Research findings from the land inequality initiative

Uneven Ground

The (shocking) state of land inequality in the world

Chapter 3
Read time:
21 minute minute read

Measuring land inequality is not easy. The literature on land distribution has long relied on estimates of Gini coefficients for land distribution, using agricultural censuses that provide data on the number of land holdings and the total area of holdings by size. These estimates face various challenges – some are related to the data used, others to the methodology applied (Box 8). Despite these challenges, the use of the Gini coefficient, as traditionally presented in the literature, remains justified as it is the longest-used methodology, based on census data which are available in most countries at a particular time, allowing for a long-term perspective of land inequality across countries. These data are now complemented by innovative methodologies developed in the framework of this project, aiming at better grasping the multi-dimensional nature of land inequality (Box 8).

Box 8: Challenges with the traditional use of the Gini coefficient to measure land inequality – towards new methodologies

Challenges include:

  • Land distribution calculated using agricultural census data captures the distribution of the size of farms rather than land ownership. Agricultural censuses do not necessarily account for multiple land holdings per owner and fail to capture the full extent of land concentration.
  • The present Gini coefficient is generally uni-dimensional, not taking into consideration the multi-dimensional complexities of land inequality.
  • Other aspects related to land (quality of land, presence of assets, other resources such as water, proximity to infrastructure and markets, etc.) are not measured in agriculture censuses.
  • Agricultural censuses generally do not distinguish between different forms of legal ownership, nor do they include corporate ownership or shareholding structures.
  • Census data focus only on agricultural and landed households and do not account for landless households; thus they do not portray actual levels of inequality.
  • The Gini coefficient is a synthetic measure of inequality that summarises the entire distribution into a single number, and it is thus less informative about where the important changes in distribution take place.
  • The coverage, methodologies, and thresholds for agricultural censuses are not uniform between countries or over time, especially in developing countries; despite efforts to bring uniformity, this reduces their comparability.

Towards new measures of land inequality:

In response to these challenges, new methodologies for measuring land inequality were developed as part of this Land Inequality Initiative. Vargas and Luiselli (2020) endeavour to integrate the multi-dimensional nature of land inequality by combining – besides the standard quantitative size of land plots indicator – tenure, quality of land, endowment, assets, and other indicators. To do this, they suggest using various additional data sources.

A second approach by Bauluz et al. (2020), based on survey data, assesses land inequality based on land owned by a household (beyond distribution of the size of farms, factoring in multiple ownership of plots) and land values (as a criterion of quality of land), and also accounts for the landless. The authors implemented this methodology using a sample of countries: India, Bangladesh, Pakistan, China, Vietnam, Ecuador, Guatemala, Brazil, Mexico, Peru, Burkina Faso, Ethiopia, The Gambia, Malawi, Niger, Nigeria, and Tanzania. The selection of countries was a result of data availability and, although some of the most populous countries were covered in this analysis, more countries need to be covered in future research to obtain a more complete picture. Nevertheless, the results represent an important attempt at innovating assessments of and deepening perspectives on land inequality.

Sources: Vargas and Luiselli (2020); Bauluz et al. (2020).

Land inequality is increasing again

Available data, despite their limitations, do permit us to look back at land inequality trends over the past 100 years.

This shows that land inequality steadily decreased from the early twentieth century, up until the 1980s. At that point, the trend reversed, and land inequality has since been increasing steadily .

From a Gini coefficient of 0.64 in the early years of the century, land inequality decreased to 0.60 in 1982, but had increased again to 0.62 by 2017 (Figure 5).

Figure 5: Land inequality over time (1910–2017), measured by the Gini coefficient

Figure 5: Land inequality over time (1910–2017), measured by the Gini coefficient

The largest 1% of farms operate 70% of farmland, supplying corporate food systems

Today, it is estimated that there are approximatley 608 million farms in the world. About 90% are family farms, which include all sizes of farm from the smallest to some of the largest, occupying 70–80% of all farmland.

About 84% of farms are smaller than two hectares, but these operate only about 12% of farmland, with little if any opportunity to be part of corporate supply chains.

Already, according to Lowder et al. (2019: v), “[t]he largest 1 percent of farms in the world operate more than 70 percent of the world’s farmland”; these farms form the core of production for the corporate food system. Unless there is substantial policy intervention, given the trends in the agriculture and food systems, land consolidation will inevitably increase further.

Although land inequality patterns vary significantly from region to region, a consistent pattern of land consolidation emerges throughout (Figure 6). After 1980, in all regions, land concentration has either been increasing significantly (North America, Europe, Asia and the Pacific) or a decreasing trend is being reversed (Africa and Latin America).

A clear trend in most low-income countries is an increasing number of farms, combined with smaller and smaller farm sizes. Across the world, and especially in higher-income countries, large farms are getting bigger.

Figure 6: Land inequality trend lines since 1975, measured by the Gini coefficient

Figure 6: Land inequality trend lines since 1975, measured by the Gini coefficient

The missing middle

North America has seen a drastic increase in land and agricultural concentration. US data show a decline from 3.7 million farms to 2.1 million between 1960 and 1990, accompanied by a steady rise in average farm sizes, from 122.6 hectares to 187 hectares over the same period. From 1990 to 2010 the number of farms and average farm sizes remained fairly stable, at 2.1 million farms with an average size of around 175 hectares (Lowder et al., 2019). While the total number of farms and their average size have stabilised, the number of farms larger than 500 hectares has risen from 1971 onwards, while the total number of the smallest farms, including those of less than five hectares, has also increased. Medium-size farms ranging from 50 to 500 hectares have significantly decreased in number. This is an increasingly polarised and unequal land distribution. Hendrickson et al. (2017: 15) argue that “the ‘agriculture of the middle’ is declining and perhaps facing extinction”.

What the figures on farm size do not reveal, however, are the even more substantial increases in the concentration of large-scale production on a shrinking number of farms. Almost 1 million farms (980,000) in the USA have less than $5,000 in sales per annum, while the largest 7% of farms account for 80% of production value (MacDonald, 2016). This leaves a situation where some 1.3 million, or 60%, of the farms in the USA produce only 6.6% of the total production value (Gollin, 2019). These include farms of below five hectares, many of which are known as “retirement farms” or “off-farm occupation farms”, with owners who do not depend on agricultural production for their livelihoods.

A very similar trend is seen in the European Union (EU). The average farm size in the EU has almost doubled since the 1960s, from 12 hectares to 21 hectares by 2010. More significantly, the number of farms across the region that are larger than 100 hectares has increased steadily from 2005 to 2013 (Lowder et al., 2019), and fewer than 3% of farms now account for more than half of the farmed land (Gollin, 2019). The Gini coefficient for the EU, which was continuously decreasing from the beginning of the twentieth century, has increased since 1980 by almost 10%, to reach an average of 0.58.

The vast majority of the smallest farms globally are in Africa and Asia, where they are essential to the livelihoods of a large proportion of the population. Figure 7 shows the distribution of farms and land in sub-Saharan Africa, South Asia, and Latin America and the Caribbean. Most farms are smaller than two hectares, and there is a significant amount of land in farms of 2–10 hectares, while a very small proportion of land appears to be part of much larger farms.

Figure 7: Distribution of land by size class in sub-Saharan Africa, South Asia, and Latin America and the Caribbean

Figure 7: Distribution of land by size class in sub-Saharan Africa, South Asia, and Latin America and the Caribbean

Figure 7 shows that these low, and in Africa’s case even decreasing, levels of inequality have given way to new trends since the 1980s. Africa’s Gini coefficient for land has stabilised at 0.54, resulting from a combination of fragmentation due to population increase at the lower levels with an increasing interest in farmland by domestic elites and national and international corporate actors. Asia, on the other hand, has seen its Gini coefficient rise significantly, from 0.56 in 1980 to 0.62 at present, an increase of 11%. In this case, it is related to consolidation within the framework of the Asian Green Revolution, the significant number of large-scale land acquisitions for agriculture and other sectors (mining, infrastructure, tourism), and a growing landless population (Djurfeldt, 2005).

Indeed, hidden behind shrinking average farm sizes in most low-income countries is the increasing number of mega-farms, each taking up thousands, even tens of thousands, of hectares (Box 9).

In Tanzania, for example, the 108 large-scale farm investments that have recently been implemented control more land than the smallest two million farm entities combined (Wegerif and Guereña, 2020).

Box 9: Large-scale land acquisitions, commercial pressures on land, and increasing inequalities

The rush for land since 2000 is a well observed trend, which has primarily affected agrarian economies in Africa and Asia. Lands that in the early 2000s were of marginal investment interest were suddenly sought out, mainly by international investors, with demand peaking in 2010.

By 2018, the Land Matrix had identified around 1,000 large-scale agricultural land deals covering 26.7 million hectares of land around the globe (Land Matrix, 2018). Africa accounts for 42% of these deals and about 10 million hectares of land, an area the size of Iceland. Even though the global land rush has slowed, new acquisitions are still being recorded, contributing to growing pressures on rural people and their land.

Figure 8: Evolution of farming units and farming area in South Africa (1918–2010)

Figure 8: Evolution of farming units and farming area in South Africa (1918–2010)

Numerous countries in Latin America and some other (often settler) countries, such as South Africa, where the unequal distribution of land formed the backbone of wealth and asset inequality during colonial times, are still characterised by extreme land inequalities. Agrarian reforms aimed at redistributing land have largely failed to rebalance inequalities (Frankema, 2009). On the contrary, the economic model in these countries based on extractivism and agricultural exports, combined with liberal market economies, is leading to significant agricultural land expansion and land concentration (Box 10).

Box 10: The 1% – extreme land concentration in Latin America and South Africa

An Oxfam analysis of 15 Latin American countries shows that the largest 1% of farms hold more than half of all agricultural land (Oxfam, 2016). In other words, this 1% of farms occupy more land than the remaining 99%. On average, the size of each of these large farms is over 2,000 hectares (equivalent to 4,000 soccer fields), although in the countries of the Southern Cone (Argentina, Chile, and Uruguay) they are much larger. For example, in Argentina the average size of farms in the largest 1% is over 22,000 hectares. The most extreme case is Colombia, where farms of more than 500 hectares – which account for only 0.4% of all farms by number – occupy 67.6% of productive land (Oxfam, 2016).

Similar trends can be found in South Africa, where years of colonial and apartheid land dispossession, combined with mainly white investment in large-scale farms, has created a skewed land and agricultural sector dominated by a small number of capital-intensive, white-owned commercial farms. Liberalisation of the agricultural sector and its integration into global markets at the end of the apartheid era only led to increasing concentration of land and control of production. While in 1994 at the end of apartheid South Africa counted about 60,000 commercial farmers, today only 34,000 remain – illustrating, despite agrarian reforms, the significant concentration trends ongoing in the country (Cochet et al., 2015). It is estimated that only around 20% of commercial farms account for 80% of agricultural production by value. Meanwhile between 2 and 2.5 million smallholders live in rural areas and produce crops largely for home consumption and occasional sales (Cousins, 2015). They contribute only a fraction of the value of crops marketed, with 98% of them being unable to support themselves from agriculture alone.

Taking into consideration all farmers in South Africa (commercial and non-commercial), it is estimated that only 0.28% of farms produce around 80% of the value of agricultural production.

This is taking place in Africa’s most industrialised and urbanised country, which is still not able to provide non-farm jobs for its adult population, leaving 30.1% of them unemployed (StatsSA, 2020).

Sources: Oxfam (2016); Wegerif and Anseeuw (2020).

The land sector is even more concentrated than we think

By assessing land inequality using survey data and factoring in multiple ownership of plots, land values, and the landless, rather than the single measure used to produce the traditional Gini coefficient for land, it becomes clear that land inequality has been significantly underestimated to date.

Overall, research carried out for this project has found that the wealthiest 10% of rural populations across the sampled countries capture 60% of agricultural land value, while the poorest 50% of rural populations, who are generally more dependent on agriculture, capture only 3% of land value (Bauluz et al., 2020).

Compared with the traditional census data and Gini coefficient generally used, this is an increase in inequality of 41% if agricultural land value and landlessness are considered, 24% if only value is considered.

Figure 9: Differences in levels of inequality when the traditional Gini coefficient is compared with inequality measures considering land values and the landless population

Figure 9: Differences in levels of inequality when the traditional Gini coefficient is compared with inequality measures considering land values and the landless population

Methodological notes: 1) The blue bar represents the Gini coefficient for land as traditionally calculated, based on census data (using the latest data available), as explained in the previous section; the red bar represents land inequality, based on the methodology developed by Bauluz et al. (2020), based on survey data focusing on land owned by a household (factoring in multiple ownership of plots) and on land values (as a criterion of quality of land); the green bar is similar to the red one but also includes the landless population. 2) Full datasets (i.e. census data; value data and data on landlessness based on survey data) were only available for India, Bangladesh, Pakistan, China, Vietnam, Ecuador, Guatemala, Ethiopia, Malawi, Niger, and Tanzania. For this reason, the following comparisons are based only on this reduced sample of countries.

These new estimates provide important new insights into international patterns of land inequality. Here, as well, regional differences are important. Although Latin America remains the most unequal region globally, land inequalities in Asia (+30%) and Africa (+74%) increase by proportionally more – leading to Gini coefficients of above 0.70 in all regions. According to these benchmark metrics of agricultural land inequality (considering land value inequality and including the landless population), South Asia and Latin America exhibit the highest levels of inequality, with the top 10% of landowners capturing up to 75% of agricultural land and the bottom 50% owning less than 2%. The African countries display land ownership patterns that are relatively less unequal, while “Communist” Asia (China and Vietnam) is the world region with the lowest levels of inequality (Figures 10a and 10b).

Asian countries that appeared to be moderately equal using traditional measures (such as India, Bangladesh, and Pakistan) have among the highest levels of inequality when land values and the landless population are included.

China and Vietnam, by contrast, display higher levels of land inequality among landowners than South Asia and Africa, but land concentration is only slightly higher when land values and landless households are considered. According to the benchmark inequality indicator created by Bauluz et al. (2020), China and Vietnam appear to be the least unequal countries in our sample.

Latin America still displays the most unequal distribution of agricultural land. However, unlike the other world regions, land inequality among landowners is substantially lower in value than in area, which is probably related to substantially less productive large holdings compared with medium- to low-sized holdings (Bauluz et al., 2020). This factor significantly reduces the difference between Latin America and other continents.

Finally, African countries occupy an intermediate position. Africa has the lowest levels of land area inequality among landowners, but this rises significantly when land values and the landless population are included.

Figures 10a (upper panel) and 10b (lower panel): The top 10% and the bottom 50% shares of land area and land value among the land owning class, and including the landless population

Figures 10a (upper panel) and 10b (lower panel): The top 10% and the bottom 50% shares of land area and land value among the land owning class, and including the landless population

These figures indicating increasing land inequality are worrying, but they are almost certainly still an understatement of the true level of inequality, as household surveys do not pick up company-owned farms. A look at the operations of corporate entities and investment funds reveals a number that are buying and controlling large amounts of land across different countries. This is a form of concentration of ownership that is currently completely missed by all surveys, and is very hard to quantify as not all investment funds are transparent about their investments.

Hidden forces in land inequality – control over land and production is driving up even more concentration in the land sector

Less visible forms of control over land create inequality in land holding itself, as well as inequality in the power over land and the appropriation of value from the land and activities on it.

First of all, a person or entity does not need to buy land in order to have control over it. For example, contract farming has been recognised as a potential route to accumulation, with the incorporation into (global) supply chains creating new dependencies and ending up perpetuating extractive models, aggravating patterns of inequality related to land (Chamberlain and Anseeuw, 2018; Sulle, 2017; Oya, 2012). Second, there is increasing corporate concentration of ownership and control throughout the agri-food sector, which influences the way that land is used to benefit those corporate entities and their investors. Third, the growing role of financial markets and actors that treat land as an asset class can significantly change the way that land is controlled and used (Wegerif and Anseeuw, 2020).

In the agri-food sector, corporate organisation is linked to industrial modes of primary production, which seek economies and other advantages of scale. This has been closely observed for several decades in the USA, with the rapid transformation of farming towards fewer large-scale industrial-style producers that are linked, through contracts or vertical integration, with processers who are required to meet uniform standards (Lang and Heasman, 2004; Martin, 2001). In this context, levels of consolidation of ownership and control have reached further and accelerated more rapidly, through a combination of two processes: 1) concentration, i.e. the exercise of horizontal ownership and control of other companies that would otherwise be competitors in the industry (a broadening); and 2) vertical integration, or just integration, which is exercised by a company taking ownership or control of the firms it buys from or sells to (a deepening).

With these processes, as Martin (2001: 13) observes, “farming is rapidly being transformed from a rural lifestyle to agribusiness with a supply chain mentality. The application of modern business principles and manufacturing approaches to agricultural production systems is commonly referred to as the industrialization of agriculture.” These changes in agricultural production and land use go hand in hand with far-reaching integration to assure efficiency and effectiveness as well as control over value and supply chains.

The control over value chains gives these actors significant control over land, as well as over the distribution of the value of what is produced on the land, which in turn contributes indirectly to land inequality.

The potential control over land and food systems at global and local levels by certain corporations and investors goes far beyond the levels of inequality detected by agricultural census data and household surveys. One example of this kind of integration and concentration in the agri-food business is the US-Brazilian investment firm 3G Capital. While the owners of 3G are hardly household names, 3G and its founding partners are major shareholders of vast global brands covering production right up to retail, including Burger King, the Kraft Heinz Company, AB InBev (the biggest beer corporation in the world), and Lojas Americanas in Brazil – a large retail group that has recently got into the grocery business.

This concentration of control is compounded by increased interest in agricultural land from the financial sector.

Parts of the world’s farmland are now considered financial assets, with no known physical owner, subject to decision-making processes that may be external to the farm and the agricultural sector.

Agricultural production is not embedded in territory anymore but depends on financial processes and actors scattered all over the world, including the use of derivative values detached from their material base, which brings greater instability to agricultural markets and puts speculative pressures on real markets and product prices (Fairbairn, 2014).

What this all comes down to is that we do not always know who owns what land. Shareholding structures and other financial constructs are mushrooming in land (and do not have to be declared in any country in the world, to our knowledge, thus remaining totally invisible), and the opacity which often surrounds the finances and activities of investment funds (Daniel, 2012) makes it impossible to assess the full extent of their impact on land concentration and inequality.

Estimations vary substantially: Buxton et al. (2012) estimate that 190 private equity firms are investing in agriculture and farmland around the world, whereas HighQuest Partners (2010) speak of 54 funds/companies that are either actively investing in funds to acquire and manage farmland or had already announced plans to raise capital to invest in the sector. Preqin lists the top US university endowment funds (Harvard Endowment Fund, for example, composed of 13,000 individual funds, distributed US$1.9 billion in 2019), showing that 10–20% of their assets are allocated to natural resources and farmland (Preqin, 2017).

The largest asset manager by value under management is the US firm BlackRock. At the end of 2010 it had US$3.346 trillion under management, very close to the US$3.4 trillion that was the gross domestic product (GDP) for 2009 of Germany, one of the top five economies in the world (BlackRock, 2009).

By the end of 2019, the funds managed by BlackRock had more than doubled in size to an incredible US$7.43 trillion, close to twice Germany’s GDP of US$4 trillion for that same year (BlackRock, 2019).

Some of this growth is coming from investments in the agri-food sector. BlackRock is now a major investor, as are some of the other largest asset management companies, in grocery retailing, with major holdings in supermarket groups including Walmart, Costco, and Target. BlackRock and other asset managers also have big investments in the largest seed companies, such as Syngenta, DuPont, Dow, Bayer, and Monsanto (ETC Group, 2019). Blackrock and Vanguard – the second biggest asset manager, with around US$5 trillion under management – are among the largest shareholders in Tyson Foods, one of the largest livestock breeders in the world (CNN, 2020; Shukla, 2019). BlackRock and Vanguard were also the two biggest shareholders in both Monsanto and Bayer, and played a key role in their merger (IPES-Food, 2017).

With complex corporate and financial structures, cross-shareholdings, and other inter-relations, clear lines of responsibility for land use and management are becoming harder to discern, just as they are becoming more important. It is also difficult to hold investors accountable for their economic, social, and environmental impacts when the primary investors are unknown or geographically and institutionally distant from the operations invested in. When corporate responsibility measures are applied (if at all), they often have a developmental or environmental aim, yet little is being done about the impacts that corporations and financial structures are having on growing land inequality and its consequences.