5 - Land Tenure and Changing Responses to the Agrarian Question

Summary

Chapter 5 examines the EPRDF’s changing approach to agricultural development and the agrarian question. The EPRDF’s initial strategy was rooted in the TPLF’s focus on the peasant majority as its political base. The government sought to secure the acquiescence of the peasantry through the distribution of land, while agricultural inputs would raise agricultural productivity and generate a surplus that could finance industrial expansion, creating mass employment and alleviating pressure on rural land. The political crises of the early 2000s forced a re-evaluation of this agriculture-first approach, however. Faced with growing land shortages, the government sought to raise productivity at the cost of inequality and differentiation. The chapter examines, first, the government’s focus on high potential smallholders; and, second, the selective expropriation of peasant producers to make way for agricultural investments. While this strategy ultimately delivered rising agricultural productivity, a combination of population growth, displacement for investments and growing market forces eroded the main means of mass distribution and political control – access to land.

The 1970s land reforms addressed the exploitation and inequality of the Imperial land tenure regime. Yet, the Derg’s Soviet-style collectivisation fell short of addressing the agrarian question. Since 1991, the EPRDF sought to transform the country’s economy and, while industrialisation was the ultimate objective, the new government was centrally concerned with the core components of the agrarian question. This chapter examines the government’s evolving agricultural development strategy and the importance of access to land and agricultural inputs as tools for mass distribution. This analysis is framed in terms of the three aspects of the agrarian question, as proposed by Terrence Byres and developed by others (Byres 1991, 1996, Bernstein 2004, Akram‐Lodhi et al. 2009). These relate to: first, industrial accumulation and the role of an agrarian surplus; second, class relations within agriculture, including differentiation within the peasantry; and third, how political forces shape processes of agrarian change.

The chapter charts the EPRDF’s evolving development strategy, beginning with the initial approach to the agrarian question, which was rooted in the TPLF’s strategy for fighting the Derg and the focus on the peasant majority as the party’s political base. As such, the government sought to secure the acquiescence of the peasantry through the distribution of land and agricultural inputs that would raise agricultural productivity and generate a surplus that could finance industrial expansion. However, the government recognised early on that rapid population growth posed a central challenge and that agricultural development that spurred industrialisation was essential to creating mass employment and alleviating pressure on rural land.

The political crises of the early 2000s forced a re-evaluation of this agriculture-first approach, however. Faced with growing land shortages in densely populated highlands, the government focused on raising agricultural productivity at the cost of inequality and differentiation. To this end, the chapter examines, first, the government’s focus on high potential smallholders and measures to protect individual tenure security; and, second, the selective expropriation of peasant producers to make way for agricultural investments. Official figures show rapid growth of smallholder agricultural productivity as a result. However, in the context of rapid population growth, the individualisation of land tenure inevitably meant that growing numbers of people – particularly young adults – had little prospect of access to land and an agricultural livelihood, while existing landholders were threatened by state-directed investments. The result has been to erode land access as a tool for mass distribution, heightening the political importance of rapid industrialisation and mass employment creation.

A Socio-economic and Political Strategy for National Government

The EPRDF’s central concern on taking national power was to secure a political base amongst the peasantry. Under federalism, the EPRDF sought to mobilise a peasantry that is homogeneous in class terms but divided along ethnic lines, through ethno-regional administrations and EPRDF-affiliated parties ‘with its own members, in its own language, using its own cultural traditions and knowledge system’ (Vaughan and Tronvoll 2003, p. 118). To do so, the government relied on state land ownership and the distribution of usufruct rights to enmesh the peasantry. However, the upkeep of these relations of dependence on the party-state required additional forms of distribution and therefore a concerted effort to expand available resources. As such, the Agricultural Development-Led Industrialisation (ADLI) strategy was conceived both as an economic development strategy and a strategy for maintaining EPRDF political control.

Land reform was a key pillar of the TPLF’s strategy for mobilising the Tigrayan peasantry, and the EPRDF viewed state ownership as essential to establishing control over the peasantry once in national office. However, the EPRDF’s commitment to state ownership was not shared by foreign donors or opposition parties. As Meles Zenawi later noted, the ‘advice we got from practically everyone was that we should privatise land … but we were not convinced’ (Zenawi 2006a). Initially, the transitional government compromised, proposing a referendum on land tenure after the transition period (TGE 1991, p. 21). As with the EPRDF’s approach to sharing political power, however, this initial openness did not last. The referendum was abandoned and the 1994 Constitution vested land ownership in the ‘State and peoples of Ethiopia’ (FDRE 1994, para. 40). Meles declared to parliament that land privatisation would only happen ‘over the EPRDF’s dead body’,¹ with the Constitutional provision intended to end the land debate,

You know the game of chicken, with the two cars coming at each other? The best way to play it if you don’t want to ‘blink’ is to remove the steering wheel and throw it out of the window … I think they [the international financial institutions] recognized that the issue was dead.

(Meles, cited in Gill 2010, pp. 93–94)

State ownership maintained the distributive role of the 1970s reforms, with the state providing land access to a homogeneous peasant class, while preventing the re-emergence of a capitalist or ‘feudal’ landed elite (EPRDF 1993, 2006). Indeed, the government framed state ownership as ‘the only social security the peasants have’ (Meles, cited in Marcus 1995). In contrast, land privatisation and market forces were a threat to the peasantry and thereby political order, with Meles repeatedly questioning the logic of privatisation,

By fully privatizing land ownership, one starts the process of differentiation. The creative, vigorous peasant farmer gets to own larger pieces of land and the less effective get to be left to live in doubt. And that is supposed to improve agricultural production and productivity in Ethiopia. We beg to differ on this issue. We do not believe this is the right approach in our country.

(Meles, cited in The Reporter 2000)

The EPRDF retained the last land redistribution before 1991, by the TPLF in Tigray and the Derg in the rest of the country, with some local initiatives to incorporate landless households through allocation of unused land during the 1990s (Rahmato 1994). The exception was Amhara, which in 1997 became the sole region to conduct land redistribution under the EPRDF. EPDM / ANDM had carried out land reforms in liberated areas of Wello during the conflict, and Amhara state officials framed the 1997 redistribution as a means of addressing Derg-era inequality and corruption in the rest of the region.² In reality, redistribution punished Derg officials who were given reduced landholdings (Ege 1997, Hoben 2001). As such, this redistribution was entirely consistent with the strategy of coercive distribution, which displaced rival elites and enmeshed the peasantry in relations of dependence.

The EPRDF acknowledged, however, that land access alone was insufficient to deliver improved livelihoods and retain mass acquiescence. In the early 1990s, agriculture was dominated by low productivity peasant production, while food insecurity remained a major problem. This was most vividly illustrated by the 1984/85 famine, which particularly affected Tigray, but also the subsequent annual appeals for international emergency assistance to avert further catastrophe. Moreover, by the early 1990s it was already apparent that population growth was resulting in growing numbers of young adults without land access. Further land redistribution could only lead to the allocation of ever smaller and increasingly unviable landholdings to a broader section of the peasantry (Bruce et al. 1994, Rahmato 1994, Habtu 1997). The government feared that ‘the weight of population pressure will decimate agriculture’ and the political order it underpinned (TGE 1994, p. 8).

ADLI was therefore conceived as a strategy for agricultural and industrial development, and thereby a political strategy for maintaining power by expanding the resources available for mass distribution. ADLI was first articulated under the transitional government (MoPED 1993, TGE 1994), building on the TPLF’s approach to mobilising the Tigrayan peasantry, but incorporating new ideas on agricultural development and structural transformation.³ According to Newai Gebre-Ab, Economic Advisor to the Prime Minister, the main influences were: Irma Adelman’s model of Agricultural Demand-Led Industrialisation (Adelman 1984) and its focus on expanding the domestic market for industrial goods; Michael Lipton’s critique of urban bias and proposal of an inverse relationship between farm size and productivity (Lipton 1977); and John Mellor’s insights regarding the links between agriculture and industry (Mellor 1995).⁴

ADLI’s short-run focus was to increase agricultural production, with state land ownership enabling the government to allocate key factors of production – land, labour and capital – in the most efficient combinations. In the highlands, where labour was plentiful and capital in short supply, labour-intensive, smallholder agriculture was prioritised, with larger-scale, capital-intensive production considered a waste of resources (Meles, cited in The Reporter 2000, MoFED 2003, p. 10). Consequently, ADLI focused on the intensification of peasant agriculture through extension services, improved seeds, fertiliser and irrigation, access to credit and basic education. In contrast, in sparsely populated lowland areas, where labour was less plentiful and the government routinely disregarded the land use of politically marginal populations employing pastoralism and shifting cultivation, ADLI aimed at agricultural extensification through ‘expansion of large-scale farms’ (TGE 1994, p. 9, see also MoPED 1993, MoFED 2003).

ADLI was based on the assumptions that, first, improved inputs are scale-neutral and, second, that there is an ‘inverse relationship’ between agricultural yields and landholding size due to the intensive cultivation practices of smallholders. As such, ADLI sought to address both food insecurity and low productivity by expanding access to improved agricultural inputs for all peasant producers (TGE 1994, p. 16). In the words of Meles’ main economic advisor and an architect of ADLI,

the commitment was to provide the maximum outreach to all smallholders. All smallholders with the right inputs can increase their productivity.⁵

ADLI therefore sought to enmesh the entire peasantry through the distribution of land and improved inputs. In doing so, ADLI also explicitly sought to limit urban migration since this would otherwise create ‘massive social problems’ and ‘be detrimental to peace and stability’ (MoFED 2003, p. 27, Office of the Prime Minister 1993, TGE 1994). According to a former State Minister of Agriculture in the early 2000s, ADLI was ‘a policy to ensure that people stay where they are, at least until there are jobs for people. The fear of migration was for this’.⁶ State land ownership tied people to the land, by preventing sales and constraining rental markets, while agricultural inputs would improve agricultural livelihoods. In the words of former deputy Prime Minister, Addisu Legesse, ‘If there is progress [in rural areas], then people won’t migrate, that was the strategy’.⁷ As such, although the EPRDF retained state land ownership from the Derg, the government’s approach to the agrarian question sharply differed. Rather than agricultural collectivisation as a means of achieving scale economies, the EPRDF sought to provide land access and improved inputs to all peasant farmers, regardless of holding size, based on the assumption that all farmers could raise productivity.

ADLI’s focus on agriculture, however, was not intended ‘as an end in itself but a means of accelerating industrial and commercial development which would then siphon off manpower from the rural areas’ (Meles, cited in BBC News 2005). The increase in agricultural production was expected in the medium to long term to support industrialisation by providing a source of foreign exchange – particularly through export of coffee and chat⁸ – to enable capital imports, cheap food to reduce industrial labour costs and inputs for agro-processing (MoFED 2003). Meanwhile, the growing wealth of smallholders would create a market for domestic industry producing farm tools and agricultural inputs such as fertiliser. Building on the emphasis on self-reliance throughout the civil war, and in the context of limited domestic and external revenues, Newai Gebre-Ab argued that ADLI,

was considered a feasible strategy because we would not need to depend on foreign assistance. ADLI would lead to an autonomous growth process.⁹

ADLI prioritised low technology industrial production that maximised employment creation by focusing on agro-processing and industry oriented to the domestic market (MoPED 1993, p. 21)(discussed in detail in Chapter 6). The long-term objective, therefore, was that ‘agriculture should be the starting point for initiating the structural transformation of the economy’ over a period of 20 years (MoPED 1993, p. 5). In that way, restrictions on urban migration could gradually be removed, enabling urbanisation in parallel with industrialisation, with expanding industrial employment serving as a distributive resource with which to maintain political order.

The government, however, made limited progress implementing ADLI in its first decade in office. To a considerable degree, these failings can be attributed to the challenging situation the EPRDF faced, with a lack of financial resources, the challenge of post-conflict reconstruction and the Eritrean War from 1998. These challenges were exacerbated, however, by divisions within the ruling party and between the political leadership, the bureaucracy and foreign donors, as discussed in Chapter 4.

The government did begin its efforts to expand the supply and utilisation of improved agricultural inputs. From 1993, the non-governmental organisation Sasakawa-Global 2000 established demonstration plots to highlight the potential of agricultural extension, credit and improved inputs. Meles’ visit to the pilot led to this model being used as the basis for the government’s National Agricultural Extension Intervention Program (NAEIP) launched in 1995 (Keeley and Scoones 2000, Howard et al. 2003). Nonetheless NAEIP was impeded by several factors. First, with just 2,500 development agents to cover the entire country in 1995 (Spielman et al. 2011, p. 26, Berhanu and Poulton 2014, p. s198), the state’s territorial reach in rural areas remained limited during the 1990s. Likewise, the EPRDF initially avoided the use of service cooperatives to provide credit, marketing and input supply until the late 1990s, as a result of the negative associations with Derg-era producer cooperatives (Kodama 2007).

Second, seed and fertiliser supply was a recurrent problem. No domestic fertiliser production was established in this period. The import and distribution of seed and fertiliser were state monopolies under the Derg, but were liberalised under structural adjustment reforms in the early 1990s. Nonetheless, EPRDF-party-owned endowment companies were given the fertiliser supply contracts for NAEIP, effectively monopolising distribution (Jayne et al. 2003).¹⁰ The government’s view, with some justification, was that the private sector would have insufficient market incentives to undertake the broad distribution to all peasant farmers required by ADLI and that party-state intervention was necessary (Spielman et al. 2011). Yet, this further routed the distribution of key resources through party-state structures, while numerous studies highlighted the inefficiency of these party-state affiliates (Jayne et al. 2002, 2003, Spielman et al. 2010, 2011).

Finally, agricultural production was impeded by limited market access. ADLI intentionally slowed urban development and thereby demand for agricultural products. Moreover, the poor state of transport infrastructure meant that agricultural markets were segmented. Overall, the result was that use of improved inputs and agricultural productivity remained stagnant, with modest, if any, increase during the 1990s (see Figure 5.12). Meanwhile, the main outcome of government efforts to promote large-scale commercial agriculture were investments in sesame production in Humera, western Tigray (Ali et al. 2017).

With stagnant agricultural production and insignificant expansion of industrial employment (see Chapter 6), the rapidly growing rural population translated into growing numbers of landless or near landless households with tiny plots insufficient for household reproduction (Rahmato 1994, Habtu 1997). Land shortages therefore presented an economic and political dilemma, to which the government’s response was ambiguous. Derg and TPLF land reforms built on the central principle of previous tenure regimes, namely that farmers have a right to a share of communal land, rather than exclusive rights to specific plots. Indeed, the Constitution states that land is ‘a common property of the Nations, Nationalities and Peoples of Ethiopia’, while peasants have the ‘right to obtain land without payment’ (FDRE 1994, para. 40), suggesting a continuation of this principle and implying that regular redistribution was required. Yet, Amhara was the only region to redistribute land, and an early statement that explicitly allowed land inheritance (TGE 1991, p. 21) suggested that individuals had rights to specific plots, not just a share of common property. Moreover, the constitution was ambiguous as to who was responsible for fulfilling this right to land: the federal government for Ethiopian citizens; ethno-regional administrations for members of ethnic groups; or kebele administrations for members of particular communities.

New Answers to the Agrarian Question under the ‘Developmental State’

The Armageddons of the early 2000s and evolving ideas within the party leadership challenged ADLI’s assumptions, leading to a change in strategy. This was not a one-off reform, but a gradual accumulation of new policy priorities that resulted in a quite distinct approach to the agrarian question, the role of land tenure and mass distribution. As shown in Chapter 4, the 2001 TPLF split concentrated power in Meles Zenawi and led to the explicit acceptance of capitalist development. Protests in Addis Ababa in 2001 and 2005, meanwhile, highlighted the political dangers of urban neglect that came with agricultural-led development. The most direct challenge to ADLI, however, was the 2002/03 food crisis.¹¹ For the EPRDF leadership, ADLI was framed as a race against population growth to raise agricultural productivity and stimulate industrial employment creation to enable industry to absorb the growing labour surplus and thereby maintain mass acquiescence and political order. The consequence of losing this race would be that,

In the nightmare scenario case, what we would have is agricultural productivity and non-farming employment would not increase significantly. New mouths to feed will obviously crop up in the rural areas in big numbers. They will have no access to non-farming employment because it is not being created in enough numbers.

(Meles, cited in The Reporter 2000)

In 2002/03, a severe El Niño disrupted rainfall across Ethiopia with the result that 14 million people required emergency assistance. Emergency appeals had continued every year since the 1984/85 famine, much to the government’s frustration. However, the 2002/03 food crisis was on a far larger scale (see Figure 5.1), and raised the possibility that Ethiopia had already reached Meles’ ‘nightmare scenario’, with modest productivity growth outpaced by rapid population growth, resulting in widespread land shortages, unemployment and hunger.

Figure 5.1 Estimated humanitarian requirements

Source: author, based on Joint Government and Humanitarian Partners’ Documents, various years.

The government’s initial response remained within the spirit of ADLI, launching a new Food Security Strategy that comprised an ‘access to land’ or voluntary resettlement programme and the Productive Safety Net Programme (PSNP). The resettlement programme that ran during the mid-2000s relocated those with insufficient land in densely populated areas of Amhara, Oromiya, SNNPR and Tigray to more sparsely populated areas within the same regions, particularly in the western lowlands (MoFED 2003). At best, such an extensification strategy could be a temporary solution, given the limits to land availability and population growth rates. Moreover, Hammond (2008) highlighted major implementation challenges. The PSNP, meanwhile, was launched in 2005 to provide regular cash and food transfers to some eight million chronically food insecure people to limit the effects of food shortages and promote graduation from assistance, primarily through the same approach as ADLI – improved smallholder productivity (see Chapter 8).

Subsequent five-year plans that outlined the ‘developmental state’ model, beginning with the 2005 Plan for Accelerated and Sustained Development to End Poverty (PASDEP), and consolidated in the 2010 and 2015 Growth and Transformation Programmes, marked a more definitive break from ADLI.¹² Like ADLI, the ‘developmental state’ sought to raise smallholder productivity and promote linkages between agriculture and industry. However, the ‘developmental state’ had quite distinct distributive implications. First, agricultural policy shifted from supporting all peasants to raising the productivity of high potential smallholders (EPRDF 2006, MoARD 2010a). Second, the government promoted domestic and foreign private agricultural investments far beyond the limited role envisaged in ADLI in an effort to accelerate agricultural growth and exports. Third, as discussed in Chapter 6, the focus of industrial policy shifted from the extraction of an agrarian surplus and the creation of a domestic capitalist class to the promotion of foreign direct investment to accelerate industrial development.

This evolving strategy changed the roles that state land ownership was expected to play. The government maintained its focus on controlling the peasantry and the distributive role of land tenure in doing so. However, population growth and the absence of redistribution meant that, over time, land access provided protection for, and thereby state control over, a smaller and smaller proportion of the population. Land access could no longer be the sole basis of mass distribution or a barrier to large-scale urban migration. Instead, as argued in Chapter 6, industrial policy was to take on the burden of mass distribution, with mass industrial employment intended to absorb surplus labour. In place of the multiple roles played by state ownership under ADLI – protecting and controlling the peasantry, and limiting class differentiation and urban migration – under the ‘developmental state’ the primary function of state land ownership was instead to centralise state control over rent allocation, enabling the state to direct productive activity in line with the national development strategy. In Meles’ words,

The decision we took in terms of the land ownership system [maintaining state ownership] is therefore one of the critical decisions that have made it possible for us to try and implement an alternative to the neo-liberal paradigm … Our strategy has thus been based on putting in place a land ownership system that curtails rent-seeking behaviour and promotes value creation.

(Zenawi 2006a)

Under the ‘developmental state’, state land allocation prioritised two forms of production, which are considered in turn in the following sections: commercial smallholders and agricultural investors.

Commercialisation of Smallholder Agriculture

From the mid-2000s, the government began to recognise that, contrary to ADLI, not all farmers could sustain investment in improved inputs and produce for the market, and that less productive farmers would need to move into off-farm activities, enabling more effective farmers to expand production. As one of the main architects of Ethiopia’s development strategies for 30 years acknowledged,

Out of 14 million farmers not all of them will produce a surplus. They are bound to move out, so we need to provide employment.¹³

By 2010, the Ministry of Agriculture’s ten-year strategy acknowledged the need for differentiation, focusing efforts on raising the productivity of those with potential, while the remainder would need to seek employment,

The primary beneficiaries of production and productivity enhancement will be smallholders adopting improved agricultural practices that increase food production and cash income generation. These will tend to be located in the higher potential areas where the prospects for improving productivity are best … Unemployed and under-employed people will enjoy improved income earning opportunities from employment in rural farm and non-farm enterprises.

(MoARD 2010a, pp. 28–29)

From PASDEP onwards there was a much more explicit focus on smallholder commercialisation, which entailed the concentration of agricultural spending on high potential areas and high potential farmers (MoARD 2010a). This push for increased smallholder productivity has entailed important changes in land tenure, and a major push to expand agricultural extension services and market access, as discussed in turn below.

The Evolution of Land Tenure and Generational Divisions in Land Access

While the Constitution guarantees the right to land, access to land had gradually diverged from this principle. The last land redistributions to accommodate new households were undertaken towards the end of the 1980s in most regions other than Amhara. Yet, as Figure 5.2 shows the rural population grew rapidly, even as the rural population declined as a proportion of the total, resulting in the declining availability of agricultural land. Figure 5.3 gives an indication of the impact of population growth, by calculating the average area of cropland per economically active person in agriculture, which approximately halved since the 1980s. Cultivated area expanded significantly in the 2000s due to resettlement, and the allocation of communal grazing and forest land in an attempt to accommodate young adults.¹⁴ However, this supply was exhausted from around 2010, with the result that land availability began to decline once more. These average figures for land availability translate into very small landholdings for most farmers. Figure 5.4 presents data on landholdings and cultivated area from a large-scale household survey conducted in 2015/16. The median household cultivated just 0.6 hectares. The data include land for housing, which likely explains why many households with very small landholdings do not cultivate any land.

Figure 5.2 Rural and urban population

Source: author, based on World Development Indicators.

Figure 5.3 Crop land per economically active person in agriculture

Source: author, based on FAOStat.

Figure 5.4 Households by landholdings

Notes: The sample for this graph is limited to rural households with some involvement in agricultural activities such as crop or livestock production.

Source: author, based on World Bank/CSA Socioeconomic Survey, Wave 3, 2015–2016.

In addition to land shortages and the early 2000s political crises, the government faced renewed pressure from opposition parties and foreign donors to revisit the land question. The Coalition for Unity and Democracy (CUD), which secured electoral gains in 2005, vocally advocated for privatisation based on the argument that state ownership undermines tenure security, investment incentives and agricultural productivity.¹⁵ Equally, foreign donors, and USAID in particular, promoted privatisation, sending Hernando de Soto, a well-known advocate of land titling, to meet Meles and the Council of Ministers.¹⁶ While the government resisted privatisation, it did undertake reforms that individualised tenure, with important distributive consequences.

The first major step towards the individualisation of land tenure – whereby farmers have rights to particular plots rather than shares of communal land – were land proclamations in Tigray in 1997 (TNRG 1997, para. 6(1)) and Amhara in 2000 (ANRS 2000). These proclamations effectively prevented future land redistributions and required land certification. Based on the Amhara experience, the Prime Minister instructed Oromiya and SNNPR to follow suit, while subsequent federal and regional proclamations from 2005 established a minimum landholding size of 0.5 hectares to prevent sub-division into economically unviable plots (FDRE 2005a, ANRS 2006, ONRG 2007).¹⁷ Land registration and new proclamations therefore marked a definitive break from past tenure regimes, giving individual landholders indefinite rights to particular plots, as well as the right of inheritance. For proponents, this move strengthens tenure security and thereby investment incentives and rental markets (Deininger and Binswanger 1999, Deininger et al. 2011).

Whatever the impact on productivity, however, in the context of rapid population growth, the inevitable result was the inability to fulfil the constitutional right to land and the gradual erosion of the government’s strategy of coercive distribution, which was based on mass land access. Indeed, land access became sharply divided along generational lines. Figure 5.5 shows the distribution of land certificates by age, while also disaggregating how those with land certificates acquired their holdings.¹⁸ The graph shows that the vast majority of those with land certificates were at least 43 in 2015 and therefore would have been adults at the time of the last land redistribution in the final years of the Derg. Indeed, for these older landholders most land was ‘granted by local leaders’ – that is, state distribution. In contrast, landholders who reached adulthood after the last redistribution are in a minority and state allocation plays a minor role for this group. By far the main source of land for younger adults is inheritance, with parents’ landholdings often divided up between multiple siblings resulting in particularly small landholdings (Bezu and Holden 2014, Adugna 2018).¹⁹ Meanwhile, gender dynamics continue to be influential, with parents often prioritising sons over daughters in inheritance (Bezu and Holden 2014).

Figure 5.5 Method of acquisition for plots with land certificates by age

Notes: The vertical line at age 43 estimates the minimum age of those eligible to receive land at the time of the last land redistribution. The survey was conducted in 2015. So, in most of the country those aged 43 in 2015 would have been 18 in 1990 and potentially eligible to receive land. In Amhara, anyone aged 36 or over in 2015 would have been 18 in 1997 and potentially eligible. The dashed line representing the proportion of land that was state allocated is a five-year rolling average.

Source: author, based on World Bank/CSA Socioeconomic Survey, Wave 3, 2015–2016.

Given that most land certificates were issued in the mid-2000s, Figure 5.5 could be misleading since younger adults that acquired land subsequently would not have a certificate. To address this, Figure 5.6 shows the age of the primary decision-maker regarding each plot of land, regardless of whether or not they have a certificate (abbreviated to ‘landholder’ in the graph), compared to the rest of the rural population (marked ‘landless’).²⁰ The graph shows that the proportion of those who are primary decision-makers on a particular plot declines sharply for those aged under 43, particularly for young adults in their twenties. Moreover, given population growth and land shortages, there is no realistic possibility of providing land to all of the current young adult population or the large numbers of children that will soon enter adulthood.

Figure 5.6 Distribution of primary decision-makers on land and rural population by age

Notes: As in the previous graph, the vertical line at age 43 is an estimate of the minimum age of those eligible to receive land in the last redistribution. The exact age differs in Amhara and depending on the exact timing of the last redistribution in each locality. The dashed line representing landholders as a proportion of total population is a five-year rolling average.

Source: author, based on World Bank Socioeconomic Survey, Wave 3, 2015–2016.

While young adults have little land of their own, there remains the possibility of accessing land and thereby an agricultural livelihood through rental markets. Landholders who lack farm labour and oxen for ploughing often sharecrop or rent out their land, potentially providing an opportunity for those without land of their own.²¹ Survey data suggest that rental markets have a limited role in land access, however. Figure 5.4 already compared households’ own landholdings with cultivated land. The very small differences between the two lines suggest that rental markets have a limited effect on land access. Indeed, this conclusion is supported by research that shows that the tenure system and legal restrictions on rental present a major barrier to rental markets (Gebru et al. 2019). In addition, Figure 5.7 shows that it is larger landholders – rather than those with small or no initial holdings – that most commonly rent in land to expand their production, while those with very small landholdings are slightly more likely to rent out their land. These data therefore suggest that rental markets lead to some land consolidation, rather than providing a mechanism by which the landless can access an agricultural livelihood.

Figure 5.7 Prevalence of land rental by initial landholding size

Source: author, based on World Bank Socioeconomic Survey, Wave 3, 2015–2016.

In sum, the land tenure regime evolved from the early 2000s to provide greater individual tenure security with a view to promoting agricultural productivity growth, but at the cost of excluding a generation or more of young adults from access to state distributed land. In effect, the government shifted its distributive strategy, accepting that land access would provide a livelihood for and thereby political control over a diminishing, albeit significant, proportion of the population, with industrial employment (Chapter 6) and social protection (Chapter 8) required to maintain mass enmeshment.

Renewed Efforts to Raise Agricultural Productivity

Alongside tenure reforms, the government also greatly expanded agricultural extension services from the early 2000s, resulting in significant increases in agricultural productivity. While the successes of these efforts have been acknowledged in recent research, existing studies have neglected two vitally important points. First, the agricultural extension system relied on party-state structures that reinforced the enmeshment of those with land access. Second, generational inequality in land access meant that the push to raise agricultural productivity had important distributive impacts, with many young adults excluded from the improvements experienced by landholders.

The renewed focus on agricultural productivity is evident in government agricultural expenditure, which rose rapidly in absolute terms and as a proportion of government spending from 2003 (see Figure 5.8). Importantly, this agricultural push was a government initiative, rather than a donor priority. Donor support for agriculture was limited during the 2000s, and only increased at the end of the decade as donors aligned with government plans. Major programmes such as the World Bank and USAID-funded Agricultural Growth Programme and the Gates Foundation-funded Agricultural Transformation Agency (ATA), both of which began in 2011, have played important roles, but built on longstanding government priorities and worked with the party-state dominated agricultural extension system. Both initiatives focused on improving market access and raising productivity through improved technology and farming techniques, particularly focusing on high potential wereda.

Figure 5.8 Government and donor spending on agriculture

Source: author, based on ReSAKSS (Regional Strategic Analysts and Knowledge Support System). 2020. www.resakss.org.

The main focus of state expenditure was the agricultural extension system and rural infrastructure (Rohne Till 2021). These investments extended the territorial reach of the party-state from the early 2000s, with this infrastructural power used to distribute agricultural inputs in line with the strategy of coercive distribution. The government trained 52,000 development agents by 2010, while constructing 9,200 Farmer Training Centres (MoFED 2010, p. 9). Within each kebele, 3–4 development agents – specialising in crops, livestock and natural resource management – were tasked with demonstrating improved farming techniques in these centres and linking farmers to the party and state-affiliated agencies responsible for agricultural input supply. The result is that Ethiopia now has 21 agricultural extension workers per 10,000 farmers, one of the highest ratios in the world and greater than China (16) and Indonesia (6) (Bachewe et al. 2018). Indeed, Figure 5.9 shows a sharp increase in the proportion of farmers receiving advice from development agents after 2003, with extension services attributed an important role in increased uptake of improved inputs (Bachewe et al. 2018).²²

Figure 5.9 Access to extension services

Source: author, based on CSA Agricultural Sample Survey, Volume III, various years.

Development agents worked particularly with the model farmers who were assigned as the leaders of male development teams and 1-to-5 networks, building on the existing gendered division of labour. Model farmers had priority access to agricultural inputs and were expected to be the first to adopt new technology and demonstrate its value to their team (MoFED 2003, pp. 64–67). Moreover, model farmers and development team leaders were enrolled en masse as party members, strengthening ties to the party-state (see Chapter 9 and Lefort 2010, 2012). Annual and five-yearly adoption and distribution targets for improved farming techniques and improved inputs were derived from national development plans for each level of state administration from federal government down to the kebele development agents and, in some instances, even development team leaders and 1-to-5 members.²³ As discussed in Chapter 9, individual farmers’ access to fertiliser, improved seeds, credit and most other state services was routed through this development team structure and mediated by the development agents, with the result that landholders were almost entirely dependent on party-state structures for their livelihoods.

After 2001, the government also massively expanded the network of agricultural cooperatives, which were tasked with a central role in distributing fertiliser – replacing the party conglomerates – and improved seeds produced by the state-owned Ethiopian Seed Enterprise (Spielman et al. 2011). By 2013 there were 13,029 agricultural cooperatives with 4.3 million members (Mojo et al. 2017, p. 58) and by 2012, 45 per cent of kebele had at least one agricultural cooperative (Bernard et al. 2013, p. 2). As with many organisations in Ethiopia, cooperatives are formally independent of the party-state, yet in practice remained under heavy state influence. Most cooperatives were established with state support and were organised in line with the state administrative hierarchy, with primary cooperatives operating at the kebele level, and cooperative unions at the wereda.

A major focus of government expenditure has also been a massive road-building programme that tripled the total length of all-weather roads by 2013, particularly focused on improving rural access (see Figure 1.1). Past studies have shown that road access reduced transport costs and improved farmers’ production incentives (Dercon, Gilligan, et al. 2009, Minten, Koro, et al. 2013, Minten, Tamru, et al. 2013). Furthermore, the massive expansion of primary education and adult literacy programmes has also been attributed an important role in supporting uptake of improved inputs and farming techniques (Bachewe et al. 2018).

The result of these initiatives was a marked increase in the use of improved agricultural inputs (see Figure 5.10). Most clearly, the proportion of land cultivated using chemical fertiliser and pesticide more than doubled after 2010. The amount of fertiliser applied per hectare also increased considerably, though at just 1.7 quintals per hectare in 2016/17, this remained well below recommended levels. Increased fertiliser use was complemented by an important ATA initiative to map soil types for the first time in order to tailor fertiliser types to local conditions (FAO 2020), a marked shift from the previous approach of issuing fertiliser quotas regardless of soil type.²⁴

Figure 5.10 Utilisation of improved agricultural inputs

Source: author, based on CSA Agricultural Sample Survey, Volume III, various years.

Research and investment by the Ethiopian Institute for Agricultural Research and international support since 2000 resulted in a proliferation of new varieties of major crops, in particular wheat, teff and maize (Moller 2015, Mellor and Malik 2017). Official data suggest that improved seed utilisation remained very low, despite more than doubling in the last ten years. Nonetheless, researchers are convinced that official figures represent a major underestimate, since these measure the purchase of seeds (Spielman et al. 2011, Bachewe et al. 2018). Improved maize seeds are hybrid varieties that must be purchased annually to maintain yield. In contrast, many improved varieties for open pollinated crops, such as teff and wheat, can be re-used for several years without significant productivity loss. Spielman et al. (2011, p. 10) argue that the majority of wheat and teff was actually farmed with improved varieties and Bachewe et al. (2018, p. 291) cite a finding that 96 per cent of wheat in Oromiya was actually improved varieties.

In contrast to the increased uptake of fertiliser and improved seeds, there was little expansion of irrigation in smallholder agriculture. The main challenge is that most highland rivers are seasonal and situated in deep valleys that make it difficult to use for irrigation. State campaigns have promoted water harvesting, though this has achieved mixed results (Segers et al. 2011). In contrast, the main irrigation potential is in lowland areas where major rivers such as the Omo and Awash descend from the highlands. Many state-promoted agricultural investments have sought to expand large-scale irrigation in these areas, often linked to multipurpose dam projects, as discussed in the following section.

The provision of credit is also important, enabling farmers to buy inputs at planting time and repay after harvest. Once again the microfinance sector expanded considerably and was dominated by party-affiliated organisations, which accounted for 89 per cent of loans in 2014/15 (NBE 2015, p. 54). Dedebit Credit and Savings Institution (DECSI) in Tigray was the largest MFI and a joint venture of the TPLF and REST (Milkias 2003), as well as being named after the village where the TPLF was founded. Likewise the Amhara Credit and Savings Institution (ACSI) was founded by the ANDM-linked Organisation for the Rehabilitation and Development of Amhara (ORDA) in the 1990s, with ANDM and ORDA retaining a majority share (Geleta 2016).²⁵ Similar relations exist between the OPDO and the Oromiya Credit and Savings Institution (OCSI) in Oromiya, and SEPDM and Omo MFI in SNNPR. These microfinance organisations have focused primarily on providing group loans based on the Grameen Bank model.²⁶ As shown in Figure 5.9, however, loan uptake remains low and has actually declined since a peak in 2008. The limited use of credit services has primarily been attributed to farmers’ concerns about the risk of rain-fed agriculture and the fear of being unable to repay (Berhane and Abay 2019).

Official figures indicate that the overall result of these efforts has been impressive agricultural growth, averaging 7.6 per cent per year in 2004–2014 (Bachewe et al. 2018, p. 286), exemplified by the sharp rise in cereal production since the early 2000s (Figure 5.11).²⁷ Rising production was partly due to an expansion of cultivated area thanks to resettlement and as local administrations sought to accommodate growing numbers of young adults. However, the stability of the total cultivated area since 2010 suggests there is little space for further expansion (Mellor and Malik 2017). Increased production was supported by a moderate increase in yields up to about 2008, consistent with improved farming techniques resulting from the expansion of the agricultural extension system and improved market access (Mellor and Malik 2017, Bachewe et al. 2018). Since 2008, however, more rapid yield growth drove total production thanks to increased utilisation of improved inputs and growth in demand from expanding urban centres (see Chapter 7) (Minten, Tamru, et al. 2013, Bachewe et al. 2018, Vandercasteelen et al. 2018). This general pattern of initially modest and then more rapid yield growth is replicated across all major cereal crops (see Figure 5.12) and all of Ethiopia’s regions.²⁸ However, the most pronounced increases and highest yields have been in the western highlands of Oromiya and Amhara, where rainfall and soil quality are most favourable.

Figure 5.11 Cereal production

Source: author, based on CSA Agricultural Sample Survey, Volume I, various years.

Figure 5.12 Yield for major cereal crops

Note: no data are available for 2003. Drought likely had a major detrimental effect on yields.

Source: author, based on CSA Agricultural Sample Survey, Volume I, various years.

Consequently, several assessments of the agricultural sector have proclaimed Ethiopia’s success in achieving a ‘Green Revolution’ (Moller 2015, Mellor and Malik 2017, Bachewe et al. 2018, Dercon and Gollin 2019, Taffesse 2019, Rohne Till 2021). Despite growing evidence of productivity growth, however, the distributive implications of these changes have received very limited attention. While improved productivity does appear to have translated into improved livelihoods for those farmers with access to sufficient land and the resources to invest in improved inputs, this has come at the cost of differentiation within the rural economy. Most significant is the generational division between those who received state allocated land under the Derg and those that reached adulthood since. Most of the latter have limited prospects of an independent agricultural livelihood and must either wait to inherit small landholdings from their parents or seek work outside agriculture.

Moreover, the concentration of improved inputs and extension services on farmers with greater potential has also led to a degree of differentiation between those farmers with modest, but adequate landholdings and a group of farmers with very small landholdings, who are struggling to reproduce themselves through agriculture. Figure 5.13 shows that the adoption of improved inputs is closely related to landholding size, with larger landholders more likely to participate in the extension programme, receive advisory services and credit, access seeds through cooperative membership, and to utilise chemical fertiliser and improved seeds. In contrast, Figure 5.14 shows that self-reported food shortages and, to a degree, off-farm employment are more common among those with very small landholdings. Indeed, research consistently finds that ‘push’ factors – land shortages and low agricultural potential – drive people into off-farm activities, usually through self-employment in low productivity activities (World Bank 2015, Schmidt and Bekele Woldeyes 2019). Adults under 35, with more limited land access are significantly more likely to engage only in off-farm activities or to mix own farm and off-farm activities than older adults (Schmidt and Bekele Woldeyes 2019, p. 117), while those with the possibility of inheriting land from their parents are much less likely to migrate for employment (Bezu and Holden 2014).

Figure 5.13 Access to agricultural inputs by landholding size

Note: data on cooperatives is based on whether the farmer got improved seeds from a cooperative, since the survey did not ask about cooperative membership.

Source: author, based on World Bank Socioeconomic Survey, Wave 3, 2015–2016.

Figure 5.14 Outcomes from agriculture

Source: author, based on World Bank Socioeconomic Survey, Wave 3, 2015–2016.

The extent of this differentiation should not be over-stated. The fact that about one-fifth of farmers with at least one hectare of land still report food shortages shows that many larger landholders also struggle to make ends meet. Moreover, the low proportion of produce sold by larger landholders suggests that increased production may be used to supplement household consumption rather than a shift towards commercialisation. Nonetheless, the data clearly indicate that recent agricultural growth has been unequal, with growing numbers excluded from agricultural livelihoods due to their lack of or extremely limited land access. While the ‘developmental state’ delivered improved livelihoods for many with sufficient land, shortages mean that many young adults have escaped the government’s strategy of coercive distribution.

State-Directed Agricultural Investment

The ‘developmental state’ entailed not just smallholder commercialisation, but also state promotion of agricultural investments, where these projects promised increased productivity and market access. Agricultural investment is legally distinct from, and sometimes in competition for land with, the smallholder sector. Though population growth presents the main challenge to the government’s strategy of using land access as a distributive resource to secure mass compliance, the government’s promotion of agricultural investments and other priorities of the ‘developmental state’ at times also necessitated the displacement of smallholders and other land users, further undermining land access.

The rapid expansion of agricultural investments from the mid-2000s was the result of both domestic and global processes. ADLI had always envisaged a spatially differentiated approach whereby the inverse relationship favoured peasant production in the highlands but large-scale, capital-intensive agricultural investments were preferred in sparsely populated lowlands. As acknowledged by Mekonnen Manyazewal, former Minister of Trade and National Planning Commissioner, increased agricultural investment was a response to ADLI’s failings and the need to boost exports to finance import of capital goods for industrialisation and food to address food insecurity (Manyazewal and Shiferaw 2019). The government particularly targeted foreign investors with the resources required to establish large-scale production in the lowlands. However, from the mid-2000s the government also strongly supported foreign and domestic investments in densely populated highlands where investors promised increased productivity, market access and foreign exchange earnings compared to existing smallholders. This shift in government policy, meanwhile, was enabled by important global dynamics in the form of the intertwined crises of food, fuel, finance and environment from around 2008 that resulted in increased demand from private and public investors from advanced and emerging economies, investing in production of food and biofuel crops in Africa (Borras et al. 2010, Zoomers 2010).

State control of land and the financial sector enabled the government to promote investments in forms and locations that met its developmental objectives. The main priorities were export crops, and local processing of agricultural products, especially cotton for the textile industry, while minimising displacement and maximising employment creation.²⁹ To this end, the state Development Bank of Ethiopia (DBE) offered subsidised loans of up to 70 per cent to investors engaged in exports and agro-processing.³⁰ Moreover, the state provided tax exemptions and incentives designed to encourage exports (FDRE 2003, para. 4).³¹ Meanwhile, state land ownership enabled the government to expropriate existing users and thereby direct investors through state land allocation. Indeed, internal party documents clearly outline this function of state land ownership,

Constructive government provides land for development. In such conditions, it provides land for free or lower price to the investors. Although such individual is privileged with lower price from what has been on the market, and such kind of property offer cannot escape from being rent. Nevertheless, the land is offered for development purpose.

(EPRDF 2006)

Agricultural investment was governed under separate investment proclamations rather than land laws.³² All investors applied for a licence from the federal Investment Commission, requiring a project proposal outlining the plans for employment creation and capital investment (FDRE 2002a). Only once granted a licence were investors directed to the relevant land administration to secure land. Meanwhile, land lease fees were set with the intention of directing investors to sparsely populated regions – particularly Benishangul-Gumuz, Gambella and the south of SNNPR – as well as remote areas of the other regions (TNRG 2000 [Ethiopian Calendar], SNNPR Investment Agency 2008). Investors were required to submit progress reports every six months while projects were being established and investment contracts had clauses specifying that the state retained the right to repossess land if investors failed to use it for the designated purpose. In principle, therefore, the land lease system provided the state with considerable leverage over investors and the ability to direct their activities to government priorities.

Moreover, state land ownership gave the government sweeping powers to expropriate smallholders’ land ‘where it believes that it should be used for a better development project’ (FDRE 2005a, para. 3(1)). Displaced smallholders were entitled to compensation of ten times their average annual income over the previous five years (FDRE 2005b, para. 8(1)). Yet, this minimal compensation was widely seen as inadequate amidst high rates of inflation, the inability to buy replacement land and the higher cost of living in urban centres. Meanwhile, landholders were well aware that their land was subsequently leased by the government at much higher rates (Abate 2019).³³

The federal government’s promotion of agricultural investment also raised important issues regarding the federal division of powers. According to the constitution, the federal government has the authority to issue land laws, while regional governments administer land. The federal government’s investment promotion encroached on regional responsibilities, however.³⁴ First, the 2002 investment proclamation required regional land authorities to provide land to investors approved at federal level (FDRE 2002a, para. 35), essentially removing regional discretion. Furthermore, in 2009 the federal government created the Agricultural Investment Support Directorate (AISD) in the federal Ministry of Agriculture and Rural Development (MoARD) to allocate land to all foreign investors and domestic investments of more than 5,000 hectares.³⁵ The AISD instructed regions to identify suitable investment land and to pass this information to the AISD, which would administer this ‘land bank’.³⁶ The weaker regions – Benishangul-Gumuz, Gambella and SNNPR – acceded to the request and handed control to the federal government, which oversaw 3.7 million hectares of supposedly ‘unused’ land in 2009.³⁷ However, the stronger regions of Amhara, Oromiya and Tigray contested what they saw as federal encroachment.³⁸ In an early sign of regional resistance to the federal ‘developmental state’ agenda, a senior manager of the Oromiya Investment Commission highlighted this tension,

The constitution does not allow this to happen. According to the constitution, land is administered by the regions, so to make the changes they [the federal government] need to change the constitution.³⁹

A representative of the Amhara Investment Promotion Agency raised similar concerns,

The idea comes from the good intention to promote development but it makes regional departments unhappy. It does not make for good relations.⁴⁰

The prioritisation of large-scale agricultural investments in sparsely populated regions was consistent with the original ADLI (TGE 1994) and, indeed, the long-term approach to agricultural development and state-building in Ethiopia. Beginning in the 1950s, Haile Selassie’s government developed and integrated peripheral areas by leasing land to investors along the Awash River and in Humera, pushing aside pastoralists in the process (Zewde 2008a, Puddu 2016). Likewise under the ‘developmental state’, large expanses of Afar, Benishangul-Gumuz, Gambella and SNNP regions were framed as ‘unutilized’ (Meles Zenawi, cited in ProKerala News 2011), ‘empty’ or ‘uncultivated’ and, consequently, deemed suitable for investment.⁴¹ The invisibility of mobile livelihoods underpinned the government’s confidence that investors could be directed to these lowlands ‘since such land has not been settled by peasant farmers, the issue of displacing farmers does not arise’ (MoFED 2003, p. 25). Indeed, the displacement of pastoralists was not merely an unfortunate by-product of investment promotion, but a specific government objective. The federal pastoralist policy presented pastoralism as unsustainable and in need of transformation (MoFA 2008), the favoured option being to settle those practicing mobile livelihoods and transition to settled agricultural production (SOZPAAB 2012). The government explicitly acknowledged the links between resettlement and the promotion of investment, given,

the necessity of implementing appropriate settlement programs and creating enabling environment for private investors to set-up large-scale commercial farms and move to the next logical stage, that is investing in agro-based industries.

(MoFED 2003, p. 58)

The result was that a federal resettlement or villagisation policy was pursued in parallel to investment promotion, clearing people from areas targeted for investment (Kefale and Gebresenbet 2014, Gebresenbet 2016). Moreover, given that land registration had not been extended to these lowland regions and the government questioned the land use of those employing mobile livelihoods, in most cases no compensation was provided (Moreda and Spoor 2015, Gebresenbet 2016). While the government placed great political importance on securing control over the highland peasantry, its approach to those employing mobile livelihoods in the lowlands was ambivalent; disregarding their existing livelihoods, while consolidating state control through sedentarisation.

In the initial rush of land leases, the government particularly targeted lowland regions, such as Benishangul-Gumuz, Gambella and South Omo in SNNPR, with many projects linked to plans for irrigation and dam projects. The failings of many of these vast, headline-grabbing projects have been well documented and reflect both the speculative nature of many investments, as well as the lack of state capacity to effectively monitor and regulate projects. A 100,000-hectare lease to Indian firm Karuturi in Gambella was the largest lease to a private investor.⁴² However, the firm was apparently unaware that much of the land was located on a floodplain, with the result that the first maize crop was flooded and destroyed. In total, only a few thousand hectares were ever developed and the government cancelled the contract in 2015 (Chandran and Gardner 2017).⁴³ This was a common pattern with several investors leasing vast expanses that they were unable to clear. Another Indian investor, Shapoorji Pallonji leased 50,000 hectares in Benishangul-Gumuz to grow biofuel feed crops, but was unable to clear more than a few thousand hectares (Keeley et al. 2014). One study found that, on average 55 per cent of leased land remained undeveloped (Ali et al. 2017, p. 62).

It was not just foreign investors who massively overestimated their capacities, however. In the mid-2000s, the state sugar company began to expand existing plantations in Wenji-Shoa, Metehara and Fincha’a in Oromiya, as well as undertaking massive new developments. The plan involved ten new sugar factories and an additional 400,000 hectares of plantations in Afar (Kessem and Tendaho), South Omo (Kuraz), Tigray (Welkait) and Amhara (Beles). The centrepiece of the strategy would be the Kuraz project in the Omo Valley with five factories and a plantation of 175,000 hectares, with irrigation enabled by the regulation of flood waters by the Gilgel Gibe III dam completed in 2016 (Gebresenbet and Kamski 2019). Like the 26,000-hectare Tendaho project in Afar, the Omo projects aimed to create large numbers of jobs, as well as turning the local pastoralist and shifting cultivator populations into outgrowers producing sugarcane. As detailed in Chapter 6, however, the sugar expansion projects quickly ran into problems and failed to deliver.

Overall, the efforts to promote large-scale, capital-intensive agricultural investments in the lowlands have been a resounding failure, with foreign private and Ethiopian state enterprises grossly overestimating their capabilities and underestimating the challenges. Moreover, despite the leverage that state control of finance and land provided, the state lacked the capacity to evaluate and monitor investors, resulting in misguided faith in investors’ capabilities, particularly those of wealthy foreigners (Lavers 2012, Gebresenbet 2016). As one investment official noted in relation to a failed biofuel project in East Hararghe, Oromiya,

Nobody dreamed that it would be such a failure. We know how they worked, how they communicated with high officials, there were more than forty foreigners here.⁴⁴

Recognition of the failings of these investments led the government to announce a moratorium on new leases in 2011 and, from 2013, to limit new leases to 5,000 hectares, with further expansion conditional on performance (Keeley et al. 2014). The result was that demand from foreign investors dried up and there have been no large-scale leases since that time (Ali et al. 2017, Dejene and Cochrane 2021).

While the poor performance of large-scale leases in the lowlands has attracted most attention, there have also been many investments in highland areas. For the most part, these investments are smaller, yet constitute an important shift from ADLI’s assumption of the inverse relationship between farm size and productivity. Under the ‘developmental state’, where investors promised new technology, expertise and access to global value chains that would significantly raise productivity and contribute to export earnings, capitalist investors were preferred to peasants. Most investments took one of two forms – either horticultural projects that expropriate landholders, or outgrower schemes, both of which are relatively capital and labour-intensive.⁴⁵ In both cases, the aim is combine investors’ expertise, global market access and capital with Ethiopian labour and land, raising productivity but limiting displacement (MoFED 2003, p. 25).

The massive expansion of the state sugar industry provides one of the main examples of outgrower schemes through the extension of the longstanding Wenji-Shoa plantation. Although the sugar factory originally wanted to displace existing landholders and directly manage the plantation, the federal and regional governments brokered a deal to establish an outgrower scheme with landholders formed into producer cooperatives to avoid large-scale displacement.⁴⁶ Wenji-Shoa built on its long experience and has managed to establish sugar production in the new project area. However, the outgrowers have benefitted little from the project. Originally forced to join the scheme by state officials against their will, they are unable to withdraw or even precisely identify their original landholdings in the vast plantation.⁴⁷ Moreover, the sugar factory, with state support, imposed low prices for the sugarcane produced, with the result that the producer cooperatives made no profit on several of their harvests, once production costs were subtracted.

Horticultural and floricultural investments, in contrast, have been concentrated in Central Oromiya and the Rift Valley, within reach of Addis Ababa airport for export (Abate 2019, p. 101) (see Figure 5.15). Following initial private sector interest, the government promoted floriculture from 2004, with access to cheap land and labour used to attract investors (Oqubay 2015). Horticulture and floriculture projects usually require the displacement of existing landholders. However, they tend to be fairly labour, as well as capital, intensive. State land ownership therefore provides a tool by which the government can re-allocate land in line with the priorities of the ‘developmental state’, expanding foreign exchange earnings by expropriating existing landholders at minimal cost and turning them into floriculture workers (MoFED 2010, pp. 46–47). The result was such that by 2014/15 there were sixty-six firms involved, primarily foreign owned, producing cut flowers for export on 1,623 hectares (Melese 2017, p. 9).

Figure 5.15 Area under commercial farms by zone

Source: author calculations based on CSA 2014, Large and Medium Scale Commercial Farms Sample Survey.

An overall assessment of the impact of these investments on the agrarian structure in Ethiopia is hampered by the incompleteness of data. Table 5.1 presents several data sources. Columns 3–4 present the land identified by the government in 2009 as being suitable for agricultural investment, showing how the lowlands were particularly targeted, with vast swathes of Gambella and other regions identified for investment based on assumptions about existing land use patterns. Columns 5–6 present Dejene and Cochrane’s (2021) calculations based on the Land Matrix, which attempts to catalogue all investments of 200 hectares or more up to 2018. These data include foreign and domestic investments, including the state sugar estates that comprise a large proportion of land in SNNPR. However, as the authors admit, the threshold of 200 hectares means that many smaller projects, particularly domestic investments, are excluded. While these figures show that reality fell well short of the government’s ambitions, land under large-scale leases concentrated in lowland regions in line with government plans. Contrasting figures in columns 7–8 are taken from the CSA’s Large and Medium Scale Commercial Farm Survey 2017/18. These data likely offer the most complete coverage of domestic and small-scale commercial farms, but exclude all horticultural projects, a number of large-scale foreign leases and the state sugar estates (Ali et al. 2017). In these data, where small-scale domestic investors are better represented, a greater proportion of land is located in the highland regions of Amhara, Oromiya and Tigray.

Table 5.1 Regional distribution of investments up to 2018

Region	Size of region (ha)	Land identified for investment in 2009 (ha)	% of region	Land leased to foreign investors (ha)	% of region	Land cultivated by commercial farms (ha)	% of region
Afar	7,205,278	409,678	5.7	20,600	0.3	69,589	1.0
Amhara	15,470,896	347,430	2.2	54,622	0.4	193,013	1.2
Benishangul Gumuz	5,069,868	691,984	13.6	261,026	5.1	78,386	1.5
Gambella	2,978,282	1,238,005	41.6	156,012	5.2	49,328	1.7
Oromiya	28,453,784	438,212	1.5	305,212	1.1	271,177	1.0
SNNPR	10,547,600	529,181	5.0	630,448	6.0	139,205	1.3
Somali	27,925,200	–	–	2,000	0.0	398	0.0
Tigray	5,363,800	0	0.0	10,300	0.2	205,765	3.8
Unspecified	–	–	–	53,069
Total		3,654,491		1,493,289		1,006,862

Source: data on land identified by the government are from Lavers (2012, p. 120), data on foreign investments are from Dejene and Cochrane (2021, p. 229) using data from the Land Matrix and land cultivated by commercial farms from the CSA Large and Medium Scale Commercial Farms Sample Survey 2017/18.

Based on these somewhat divergent figures, some elements of a pattern emerge. First, most investments and land leased took place in the period between 2007 and 2012, coinciding with the rise in global commodity prices. Following the government’s limits on large-scale leases in 2011 and the fall in commodity prices, investment has reduced considerably. Second, estimates of the total land leased at 1.33–1.49 million hectares are much lower than the government’s original plans and also well below past claims.⁴⁸ Third, foreign investments tended to be considerably larger in terms of landholdings and capital investment than domestic investments, with many highly speculative foreign investments following government incentives to lowland regions. However, these projects have, for the most part, failed to meet their objectives, with many collapsing entirely. Meanwhile, smaller domestic investors, as well as some of the more successful foreign investors involved in horticulture, have focused on highland areas, often competing for land with the smallholder sector.

Based on CSA data for 2017–2018, large and medium-scale commercial farms accounted for roughly 7 per cent of cultivated area compared to 93 per cent for smallholders, although given the omission of several large farms from their sample, this may be a slight underestimate. Clearly the promotion of agricultural investments is in tension with the government’s political strategy of using land access as a means of mass distribution to secure political control. Indeed, the promotion of large-scale investments in lowland areas has systematically disregarded the land use of pastoralists and shifting cultivators, while small-scale investments in highland areas have frequently led to smallholder displacement. Understandably, this displacement was a source of great resentment for those affected. However, the impact of investment on landlessness in Ethiopia as a whole is an order of magnitude smaller than that of population growth that has left a generation or more of young adults with little hope of accessing land.⁴⁹

Conclusion

From 1975 onwards, relevant authorities – the Derg and the TPLF – used state land allocation as a strategy of ‘coercive distribution’, cultivating political dependence and mass acquiescence. Under the EPRDF, ADLI built on this approach to craft a national development strategy that would maintain coercive distribution over time and in the face of rapid population growth, first by improving agricultural livelihoods through the distribution of agricultural inputs, then by using the agrarian surplus to finance industrialisation and mass industrial employment creation.

The political crises of the early 2000s consolidated the longstanding view within the EPRDF that mass distribution was required to secure popular acquiescence and maintain political order, yet raised questions about the feasibility of ADLI and the utility of land as a distributive resource. Under the ‘developmental state’ from the mid-to-late 2000s, the government sought to accelerate economic development, even where this further eroded enmeshment, focusing attention on high productivity producers – both commercial smallholders and agricultural investors. As such, upkeep of the system of coercive distribution would also need to evolve, with the government’s distributive strategy shifting from guaranteeing land access to industrial policy aimed at creating and distributing access to jobs, discussed in Chapter 6.

This strategic shift necessitated a change in the central principle underpinning land tenure. From the rist system in the Imperial era to the land redistributions undertaken by the Derg and TPLF, the central guiding principle was that peasants had a right to a share of communal land, rather than rights to a particular plot. In the context of rapid rural population growth, the EPRDF was faced with a choice – either maintain this right to a share of community property by carrying out regular redistributions of ever smaller, economically unviable plots, or to exclude sections of the population from land access and enhance tenure security for existing landholders. The government chose the latter, with land tenure shifting from the early 2000s to favour economic productivity and individual rights through prohibitions on redistribution, the inheritance of land rights and land certification. The result has been a process of differentiation within rural areas, excluding more than a generation of young adults from state-distributed land, eroding the government’s mass distributive strategy and heightening the political importance of off-farm employment.

Ultimately, the EPRDF arrived at a quite different response to the agrarian question to that original envisaged in ADLI. In contrast to the homogeneous peasant farmers and domestic production linkages envisaged under ADLI, agricultural policy increasingly envisaged diversity of production and, as discussed in Chapter 6, industrialisation through foreign direct investment. Within agriculture, there was a growing acceptance of differentiation within the peasantry to provide opportunities for high potential farmers to expand and commercialise, while less effective farmers move out of agriculture. Alongside this, the government promoted capital-intensive investments, particularly large-scale mechanised production in lowland areas, and capital and labour-intensive production in highland areas in an attempt to accelerate agricultural growth. In doing so, the state promoted ‘capitalism from below’ – involving differentiation within the peasantry – as well as ‘capitalism from above’ through the selective promotion of agricultural investments in highland and lowland areas (Byres 1996).

The result of this change was both rapid growth in smallholder agricultural productivity, but also growing inequality and differentiation along generational lines. While progress has rightly attracted attention, what is often neglected is that this strategy restricted access to land and thereby an agricultural livelihood. Consequently, the ‘developmental state’ marked a significant change in the government’s distributive strategy, with land access providing a means of distribution for a still large, but progressively smaller and smaller section of the population. The perpetuation of a strategy of coercive distribution therefore necessitated the creation of new distributive resources beyond agriculture, with the government focusing on industrial development as the solution to this growing challenge.