Agrarian Structure and Underdevelopment in Latin America: Bringing the Latifundio “Back In”

Latin American scholars in the twentieth century often pointed to the agrarian question and the dominance of the large latifundio estates as the key to explaining their countries’ underdevelopment. Yet, in more recent years, the emphasis on agrarian structures has declined and the latifundio is no longer seen as a relevant factor to economic development. In its place have emerged explanations that emphasize international trade relations, institutions, or the capacity of local states. This study makes a case for bringing the latifundio “back in” to explanations of underdevelopment in Latin America, and points to the persistence of particular property relations, low-intensity land use, and low-risk productive strategies throughout the region as barriers to development. Agricultural census data and various other sources on land use suggest that large latifundio estates are still a relevant factor throughout the region, and I argue that they continue to represent a significant barrier for the deepening of industrialization.


Introduction
Shortly after Brazilian president Dilma Rousseff was reelected in 2014, her newly appointed minister of agriculture claimed that the large unproductive landholdings known as latifundios "no longer exist[ed]" in Brazil (Passos 2015). The comment ignited a controversy among social movements and sparked a flurry of responses from members of the Worker's Party (Partido dos Trabalhadores), who accused the minister of bowing to the interests of the landowning class and seeking to sabotage efforts for agrarian reform. The strongest reaction came from the Landless Workers Movement (Movimento Sem Terra), who launched protests around the country and blocked roads to demand the removal of the minister. The MST insisted that latifundio not only was still prevalent throughout much of the country but that the problem continued to be a central issue for Brazil's economic development.
Indeed, for much of the twentieth century, the agrarian question and the role of the latifundio was a major focal point in the debates on Latin American development. This was especially the case among Latin American scholars, who wrote numerous studies on the large rural estates that dominated their countries' agrarian structures. Authors like Alberto Passos Guimarães (1964) of Brazil and Miguel Acosta Saignes ([1938] 2009) of Venezuela placed the problem of latifundio at the center of their explanations for economic backwardness in their countries, focusing on the coercive labor regimes that predominated on these estates, and the consequences for the formation of domestic markets for industry.
However, in more recent years emphasis has shifted to other explanations, and the focus on agrarian structures has declined. Since the 1970s, dependency-inspired approaches turned the focus toward international trade relations and countries' particular integration into the world economy, whereas more recent approaches focus on the state, state capacity, and the qualities of a country's political and economic institutions. Latifundio is often seen as a thing of the past, no longer a relevant factor in economic development. Indeed, some have questioned whether we can even speak of an agrarian question in the globalized world of the twenty-first century (Bernstein 2006).
A look at the data on Latin American agriculture, however, would seem to suggest otherwise. Not only have the infamous latifundio landholdings not disappeared, but they continue to dominate the landscape throughout much of the region. Though the labor regimes that once predominated on these estates have been transformed, and many large estates have modernized, they continue to greatly underutilize the land that they occupy, often grazing cattle on massive tracts of fertile land instead of carrying out intensive cultivation. In other words, the continued dominance of the large, underproductive landholdings still contributes to underdevelopment, not because of the persistence of certain social relations, as once argued, but rather because of systematic underinvestment and underutilization of the land and the consequences this has for the deepening of domestic markets.
In this study, I make a case for bringing the latifundio "back in" to explanations of underdevelopment in Latin America. I argue that the tendency for scholars to focus on the labor regimes in agriculture has led them to misunderstand the logic of the latifundio, both past and present, and to overlook what is more important: the property relations. Widespread portfolio-based land appropriation has meant that the property relations characteristic of large landholdings have often changed little from those of the traditional latifundio, and thus a similar productive logic has remained in place. This has led to a short-circuiting of the process of capital accumulation in agriculture, which in turn has cut short the process of industrialization and capitalist development. Using data from agricultural censuses in various countries along with various sources on land appropriation and land use, I attempt to show that the persistence of the latifundio agrarian structures means there is still an unresolved agrarian question throughout much of Latin America, and this is still a major impediment to sustained processes of capitalist development.

The Focus on Labor Regimes
For many Latin American scholars seeking to explain their region's backwardness in the first half of the twentieth century, the prevalence of latifundio-dominated agrarian structures was key. The latifundio was seen as a fundamental impediment to economic development due to its feudal-like social relations, its tendency for monocropping, and its negative impact on the formation of domestic markets. However, it was these scholars' emphasis on the labor regimes that were characteristic of the latifundio that prevented them from fully grasping the nature of the problem.
In his pioneering 1938 book, Venezuelan anthropologist Miguel Acosta Saignes stressed the role that the rural economy played in blocking the formation a domestic market for industry: "The primary problem of Venezuelan industrialization centers on the meager domestic market. The great majority of peasants that have absolutely no capacity for consumption must be converted into masses capable of consuming products" (Acosta Saignes [1938] 2009, 144).
The reason for this situation, according to Acosta Saignes, was that the labor regimes on the latifundios, such as labor tenancy and sharecropping, kept workers mired in poverty and unable to consume manufactured goods. Although he recognized the pattern of low investment and extensive land use, Acosta Saignes's primary focus was on ending the exploitation of the peasantry and rural laborers. This could only be achieved, he argued, through a land reform that would divide up the large estates and redistribute them among the peasants, thus ending their dependence on various forms of tenancy and dissolving precapitalist productive relations. "The semi-feudal forms of production and labor relations must be replaced by others of a capitalist form," he argued (Acosta Saignes [1938] 2009, 143).
This was also the view of the so-called feudal current of scholars in Brazil, who were engaged in a debate about whether the latifundio represented the vestiges of a feudal system or had always been thoroughly capitalist (Sodré 1944;Guimarães 1964;Furtado 1964;Prado Júnior 1966). For Guimarães, the primary objective of any land reform would be to "uproot and destroy, in our agriculture, the feudal type of relations of production" (1964,32). These included the latifundio's monopoly on land, the extraeconomic power of landowners over peasants, and the backward labor regimes that predominated in the countryside. Much like Acosta Saignes, Guimarães saw these productive relations as limiting the domestic market needed for industrialization: "all of this pre-capitalist apparatus of production and distribution … decapitalizes the country and limits the industrial development; and to the extent that it constricts the purchasing power of the rural masses, it limits the expansion of the domestic market" (Guimarães 1964, 34). 1 Many of Latin America's communist parties adopted this view as well-indeed many of these scholars were communist party members themselves-and the struggle against feudal relations became an essential part of their strategy in the postwar period. This led them to seek alliances with their respective national bourgeoisies, seen as key agents in eradicating feudal relations, who could assist them in their aim of ushering in capitalism and allowing for an eventual transition to a postcapitalist order.
But, ironically, just around the time political activists and scholars began emphasizing the backward social relations in agriculture, those very relations were beginning to disappear throughout Latin America. By the middle of the twentieth century, the increasing mobilization of peasant organizations demanding "all land to the tiller," and the growing calls for land reform among scholars and political leaders, were forcing landowners to legitimate their holdings and avoid expropriation by shifting away from the various forms of tenancy toward more direct control over production. In addition, a decline in labor shortages meant there was a decreasing need to secure laborers through tenancy, and so tenants could be replaced by wagelaborers (de Janvry 1981, 82). This resulted in the widespread substitution of wage laborers for tenants, and, in the eyes of many, a transition to a thoroughly capitalist mode of production.
Around this same time, the debate about the role of feudal or noncapitalist forms of production provoked critical responses from scholars influenced by the emerging dependency school of thought. Authors like Andre Gunder Frank (1967) andCaio Prado Júnior (1966) argued that the latifundio had never represented feudal relations at all, but rather had been thoroughly capitalist from the beginning. Frank pointed to the commercial nature of the large estates and the fact that they were often engaged in commodity production for international markets, not closed off subsistence production. This was seen as proof that capitalism had completely penetrated the periphery going back to the colonial period.
This claim from the dependentistas was partly due to their desire to explain underdevelopment as a product of integration into a capitalist world economy. They argued that countries on the periphery were progressively underdeveloped by the penetration of capitalism and their incorporation into the capitalist world system as primary exporters. Underdevelopment was largely a product of external forces, the relations between the periphery and the metropolis, and the periphery's dependence on the metropolis for export markets, capital, and technology. Therefore, attempts to explain underdevelopment based on peripheral countries' internal dynamics-the presence of certain relations of production-presented a challenge to dependency's emphasis on the capitalist world system.
As dependency's influence declined in the final decades of the twentieth century, more recent approaches began turning attention back to various factors internal to these countries. Yet, for the most part, the focus shifted away from the agrarian question. Given that the labor regimes in agriculture had been transformed along capitalist lines in the second half of the twentieth century, it seemed there was no longer an incomplete transition that needed to occur in the countryside. With capitalist labor relations fully in place, much of the focus turned to the state, what role the state should play in promoting economic development, and what should be left to market forces. Free market advocates called for greatly reducing the role of the state in the economy, while other theorists responded by emphasizing the importance of state intervention, the need for greater state capacity, and the type of interventions the state must make to be successful (Evans 1995;DiJohn 2009).
By the twenty-first century, it was argued that we had reached the "end of the agrarian question" in the developing world (Bernstein 1996(Bernstein , 2006. Globalization and the integration of southern agriculture into global commodity chains and networks of corporate agribusiness had rendered the agrarian question obsolete, as the linkages between domestic agriculture and industry had been replaced by international circuits and world markets. In addition, the disappearance of past forms of "predatory landed property," such as the feudal-like latifundios, and the "generalization of commodity relations" in the countryside meant that capitalism had fully permeated agriculture in the global South, thus ending the debate about precapitalist obstacles to development (Bernstein 2006).
Much like earlier debates, more recent debates focus largely on the labor regimes in agriculture, with little attention given to rural property relations. Though some still argue for the relevance of agriculture in industrialization (Byres 2012;Kay 2002), both sides of the debate tend to agree that the defining characteristics of agrarian capitalism are the proletarianization of rural labor and the market integration of rural productive units. Akram-Lodhi and Kay (2010, 256) note in a review of the debates that "a central moment in the development of rural capitalism" was "the emergence of generalized rural wage labor and, as a corollary, the emergence of agrarian capital." Meanwhile, for Bernstein (2006, 454), capitalist relations are denoted by "how economies are located in international divisions of labor, and circuits of capital and commodities." The globalization of agribusiness and the increasing integration of southern agriculture into global markets are seen as proof that capitalist imperatives and market forces have thoroughly penetrated the global South. According to Akram-Lodhi and Kay (2010, 271), "the productive subsector that produces to sell must strive under the market imperative to improve its competitive profitability within a law of value that is global in scope." This means that those who produce for the market are "strongly bound by the logic of the market imperative, in that [they] must continually strive to improve competitiveness" (Akram-Lodhi and Kay 2010, 271). In other words, integration into world markets has created the universalization of capitalist imperatives on a global scale. Producers in the South are now, as Bernstein (2004, 139) puts it, exposed "to the same fundamental forces of capitalist class relations and differentiation as their counterparts in Europe or North America." However, the problem with these characterizations is that they do not take into account the specific property relations in agriculture in Latin America and other parts of the global South. That is, they do not account for the specific relationships that exist between producers and their land, and whether access to land is mediated by market competition. As has been argued by Brenner (1982), the existence of specific property relations is essential to the functioning of the market imperatives characteristic of capitalist society. Regardless of the presence of wage labor or market integration, in the absence of capitalist property relations the basic forces of capitalist competition and market discipline will not function, and thus producers will not be subject to market coercion.
The persistence of large, underproductive landholdings in Latin America provides ample evidence of this. Those who claim that the forces of capital are universal and that all production is subject to the same basic market forces cannot account for what is still today the most dominant sector in many Latin American countries-the large landed estate. These estates are widely recognized for maintaining uncompetitive production methods and greatly underutilizing the land they occupy. And though they operate with wage labor and are often highly integrated into world markets, they tend to be immune to market pressures, maintaining low-investment, low-risk production strategies and extensive as opposed to intensive uses of the land.
The reason for this, I argue, is that these productive units maintain distinct property relations that make them not subject to the same fundamental forces of capitalist competition as producers in the global North. The long-standing focus by scholars on the labor regimes in agriculture has meant they have largely ignored this important aspect of the rural economy, and thus have failed to understand the underlying dynamics of Latin American agriculture and its role in the region's underdevelopment. As I argue in the next section, the presence of specific property relations is the key mechanism behind sustained processes of capital accumulation, and this mechanism is largely absent in much of Latin American agriculture.

The Primacy of Property Relations
If much of the analysis on Latin America has tended to associate capitalism with the presence of wage labor and market integration, scholars have also tended to see the relationship between capital and labor as the driving force behind the process of capital accumulation. As Alain de Janvry states in his well-known work on the agrarian question in Latin America: "The accumulation of capital is the objective force that guides the historical development of capitalism, and the exploitation of labor by capital is the mechanism by which this accumulation occurs" (de Janvry 1981, 8).
Thus, the transition to wage labor on Latin America's estates is often associated with a certain modernization and development of the productive forces in agriculture (Bernstein 2006;Kay 2002). Indeed, it is often argued that the transition to capitalism in Latin America was a transition "from above" in which the precapitalist estates transformed themselves internally along the lines of Lenin's "junker road" as they gradually shifted to wage labor and expelled tenant laborers (de Janvry 1981, 111). As the latifundios adopted more modern labor regimes, they also began to adopt more modern methods of production, and became increasingly capitalized in the second half of the twentieth century. In other words, the shift in labor regimes meant these estates became fully capitalist, and this new relationship between labor and capital was the mechanism that drove capital accumulation.
However, what this view ignores is that the market forces that drive accumulation in capitalism depend much more on the commodification of the means of production, such as land, than on the commodification of labor (Brenner 1982;Wood 2002). Regardless of whether or not wage labor is employed, it is only when land becomes a commodity that is traded on the market that producers may be forced to make competitive use of their land and to sell competitively on the market in order to maintain possession of the land. Indeed, some of the most advanced and dynamic agriculture in the world, such as that of the United States, has generally made little use of wage labor in its historical development. Instead, it was the generalized commodification of land that forced producers to plow surpluses back into production, innovate, and transform the productive process to match the competition and stay in business (Post 2011). 2 It is for these reasons that the key mechanism behind capital accumulation is actually not the labor relations in agriculture, as de Janvry suggests, but rather specific property relations. These relations, when predominant in a given society, facilitate the functioning of market imperatives and impose market discipline on producers to maximize the output of their land, making capital accumulation become generalized and systematic (Post 2011;Brenner 1982). Those that do not systematically reinvest surpluses back into production may risk being unable maintain a competitive rate of profitability, be unable to remain solvent, and thereby lose possession of their land. In this way, it is the property relations that create the market discipline characteristic of capitalist society and thus the continuous drive to maximize profits.
In many Latin American countries, these specifically capitalist property relations have seldom materialized on the large estates, despite the transition to wage labor and a high level of market integration. Indeed, scholars have seldom focused much attention on the specific relationship between landowners and their land in Latin America. Yet, the specific ways in which land has been appropriated throughout history has meant that landowners are often under little pressure to produce their land effectively or to maximize profits in order to maintain possession.
Going back to colonial times, land in Latin America has often been acquired outside of market mechanisms. This typically occurred through massive land grants from the crown such as the merced, or the sesmaria in Brazil, or, after independence, through free or low-cost land concessions from national or local governments (Furtado 2003, 68-80). Moreover, the illegal or informal occupation of public lands-what are known as tierras baldías, or terras devolutas in Brazil-has also been a common form of nonmarket land appropriation that has continuously occurred up to the present day. All of this not only led to the prevalence of massive landholdings and an extreme concentration of land, but also, more importantly, to specific property relations in which a lack of competitive constraints meant large landowners did not need to engage in systematic capital accumulation and could instead maintain low-intensity, low-investment production strategies that provide rents from the land without the risk of large investments (Soto Baquero and Gómez 2012, 11). As Edelman (1992, 22) explains: "the important point is that the dynamics of accumulation are radically different than those of classical capitalist development. Rather than investing heavily in improved technologies, employing productive human labor, attempting to capture increased market shares, or developing linkages with other production processes, latifundistas could become wealthy from harvesting natural and quasi-natural products of the land." In other words, instead of maintaining a competitive use of the land, producers could minimize risk through more extensive uses of the land such as cattle grazing or forestry, which provide a constant income but require only a fraction of the investment. When possession of land is not mediated by market competition, producers are under few market constraints to utilize their land competitively, allowing them to ignore productivity and channel investment away from production toward the diversification of their economic activities. Surpluses from agriculture can be invested into other more lucrative businesses, commercial endeavors, or the acquisition of more land, with little need for systematic reinvestment.
This productive logic has continued up to the present day on large estates throughout much of Latin America. Though land was widely commodified by the twentieth century, and land markets became the principal means of acquiring land, the property relations on large estates were often not significantly changed. This is because the general pattern of land appropriation came to be characterized by a logic of portfolio diversification (Bicalho and Hoefle 1989;Richani 2012). That is, land is acquired by individuals who have accumulated wealth from various economic activities, usually commerce or other urban endeavors, and who seek to invest in land as a speculative investment, or as a store of wealth in the context of high inflation. This type of land appropriation is similar to what de Janvry (1981, 109) described as the "merchant road" to capitalism in Latin America. Yet, unlike de Janvry's characterization, the result has not been conducive to capital accumulation.
Because land is appropriated under a logic of portfolio diversification, it means the owners possess various other sources of wealth and income and therefore do not depend on the output of the land. Regardless of the fact that the land is acquired on the market, the owners are under little market pressure to use the land competitively in order to maintain possession of it. Instead, land can be paid for with wealth generated from other economic activities, and profits from the land can be channeled back to those economic activities or into an even greater diversification of investments. Instead of being used as a productive asset, land is treated as a medium for storing wealth that otherwise would be eroded by inflation or taxed by the state, with the actual production of the land playing only a secondary role. In other words, this specific relationship between producers and their land results in a short-circuiting of capital accumulation in agriculture.
In fact, the so-called modernization that occurred on large estates in the second half of the twentieth century has often been characterized by scholars as a "conservative" modernization, or "pseudo-modernization" (Bicalho and Hoefle 1989). Many landowners began to implement modern technologies and methods on their farms such as improved crop varieties, animal breeds, and fertilizers, yet they often continued to employ low-investment production strategies that greatly underutilized the land. Extensive activities like cattle grazing continued to operate largely as before, as Bicalho and Hoefle (1990, 57) explain for northeast Brazil: "While the new system of cattle raising uses such technical innovations as planted pasture, pasture divisions with rotation of use, purchased animal feed, improved breeds and the greater use of vaccines, which together with the use of waged labour, satisfy the most demanding definitions of capitalized agriculture, the productivity per hectare has not increased significantly. Mere pseudo-modernisation has occurred. The ranches have all the trappings of being highly productive but the pastures only have one or two steers per hectare." Even in the context of neoliberalism and the opening up of markets, landowners who have acquired land as portfolio diversification are often not under market pressure to make significant investments or improve productivity, as their continued possession of the land does not depend on competitive production. Neoliberal reforms to land markets have done little to change this, since it is not the functioning of land markets that creates this scenario, but rather the predominant forms of land appropriation among elites that generate specific property relations. Therefore, even in the context of dynamic land markets, elite land can often remain underutilized or entirely unproductive without the threat of foreclosure or bankruptcy.
In what follows, I attempt to demonstrate how widespread this productive logic is in Latin America today. Although it is difficult to determine the true extent of certain property relations, agricultural census data can demonstrate how prevalent large landholdings are throughout the region and how they utilize the land that they occupy. As I will show, the data on large estates reveals the prevalence of low-risk, lowinvestment production strategies, and there is evidence from various countries of widespread portfoliobased land appropriation that gives rise to this type of productive logic. In the final section, I address the possible consequences of the region's particular agricultural logic for larger processes of industrialization and economic development.

The Prevalence of the Latifundio
For this section, recent agricultural census data was collected from six major Latin American countries: Argentina, Brazil, Colombia, Mexico, Peru, and Venezuela. The type and amount of data collected in each country's census varies considerably, and therefore the depth of analysis is limited by the availability of the data. For this reason, the focus will be mostly on Brazil and Peru, which have the most detailed data, and will include only limited data from Colombia, Venezuela, Argentina, and Mexico.
The first thing to note from the data is the continued dominance of large landholdings in these countries. Despite frequent claims over the years that the large estates would disappear due to their inefficiency and inability to compete with the more productive small and medium producers, this does not appear to be the case. As Table 1 shows, large holdings of 500 hectares or greater occupy the majority of agricultural land in all of the countries studied, 3 and the largest holdings of greater than 1,000 hectares occupy nearly half of all agricultural land in Brazil and Venezuela, more than two-thirds in Peru, and three-quarters of all agricultural land in Argentina and Colombia. 4 In other words, though land concentration has decreased in some places in recent years, it is still extreme in most major Latin American countries, and the countryside continues to be dominated by large landholdings. However, it should be noted that there are significant differences both between and within countries. In some countries, such as Chile, large unproductive estates are practically nonexistent (Bengoa 2013), whereas in other countries, such as Colombia, they are ubiquitous. Likewise, there are often significant differences from one region of a country to the next. Northeast Brazil, for example, has a significantly different agrarian structure from southern or southeastern Brazil, and these differences should not be overlooked.
Moreover, concentrated ownership and the dominance of large holdings does not necessarily mean the land is underutilized, nor that it is operated under a logic of risk minimization. In some regions, large farms are used for intensive mechanized agriculture, and areas of highly productive agriculture have developed in many Latin American countries. Indeed, countries like Brazil and Argentina have experienced a major boom in soybean production in recent years, and much of this production is carried out on very large, mechanized farms.
Therefore, to get an idea of the relative weight of large mechanized farms as compared to other less intensive forms of production I look to data on land use specific to large farms. Figure 1 shows land use on large holdings in Brazil, Colombia, and Peru, the only countries for which land use data specific to farm size was available. These holdings are almost entirely dominated by extensive activities like grazing and forest, with over 80 percent of the land they occupy dedicated to these two activities, while only a small portion is used for cultivation. This appears to be the case for the other countries as well, as data on overall land use 4 Detailed data on land distribution by holding size was not available for Mexico.  reflects this same general pattern. In other words, while it is true that in some regions large holdings are used for intensive cultivation, the overall trend is clearly the opposite-extensive activities like grazing and forest predominate on Latin America's large farms. Not only do extensive activities predominate, but land use statistics reveal the low-investment and lowproductivity nature of these types of production. On grazing land, for example, if we divide the total number of animals on large farms by total hectares of pasture, Brazil's large farms have only 0.65 animals per hectare of grazing land, while in Peru it is an incredibly low 0.06 animals per hectare (IBGE 2012;INEI 2012). This means that these landowners average far less than one animal for every hectare of grazing land, whereas smallholdings may have anywhere from 2 to 10 animals per hectare. In Peru, small farms of under fifty hectares average 12 animals per hectare of grazing land, about two hundred times greater than the country's large estates (INEI 2012).
In addition, much of the land dedicated to grazing and forest is entirely unimproved, meaning that the landowners have invested little to nothing in the land and simply harvest it for its natural and seminatural products. On Brazil's large ranches, 33 percent of all grazing land is unimproved, whereas on Peru's nearly 95 percent is unimproved (IBGE 2012;INEI 2012). This means there has been no investment in such things as planted or irrigated pastures, new grass varieties, or fertilizers that can raise the productivity of grazing land. This data was not available for large holdings in the other countries, but the situation is likely to be very similar. In Venezuela, 73 percent of all grazing land is unimproved (GBV 2010).
In other words, much of the livestock production in these countries takes place on vast expanses of largely unimproved land with a very low density of livestock per hectare. This is in stark contrast to livestock operations in North America or Europe, where nearly all production involves planted and fertilized pasture, as well as intensive, concentrated factory farms such as feedlots or other types of confined feeding operations. In cattle production, feedlots have only very recently begun to be used in some Latin American countries, with about 5 percent of cattle production in Brazil now passing through feedlots, and about a third in Argentina due to government subsidies that began in 2007 (Guevara and Grünwaldt 2012).
Additional data on agricultural methods and technologies further reveals the limited nature of the modernization of Latin America's large estates. As various scholars have pointed out, the large farms often adopt some modern technologies while still maintaining low-intensity activities that underutilize the land. Table 2 shows the percentage of large landholdings in Brazil and Peru that use various methods and technologies associated with modern agriculture. Only a minority of these farms invests in fertilizers, insecticides, or irrigation, and very few use harvesters or other agricultural machinery. Although a majority of large farms in Brazil have tractors, these are often used as a means of transportation and clearing of land, not for cultivation. And while vaccinations are widely used in both countries, very few large cattle farms in either country employ artificial insemination.
Given the fact that Brazil has one of the more dynamic agricultural sectors in the region, it is likely that the rest of the countries are even less modernized. The data for all holdings in Mexico and Venezuela gives an idea of how prevalent the various technologies are throughout the agricultural sector in these countries. In both cases, a minority of landholdings employs technologies associated with modernized agriculture, and we could likely expect similar or perhaps lower rates among the large estates that occupy most of the land.
Brazil's census data on production and investment provides another indication of the low-productivity logic of large holdings. While large farms of five hundred hectares or greater occupy 56 percent of all agricultural land in Brazil, they account for only 36 percent of the total value of production and make up only 43 percent of total investment (IBGE 2012). This is in sharp contrast to small farms of less than fifty hectares, which occupy only 13 percent of the land yet account for 35 percent of the total value of production and 28 percent of total investment. In other words, the large farms occupy more than four times as much land as small farms, yet have only 1 percent greater total value of production and only 15 percent greater total investment. These findings are further supported by information from Brazil's agrarian reform agency, which reported in 2010 that more than 50 percent of all large landholdings and 72 percent of all land occupied by large holdings was considered "unproductive" according to agency parameters (INCRA 2011). 5 Thus, despite the limitations of agricultural census data, we can confirm that the agrarian structures in these countries continue to be characterized by extreme land concentration and are dominated by large estates that employ low-investment, low-productivity production strategies. What gives rise to this situation, I argue, is the prevalence of portfolio-based land appropriation and the resulting property relations, which exempt large landowners from market pressures to maximize the productivity of their landholdings. However, detailed information on land appropriation can seldom be found in agricultural census data, and so I turn to secondary literature for evidence of this.

Portfolio-based land appropriation
There is extensive evidence of portfolio-based land appropriation and other similar types of speculative land investment throughout much of Latin America. De Janvry (1981, 109) described this "merchant road" to capitalism as one of the principal ways in which capitalism developed in the Latin American countryside in the twentieth century. Capital generated in various business activities is invested in rural landholdings for speculative purposes or for purposes of agricultural production, often under absentee management and with the use of wage labor. But while de Janvry associated this urban investment in land as a source of dynamism in agriculture, subsequent studies have associated it with extensive, low-productivity, lowinvestment productive strategies. In Brazil, Hoefle (1989, 2008) note that land appropriation by urban capital is common throughout much of the country but especially in the underdeveloped Northeast and Amazonian frontier. Their research in one region of Paraíba State showed that nearly two-thirds of large landowners in this region earned more than 75 percent of their income from urban sources, and that they typically used the land for extensive cattle grazing with little concern for raising productivity. Despite the fact that substantially higher profits could be earned through cultivation of various cash crops, or that the use of fodder or irrigated pasture could double or triple the productivity of cattle raising, survey data showed that landowners typically preferred to utilize land as a "low-risk, low-income investment for portfolio diversification," diverting surpluses to "other more lucrative investments like urban and rural land speculation" (Bicalho and Hoefle 1989, 43).
Extensive cattle ranching is preferred by the landowners because it is "subject to less market risk" and "operating costs are lower and seasonal price fluctuation is not as pronounced as in cropping" (Bicalho and Hoefle 1989, 41). A similar logic exists with the widespread cattle ranching on the Amazonian frontier as well. Much of the investment in large cattle ranches in the Amazon comes from wealthy investors from southern Brazil or from local urban-based merchants seeking to diversify their portfolios. Scholars in Brazil have long recognized this type of land appropriation to be associated with unproductive holdings or an underutilization of the land (Sales 1982;Barbosa 1998), and indeed this is the case with cattle ranching in the Amazon. The process has become known in Brazil as pecuarização, or "cattleization," as massive areas of potential cropland are converted into extensive pastures for cattle.
In Colombia, scholars have pointed to a very similar scenario in which much of the agricultural land has been bought up by wealthy investors as a hedge against inflation and taxation and then used for extensive cattle ranching or other low-investment activities (Heath and Binswanger 1998). Between 2001 and 2013, the amount of land occupied by large farms of five hundred hectares or greater increased from 61 percent to 77 percent, the vast majority of which is used for grazing (Fajardo Montana 2008; Vergara Vergara 2010). Richani (2012) argues that a good portion of the investment in cattle ranching comes from the "narco-bourgeoisie"-individuals involved in the drug trade that use land as a mechanism for laundering money. However, other sectors of the bourgeoisie are equally involved, as individuals with various business activities see it in their interest to invest in land to shelter capital gains or for speculative purposes.
The result is that despite its lower economic returns, cattle ranching is the preferred activity for Colombia's landholding elite. Regardless of whether the land is fertile, landowners minimize risk by using it for grazing, resulting in a highly grazing-dominated agricultural sector throughout the country. According to Kalmanovitz and López (2009), Colombia has at least 21.5 million hectares of land suitable for cultivation, yet only about 20 percent is actually used for cultivation, mostly by smallholders. Meanwhile, around 40 million hectares are currently used for grazing, whereas only about 14 million are deemed suitable for such use. In other words, about two-thirds of the land currently used for grazing is actually suitable for other, more profitable activities like intensive cultivation.
In Central America, the same logic has been uncovered, despite the fact that cattle ranching and beef production have been increasingly integrated into US markets since the 1960s. Williams (1986) traced the beef export boom in Central America in the second half of the twentieth century and found that although cattle production underwent modernization and began using a number of new technologies and inputs, it remained dominated by the old landowning elite and urban-based investors, who tended to utilize land for extensive grazing and often used their influence to expand onto untitled peasant and forest lands. Edelman (1992) detailed the logic of these large estates in one region of Costa Rica and found that large, underutilized landholdings persisted there throughout the beef export boom. Cattle production increased during the boom largely on the basis of extensive growth and the expansion of grazing lands "rather than from any significant intensification of production on existing grasslands" (Edelman 1992, 247). Though certain technologies and improved practices were adopted, the number of head per hectare of pasture hardly increased since the 1950s, and in many cases actually declined. This was because the landowning elite were "under little pressure to modernize" since their families had acquired the land generations ago, but also because of the "possibilities for speculative gain," as landowners could obtain cheap credit and divert surpluses toward other, more lucrative activities (Edelman 1992, 252).
As a result, much of Central America's fertile land is now used for cattle grazing, and vast tracts of the tropical forests in countries like Honduras, Costa Rica, and El Salvador have been burned and converted into pasture in recent years. In Honduras, more than 40 percent of the nation's fertile land had been occupied by cattle ranching by the 1990s (Gradwohl and Greenberg 1988, 42). Likely a greater portion is occupied today as producers continue to convert land to cattle production to supply the United States' demand for beef.
Finally, research in Venezuela confirms that this same general pattern of land appropriation and lowinvestment productive logic is present there as well. Extensive cattle ranches dominate much of the country, even in regions that contain some of the country's most fertile agricultural lands, where there is potential for much higher profits with intensive cultivation (World Bank 1993). Because much of the land was bought up by wealthy investors from outside of agriculture with diversified investments in various economic activities, there is a virtual lack of market discipline to force them to maximize the productivity of their holdings. Instead, they use the land for extensive cattle grazing while channeling profits into banking, commerce, and agribusiness (Carlson 2017).
The prevalence of this productive logic among large holdings throughout much of Latin America has led to a situation in which a majority of the agricultural land in these countries is underutilized or completely unproductive. Census data show that most countries use less than one-fifth of their agricultural land for cultivation, while the rest is used for extensive activities. Colombia is the most extreme example, with more than 80 percent of all agricultural land used for grazing, and only 8 percent dedicated to crops (DANE 2015). Other countries are not far behind, as Venezuela cultivates only 13 percent of its agricultural land, and Brazil and Peru only 18 percent (GBV 2010;IBGE 2012;INEI 2012).
This does not appear to be due to a lack of land suitable for cultivation. According to the FAO (2000), South and Central America have vast amounts of potentially arable land that could support the expansion of rain-fed agriculture. Yet these regions have among the lowest level of utilization of their agricultural potential in the world, second only to sub-Saharan Africa (FAO 2000, 39).
As Figure 2 shows, most of these countries appear to have relatively large amounts of arable land, yet they tend to cultivate only a very small portion of that land. Meanwhile, in every case analyzed, grazing takes up a disproportionately large amount of agricultural land. In other words, most of these countries would appear to be greatly underutilizing their agricultural potential not only in terms of the types of agricultural activities that dominate agricultural land but also in terms of the amount of land they use for agricultural purposes. In the next section, I argue that this situation has important consequences for industrialization and overall economic development.

Agriculture and Industrialization
Regardless of the prevalence of low-productivity latifundios in Latin America, there is still the question of how relevant this is for overall economic development in these countries. When Latin American scholars were writing in the mid-twentieth century about the importance of the latifundio, agriculture still made up a substantial portion of Latin American economies. Today, however, the agricultural sector accounts for only about 5 to 10 percent of GDP in most cases. This would seem to suggest that the dynamics of the agricultural sector are no longer very important for the overall economy and therefore that agriculture no longer really "matters" for economic development in Latin America.
But despite its reduced share in the economy, there is reason to believe that agriculture still plays an important role. Studies have shown that when forward and backward linkages are included, the share of agriculture in Latin American economies can increase to over 30 percent of GDP (IICA 2004). Meanwhile, growth in the agricultural sector has been associated with positive developmental effects that are roughly twice as large as agriculture's actual GDP share (De Ferranti et al. 2005, 61-72). In other words, even though the agricultural sector in Latin America tends to be rather small relative to the size of the national economy, it can still have large impacts on the rest of the economy through "spillover" or "multiplier effects" and through linkages between agriculture and other sectors.
The predominance of low-productivity latifundio agriculture in many countries, however, suggests that these linkages and "multiplier effects" are not nearly as strong as they could be. Low levels of investment and utilization of industrial inputs mean that the agricultural sector is providing a much weaker source of demand for upstream industries than would otherwise be the case, and, as a result, backward linkages to agriculture tend to be relatively small. Moreover, low agricultural productivity negatively impacts wages, which then reduces demand for manufactured goods from consumers. Studies have shown that rural wages are positively correlated with the level of sophistication of agriculture, in terms of use of fertilizer and irrigation, through higher labor productivity (Gasparini, Gutiérrez, and Porto 2004). Yet, as I have shown, the adoption of these methods is very low on the large estates that occupy most of the land in Latin America. Meanwhile, agricultural productivity affects wages, both rural and urban, by virtue of the impact on food prices; low productivity forces consumers to spend a large portion of their income on food, thus eroding consumer demand for other goods (Irz et al. 2001).
All of this means that low agricultural productivity can have a negative impact on domestic demand for manufactured goods in these countries and, thereby, can affect the development of local industries. Figure 3 shows the relationship between labor productivity in agriculture and level of industrialization for Latin American countries compared to high-income OECD countries. As it shows, Latin American countries remain far less industrialized than most of their counterparts in the developed world and, likewise, have much lower agricultural productivity. Moreover, there is a fairly strong relationship between the two variables, where increasing agricultural productivity is strongly associated with an increasing level of industrialization. This would seem to support the argument that low productivity agriculture continues to be an important barrier to industrialization. Of course, it could be argued that the direction of causality is the reverse; that it is industrialization that drives increasing productivity in agriculture, not the other way around. However, this does not align well with much of the historical evidence. As Dieter Senghaas notes in his review of European development processes, "nowhere has industrial development reached the state of self-sustaining growth unless an increase in agricultural productivity preceded or accompanied industrialization" (Senghaas 1985, 52, emphasis in original). In Europe, there was "a clear correlation between agricultural modernization and successful industrialization, or else between the lack of agricultural modernization and the failure of industrialization," and this is due to the fact that "the size of the industrial sector depends directly on agricultural productivity" (Senghaas 1985, 47). Others have made similar arguments for the developing world, where the size of a country's domestic market for industrial goods is seen as largely dependent on the level of productivity of local agriculture, and thus the growth possibilities for local industries depend on the level of demand generated by agriculture (Lewis 1978;Mundle 1985).
Indeed, for much of the developed world, the historical record supports the notion that industrialization was achieved largely on the basis of expanding domestic markets rather than via trade or exports, and this expansion was fueled by increasing agricultural productivity. This occurred first in England, where industrialization was driven by important changes in the agrarian structure that led to systematic gains in productivity through the widespread adoption of new innovations, methods, and technologies, which increased output and improved efficiency (Zmolek 2014, 165-214). These changes spurred industrial growth as rural producers began systematically investing surpluses in inputs and capital goods, thereby creating expanding markets for local industries, while increasing agricultural productivity raised rural incomes and thereby rural consumption of manufactured goods. Foreign markets, on the other hand, are estimated to have only played a minor role, accounting for only 10 percent of gross national product in the period leading up to the Industrial Revolution, and even less for much the rest of Western Europe (O'Brien 1982).
Japan is another important case, given that it is commonly seen as a case of export-led development. In fact, domestic sources of demand are estimated to have made up anywhere from 75 to 80 percent of aggregate demand throughout the period of industrialization (Ohkawa and Rosovsky 1973). And here too it is argued that an agrarian revolution starting as early as the seventeenth century laid the foundation for that industrial growth (Mundle 1985). In other words, industrialization in the developed world was fueled much more by internal rather than external demand, with much of it originating in agriculture. Perhaps the only true cases of export-led development are South Korea and Taiwan, and even here it is argued that important changes in the countryside, such as the transformation of agrarian class structures, were important factors for industrialization (Kay 2002;Davis 2004).
Therefore, if Latin American countries are to move up the scale to higher levels of industrialization like their more developed counterparts, it would seem that industrial growth will need to be fueled by increasing domestic demand for manufactured goods, and this will likely need to have a large agricultural component. As evidenced by both the historical record and statistical data cited above, increasing agricultural productivity stimulates demand for local industries through a variety of linkages to the rest of the economy and can therefore allow for a deepening of the industrialization process by expanding domestic markets. Indeed, the size of local markets can be a crucial factor in the viability of many industries, especially those industries that rely on economies of scale or have significant transport costs.
It is in this sense that the persistence of a stagnant, low-productivity agricultural sector in Latin America should be seen as an important barrier to industrialization. The fact that the vast majority of the countryside in these countries is dominated by large landholdings that invest relatively little in industrial inputs or productivity-enhancing technology means that demand for manufactured goods is significantly constrained. This is not only because landowners purchase relatively few industrial inputs that would stimulate local industries, but also because low levels of investment in productivity translate to low wages in the countryside and higher food costs for consumers, which restrict purchasing power and overall consumption. Extreme land concentration and the dominance of extensive activities with low labor requirements are also contributors to unemployment and rural exodus and thereby contribute to depressed wages in both rural and urban zones. All of these factors combine to create relatively weak domestic markets for many basic industries and, therefore, impede further industrialization.

Conclusion
Latin American scholars from the "feudal current" were on the right track when they pointed to a certain "feudal" or precapitalist logic on their countries' latifundio estates as the root of their development problems. Indeed, I have argued here that a certain productive logic on large holdings was, and continues to be, a major factor underlying underdevelopment in Latin America. But what led many of these earlier scholars astray was the emphasis they placed on the labor relations in agriculture, and the view that it was a certain relationship between capital and labor that drove capital accumulation. This meant that once precapitalist social relations were converted to wage labor in the second half of the twentieth century, and the agricultural labor force was proletarianized, many scholars came to see Latin American agriculture as fully capitalist, with no further agrarian transition in store in the countryside. Much of the subsequent analysis turned the focus away from the agrarian question, and the issue of the latifundio was seen by many as relatively unimportant by the end of the twentieth century.
I have argued here, however, that we must bring the issue of the latifundio "back in" to our explanations of underdevelopment in Latin America. In my view, the evidence suggests that the agrarian question has not been resolved throughout much of Latin America, as large landholdings continue to control and underutilize vast amounts of the region's agricultural lands, with little to compel them to maximize productivity. This is because the market imperatives that drive capital accumulation depend on specific property relations in agriculture, and these property relations have not been generalized throughout much of the Latin American countryside. A long history of nonmarket appropriation of the land, together with the continued logic of portfolio-based land appropriation, means that much of the landowning elite that controls the land in these countries is under little market pressure to improve productivity and increase investment, and so a productive logic of low-risk and low-investment prevails. The end result is a short-circuiting of capital accumulation in agriculture, a significant weakening of the industrialization process, and few prospects for sustained capitalist development.

Author Information
Chris Carlson is a PhD candidate in sociology at the CUNY Graduate Center. His research is on the sociology of development, with a focus on agrarian relations and structural transformation in the global south.