic environment between state and civil society, illustrated by a series of anecdotal tales from colonial Brazil. The most celebrated independence movement before 1822 was the Inconfidência Mineira, a movement organized by the local elites in 1789 in the state of Minas Gerais, and planned to burst out on a day scheduled for tax collection. The movement was frustrated by the early warnings of a traitor who was rewarded with the cancellation of his debts to the crown. The efforts to smuggle gold out of the mining areas involved entire local communities. A figure that became entrenched in Brazilian popular culture is the “hollow saint” (santo do pau-oco, a hollow wooden image of a Catholic saint). “Hollow saints” were used by various religious orders to smuggle powder gold out of the colony. Because of their widespread use, in 1711 the Portuguese Crown forbade the entrance of priests without explicit authorization and the establishment of religious orders in the state of Minas Gerais.Footnote ²²

The height of this tension materialized when the derrama, an additional tax raised when the aggregate amount of gold collected by the crown as one-fifth of total production did not reach a preestablished goal.Footnote ²³ The derrama was raised from the entire population as a tax on virtually any type of activity—trade, ownership of business or slaves, or use of public roads—and also as a tax on wealth, based on more or less arbitrary assessments of personal assets. Tax collection when the derrama was actually installed, which might last for several months, made extensive use of the army and involved prisons, apprehension of personal assets and durable goods, and raids into private households.Footnote ²⁴

The mining areas were peculiar in other ways. Despite the widespread use of slaves, society was not as polarized in gold areas as it was in the sugar cane areas. Technology and the scale of production implied a certain degree of horizontality in its social organization.Footnote ²⁵ Celso Furtado noted: “(…) the mining economy in Brazil opened possibilities to people with limited resources, since exploration was not based on large and deep mines—as in the case of silver in Peru and Mexico—but on alluvium metal deposited in the bottom of rivers.”Footnote ²⁶ Also, slaves enjoyed some degree of bargaining power because of informational advantages derived from the nature of the production technology. They could steal or hide whatever they found, and miners depended on their willingness to report discoveries for the enterprise to be profitable.Footnote ²⁷ It became relatively common for slaves to buy their own freedom.Footnote ²⁸

Hence, although mining also led to slavery and inequality, the social structure was less rigid and the distribution of endowments across the civil population was relatively more equal. In contrast, the availability of rents from gold coupled with the tremendous efforts of the Portuguese Crown to monitor every single step of production and trade of precious metals led to an overgrown, extremely focused and inefficient government apparatus. This apparatus was constructed entirely around the supervision and extraction of rent from one single activity: mining. The political science literature on the resource curse has repeatedly stressed the perverse effects that the possibility of rent extraction decoupled from productive activities and political representation may have. In the context of colonial Brazil, where the governing state (Portugal) was in fact formally detached from local population, these effects are likely to have been even stronger. As a consequence, the state in the gold mining regions did not function effectively in other areas and was involved in a constant struggle against civil society, leading ultimately to the development of a culture of detachment between population and state. Therefore, the implications of the gold boom are likely to be related to the effectiveness of government, the provision of public goods, and the relationship between local population and state power, which may have persisted despite being inefficient.

Historical Variables

We study the colonial determinants of current institutions based on the location and timing of the sugar cane and gold episodes and the distance to Portugal, which was an important determinant of the actual degree of intervention imposed by the metropolis. Municipalities involved in the gold boom are easily identified in historical accounts through the specific location of mines.Footnote ³⁰ They were located in the states of Bahia, Goiás, Mato Grosso, Minas Gerais, and Rondônia, as represented by the darker points in Figure 1B. Municipalities involved in the sugar cane boom correspond to the coastal areas of the current states of Rio Grande do Norte, Paraíba, Pernambuco, Alagoas, Bahia and Espírito Santo, and to the region of Campos dos Goytacazes, in the state of Rio de Janeiro (darker points in Figure 1A).Footnote ³¹ The specific areas identified as sugar boom areas are those located less than 200 kilometers from the coast in one of the regions enumerated above and founded before the end of the sugar boom in 1760.Footnote ³²

We construct two variables, gold and sugar with a value of one if a municipality was directly affected by the respective colonial resource boom. Sugar cane and gold production and processing were not entirely concentrated on these locations, and the political and social structures associated with the episodes most likely had a broader geographic influence. Therefore, in nearby districts within 200 kilometers the variables take a value of between 0 and 1 based on distance according to the functional form

(1)

where d_i is the distance from municipality i to the closest municipality directly involved in the respective resource boom. Municipalities beyond 200 kilometers are assigned a value of zero.Footnote ³³

In addition to the indices of influence of the sugar cane and gold episodes, we use distance to Portugal (measure in 1,000 kilometers) to incorporate the active interference of the metropolis during the colonial period. To capture how Portugal's interference affected the intervention of the metropolis during the extractive episodes, we use the interaction terms between distance to Portugal and the sugar cane and gold influence functions (sugar × distance to Portugal and gold × distance to Portugal).

Institutional Variables

We examine the influence of the colonial booms on three modern indicators related to institutional quality and distribution of economic and political power. The first variable captures the concentration of land within municipalities. Acemoglu and Robinson point out that individuals may exert political power inside or outside political institutions.Footnote ³⁴ Thus the concentration of economic resources in the hands of the elite acts as a source of de facto political power. Our measure of concentration is the Gini coefficient of the distribution of land in 1996 in the Brazilian Agricultural Census (land gini).

The second variable measures the quality of local administration. It captures the efficiency of the local executive in terms of administrative capacity, as reflected by its ability to provide public goods and its potential responsiveness to demands of the local population. We use an index of governance practices calculated by the Brazilian Census Bureau (IBGE) and used by the Ministry of Planning as a tool to monitor the administrative performance of municipalities. The index has various components representing different aspects of administrative capabilities at the municipality level, from efficiency in tax collection and information gathering to adoption of administrative and planning instruments.Footnote ³⁵ It is constructed as the simple average of four qualitative indicators, normalized from 1 to 6: the year in which the database of the tax on urban property (“IPTU”) was updated, the IPTU payment rate, the number of administrative instruments,Footnote ³⁶ and the number of planning instruments.Footnote ³⁷ The index (governance) is calculated using data between 1997 and 2000. This variable comes close to a measure of the overall efficiency and organizational capability of the local administration. According to the literature discussed before, governments in the sugar and gold boom areas would be expected to be weaker precisely along these dimensions.Footnote ³⁸

The third variable measures access to the justice system at the local level. There is no formal municipal judicial system in Brazil. Any court or justice commission in a municipality is either related to the local executive or to the state judicial system. Nevertheless, it is not clear why some municipalities provide different access to justice even if located in the same state and when certain types of courts are mandatory at the municipal level. In theory, mayors can be prosecuted if a mandatory court is not present in the municipality. For this reason, the existence of courts at the local level represents a de facto dimension of rule of law and of provision of access to justice. In order to capture this notion, we use an index based on the definition of access to justice proposed by the Brazilian Census Bureau.Footnote ³⁹ The index (access to justice) varies from 0 to 3 according to the existence of courts or justice commissions. It is calculated with data from 2001, as the sum of three binary variables indicating the existence of Special Civil Tribunals (Juizados Especiais Cíveis), Youth Councils (Conselho Tutelar), and Consumer Commissions (Comissão de Defesa do Consumidor).Footnote ⁴⁰

Empirical Estimation

Our main goal is to analyze whether the historical episodes explain current variation in local institutions. The summary statistics in Table 1 offer comparisons of the boom areas with the rest of the country without controlling for other correlates. The sugar boom areas in 2000 had more unequal income distributions and lower quality measures of governance and access to justice. The differences in means between the sugar boom and other areas are all statistically significant. The sugar boom areas are also among the poorest in Brazil, with less educated populations. The gold boom areas generally fared better than the sugar areas relative to the rest of the country. The gold boom areas had similar shortfalls in access to justice in 2000 but had slightly more equal land Gini's and similar quality of governance to the rest of the country. The gap in schooling was not as large and there was no income gap.

TABLE 1

DESCRIPTIVE STATISTICS: BRAZILIAN MUNICIPALITIES, 2000

+(39 obs. directly affected + 1,021 obs. indirectly affected)

++(354 obs. directly affected + 1,156 obs. indirectly affected)

* = Statistically significant at the 10 percent level.

** = Statistically significant at the 5 percent level.

*** = Statistically significant at the 1 percent level.

Notes:

Variables are index of quality of local administration (governance), indicator of the local presence of three types of courts (access to justice), Gini coefficient of the distribution of land (land Gini), distance to Portugal in 1,000 km, income per capita (ln), average years of schooling in the population aged 25 and above, and absolute value of the latitude. Municipality-level observations. Variables measured with data available between 1996 and 2000, unless otherwise noted. Areas affected by the sugar cane and gold booms defined in section, “Data and Empirical Estimation.”

Sources:

See the Appendix.

To examine the differences after controlling for other potential explanatory variables, we estimate the following equation

(2)

where i indicates municipality and Z_i is the measure of governance, access to justice, land Gini, or the public good characteristic. S_i is the sugar cane boom variable, G_i is the gold boom variable, P_i is distance to Portugal, X_i is a vector of geographic attributes (distance to equator, distance to coast, rainfall, sunshine, altitude, temperature, and soils, defined in the Appendix), and ε_i is an error term. Since municipalities in the same state can be correlated in a systematic way, standard errors in all regressions are clustered at the state level.

The vector of geographic attributes (X_i) plays an important role in our analysis. As discussed in the introduction, there is a significant correlation between institutional characteristics and geography. When compared to cross-country analyses, municipality-level data allow us to better control for the role of geographic characteristics because the geographic variation within a country is typically much larger than within a municipality.

The effect of the sugar cane episode on the institutional indicator is measured asγ^S + γ^SP P. For the governance and access to justice measures, we expect γ^S < 0, indicating that the sugar cane boom had a negative influence on current characteristics when the distance to Portugal is set to zero. We anticipate that the coefficient will be positive for the Gini measure. The parameter γ^SP captures how this effect varies with distance to Portugal. For example, when γ^S is negative and γ^SP is positive, it indicates that the sugar cane episode had a negative effect and that it became less negative for municipalities further from Portugal. Similarly, γ^G + γ^GP measures the effect of the gold boom.

The Coffee Expansion: A Postcolonial Resource Boom

The coffee boom is often seen as the third main “commodity cycle” in Brazilian history. However, the coffee experience was very different from the sugar cane and gold booms. First, the coffee expansion took place at a much later stage (nineteenth century) and over an area previously occupied. Second, its expansion extended well after Brazilian independence.Footnote ⁴¹ And third, the organization of production followed more closely a spontaneous and entrepreneurial logic, and not an extractive one. Therefore, the coffee boom is not the same type of initial institutional shock as the sugar cane and gold episodes. Nonetheless, coffee became a major economic activity by the end of the nineteenth century and changed the economic and political landscape of the newly independent country, and coffee is closely associated with the industrialization process and the political changes brought by the republic in 1889.

In this section, we do not intend to fully examine the long-term consequences of the coffee boom in Brazil.Footnote ⁴² We simply try to account for the coffee influence in our regressions as a robustness check for our main results, and contrast the effects of this postcolonial activity with the effect of the colonial resource booms discussed before.

In the beginning of the coffee expansion—during the monarchic period (1822–1889)—the neighboring areas in the states of Rio de Janeiro and São Paulo along the Paraiba Valley were the main coffee producers. Until the mid-nineteenth century, São Paulo occupied a secondary position but became increasingly dominant from 1850 onwards.Footnote ⁴³ In the early decades of the twentieth century, along with the expansion of railroads, São Paulo expanded production towards the west, where coffee flourished. All these coffee regions were initially either sugar cane or food crops producers. In particular, one of the reasons why the West portion of São Paulo did not grow coffee until the twentieth century is that it became an important sugar cane producing area at the end of the colonization period.Footnote ⁴⁴

Unlike sugar, coffee could be grown along with other food crops, and so the initial entry costs were substantially lower. In fact, given the length of time that coffee trees took to produce berries, the existence of alternative sources of income was essential to the survival of the new crop.Footnote ⁴⁵ Also, it was a spontaneous activity as opposed to an explicit endeavor of the Portuguese Crown.

The production of coffee until 1850 was an unusual slave-based multi-culture and relied on very poor technology. All growth in production came from expansions at the extensive margin.Footnote ⁴⁶ Only labor scarcity due to the end of the Atlantic slave trade and rising coffee prices induced a higher concentration of production, leading to a monoculture with improved technology in the second half of the nineteenth century.Footnote ⁴⁷ In 1888 Brazil abolished slavery and the labor force of European immigrants became increasingly prevalent in coffee fields, and coffee production expanded to the West of São Paulo following the expansion of railroads.

We build two measures of the coffee influence. The first identifies municipalities directly affected by the coffee boom as those founded up to 1886, which accounts for the main producing areas before the end of slavery and proclamation of the republic. These are the center and South portions of the state of Rio de Janeiro and the East areas of São Paulo according the coffee expansion map contained in Sérgio Milliet.Footnote ⁴⁸ The second measure identifies the municipalities directly affected by the coffee boom up to 1935, including areas in the West of São Paulo state that became the most important coffee producers during the first half of the twentieth century. We define both coffee influence variables in the same way as we defined the influence area of the sugar and gold booms, according to the function in equation 1.

In columns 1–3 of Table 3, we show the main specification from Table 2 controlling for our first measure of the coffee influence area and its interaction with distance to Portugal. The early coffee boom experience seems to have very similar effects to the colonial sugar boom: it is associated with more land concentration, and this effect is aggravated for municipalities located closer to Portugal.Footnote ⁴⁹ These results suggest that the coffee boom in the nineteenth century was somewhat similar to a colonial resource boom, which can be attributed to the fact that Brazil was recently independent and was still a slavery based society.

In columns 4–6 of Table 3, we use a broader measure of the coffee boom (up to 1935). Interestingly, this coffee boom measure and its interaction with distance to Portugal have no effect on any of the three measures of institutions. This definition of the coffee boom includes a period well after the colonization era ended, when no slave labor was used. This result suggests that the postcolonial resource boom had no negative impact on local institutions and the interaction with distance to Portugal plays no role.

It is also important to note that the sugar cane and gold results and the analysis of the effect of distance to Portugal from the previous section remain unchanged when both measures of the coffee influence variable are included in the regressions.

Public Good Provision as Institutional Development

We now extend our analysis, considering the provision of public goods as indicators of overall institutional and political development, obtained from the 2000 National System of Urban Indicators, from the Brazilian Ministry of Cities. These are percentage of households connected to the public sewage system, number of health centers per 10,000 inhabitants, per capita public spending on education and culture (ln), whether the municipality has at least one public library, and whether the municipality has at least one local radio station.Footnote ⁵⁰

TABLE 3

ROBUSTNESS OF THE EFFECTS OF COLONIAL BOOMS ON INSTITUTIONAL DEVELOPMENT: THE COFFEE BOOM – BRAZILIAN MUNICIPALITIES, 2000

Notes:

* = Statistically significant at the 10 percent level.

** = Statistically significant at the 5 percent level.

*** = Statistically significant at the 1 percent level.

Notes:

Robust standard errors are clustered by state in parentheses. Dependent variables are the index of quality of local administration (governance), indicator of the local presence of three types of courts (access to justice), and Gini coefficient of the distribution of land (land Gini). Independent variables are the sugar cane boom influence index, gold boom influence index, coffee boom influence index, distance to Portugal (in 1,000 km), and interaction between the resource booms influence and distance to Portugal. Columns 1–3 coffee boom influence area is defined up to 1886, in columns 4–6 it is defined up to 1937 according to the map in Milliet, Roteiro do Café e Outros Ensaios. All regressions include as independent variables (not shown): constant, absolute value of latitude, distance to the coast (in km), average water precipitation between 1931–1990 (100 millimeters per year), average sunshine during the day between 1931–1990 (100 hours per year), average altitude, average monthly temperatures in the 12 months (degrees Celsius), and 12 dummies indicating the types of soil present in a 0.1 degree ray from the municipality's center. Municipality-level observations. Colonial resource booms variables described in section, “Data and Empirical Estimation” and coffee resource boom measures are described in section, “Coffee Boom.” Variables are measured with data available between 1996 and 2000, unless otherwise noted.

Sources:

See the Appendix.

We estimate the main specification from Table 2 for these five dimensions of public goods.Footnote ⁵¹ The results are presented in Table 4. The sugar cane episode is associated with lower number of health centers per 10,000 inhabitants. At the mean distance from Portugal, the effect of the sugar boom on the number of health centers is −1.8 (−16.24 + 2.123 * 6.8), which is a substantial effect considering that the average of this variable is 5.72. This result also varies substantially with distance to Portugal—one standard deviation closer to Portugal increases the effect to −3.18 (−16.24 + 2.123 * (6.8 − 0.65)).

The gold boom, on its turn, is associated with worse provision of different dimensions of public goods—health centers, public libraries and radio. At the average distance from Portugal, the estimated effects for health centers, libraries and radio are −0.01, −0.03, and −0.04, respectively. In all cases, a decrease in one standard deviation in the distance to Portugal also increases the effects substantially to −1.16, −0.12, and −0.13, respectively.

These results are suggestive that the deterioration in institutional quality, caused by the colonial episodes, is related to worse provision of public goods. In particular for the case of the gold boom, the results are consistent with the theoretical characterization of the rentier state as nonresponsive to public demands and detached from the local population.

TABLE 4

EFFECTS OF COLONIAL RESOURCE BOOMS ON PUBLIC GOOD PROVISION – OLS ESTIMATION: BRAZILIAN MUNICIPALITIES, 2000

* = Statistically significant at the 10 percent level.

** = Statistically significant at the 5 percent level.

*** = Statistically significant at the 1 percent level.

Notes:

Robust standard errors are clustered by state in parentheses. Dependent variables are dummy variables for presence of radio station and public library, number of health centers per 10,000 inhabitants, percent of households connected to public sewage system, and ln of municipal spending on education and culture per capita. Independent variables are sugar cane boom influence index, gold boom influence index, distance to Portugal (in 1,000 km), and interaction between the resource boom influence indices and distance to Portugal. All regressions include as additional independent variables (not shown): constant, absolute value of latitude, distance from municipality center to coast (in km), average water precipitation between 1931–1990 (100 millimeters per year), average sunshine during the day between 1931–1990 (100 hours per year), average altitude, average monthly temperatures in each of the 12 months of the year (degrees Celsius), and 12 dummies indicating the types of soil present in a 0.1 degree ray from the municipality's center. Municipal-level observations. Colonial resource booms variables described in section, “Data and Empirical Estimation.” Variables are measured with data available between 1996 and 2000, unless otherwise noted.

Sources:

See the Appendix.