Bumpy Road

In 1900, railway trackage in the United States was almost
20 times as great as in Brazil. Even after the large post-1900
increase in Brazil's railway construction,
in 1914 the country had only 26,060 kilometers of track.
This was a figure that the United States
had surpassed by the 1850's.

Nathaniel H. Leff

In the United States, real per capita income grew at a long-term rate of approximately 1.5 percent per year in the nineteenth century. Long-term growth at that annual rate for a period of 91 years (the time interval between 1822 and 1913) implies a cumulative increase of per capita income from an initial level of 100 to an index of 388 at the end of the period. By contrast, the data available for Brazil suggest a very different economic experience in the nineteenth century. Although real output was able to keep pace with Brazil's rapid population growth (1.8 percent per year), real per capita income seems to have grown very little between 1822 and 1913. Further, most of the per capita income growth that took place in Brazil between 1822 and 1913 seems to have occurred in the period 1900-13. Those years were indeed a period of rapid economic progress. By the same token, the years 1822-99 seem to have been a long period of disappointing economic achievement in Brazil.

Disaggregation of these figures in terms of geographic regions is also helpful in understanding Brazil's economic experience during the nineteenth century. The country's overall performance masks a significant differential in the pace of development between regions. In part, the poor aggregate experience of the Brazilian economy during the nineteenth century reflects the especially dismal performance of the country's large northeast region, where almost half of Brazil's population resided. A rough estimate suggests that real per capita income in the northeast fell, by approximately 30 percent between 1822 and 1913. Our first task, then, is to try to understand why the large northeast region did so poorly. We then proceed to a more general analysis, encompassing the rest of the country as well.

Exports were the main source of productivity growth in nineteenth-century Brazil. International trade was important both for permitting higher income from available resources and for stimulating capital formation, including public-sector and foreign investment, in economic infrastructure. The northeast's negative economic experience during the nineteenth century stemmed largely from the poor export performance of the two products in which the region had an international comparative advantage: sugar and cotton. In 1822, sugar and cotton accounted for 49 percent of Brazil's aggregate export revenues, while coffee (produced in the southeast) accounted for 19 percent. In the course of the nineteenth century, Brazil's export receipts from sugar and cotton showed little long-term growth and actually declined in terms of receipts per capita. In 1913, sugar and cotton provided only 3 percent of Brazil's total export revenues. By contrast, real income from coffee exports increased at a long-term annual rate of approximately 5 percent. By 1913, coffee accounted for 60 percent of Brazil's aggregate export revenues.

The decline of the northeast's sugar and cotton exports reflected the fact that nineteenth-century Brazil had a stronger comparative advantage in coffee than in sugar or cotton. That is, a unit of foreign exchange could be earned with fewer domestic resources in coffee than in sugar or cotton. Because the domestic-resource cost of foreign exchange was much lower in coffee than in sugar or cotton, the northeast experienced a nasty case of the "Dutch disease." As the foreign currency provided by coffee exports grew as a source of supply in Brazil's foreign-exchange market, the country's overall exchange rate increasingly reflected the importance of coffee and its pressures for real-currency appreciation. Revenues of the producers of Brazil's various export commodities and the volume of output that they supplied varied in function of changes in the mil-réis (the Brazilian currency) price that producers received. Mil-réis prices for individual commodities, in turn, varied both with changes in the specific commodity's international price and with changes in Brazil's overall exchange rate. In fact, much of the variance in the mil-réis prices for Brazilian cotton and sugar resulted from changes in the mil-réis sterling exchange rate. Therefore, the coffee-dominated exchange rate squeezed factor returns and priced ever-larger quantities of the northeast's sugar and cotton out of the world market.

In introductory economics textbooks, when a new export activity emerges with a stronger comparative advantage than that of the country's traditional export activity, factors are reallocated to earn the higher returns available in the new activity, and income rises. By contrast, in the Brazilian historical context, the northeast's adjustment was constrained by some rigidities imposed by geography. The northeast's specific types of land (arid climate) were not well-suited to coffee production and were therefore not reallocated to coffee. Consequently, transfer of other productive factors from sugar and cotton to coffee required interregional migration. The large distances between Brazil's regions, however, meant high transportation costs, such that migration involved an investment. Brazil's slave market financed the transfer of slaves, and most of the northeast's stock of slaves was indeed bid away to the southeast. But much of the northeast's labor force was free, and large-scale transfer of free labor was precluded by the absence of a capital-market institution to finance free workers' investment in interregional migration.

Economic theory points to a key condition that must be satisfied if the integration of multiple geographical regions in a single political unit is to constitute an economically optimal currency area, That condition is inter-sectoral factor mobility. As we have seen, nineteenth-century Brazil did not satisfy that condition. Under these circumstances, one may wonder whether the northeast might not have been better off economically as a separate political entity, with its own exchange rate. The northeast's trade and development would then have been governed by its own (rather than by Brazilian) comparative advantage. In fact, the northeast's political elite did attempt to secede from Brazil during the nineteenth century, but maintenance of the country's territorial integrity was a key priority for Brazil's political leadership, which used military force to repress secession. The northeast therefore remained within Brazil, and the region's monetary and trade conditions were greatly aggravated by its being part of a political entity that did not meet the conditions for an optimum currency area. The northeast's dismal economic experience was an important part of Brazil's overall poor record in the nineteenth century.

The Elastic Supply of Labor

In the southeast, coffee exports grew rapidly, with major linkage effects on the regional economy. But long-term increase in real wages, and hence in income for much of the population, was constrained by two labor-market institutions that provided an elastic supply of labor to Brazil's "advanced" sector throughout the nineteenth century. Accordingly, output and the demand for workers in the Southeast coffee region could increase rapidly without generating an increase in real wages.

First, importation of slaves from Africa enabled Brazil's plantation owners to satisfy their growing demand for labor with relatively little utilization of workers from the country's domestic agricultural sector. Consequently, the export activities could expand their output substantially without bidding up wages within the Brazilian economy. In the first half of the nineteenth century, the British government attempted to stop the importation of slaves from Africa. The economic advantages that importation afforded Brazil's planter class were so great that the Brazilian state resisted British interventionism for half a century. Between 1800 and 1852 (when the British navy finally forced suspension of slave imports), approximately 1.3 million slaves were imported to Brazil. This amounted to more than one-fifth of the growth of the country's total population and an ever larger share of the increase in the Southeast's labor force.

The increase in the supply of slave labor to the coffee sector seems to have been sufficiently great that the real cost of labor did not rise over the century. Pedro Carvalho de Mello has collected data on nominal slave purchase prices and rental rates between 1835 and 1888. (the year of abolition) in Rio de Janeiro. Regression equations estimated with Mello's undeflated data show an annual trend rate of increase of 2 percent for the slave-purchase time series and an annual rate of increase of 1.8 percent for the slave-rental series between 1835 and 1888. Deflated with observations for the price of coffee, Mello's time series can be regressed against a time trend to ascertain the rate of change of real labor costs in nineteenth-century Brazil's "advanced" sector. Over the years 1835-88, the regressions show an annual trend rate of change of - 0.1 percent (with a t-ratio of 0.39) for the deflated purchase-price series, and an annual trend rate of change of - 0.3 percent (with a t-ratio of 1.43) for the deflated rental-rate series.

These regressions, in which coffee prices are used as the deflator, indicate that real labor costs for Brazil's coffee producers did not increase in the half century between 1835 and 1888. We lack an annual index of consumer prices that could be used to deflate the nominal slave price and rental series in order to assess rigorously the time trend in real consumption wages for coffee workers. As discussed elsewhere, however, data on the medium-term rate of increase of consumer prices in Rio de Janeiro suggest that consumer prices rose at least as rapidly as the current-price series for coffee labor. Thus despite the great growth of coffee production and the rapid expansion of the southeast's economy, the supply of labor apparently kept pace with the demand for labor, obviating upward pressure on labor costs or worker incomes.

The second labor-market institution involved immigration. As noted, in 1852 the British government stopped Brazil's importation of slaves from Africa. Following a long-term interaction between domestic economics and politics, slavery was abolished within Brazil in 1888. From the view-point of maximizing coffee-planter returns, the mounting pressures for abolition posed a potential problem. Unless accompanied by other changes in the labor market, abolition would bring a sharp rise in labor costs. Accordingly, some of the coffee sector's political leadership sought a monopsonistic, class solution to protect planter interests. Their approach to the impending problem was, in effect, to shift downward the supply schedule of labor in anticipation of the planters' growing demand for workers. To achieve this objective, they developed a new labor-market institution that would maintain an elastic supply of low-cost labor from overseas.

To endogenize the supply of labor, the coffee planters pressed Brazil's central government and the government of São Paulo province to pay the transportation costs of immigrants from southern Europe. Such subsidies had two important consequences for potential European immigrants. First, without raising Brazilian wages, transportation subsidies increased the net private returns from immigrating to Brazil. In addition, the subsidies overcame the capital-market imperfection that might otherwise have prevented destitute Europeans from immigrating at all. By paying transportation costs, Brazil could attract immigrants who, if they could have financed their own immigration, might have gone to the United States or to Argentina, where wages were higher.

The Brazilian policy intervention to attract European immigration achieved its objective. Immigration, mostly from Southern Europe, accelerated sharply. The increase was most dramatic in the case of the province where coffee production was expanding most rapidly, São Paulo. Between 1880 and 1885, an average of 4,300 immigrants entered São Paulo annually. In 1886, the figure was 9,500, and in 1887, the year before abolition, the figure was 33,000. Overall, between 1885 and 1909 some 2.8 million European immigrants entered Brazil. Almost all of these people went to the southeast. Between 1890 and 1913, the stock of coffee trees in São Paulo province (a proxy for the demand for labor) increased at a rate of approximately 6.5 percent per year. In addition, the demand for workers also rose in manufacturing as well as in other activities in the booming southeast. Despite these pressures on the demand side of the labor market, however, real wages apparently did not increase.

One may wonder why the supply of labor to the advanced sector in the southeast came from overseas rather than from within Brazil. In principle, workers from within Brazil might have come either from the domestic agricultural sector in the southeast or from the declining northeast. The inability to attract many workers from the domestic agricultural sector in the southeast is not surprising. Incomes earned in that sector made for an opportunity cost that was apparently well above the labor costs offered by subsidized immigration.

The failure to draw on labor supply from the northeast is more puzzling. It seems unlikely that transportation costs for would-be immigrants from the northeast to the southeast exceeded the cost of transporting workers from Southern Europe to Brazil. Another possibility is that supply constraints (perhaps reflecting sociocultural rigidities or political restrictions) limited labor mobility in the northeast. In fact, supply constraints do not seem to have been a problem. There is evidence of considerable labor-mobility in the northeast. And as regards extraregional labor mobility, between 1872 and 1910 hundreds of thousands of northeasterners emigrated to the booming Amazon region. Migration to the southeast, however, involved greater distances, higher costs, and a larger investment. As noted earlier, the absence of a capital-market institution to finance those investments seems to have been important in limiting migration from the northeast to the southeast during the nineteenth century. Hence, our question reduces to, Why were the coffee planters in the southeast more willing to finance immigration from Europe than from the northeast? Part of the answer may have been then prevalent racial attitudes on the part of the coffee planters, which led them to prefer European to mulatto workers.

The consequences of large-scale subsidized immigration from overseas are clear. The program continued through the beginning of the twentieth century the economic structure that importation of slaves from Africa had provided earlier. The highly elastic supply of labor from overseas meant that output could expand at a rapid pace in Brazil's advanced sector without raising the wages of workers in the rest of the economy. The similarities between Brazil's historical experience in the nineteenth century and W. A. Lewis's celebrated model, "Economic Development with Unlimited Supplies of Labour," are evident. There were, however, two important differences between Brazil's historical experience and the Lewis model. In the Brazilian case, the elastic supply of labor came from overseas. Also, in Brazil the elastic supply of labor continued "forever"—with ensuing long-term consequences for capital-labor ratios, wages, and technical progress. Continuing importation of labor from abroad enabled Brazil's planters to maintain their returns but had adverse effects on the rest of the population. This experience suggests that conclusions concerning the welfare effects of population growth in nineteenth-century Brazil may be a function of the observer's class perspective. Explicitly or implicitly, historians often discuss welfare effects over time, and their unit of study is usually "the nation." In the Brazilian case, class interests were so obviously disparate that it raises questions concerning the validity of using the nation as the unit of analysis.

The Domestic Agricultural Sector

Like most studies of Brazil's economic history before the twentieth century, we have focused thus far on conditions in the country's export activities. The greater availability of data for those activities should not lead us to exaggerate their quantitative importance. In fact, most of Brazil's labor force was engaged in the domestic agricultural sector: the production of food for local consumption and the internal market." Brazil's domestic agricultural sector in the nineteenth century has been little studied. In the words of two scholars (Reigelhaupt and Forman), the sector usually appears only "between the lines" of the country's historiography. Consequently, detailed information on this sector is scanty. Nevertheless, the domestic agricultural sector was too important a feature of Brazil's economy during the nineteenth century to be ignored. As a first approximation, the following statements can be advanced concerning the sector's size and composition.

Socially, this sector comprised many of the people in Brazil's population who, in the words of historian Caio Prado, "were not slaves, but could not afford to be masters." This observation suggests one way of forming an impression of the quantitative importance of the domestic agricultural sector in the Brazilian economy: an examination of the proportions of free people and of slaves in Brazil's total population. This procedure obviously provides only a very approximate picture. All of Brazil's slaves were not engaged in exports or in urban-based activities; many free people did work in those activities and in roles other than plantation owners. Bearing this caveat in mind, let us see what analysis of the population in terms of its slave and free portions suggests.

Brazil's social structure has often been conceptualized in terms of a master/slave dichotomy. That approach ignores the presence of a very large intermediate stratum of squatters, sharecroppers, and small farmers. At the very beginning of the nineteenth century, at least one-half and perhaps as much as two-thirds of Brazil's population was free. Relatively few of these people—poor whites, mulattoes, freedmen, and caboclos (peasants of mixed Indian and white ancestry)—were large slave owners engaged in production for the export market. Lacking alternative opportunities in a predominantly agrarian economy, many people in this intermediate social stratum were engaged in production of food for domestic consumption.

In 1820, some 70 percent of Brazil's population was free. Until 1852 (when importation of slaves from Africa was stopped), only a small percentage of these people was employed in export activities, which relied heavily on slaves for most occupations. With the decline of slavery, free people were increasingly employed in export activities, but by that time the free population had grown rapidly as a result of high rates of natural increase. Consequently, the number of people in the domestic agricultural sector remained large relative the country's total labor force.

The impression that much of Brazil's labor force was not engaged in export production is corroborated if we consider disaggregated population surveys for specific locales during the nineteenth century. Further, the limited information available on the sectoral composition of Brazilian output also suggests that a large fraction of the labor force was engaged in the domestic agricultural sector. In 1911-13, exports accounted for approximately 16 percent of gross domestic product (GDP) in Brazil. During the nineteenth century, exports had grown at a higher rate than output in the rest of the economy. Consequently, earlier in the century, the share of exports in aggregate economic activity had been even lower than 16 percent. Further, labor productivity was generally higher in exports than in other activities of the Brazilian economy. Hence, the export sector's share in the total labor force was even smaller than its share in GDP. There were of course other activities in this economy besides exports and the domestic agricultural sector. In absolute terms, many people were employed in transportation, commerce, crafts, manufacturing, and government. Those activities were located to a great extent in the cities, however, and as late as 1890 only 11 percent of Brazil's population resided in urban centers of 10,000 or more inhabitants. These considerations suggest that a large fraction of Brazil's labor force was engaged in the domestic agricultural sector during the nineteenth century.

This sector seems to have consisted of two parts. First, there were people who lived as sharecroppers, smallholders, or squatters in or near the areas of export production. Because of factor-market imperfections, these people rarely engaged in production of the principal export crops. Their main products were such foodstuffs as manioc, beans, and maize. In addition, the observations of contemporaries suggest that these people took much of their total income in the form of leisure. Second, part of the labor force in the domestic agricultural sector was engaged in farming on the abundant lands in Brazil's interior, relatively far from the areas of export production. Output consisted mainly of cattle ranching and of semisubsistence agricultural cultivation. In the latter case, production was mainly in the form of small-scale family farming under the overlordship of a large local landowner. With labor scarce relative to land, cultivation was land-extensive. Population in this sector was increasing rapidly, while abundant lands existed further in the interior. Consequently, the production frontier shifted ever farther from the markets and centers of consumption. As marginal physical productivity fell with soil depletion on the intensive margin, incremental production shifted, with rising transport costs, to the extensive margin. Until the end of the nineteenth century, it is hard to believe that the value of output per worker in the domestic agricultural sector was more than, at best, constant over time.

Transportation Costs and the Slow Pace of Economic Development

Because a large portion of Brazil's labor force was employed in the domestic agricultural sector, the modest rate of per capita output growth in that sector weighed heavily on the pace of aggregate development. We noted earlier that exports were the main avenue to economic development in nineteenth-century Brazil. The central importance of the export activities reflects a default, the poor performance of the rest of the economy.

High transportation costs affected both the level and the growth of productivity in Brazil's domestic agricultural sector, limiting the access of many agricultural producers to markets beyond their immediate locale. As a result, the volume of intraregional, interregional, and international trade was curtailed. Because of the high ratio of land to labor, cultivation was land-extensive, and distances to the markets were large. Low-cost transportation facilities were therefore crucial for developing a high-productivity agriculture. Unfortunately, the country's geographical and topographical conditions made for relatively high transport costs from the production areas to the market centers.

Rivers and coastal shipping were used for transportation, but some of the country's rivers (the Amazon, for example) were poorly located from the viewpoint of promoting economic development. Other rivers flowed in a direction that was not advantageous from the perspective of production for markets. Geographical conditions also imposed another problem that hampered low-cost shipments of bulky commodities from deep in the interior. Unlike the United States with its Mississippi and Great Lake systems, Brazil did not have an extensive network of navigable, interconnecting waterways. Further, road conditions were also poor, to the extent that at the beginning of the period wheeled vehicles could seldom be used in the interior. Transport costs were so high that they absorbed a third of the value of coffee shipments during the prerailroad era. Similar conditions prevailed in the northeast. Thus the cost of shipping cotton from the São Francisco Valley to Bahia in the 1850's amounted to some 50 percent of the prices received. Under these conditions of high-cost transportation and poor access to markets, abundant land was not associated with a high value of output per worker in agriculture.

The combination of high transportation costs and a large domestic agricultural sector also had other consequences for the Brazilian economy. Because of Brazil's poor internal transportation facilities, food produced on more distant land involved higher supply prices. Inelasticity in the supply of foodstuffs meant that when income and demand in the economy's advanced sector increased, prices rose. Unlike many other countries, Brazil experienced a long-term inflation during the nineteenth century. Price inflation was a feature of the Brazilian economy that had its own welfare costs, both direct (higher uncertainty) and indirect (presumably, lower cash balances and lower investment). In addition, conditions in the domestic agricultural sector constrained Brazil's industrial development. Low income levels and high costs for transporting goods to the hinterland limited the size of the market for manufactured goods in Brazil. The Brazilian government imposed protective tariffs on many industrial products during the nineteenth century, but protection against imports could not assure would-be Brazilian industrialists access to a market that did not yet exist. Industrialization based on the internal market clearly required the prior emergence of a domestic market.

More generally, high transport costs diminished the net receipts that producers obtained from shipment of bulky, low-value foodstuffs to the market. As a result, income in the domestic agricultural sector was reduced—both because of the low value received for output and because of the disincentive effect that unfavorable relative prices had on the quantities produced. Low prices in the domestic agricultural sector were reflected in a small marginal value product for labor and, as a consequence, in widespread substitution of leisure for monetary income. Finally, high transport costs for foodstuffs also had an important intersectoral effect. The country's steep price-distance gradients in regional markets meant rising incremental costs for food, the economy's wage good, in the face of buoyant demand conditions. Expanding aggregate demand therefore reduced the returns to capital and the rate of expansion in the advanced sector, with little impact on higher real output levels in the economy's backward sector.

Efforts at modifying geographical conditions and lowering costs by construction of transportation infrastructure were slow to materialize in nineteenth-century Brazil. In contrast with the United States, there was virtually no canal construction. The country's rivers also remained largely without improvements. Consequently, the boats used for internal transportation were small and entailed high unit costs. The country's first railroad legislation was promulgated in 1835, but actual railway construction was late in coming to Brazil. The country's earliest railway, extending some 15 kilometers, was built in 1854. Ten years later, approximately 424 kilometers of track were in operation. As late as 1890, however, the country had only 9,973 kilometers of operating trackage. This did not amount to much in terms of Brazil's overall expanse of some 8.1 million square kilometers. Furthermore, the country's road network was extremely limited. As late as 1923, São Paulo state, one of the largest and most developed in the country, had only 1,025 kilometers of highways (of which 55 kilometers were macadamized) suitable for automobile use.

Railroads might have helped this situation by lowering transportation costs. This would have provided a necessary condition for linking part of the domestic agricultural sector with the rest of the economy and permitting it to shift from subsistence to market-oriented production (for the domestic market or for exports), whether in the family farms or in large-scale agriculture. Lower transportation costs would also have provided producers with the stimulus of market demand and might thereby have induced higher output levels. On the supply side, producers would have been able to reap the gains from specialization and local comparative advantage. Hence even with unchanged physical productivity, lower transport costs might have raised the value of production in the domestic agricultural sector, both by increasing the quantities produced and, with new relative prices, by altering the composition of output.

Notwithstanding these potential benefits, nineteenth-century Brazil was late in initiating large-scale railroad construction. Thus despite Brazil's vast territorial expanse, as late as 1884 the country had only 6,240 kilometers of track. This amounted to approximately 0.7 kilometers of track per 1,000 square kilometers of territory. Further, in terms of timing, the great increase in railway construction toward the interior began only in the 1890's. Indeed, the largest absolute rise in railway track occurred only in the twenty years before 1914. To gain some comparative perspective, note that in 1900, railway trackage in the United States was almost 20 times as great as in Brazil. Even after the large post-1900 increase in Brazil's railway construction, in 1914 the country had only 26,060 kilometers of track. This was a figure that the United States had surpassed by the 1850's.

Why were the railways built so late in Brazil? The difficult terrain often led to high construction costs, but these would have been no obstacle if the railroads had also generated substantial benefits. Capital immobilities were also a problem. Although some of the first railroads in the coffee region were built with local capital participation, construction of Brazil's railways in general depended heavily on foreign investment. In the nineteenth century, this was largely British, and British investment was directed away from Brazil by such non-market considerations as imperial policy. In addition, private rates of return on Brazilian railway investments were apparently not high enough to attract substantial British capital from its alternative opportunities during most of the nineteenth century.

Brazil's limited attractiveness to foreign investors, however, is not a sufficient explanation of the long delay before large-scale railway construction began. If private returns were low but investment in low-cost transportation facilities were justified in terms of external economies and high social returns, another approach might have been followed. In principle, the task of providing Brazil with an adequate transportation system might have been undertaken by government—central, provincial, or local. That was the course followed with many of the "public improvements" that were supplied in the nineteenth-century United States. In fact, Brazil did not follow that approach until the end of the period. For reasons discussed below, during most of the century Brazilian governments failed to provide on a sufficient scale the infrastructure investment needed for the country's economic development.

Railroads and the Acceleration of Economic Development

Once the railways were extended, economic development seems to have proceeded along the lines outlined above. Even with unchanged output levels, the higher ex-farm prices made possible by low-cost transportation would have raised producer incomes. In addition, producers in the domestic agricultural sector responded to the new market opportunities opened by lower transport costs. Producers increased the volume of their output for the market, while the fall in transportation costs also led to new patterns of intraregional specialization. Another feature was a rise in the price elasticity of the food supply.

Some numerical information on these developments is available for Minas Gerais. This large state had approximately 21 percent of Brazil's population in 1900. Despite its geographical proximity to São Paulo and that province's large regional market, Minas Gerais was not economically well-developed. In the 1890's, however, the province "caught railroad fever": half of Minas Gerais's pre-1899 trackage was laid in that decade.

The process through which railroads promoted economic growth in the domestic agricultural sector had some special features. The railways helped domestic agricultural producers not only by reaching the distant interior, but also by existing in the zones of export production. Part of the country's food supply was produced in and around the plantation areas.

Food producers in those areas benefited directly from the new availability of low-cost transportation to the regional market. In addition, the lines opened in the export zones lowered the cost of shipments that originated in the far interior and proceeded, via the railroad, to the markets. Under these conditions, even railways that had been built primarily to carry export commodities came to transport large volumes of products from the domestic agricultural sector.

The growth in shipments of domestic agricultural products was facilitated by the Brazilian government's tariff and rate-setting policies. Brazil experienced considerable price inflation in the decades before 1913. As a consequence both of normal regulatory lag and of hostility to the foreign railway companies, however, the government's rate-setting authorities resisted efforts to raise transportation charges to keep pace with the country's inflation. Thus, not only did shipping costs fall when the railways were opened but, in addition, the price of railway transportation declined thereafter relative to the general price level. This rate-setting policy led to government subsidies for the railways and, eventually, to nationalization. What is important in the present context is that government regulation further lowered real freight charges for producers in the domestic agricultural sector.

The structure of railway rates also discriminated in favor of the domestic agricultural sector. Between 1874 and 1900, the rates charged for shipments of foodstuffs on the railways ranged between 26 and 49 percent of the rates charged for coffee. For livestock and timber the rates were even lower. Moreover, in 1899 the government implemented a general policy that obliged the railway companies to lower their charges on domestically produced foodstuffs. As a consequence, the domestic agricultural sector drew special and disproportionate advantage from the fall in transport costs that the railroads made possible. For this reason, it is difficult to make meaningful comparisons with shipping costs in the prerailroad era, which might serve as a basis for comparative welfare analysis. In the earlier period, freight charges for the domestic agricultural sector's highweight/low-value commodities had often been so high in many areas that these products had not been shipped at all.

The government's policy with respect to import duties also promoted economic growth in the domestic agricultural sector. At the turn of the century, the government imposed protective tariffs on many foodstuffs produced in Brazil. The fact that politicians from Minas Gerais took a prominent role in this policy initiative suggests that the new measures should not be viewed as determined randomly or by a process that was completely exogenous. The advent of low-cost transportation had greatly increased the potential economic returns that protective tariffs offered to domestic food producers. Political returns rose correspondingly for the political entrepreneurs who would implement the necessary policy measures. From this perspective, provision of the import tariffs can be regarded almost as endogenous to the process.

The economic consequences of the new import duties were clear-cut: reduced uncertainty and a larger market for the domestic agricultural sector. Moreover, the fact that part of the sector's market growth came at the expense of imports helped avoid a potential pitfall. That would have been a situation in which large increases in domestic food supply pressed on stationary, price-inelastic demand and thus reduced aggregate revenues for producers. The policy initiative also had broader economic effects. As noted, the new tariffs were implemented in conjunction with the heightened domestic supply response that low-cost transportation made possible. Under those conditions, the import tariffs led to import substitution in many food products and intensified intersectoral linkages within the Brazilian economy.

The northeast also benefited to some extent from a decline in transport costs. In areas where railways were built, internal freight charges for sugar and cotton appear to have fallen some 50 percent from their level in the prerailroad era. Railways could promote economic development only when they were built, however, and because of the poor economic prospects of the northeast's export activities, little railway construction took place in the region. In the southeast, however, extension of the railways seems to have opened a new period of generalized economic development.

Prior to the extension of the railways, a rising value of output per capita in Brazil had been limited mainly to the export sector. By lowering transport costs in a vast, land-rich country, railways permitted more rapid growth of income in the large domestic agriculture sector. The downward shift in internal freight charges also led to other structural shifts and new intersectoral linkages within the Brazilian economy. Thus the internal market for manufactured products also expanded. Supported by ample tariff protection, Brazil's cotton textile industry increased its output at an annual geometric rate of 11 percent between 1885 and 1915. As noted earlier, Brazil's economic development proceeded much more rapidly after 1900 than in the preceding century. For the reasons discussed, the extension of the railways seems to have played a key role in the shift to the new development trajectory. This experience is also consistent with interpreting Brazil's slow economic development during the earlier period as stemming largely from an absence of the external economics that railways would have provided. Because of the country's factor endowment and geographical features, the availability of low-cost transportation was of special importance for economic development in nineteenth -century Brazil.

The Brazilian State and the Public-Finance Constraint on Public Investment

The preceding discussion raises an obvious question. We can well understand the failure of private entrepreneurs to invest in railways in the Brazilian interior. Much of the economic benefits of that investment came in the form of external economies, such that the railroads' social returns exceeded their private returns. But why did the Brazilian state not provide the resources—either through direct investment or through subsidies—to equip the country with railways earlier, so that Brazil could have been launched on its path of long-term economic development much sooner in the nineteenth century?

One possibility is that the vision of implementing a rational public-investment policy was distorted by the lens of Brazilian politics. The large landowners had considerable influence in Brazilian politics during the nineteenth century, and they are generally not considered to have been a very "progressive" or "development -oriented" group. In fact, what was needed in this context was not an interest in development but an interest in wealth maximization. Brazil's landowners displayed ample evidence of such an interest. Thus, responding to the prospect of favorable returns, Brazilian planters allocated sufficient resources—even to products with a long gestation period, for example, cocoa in Bahia and coffee in São Paulo—to make possible sharp increases in output. Further, far from explaining the failure of Brazil's governments to provide large infrastructure investments, an interpretation that emphasizes the role of the large landowners in Brazilian politics only sharpens the question. For, following Joseph Schumpeter's insight concerning the convergence of monopoly and socialism, one would expect large landowners to be especially energetic in pressing for public investment. This is because landowners with extensive holdings and market power can internalize and appropriate most of the social benefits of infrastructure investment. Therefore Brazil's internal political conditions should have led to large government investment in economic infrastructure.

Another possibility is that ideology inhibited a rational public-investment policy. In principle, Brazil's political leadership may have been constrained by nineteenth-century doctrines of laissez-faire. Voices of economic liberalism were heard in nineteenth-century Brazil, but, in practice, Brazilian governments did intervene in the economy, imposing protective tariffs as well as providing subsidies—for example, for European immigration and for technological modernization of the northeast's sugar industry—when these did not require a large financial input. Likewise, the Brazilian state was so little bound by the canons of nineteenth-century economic orthodoxy that it ran frequent fiscal deficits and maintained economic policies that led to chronic inflation and long-term exchange-rate depreciation.

Another possible explanation for the government's lack of support for new railroads suggests that it would be naive to expect the Brazilian state in the nineteenth century to demonstrate an interest in promoting economic development. The country's political and administrative elites are generally considered to have been more interested in self aggrandizement and bureaucratic expansion than in economic development. Such concerns, however, are perfectly consistent with a large promotional role for the public sector. Expanded state investment and subsidy programs would have meant more government jobs and greater control over society's economic resources. Thus the existence of self-seeking motives is hardly an adequate explanation of the Brazilian state's failure to pursue a more active public-investment policy.

One set of conditions does seem to have constrained the Brazilian state's developmental activity: public finance. Through most of the nineteenth century, the fiscal resources the Brazilian state had at its disposal to pay for infrastructure investment and subsidy programs were small relative to the country's development needs.

During most of the century, Brazil's fiscal system was highly centralized. Until the 1880's, the tax revenues of the central government were approximately 4.5 times larger than those of the provincial governments. The central government's share in total public-sector expenditure was even larger, for the central government had much greater access to foreign and domestic borrowing. Likewise, the tax revenues collected by local governments in nineteenth-century Brazil were a small fraction of total public-sector revenues.

For the first four decades after independence, the central government's expenditures per capita were well below £1. It was only with the Paraguayan War (1864-70) that per capita expenditure exceeded £1. And it was not until the first decade of the twentieth century that central-government expenditure in current prices approached £1.5 per capita. Different measuring rods may be used to assess these expenditure levels. In the present context, the most pertinent comparison is with the magnitude of the development task that Brazil faced in the nineteenth century. As noted earlier, the country's initial conditions with respect to social overhead capital were poor. In addition, difficult geographical conditions meant that the costs of providing the country with a low-cost transportation system were high. Viewed in terms of providing infrastructure investment adequate for the country's development needs, the fiscal resources available to the Brazilian state until the end of the nineteenth century seem to have been relatively small.

The central government's low expenditure levels reflected basic features of the fiscal situation that confronted the Brazilian state in its efforts to raise tax revenues. Public finance was constrained by the paucity of tax bases that would yield revenues commensurate with the costs of tax collection. Consequently, government expenditure levels did not reach the scale that would have been socially optimal if such transaction costs did not have to be considered. As noted earlier, the Brazilian state had major incentives (if only for its own self-aggrandizement) to enlarge the volume of economic resources at its disposal. The country's landowners, who would have appropriated most of the benefits of expanded public investment, also stood to gain. But a large increase in fiscal penetration (increase in the size of the tax base) within the broader society also involved significant economic costs. The net marginal social benefits of public-finance expansion were thus low. As a result, such fiscal expansion understandably (and rationally) encountered resistance on the part of Brazil's socioeconomic elites.

Because of the great distances, poor communications, and low literacy rates present in nineteenth-century Brazil, the costs involved in tapping most potential tax bases were high. By contrast, the administrative costs of collecting taxes on imports and exports were relatively low. Accordingly, the Brazilian state's revenues and expenditures depended heavily on foreign-trade duties. Between 1830 and 1885, some 70 percent of the government's revenues came from taxes on imports and exports. As this number indicates, generalized taxes on agricultural land were not an important source of government revenue in nineteenth-century Brazil. In this respect, Brazil contrasted notably with countries otherwise as diverse as India and Japan in the nineteenth century. Not only would the administrative costs (including a cadastral survey) of generalized land taxation have been high, but the revenue prospects of such an effort were meager. An important difference with India and Japan was Brazil's abundance of land and the ensuing low ratios of labor to land in the domestic agricultural sector. With little pressure of population on land, Ricardian rent, the basis for land taxation, was small. These conditions, which made for high transactions costs and a low economic surplus in the domestic agricultural sector, meant that the net fiscal yield of generalized land taxation would have been small.

Fiscal prospects in Brazil's foreign-trade sector were more attractive. There, transactions costs were not so large relative to the size of the economic surplus as to lower sharply the net social gains of taxation. Because of these differences between the foreign-trade sector and the domestic agricultural sector, government revenues and expenditures depended heavily on the value of Brazil's foreign-trade receipts. The tax rates imposed on this base, however, could not be set at arbitrarily high levels lest exports and imports diminish to the point where tax revenues would fall. Unfortunately, through most of the nineteenth century, Brazil's foreign trade volume was too small to provide the fiscal resources needed to finance infrastructure development. A comparative perspective from the United States is useful in this context. The central government in the United States also relied heavily on foreign-trade duties as a source of tax revenue in the nineteenth century, but foreign trade provided a much larger tax base in the United States. From the 1820's through the 1850's, U.S. export receipts were approximately five times larger than those of Brazil. In the subsequent four decades, the ratio was even higher, 6.8 to 1. As these numbers indicate, the central government in the United States could draw on a much larger tax base to support its expenditure programs.

The Brazilian state attempted to supplement its revenues by borrowing, both at home and abroad. In 1864, before the sharp rise in government expenditure that came with the Paraguayan War, government debt (including money issued by the government) amounted to £5.5 per capita. Moreover, the Brazilian state's borrowing was not limited to foreign sources. Between 1841 and 1889, the share of domestically held obligations in the government's total debt-service payments ranged from 42 to 62 percent. Although borrowing afforded the Brazilian state a welcome short-term addition to its fiscal resources, it did not solve the country's public-finance problem. The scope for borrowing was set ultimately by debt-service capacity and hence by tax revenues. Until the end of the nineteenth century, the volume and growth of Brazil's foreign trade were too small to permit a high level of government expenditure.

This text was excerpted from How Latin America Fell Behind—Essays on the Economic Histories of Brazil and Mexico, 1800-1914, Edited by Stephen Haber, Stanford University Press, 1997, 316 pp

Send your
comments to
Brazzil