Agricultural Development-Led Industrialisation was a strategy for broad-based growth that was expected to raise agricultural productivity, address food insecurity and drive the transition to an urban industrial economy. As such, this broad-based growth would obviate the need for specific programmes to address poverty and food insecurity. ADLI’s limitations in practice and the increasingly acute shortage of land and employment, however, constituted a growing distributive crisis during the 2000s. This chapter focuses specifically on the government’s use of social protection to respond to this emerging political threat.

Social protection has long been recognised as a means of reducing poverty, inequality and insecurity, but also as a fundamentally political tool which governments have used to maintain political order in the face of inequality and injustice (Seekings and Nattrass Reference Seekings and Nattrass2005, Ravallion Reference Ravallion2015, Saad-Filho Reference Saad-Filho2015). Increasingly, social protection in developing countries has come to be associated with the global spread of cash transfers that de-link support from prior financial contributions and tend to be targeted to the poorest in society. In Africa, donor-promoted cash transfer schemes have secured greatest government support where population pressure has undermined land access and the ability of government to rely on ‘agrarian welfare regimes’ (Seekings Reference Seekings2012), and where distributive crises present a political threat to incumbents (Lavers and Hickey Reference Lavers and Hickey2021).

The analysis that follows shows that the Ethiopian government did indeed respond to distributive crises with an expansion of social protection, drawing on significant donor support to do so. However, resource constraints and the government’s ideological resistance to state handouts meant that this response was highly ambiguous. A consistent theme of party policy since the civil war has been its attempt to utilise social spending as a productive investment in economic development, while also limiting coverage to as small a section of the population as possible to avoid damaging work incentives. The result was that the government saw social protection as little more than a sticking plaster, temporarily supporting the most vulnerable while buying time for implementation of the national development strategy. Consequently, social protection did little to address the growing distributive problems highlighted in previous chapters.

The chapter begins by analysing how distributive crises and EPRDF ideology have shaped social protection in urban and rural areas. The remainder of the chapter then examines the main social protection programme, the rural Productive Safety Net Programme and how it has been shaped by the government’s dual objectives: to expand provision in response to distributive crises, while limiting and ultimately eradicating the programme in line with ideological commitments and resource shortages. To do so, this analysis focuses on the geographical distribution of the programme, attempts to promote graduation from support and the relationship between the PSNP and the humanitarian system.

Distributive Crises and Productivist Ideology in Social Protection

The only formal social protection programme when the EPRDF assumed office in 1991 was the public sector pension scheme established under Haile Selassie. In terms of the number of people covered, however, the humanitarian emergency system put in place during the 1984/85 famine was at least as significant and provided support for several million people each year (see Figure 5.1). This emergency system had gradually been institutionalised into annual needs assessments and appeals for support from international donors. Meanwhile, much of the aid received was distributed through the World Food Programme (WFP) and a range of non-governmental organisations.

Not only was state social protection extremely limited, but the EPRDF was also resistant to expansion and desperate to eliminate the emergency food system. As outlined in the foreign policy written by Meles Zenawi, annual appeals for food aid were a particular source of shame,

During the civil war, the TPLF placed a strong emphasis on self-reliance, partly borne of necessity given resource shortages, but also related to core ideological commitments (see Chapter 3). As such, services such as healthcare and food aid were made conditional on contributions of labour or other resources (Adhanom et al. Reference Adhanom, Alemayehu, Bosma, Hanna Witten and Teklehaimanot1996, Barnabas and Zwi Reference Barnabas and Zwi1997). This approach was taken forward by the EPRDF government in a context in which state resources were extremely limited. The government adopted a ‘productivist’ approach, which sought to utilise social spending as a productive investment, while also mobilising additional resources in the form of labour and financial contributions.Footnote ¹

ADLI was explicitly framed as a strategy to promote self-reliance at the individual and national level, with land access and rising agricultural productivity expected to improve the livelihoods of all peasants and thereby resolve the problem of food insecurity once and for all (FDRE 2002c, MoFED 2003).Footnote ² The 1996 Developmental Social Welfare Policy is illustrative of the EPRDF’s approach to social protection, both in its developmental framing and the emphasis on community mobilisation, with people expected to contribute labour while the state was limited to supporting local initiatives (MoLSA 1996).Footnote ³ While the government accepted that food aid would be required for the time being, it nonetheless sought to ensure that ‘no free distribution of aid be allowed to able-bodied affected population’ (TGE 1993, p. 3). The vast majority of those receiving emergency assistance were required to participate in employment generation schemes, while ‘gratuitous’ relief, free of work requirements, was reserved for a minority that was unable to work.Footnote ⁴ For Addisu Legesse, subsequently deputy Prime Minister and Minister of Agriculture, a major concern was that people were becoming dependent on support and therefore had to be pushed into productive activity,

While several donors became frustrated with the emergency system during the 1990s, reform efforts gained little traction with the government, which was intent on eliminating the system entirely through ADLI’s successful realisation. Indeed, the 2002 national development strategy made it quite clear that the government’s intention was to address poverty and food insecurity through broad-based economic development, not a targeted programme,

As such, the government’s focus remained supporting peasant production and a resettlement programme to relieve pressure on land in densely populated highlands by relocating people to the more sparsely populated lowlands (FDRE 2002c). This focus on broad-based growth and the distribution of opportunity through access to land and employment was only altered in moments of distributive crisis that presented a clear political threat. This was the case with the 2002/03 food crisis, the 2005 electoral crisis and subsequent periods of high inflation. In each case, the government adapted its strategy, adopting limited forms of social protection that built on the government’s longstanding ideological commitments and supplemented the main focus on broad-based development.

The first example is the 2002/03 food crisis in which some 14 million people required emergency assistance (see Figure 5.1). This crisis led the government to re-assess its agricultural development strategy, but also forced it to confront the limitations of the emergency system. The ultimate result was the adoption of the Productive Safety Net Programme (PSNP) in 2005. The PSNP provides regular food and cash transfers for approximately 8 million people, mostly through public works, but with unconditional ‘direct support’ for those that are unable to work. The PSNP distinguishes between ‘chronic’ and ‘transitory’ food insecurity, both of which had previously been covered by the emergency system. The PSNP was intended to meet the needs of the chronically food insecure – those who, regardless of the agricultural season, are unable to meet their household food requirements – by providing regular and predictable medium-term support. Meanwhile, a scaled back emergency system would provide annual support to the transitory food insecure – those who are able to make ends meet in a normal year, but are affected by annual production shocks. This distinction – plausible in theory, but blurred in practice – has been the basis of both geographical and household targeting in the PSNP. The PSNP covers only chronically food insecure wereda, namely those which have received emergency assistance for the previous three years. Meanwhile, the PSNP targets chronically food insecure households, those having faced food shortages of at least three months every year for three years. Furthermore, there is a preference for vulnerable groups, such as female-headed households, those with chronic illness, the elderly caring for orphans, and poor pregnant and lactating women (MoA 2014, pp. 3–7). Young landless and unemployed adults are therefore largely excluded.

While Ethiopia’s foreign donors funded much of the PSNP and were intimately involved in the design process, it would be a mistake to view the PSNP as a donor-driven programme. Donors promoted certain key design features – such as cash transfers and the move away from annual appeals towards medium-term support. However, the government also exerted its influence, ensuring that the PSNP addressed the EPRDF’s distributive and political challenges and in such a way as to complement existing ideological commitments. The government’s efforts to shape the programme were achieved by selectively aligning with donor interests and playing donors off against each other where there was resistance.Footnote ⁶

The PSNP was seen within government as a temporary response to the distributive crisis in rural areas, with the PSNP used to supplement existing distribution through land and agricultural inputs, buying time for the government to implement its development strategy. In the words of one of the main government members of the design team, the PSNP’s purpose was ‘to ameliorate and to keep things calm as this policy direction [agricultural productivity] is being implemented’.Footnote ⁷ Addisu Legesse, who as Minister of Agriculture oversaw programme design, expected the programme to last no more than five years.Footnote ⁸ Moreover, by supplementing land distribution in rural areas, the PSNP would limit urban migration and the growing challenge of urban unemployment. Indeed, for one regional government official, ‘the highest role of the PSNP is to protect migration from rural to urban areas’.Footnote ⁹ The government therefore sought both to respond to the distributive crisis and maintain political order, but also to limit support only to those who are most in need and who, in time, could be pressed to achieve self-sufficiency. In doing so, this objective certainly resonated with the donor focus on the reduction of extreme poverty as part of the Millennium Development Goals, albeit for somewhat distinct reasons. The result was the design of a programme that was targeted geographically and at household level.

Meanwhile, the government’s longstanding commitment to self-reliance and its desire to phase out the programme within five years also shaped the programme’s ‘productive’ focus.Footnote ¹⁰ First, and in line with the government’s past approach to food aid, most participants were required to engage in public works in exchange for support. In this, the government found allies in key donors such as the World Bank, WFP and USAID, which was sufficient to overcome concerns from other donors. As Meles argued at the time, the PSNP’s work requirements were seen as a vital means of addressing ‘dependency’,

Moreover, the programme aimed to build community infrastructure, including terracing, irrigation systems and roads that would contribute to economic development, with the government designing ‘the project to mobilise the labour of the peasant to achieve these goals’.Footnote ¹¹ Investment in community infrastructure was to be complemented by credit provision, enabling participants to invest in improved agricultural production and off-farm activities with a view to ‘graduating’ from the PSNP within five years. Finally, where households lacked the resources – particularly land – required to graduate, the resettlement programme would relocate households to lowland areas where the government considered land to be plentiful. In doing so, the PSNP and Food Security Programme extended the government’s strategy of coercive distribution, with access to land, credit and now social transfers routed through the party-state. Indeed, officials were clear that one of the PSNP’s objectives was to bring the uncoordinated activities of NGOs providing emergency assistance into one consolidated programme, with distribution placed under government control.Footnote ¹²

Very similar dynamics drove social protection in urban areas, with the government expanding provision through party-state structures in response to distributive crises, while seeking to limit spending and coverage, and utilise spending for productive purposes. As discussed in Chapter 6, one of the main responses to the 2005 election upheaval – attributed by the ruling party to urban unemployment – was the creation of Micro and Small Enterprise programmes to incorporate young adults through provision of credit and training. Urban social protection was also expanded when high inflation and particularly food inflation threatened to erode urban living standards in 2008 (Figure 8.1). Inflation was driven by the government’s expansion of the money supply in order to provide credit for state investment, and rising global commodity prices for food, oil and fertiliser (Admassie Reference Admassie and Pinstrup-Andersen2014). In response, the state Ethiopian Grain Trade Enterprise began importing grain, which was distributed at subsidised prices through new Urban Consumer Cooperatives (UCC) based on ration cards (discussed in Chapter 9). Subsidies were initially introduced in Addis Ababa – the main political threat in 2005 – but subsequently extended to other regional cities after another inflation peak in 2011. Over time, subsidies for wheat, energy and fuel came to account for 20 per cent of government social protection spending (World Bank 2015b, p. 6).

Figure 8.1 Inflation rates (CPI growth rate)

Source: author, based on National Bank of Ethiopia annual reports, various years.

The political threat posed by inflation and unemployment were also key factors leading to the creation of an Urban Productive Safety Net Programme (UPSNP) in 2016. The idea of an urban programme along the lines of the rural PSNP had been discussed between the government and the World Bank since 2008, when the government originally introduced food subsidies. However, it was not until 2014 that both government and the World Bank committed funding.Footnote ¹³ For the World Bank and the Ministry of Urban Development, if not necessarily the government as a whole, a major motivation for the UPSNP was to replace the food subsidy programme, which was seen as a major drain on resources, with targeted support.Footnote ¹⁴

The government’s main motivation for the UPSNP was to respond to the growing crisis of rural landlessness, urban migration and unemployment. As one former government official, now working for a donor agency, emphasised,

The UPSNP is closely based on its rural equivalent, providing targeted support to the urban destitute, primarily through public works though with unconditional support for those unable to work. Like the rural PSNP, the programme is strongly focused on graduation, targeting 80 per cent graduation within three years.Footnote ¹⁶ While young, able-bodied adults are expected to engage in the MSE programme, the UPSNP acknowledges that many of the poorest people are not able to take on credit to set up their own business. Instead, the UPSNP provides three years training and a livelihood grant to enable establishment of their own business with a view to engaging with the MSE programme subsequently.Footnote ¹⁷ As with all other support, the UPSNP is routed through party-state structures.

The UPSNP began rolling out in 2016 and made the first payments to some 190,000 participants in 2017. By 2018, the programme covered nearly 450,000 people, of which 70 per cent were in Addis and the remainder in eleven cities spread across the ethnic regions (World Bank 2018a). Though the UPSNP was initially modest in comparison with the rapidly growing urban population, the Ministry of Urban Development developed plans for a major expansion to be funded by the removal of subsidies. The Ministry aimed to expand the UPSNP to 4.7 million people, covering all of those below the urban poverty line in nearly a thousand towns and cities with a population greater than 2,000.Footnote ¹⁸ The proposal was explicitly framed as a means of addressing urban poverty, but also limiting migration to major cities,

The result then is that the government had a rather ambiguous approach to social protection. On the one hand, growing distributive crises in rural and urban areas resulting from the shortage of access to land and employment were seen as major political threats that warranted a distributive response, leading to the expansion of social protection. As shown in Chapter 9, these distributive resources were channelled through party-state structures with a view to strengthening the party’s political control. On the other hand, however, resource constraints and deep-rooted ideological aversion to handouts led the government to limit support and to use spending to promote productive activity and self-sufficiency. These somewhat contradictory priorities are a key dynamic shaping the distribution of social protection that is the focus of the rest of the chapter. Given that the rural PSNP is by far the largest and long-running programme, this analysis focuses on the evolution and implementation of that programme.

Geographical Targeting of the PSNP

The PSNP was launched in 2005 and initially covered some 4.5 million people in chronically food insecure wereda in the four main highland regions – Amhara, Oromiya, SNNPR and Tigray. The programme initially excluded Afar and Somali regions, due to limited state capacity and the prevalence of pastoralist livelihoods, which were thought to be a problematic fit for a programme designed for highland peasants. Nonetheless, donor enthusiasm about early implementation and major problems with food insecurity in these regions led to phased expansion from 2006/07.Footnote ²⁰ Other than this expansion into Afar and Somali, the PSNP wereda have been remarkably consistent since 2005, focusing on the eastern half of the country, despite growing evidence of widespread poverty and food insecurity in other parts of the country (Figure 8.2). Though the number of wereda covered has increased over time, this is primarily due to the division of existing wereda.

Figure 8.2 Geographical coverage of the PSNP

Source: author.

This geographical distribution is the legacy of past food crises, the emergency system that evolved to address them and the government’s ultimate aim of eliminating the programme altogether. The PSNP not only draws heavily on the ideas that shaped the food aid distribution system during the 1980s and 1990s, but also the administrative machinery used to deliver emergency support. During the 1990s, several studies asked whether food aid was being directed to the right areas of the country and the right people within them. The answer, particularly regarding geographical targeting, was that there was little to no relationship between wereda food insecurity and food aid distribution (Clay et al. Reference Clay, Molla and Habtewold1999). Certain areas – Tigray and North Wello in Amhara – received far more food aid than justified by present needs or household income, while most other areas received less (Clay et al. Reference Clay, Molla and Habtewold1999, Jayne et al. Reference Jayne, Strauss, Yamano and Molla2000). This imbalance was largely attributed to ‘inertia’ in the system. Tigray and North Wello were the areas most affected by the 1984/85 famine and they received a disproportionate share of food aid ever since. The authors hypothesise that,

The prioritisation of Tigray inevitably also raises the question as to whether the TPLF’s dominance in the government throughout the 1990s shaped the distribution of food aid. Clay et al. (Reference Clay, Molla and Habtewold1999, p. 406), for example, find that ‘only about half of Tigray’s success in attracting food aid can be accounted for by such built up capacity and infrastructure’, leading them to speculate that ‘regional biases may also be the result of political influences and other pressures on the food aid delivery system’. While certainly plausible, there are good reasons for caution, however. As Jayne et al. (Reference Jayne, Strauss, Yamano and Molla2000) find, the strongest determinant of food aid distribution in the mid-1990s was actually the Relief and Rehabilitation Commission’s needs assessment during the 1984/85 famine. Not only does this highlight the importance of ‘inertia’ in the system, but it also raises significant doubts about any TPLF influence. In the mid-1980s, the Derg was in power and was fighting the TPLF – then a regional insurgency in Tigray. Far from favouring Tigray, the Derg used food as a weapon of war – destroying crops and denying food aid to the Tigrayan peasantry to undermine TPLF support (De Waal Reference De Waal1997). On coming to power, the EPRDF maintained an existing imbalance in food aid distribution, rather than diverting additional flows to Tigray. Furthermore, as discussed in the previous section, a central focus for the party from its days as a rural insurrection was to limit dependence on external support in a push to achieve food security and development. The EPRDF’s ideological influence in this regard has always been stronger in Tigray than anywhere else. The diversion of large amounts of food aid to Tigray – presumably the implication being that it was used to favour ethnic Tigrayans and build support for the TPLF – is completely at odds with this ideological commitment to self-reliance.

Regardless of the reasons for the apparent imbalance in food aid distribution in the late 1990s, the key point is that the PSNP relied ‘entirely on past food aid assessments to gauge the extent and distribution of chronic food insecurity’ (Sharp Reference Sharp2004, p. 6). Any wereda that had received food aid for the previous three years in 2004 was classified as chronically food insecure and included in the PSNP, while the rest were limited to the scaled back emergency system.Footnote ²¹ Effectively this decision hard-wired the existing geographical distribution of food aid – and any inequalities that this entailed – into the PSNP, as outlined by a consultant involved in the design process,

Decision making regarding wereda classification and the allocations of PSNP quotas to wereda were highly centralised within the federal government, with little to no involvement of either donors or regional governments. Indeed, this was an explicit decision intended to forestall any debate and bargaining with regional administrations. According to the federal head of food security,

The fixed classification of chronically food insecure wereda did latterly lead to growing concerns among donors that many of the poorest households may live outside the PSNP wereda identified in 2005.Footnote ²³ The result was a series of donor-commissioned studies from 2013 onwards that examined the needs outside PSNP wereda. One study, focusing on vulnerability to absolute poverty, rather than chronic food insecurity specifically, underscored the extent to which the PSNP’s geographical distribution excluded large numbers in need,

The government contested these findings, leading to long discussions with donors about the PSNP’s geographical distribution. The government’s central objective, as discussed previously, was to eliminate food insecurity and therefore the PSNP. Indeed, progress towards this target was increasingly seen as a bellwether for the EPRDF’s developmental project. Consequently, any suggestion that the PSNP was significantly under-sized and that a major expansion was warranted was met with a frosty reception. Instead, the government stuck with its longstanding view that food insecurity was primarily related to rainfall patterns, justifying the focus on the eastern half of the country and avoiding more problematic questions about the PSNP’s coverage. In the words of a senior official in the Ministry of Agriculture’s Food Security Directorate, the ‘principle [of the PSNP] is vulnerability to drought. We do not cover moisture sufficient areas’.Footnote ²⁴ The result is that many of the poorest people were excluded on rather arbitrary historical grounds, with many instead receiving ad hoc emergency assistance.

The Push for Graduation as a Sign of Progress

The government’s focus on eliminating the PSNP had important implications not just for geographic distribution between wereda, but also within them. From its inception, the PSNP was expected to promote graduation of participants out of the programme and into self-sufficiency. Progress with graduation, however, was limited in the early phases. The government did start graduating households from 2007/08, but just 500,000 individuals out of a programme reaching more than 7.5 million had graduated by 2010. The PSNP proved an important means of preventing drought from translating into famine in 2008–2010 and the reality was that neither government nor donors had much option other than to extend the programme for another five years. However, the government’s willingness to extend the programme did not amount to a change in its ideological commitment to self-reliance, the productive contribution of the programme or graduation from state support. If anything, developmental successes over the previous five years – with double-digit economic growth and rising agricultural productivity – only reinforced the government’s commitment.

The new national development strategy, the 2010 GTP1 reflected this step change in ambition, setting a range of enormously challenging targets across sectors. With respect to agriculture and food security, Prime Minister Meles, on presenting the strategy to an EPRDF meeting proclaimed that,

In doing so, the government reiterated its commitment to reducing the size of the PSNP,

Reduced PSNP spending was, in turn, hoped to free up significant resources that could be redirected from ‘areas of lower agricultural potential’ covered by the PSNP towards ‘high-potential rainfed areas and irrigation development, in order to accelerate productivity growth, agricultural led industrialisation and long-term food security’ (MoARD 2010a, p. 13).Footnote ²⁵ The result was that the government used annual PSNP graduation rates as one of the performance indicators for its agricultural development strategy (MoARD 2010a, p. 21), with the GTP1 aiming to reduce PSNP coverage from 8 million to just 1.3 million by 2015 (MoFED 2010). In effect this meant the graduation of all able-bodied participants, leaving just the elderly and those with disabilities receiving direct support.

Though donors were always sceptical about the feasibility of this target, they also became interested in graduation at this time, considering 50 per cent to be more realistic.Footnote ²⁶ The donors had always refused to get involved in the government’s resettlement programme that proved to be as problematic as many had initially feared (Hammond Reference Hammond2008). Moreover, in the early phases they declined to engage with the government’s attempts to improve agricultural livelihoods through revolving credit known as the ‘Other Food Security Programmes’, instead focusing support on the PSNP itself. Growing donor interest in graduation, however, resulted in financial support for a re-design of these supplementary programmes, which were separated into the Household Asset Building Programme (HABP) and Complementary Community Investments (CCI). Donors provided technical assistance to HABP, which aimed to establish revolving credit funds to enable PSNP participants to borrow money and invest in increased production. In contrast, donors remained reticent about the CCI, which provided capital investment in infrastructure, particular irrigation schemes and water infrastructure.Footnote ²⁷ The design process also resulted in a concerted attempt to define graduation based on a distinction between graduation from the PSNP itself based on attainment of ‘food sufficiency’ – where a household is able to ‘meet its food needs for all 12 months and is able to withstand modest shocks’ – and the higher bar of graduation into food security and out of the broader Food Security Programme, including HABP (MoARD 2010b, p. 21). To operationalise these somewhat imprecise categories, graduation was to be assessed based on the asset ownership of PSNP participants, comparing levels of landholdings, livestock and education with regionally defined graduation thresholds.

The government maintained its focus on graduation after Meles Zenawi passed away in August 2012 and was replaced by Hailemariam Dessalegn. Moreover, the new Prime Minister announced in 2014 – at the annual celebration of the EPRDF’s removal of the Derg from office – that Ethiopia had achieved national self-sufficiency, a longstanding objective that Meles had repeatedly promised,

It is far from clear quite what this announcement was based on. While 2014 was a comparatively good year for agricultural production, Ethiopia nonetheless appealed for emergency food aid to feed some 2.7 million people beyond those covered by the PSNP. Moreover, Hailemariam acknowledged that household self-sufficiency remained a work in progress. What it does signify though was an attempt by a new Prime Minister, with limited authority, seeking to maintain a narrative of progress with respect to food insecurity and to deliver on the promises and plans laid out by his predecessor. This likely added to the pressure for a similar demonstration of progress from the PSNP itself.

The federal graduation target in the GTP was cascaded down through the tiers of state administration at the regional, wereda and kebele levels, with officials’ performance in relation to these and other targets assessed through gim gema. The result was a major wave of graduation, with some 10 per cent of PSNP participants graduating each year in 2010–2014 (Figure 8.3), resulting in a sharp reduction in programme coverage (Figure 8.4). While the target of reducing the PSNP to 1.3 million by 2015 was not met, somewhere in the region of 3.5 million people were removed from the programme, reducing enrolment to less than 5 million, with many state officials of the genuine belief that the PSNP would come to an end.Footnote ²⁸

Figure 8.3 Graduation from the PSNP by region

Source: author, based on DFID staff calculations. For legibility the least populous regions (Dire Dawa and Harari) are omitted.

Figure 8.4 PSNP caseload by region

Source: author, based on DFID staff calculations.

The federal government’s pressure for graduation had quite different impacts between regions. A major drought and food crisis in 2009/10 across the Horn of Africa meant that there was little attempt at graduation that year. However, graduation resumed in earnest from 2010/11, with large numbers graduating in Amhara, SNNPR and Tigray. Numerous respondents across the tiers of state administration were clear that this was a top-down process driven by performance targets, rather than a meaningful reflection of improvement in household food security.Footnote ²⁹ As one local government official in Tigray put it,

There was initially comparatively little progress in Oromiya before graduation increased substantially in 2013/14. During this period, the Oromiya regional government was under pressure from the federal government to make progress, with an Oromiya government official admitting that, ‘We frequently get questions from the federal level – “where is your progress on graduation?”’Footnote ³¹ Indeed, this resulted in some competition between regions, with the Oromiya government embarrassed to fall behind the rest,

The regional government had no input into the distribution of PSNP places within their region, rather officials in the federal government ‘decide and send us numbers, wereda by wereda’.Footnote ³³ Nonetheless, the regional administrations were held responsible for graduation targets and the performance of lower-level officials. The result is that lower-level officials were forced to make arbitrary selections to remove people from the programme. As one wereda agricultural officer described,

Indeed, these targets were passed on to kebele officials who were expected to implement them,

The fact that much of this graduation was a mere administrative fiction to justify a political narrative was well known. Indeed, it was readily admitted by one senior official in the Ministry of Agriculture’s Food Security Directorate, who underscored the tension between the imperative of developmental progress and the reality of vast unmet needs,

Here the tension between the programme’s protective rationale and its developmental pretensions is laid bare. The necessity of demonstrating progress with graduation and a somewhat arbitrary target in the national development strategy overrode any concern with the protective role of the programme.

In contrast to the pressure for graduation in the highlands, there was no attempt to promote graduation in Afar or Somali. Indeed, the key mechanisms underpinning the PSNP’s graduation model have been absent in these regions due to concerns about the implementation capacity of the regional states and the programme’s suitability to pastoralist contexts. Initially the PSNP only provided direct support in Afar and Somali, with no public works requirement or community infrastructure development until 2010. Moreover, there are no livelihood programmes in either Afar or Somali, unlike the other regions. Nonetheless, the government has not accepted the PSNP as a permanent mechanism of support in these regions either. Instead, the longstanding belief within the state, under the EPRDF and past regimes, is that pastoralism is inherently backward and unsustainable. From this perspective, the only feasible option for future development is that those engaged in mobile livelihoods begin practicing settled agriculture, bringing them into line with highland areas (MoFA 2008). Rather than adapting the PSNP to the needs of pastoralist livelihoods, the approach has been to await the implementation of villagisation that places pastoralists in planned settlements.

The result of this divergent pattern of graduation has been to change the PSNP’s regional distribution. In 2005, Amhara was home to by far the largest number of PSNP participants with some 2.5 million, more than 40 per cent of the programme, while Oromiya, SNNPR and Tigray each received support for 1.3–1.4 million or 15–20 per cent (Figure 8.5). Over time, the main change has been the PSNP’s expansion in Afar and Somali. This shift has entailed a reduction for the four highland regions as a proportion of the total, but also in absolute terms thanks to large-scale graduation in 2010–2015. A distinct pattern also emerges when PSNP coverage is viewed as a proportion of the regional population (Figure 8.6). Initially, Tigray received the highest level of support as a proportion of its population, with more than 30 per cent covered. This changed markedly from 2010 with a sharp decline due to large-scale graduation. Amhara, Oromiya and SNNPR exhibit similar, though less pronounced declines. In contrast the proportion covered in Afar and Somali has massively increased.

Figure 8.5 PSNP distribution by region

Source: author, based on DFID staff calculations.

Figure 8.6 PSNP coverage as a percentage of the regional population

Source: author, based on DFID staff calculations and World Development Indicators. For legibility the least populous regions (Dire Dawa and Harari) are omitted.

The result is that any bias towards Tigray and parts of Amhara has been significantly eroded through graduation.Footnote ³⁷ A recent World Bank report that assessed the regional distribution of the PSNP concluded that SNNPR – whose quota was significantly reduced through graduation – is the most under-represented region, accounting for 40 per cent of those living in chronic poverty but just 13 per cent of the PSNP case load (World Bank 2020, p. 25). Amhara is also slightly under-represented, while all other regions receive a greater share of the PSNP than poverty rates warrant.

Developmental Progress and the Persistence of Food Insecurity

The PSNP was intended to end the ‘humiliation’ of annual emergency appeals by establishing a medium-term mechanism for supporting food insecure households while raising agricultural productivity to enable mass graduation. Indeed, political leaders repeatedly claimed that Ethiopia was on the brink of reaching national self-sufficiency and even, under Hailemariam, that it had achieved it. The reality is that Ethiopia has yet to come close to achieving these goals.

Despite reduced PSNP coverage by 2015, the government was forced to accept that food insecurity had not disappeared and that the programme could not end. There was little alternative but for the government and donors to agree a new five-year phase. Indeed, no sooner had the government staged the graduation of several million PSNP recipients in the lead-up to 2015 than Ethiopia faced another major food crisis in 2015–2016 as a result of an extreme El Niño that disrupted rainfall. The result is a growing acceptance in government that the food insecurity problem is not going away anytime soon and that the elimination of the PSNP is simply infeasible for the foreseeable future. The new phase of the programme from 2015 expanded to cover roughly 8 million people, with a contingency fund to cover an additional 2 million people based on annual needs. Moreover, the government, which had previously limited its financial contribution to paying the salaries for the officials that administer the programme, relented to donor pressure and agreed to take on a progressively larger responsibility for financing the PSNP in the following years. Donor officials involved in the negotiation of the new phase argued that the government had itself recognised the failings of its graduation efforts and was increasingly open to incorporating social protection into its ‘developmental state’ model,

The change in approach from a short-term safety net to a long-term mechanism for supporting the poorest 8 million or so people in the country might have been expected to end the emphasis on graduation. This was not the case, however, and a strong commitment to graduation remained, largely as a result of the ideological aversion to ‘dependency’ and the concern that donor funding was unreliable and therefore there needed to be a plan in place to wind down the programme, if necessary.Footnote ³⁹ As such, GTP2 stated the target of graduating 5,001,116 individuals by 2020 (NPC 2016, p. 126). Rather than focusing on graduation as a means of eradicating the programme, according to the head of food security, ‘when some people graduate, other vulnerable people replace them’, maintaining a consistent caseload of approximately 8 million people.Footnote ⁴⁰ As such, the majority of participants should strive to graduate and, as they do so, they will be replaced by other individuals in need. In effect, such an approach acknowledges the insufficiency of PSNP coverage compared to poverty levels, with a plentiful supply of new eligible participants available to replace any that graduate.

The PSNP is based on a dubious distinction between chronic and transitory food insecurity that divides the PSNP and its chronically food insecure caseload from emergency assistance for those facing transitory food insecurity. As shown in previous sections, however, this conceptual distinction breaks down in practice with geographical targeting arbitrarily dividing the country into PSNP wereda and those limited to emergency assistance, regardless of need, while forced graduation routinely removed chronically food insecure households from the PSNP, with many subsequently receiving emergency assistance instead.Footnote ⁴¹ Figure 8.7 provides further evidence. If emergency assistance really was addressing transitory food insecurity, then in some years of adequate and timely rainfall, the annual emergency appeals should fall to, or very close to, zero. The graph shows, however, that emergency appeals have never come close to zero throughout the PSNP period. Combining the caseloads of PSNP and humanitarian support and disaggregating these figures by region is also illustrative (Figure 8.8). The total number of people supported is on a clear upward trajectory in every single region, while the caseload as a proportion of the regional population is on an upward trajectory in every region except Amhara and Tigray, where, thanks to large-scale graduation, the trend is flat (Figure 8.9).Footnote ⁴²

Figure 8.7 Food insecurity over time

Source: author, based on DFID staff calculations and Humanitarian Requirements Documents, various years.

Figure 8.8 Combined PSNP and humanitarian caseloads by region

Source: author, based on DFID staff calculations and Humanitarian Requirements Documents, various years. For legibility the least populous regions (Dire Dawa and Harari) are omitted.

Figure 8.9 Combined PSNP and humanitarian caseloads as a proportion of regional population

Notes: regional populations are approximated by applying the region’s proportion of the national population at the time of the 2007 census to population totals for subsequent years. For legibility the least populous regions (Benishangul-Gumuz, Dire Dawa, Gambella and Harari) are omitted.

Source: author, based on DFID staff calculations, Humanitarian Requirements Documents, various years, and CSA (2008).

The PSNP itself was prompted by the severity of the 2002/03 food crisis in which food aid was sought for 14 million people. Yet what was considered an exceptional event at the time was matched or surpassed by the combined PSNP and humanitarian caseload in 2007/08, 2008/09 and every year since 2015/16. The PSNP clearly plays an important role in delivering support and responding to crises, but the problem of food insecurity at national and household level is not going away anytime soon. Moreover, it has become increasingly clear that the emergency system does not just address transitory food insecurity, but also compensates for the limited size of the PSNP. In certain respects, the situation is therefore not so dissimilar to that prior to the adoption of the PSNP, with the humanitarian system used to address a structural problem to which it is poorly suited. Ultimately, this was the result of the restrictive geographical focus of the PSNP, which excluded significant numbers, and the limited quotas provided to PSNP wereda, leading to a shortage of PSNP places. Government officials increasingly acknowledged the reality that the PSNP was significantly smaller than the number in need, with a more realistic figure considered to be 10–12 million by one estimate or as much as 30 million people according to another.Footnote ⁴³ Yet such an expansion would have contradicted the narrative of developmental progress and efforts to eliminate the ‘national humiliation’ of food insecurity. Moreover, there were simply insufficient resources between donor and government contributions to expand the PSNP, much less to nearly quadruple the programme’s coverage.

Conclusion

As outlined in previous chapters, alongside rapid economic development Ethiopia faced a growing distributive crisis from the early 2000s, with acute shortages of access to rural land and urban employment. This distributive crisis, which manifest in widespread food shortages in 2002/03, urban protests in 2001 and 2005, and food inflation in 2008 and 2011, was the principal driver of social protection expansion. However, the government’s approach to social protection was highly ambiguous due to resource shortages and the government’s ideological commitment to self-reliance and resistance to ‘dependency’. The result was that the government expanded provision to mitigate politically problematic distributive issues, while limiting support and attempting to eliminate programmes as quickly as possible.

The government’s approach to social protection is also consistent with the argument that an important driver of state-led development is elite perceptions of the threat posed by mass distributive pressures amidst resource constraints. As argued in Chapters 4–6, mass distributive pressures were an important motivation for government attempts to promote broad-based development in the form of access to land, rising agricultural productivity and access to employment. The government prioritised distribution in the form of land, agricultural inputs and jobs due to their perceived compatibility with economic growth. In contrast, the government lacked the resources required to address the growing distributive crisis through social spending alone and was concerned that social protection might undermine work incentives and thereby the government’s developmental push. The distributive limitations of land and employment eventually led the government to expand social protection temporarily, but this was insufficient for politicians to question the core idea that broad-based economic growth was the only long-run solution to the distributive crisis.

As such, the government’s social protection response was ambiguous and insufficient to address the growing problem. In particular, the government intentionally excluded large numbers of poor and vulnerable people through geographical targeting and the push for graduation. Meanwhile, the launch of an urban PSNP was both late and insufficient to address the growing distributive problems. Both the rural and urban programmes prioritised the destitute, largely excluding landless and unemployed young adults who were instead expected to seek employment – for which demand was extremely limited – or to take out loans to establish MSEs.