The politics of ‘institutionalising’ social protection in Africa: The retrenchment of social cash transfers in Zambia, 2015–2021

Abstract Social protection has expanded unevenly across Africa because of variations in both the initial adoption of programmes and their subsequent ‘institutionalisation’ through government-funded expansions in coverage. The case of Zambia illustrates how policy coalitions promoting the institutionalisation of social protection compete with other claimants over prioritisation in public spending. Even when faced with competitive elections, incumbent governments may prioritise other programmes over social protection. In Zambia, the incumbent government announced and budgeted for a massive government-funded expansion of social protection but failed to allocate the necessary funding – with the result that benefits were not paid to registered beneficiaries. If ‘institutionalisation’ is understood as entailing the political irreversibility of expansion, then the rhetoric of institutionalisation belied the reality (for several years) of retrenchment. The weakened policy coalition supporting social protection was unable to prevent government defunding as scarce government resources were allocated to competing programmes.


Introduction
Social protection has steadily expanded across much of Africa, but its reach continues to lag behind most other regions.Social protection programmes have a long history in Africa, especially in South Africa, Mauritius, and North Africa.Prior to the 2000s, however, social protection in the Sahel and West, Central, and East Africa was generally limited to 'humanitarian' provision in times of drought, other natural disasters, or war.With strong support from international and donor agencies from the 'Global North', provision expanded across most of the continent since about 2000.Both the number and the coverage of major social cash transfer programmes in sub-Saharan Africa are estimated to have expanded fourfold between 2000 and 2015 (Hickey et al., 2019).Yet the reach and adequacy of existing programmes remain very uneven, and many programmes continue to rely on donor funding as international organisations have documented repeatedly (Beegle, Coudouel, & Monsalve, 2018;ILO, 2021;UNDP, 2019).
The scholarship on social protection reform in Africa has expanded rapidly also, focusing primarily on how and why national governments have embraced or resisted proposals for new programmes.Recent research has extended this analysis to consider how and why national governments implement policies or expand and 'institutionalise' programmes after their initial, often very limited, introduction.This article builds onand challengesthis literature through an analysis of one African country -Zambiawhere, contrary to assessments that social protection had been 'institutionalised', the government partially retrenched its social protection programmes for a period of five years.Between 2015 and 2020, the Zambian government failed to pay most of the benefits due to registered beneficiaries of its social cash transfer programme.Faced with competitive elections, the government chose to allocate scarce public resources to other programmes in order to shore up its support.This reflected the nature of the regime and its support base as well as the weakening of the 'policy coalition' that had previously placed social protection on the national agenda and lobbied successfully for its initial expansion.Through an analysis of this case study of partial retrenchment, this article shows that the process of social protection reform is messier, more contingent, and more vulnerable to reversal than recent studies have suggested.
This article first considers the meaning of 'institutionalisation' and the role of 'policy coalitions' within this.The article then turns to the Zambian case, documenting the shift from apparent institutionalisation to deinstitutionalisation.It charts and explains the prioritisation of other government programmes (especially infrastructure and agricultural support).The article concludes with a discussion of the implications of this case study.

Explaining the uneven expansion of social protection in Africa
The scholarly literature on the politics of social protection in Africa has grown rapidly (Hickey et al., 2019;Seekings, 2020).Comprising primarily case studies that employ some form of process tracing, this literature documents how international organisations (including aid donors, UN agencies, and international NGOs) have combined with local activists and sympathetic government bureaucrats to recruit national political champions for their preferred reforms.Scholars have paid some attention to negotiation and contestation within and between international organisations, but their primary focus has been on the interactions between them and national governments (Foli, 2016;Hickey et al., 2019;Ouma, 2019).More recently, scholars have turned to what happens after national governments introduce social protection programmes (or allow donors to do so in their territories).Some recent research focuses on the implementation of social protection programmes (Lavers, 2022).A second strand of research focuses on the process of 'institutionalising' social protection after the initial adoption of programmes.As Lavers and Hickey (2021) note, the political dynamics may change as countries proceed further down one or other pathway of reform.Related research focuses on the consequences of social protection for relationships between citizens and states (see Social Policy and Administration, 2023).Lavers and Hickey (2021), reviewing case studies of eight African countries, argue that international organisations are most influential in the early stages of the process through the formation of 'transnational policy coalitions'.International organisations can combine with local allies in both civil society and the state to lobby for the introduction or adoption of generally small-scale (or 'pilot') and donorfunded programmes, often 'in the face of resistance or ambivalence from ruling elites'.Lavers and Hickey argue that these 'initially donor-dominated policy coalitions' have much less influence over the subsequent expansion or 'institutionalisation' of social protection.By 'institutionalisation', Lavers and Hickey mean 'the extent to which social transfer programmes are implemented and financed by national governments, grounded in legislation [and] national in reach'.A 'fully institutionalised programme is one that provides eligible recipients with a right to support that the government is obligatedfinancially and administrativelyto deliver'.Lavers and Hickey measure institutionalisation using five criteria: the statutory basis of programmes, the government's share of financing, state responsibility for implementation, whether the programmes are national in scope, and overall coverage (ibid: Table 1).Using these criteria, Lavers and Hickey point to periods of rapid institutionalisation in six of their eight case studies, including Zambia in the early 2010s.In some cases, institutionalisation was 'a top-down response to competitive elections'; in more authoritarian cases, it was driven by 'elite perceptions of vulnerability in the face of distributional crises' (ibid: 1).
Distinguishing between phases of (initially) agenda setting and (later) government financial and legislative commitment usefully points to the dynamic nature of the policy reform process.But the Lavers and Hickey's analysis suffers from three weaknesses.Its first weakness is its unambiguously state-centric assumption.For Lavers and Hickey, 'institutionalisation' entails the construction of a welfare state through the central state's acceptance of statutory, financial, and institutional responsibility.Provision by non-state institutionsincluding religious institutions and other provision at the local level, as well as external financing through aiddoes not count towards 'institutionalisation'; social protection is 'institutionalised' in Africa insofar as states accept the responsibilities accepted in the past by states in north-west Europe.
Secondly, the Lavers and Hickey's explanation of why incumbent political elites conclude that they have a political interest in the 'institutionalisation' of social protection focuses on the threat and not the choice of remedy.They do not consider why elites prioritise social protection specifically as the key to their political survival.Even when national governments are both politically vulnerable and sympathetic to the expansion of social protection, they might worry about losing support if spending on social protection is (or is seen as) a diversion from more important priorities (such as support for small farmers or investment in public goods) or, worse still, is seen to encourage unpopular behaviour such as laziness or disrespect, especially among young people.They might also worry that diverting resources from patronage networks to programmes might undermine support.In Malawi and Tanzania, for example, the major parties in highly competitive elections have eschewed the expansion of state-funded social protection in favour of other programmes, leaving social protection heavily dependent on donors (Hamer & Seekings, 2019;Masi, 2023;Ulriksen, 2019).
Thirdly, Lavers and Hickey appear to assume that their empirical measures of institutionalisation reflect an underlying irreversibility of reform.In none of their eight cases did their aggregate measure of institutionalisation ever decline significantly in the period between 2000 and 2015 (with only a negligible decline in Ethiopia between 2009 and 2013).One of their indicators of institutionalisation is the existence of enforceable legal rights, although these have rarely been a significant factor in Africa.More importantly, the various indicators of institutionalisation presumably reflect the enduring or deepening power of a supportive policy coalition (including beneficiary citizens, politicians, and bureaucrats) that is strong enough to prevent the retrenchment of public provisionincluding the efforts of rival policy coalitions to redirect scarce state expenditures to other programmes.In other words, institutionalisation entails passing some political threshold, after which governments cannot retreat back down the path they came along.
If institutionalisation is understood in terms of political irreversibility, it requires more attention to the policy coalitions defending or resisting reform.Lavers and Hickey are quite right to view the transnational coalitions promoting publicly the idea of social protection as policy coalitions in the sense of 'advocacy coalitions', but they are too quick to dismiss the role of other kinds of policy coalitions in the subsequent expansion of social protection.At this point, the crucial coalitions are likely to comprise actors who are resisting expansion in order to reallocate scarce public resources to their preferred, alternative programmes.These coalitions are unlikely to be as visible as the initial pro-social protection policy coalition, but their apparent invisibility should not be confused with weakness.
This study examines the case of Zambia, focusing on the Patriotic Front (PF) government led by Edgar Lungu between 2015 and 2021.Under the preceding PF government, under Michael Sata (from his election in 2011 until his death in office in late 2014), social protection appeared to have been 'institutionalised', as Lavers and Hickey (2021) find on the basis of their measures.In 2013-14, Sata's PF government had expanded coverage countrywide and committed to funding most of this from its own revenues (rather than relying on donors, as hitherto), although it did not incorporate the expanded programmes in legislation.Under Lungu, however, this supposed institutionalisation of social cash transfers was quickly reversed.The Ministry of Finance (MoF) failed to transfer sufficient funds to the Ministry of Community Development and Social Services (MCDSS) to pay most of the households that had been registered for social cash transfers.By the end of 2018, payments were a year in arrears.The state's assumption of responsibility for social protection proved to be reversible.Using a strongand politicalunderstanding of institutionalisation, social protection was yet to be institutionalised in Zambia.At the same time, Lungu's government spent heavily on other programmes, including on 'developmental' infrastructure and programmes supporting small farmers.
The case of Zambia is not in any sense typical of countries in Africa (or elsewhere).Whilst it is not unique (as we shall show later), it is an unusually clear case of substantialif perhaps short-livedretrenchment or deinstitutionalisation through defunding.The value of this case study lies not in its typicality but rather in what it reveals about the meaning and politics of institutionalisation.The institutionalisation of social protection requires not only that governing elites identify social protection as a means to their political survival (as Lavers and Hickey argue) but also that it is a high priority among the various programmes that might serve this purpose.The coalition defending public expenditure on social protection must retain or gain power relative to the competing coalitions that lobby for government expenditure on other programmes or activities.
The literature on the politics of taxation in Africa has grown (including in Zambia), but much less attention has been paid to the politics of budgeted or actual public expenditure.Across Africa, national governments spend much less on social protection than is recommended by international organisations including the World Bank, International Labour Organisation, and bilateral donors.This is not simply because social protection is unaffordable.Rather, governments choose to commit their scarce resources to alternative programmes.The 'affordability gap' between the recommended levels of spending by international organisations and what national governments are willing to spend reflects political prioritisation (Seekings, 2017) and the power of competing political actors or coalitions.

Data
This study examines public policy in Zambia through process tracing (Beach & Pedersen, 2016), as is widespread in the study of social protection in Africa (Hickey et al., 2019).The study draws on in-depth interviews conducted between 2015 and 2023 with more than sixty informants.The sample was purposive: Individuals were approached because of their positions in the institutions within and outside the state that were involved in policymaking.The informants included 11 political leaders from the PF (in power from 2011 to 2021); 19 officials from international organisations (including the UK's Foreign, Commonwealth and Development Office [FCDO], the United Nations Children's Fund (UNICEF), the World Bank, the World Food Programme, and the Friedrich Ebert Stiftung); 10 officials in district-level social welfare offices; and 13 technocrats from the MCDSS, MoF, and Ministry of Labour and Social Services (MLSS).Other informants included four civil society actors, two local consultants, and two executives of Zambian think tanks.Most of these interviews were conducted between 2022 and 2023.The defeat of the PF in elections in August 2021 provided an opportunity to interview individuals who had not been easily accessible whilst they were government ministers, including former ministers of Finance, Agriculture, and Housing and Infrastructure.All informants consented to participation in the study and (in most cases) the recording of interviews, but they have been anonymised to protect their identities.Whilst respondents are identified by their official positions, we do not specify the periods during which they served.We also examined a wide range of documents.The combination of sources facilitated the corroboration of 'facts' and interpretations.Almost all interviews were conducted by only one of us.The fact that one of us is Zambian (and junior researcher) and the other an outsider (and professor) facilitated access to different informants and allowed us to compare responses.

Adoption to apparent institutionalisation, 2003-2014
Zambia was a case of early adoption but then slow expansion of social protection.Social protection programming in Zambia was initiated by the German government (through its development agency GTZ) in 2003 and later financed primarily by the UK (through the then Department for International Development), UNICEF, and to a lesser extent Irish Aid.These international organisations forged a social policy coalition with technocrats in the MCDSS and civil society organisations and 'piloted' several cash transfer models (including an old age pension and a child grant).Initially, in the mid-2000s, the governing Movement for Multiparty Democracy (MMD) had opposed SCTs, preferring rather subsidies for farming inputs to small-and medium-scale farmers (through the Farmer Input Support Programme, FISP) as well as price support (through the Food Reserve Agency, FRA), which controversially benefitted constituencies that were perceived to be loyal to the MMD, local officeholders, and suppliers of fertiliser and other inputs (Andrew, 2021;Kabandula & Seekings, 2016;Kuss, 2015;Nkinke, 2023).Although the government did begin to commit modest funds for SCT pilots from 2008, it consistently overspent its budget for both FISP and the FRA (Pruce & Hickey, 2019;Siachiwena, 2020).
Farm subsidies were defended by the 'maize mafia' (Pruce, 2019) comprising not only the large-and medium-scale farmers who benefitted disproportionately but also the minister and civil servants in the Ministry of Agriculture, suppliers of farming inputs (who imported inputs into Zambia), transporters (who distributed inputs across the country), and agro-dealers (who stocked inputs in retail outlets), with the support of officials in the presidency, Members of Parliament (MPs), district commissioners, and traditional leaders who all believed that farm subsidies were important for addressing both household food security and incomes.It is widely believed that senior political leaders had vested interests in the supply chain of farming inputs or received funding from stakeholders in the agriculture sector (Simutanyi, 2021). 1 The policy coalition supporting farm subsidies was thus in part also a patronage machine.
After defeating the MMD in the 2011 elections, the Sata/PF government instituted reforms to reduce the subsidy on fertiliser and improve targeting of small-scale farmers (Kuss, 2015, p. 16) as well as expanding SCTs.This was in line with the World Bank's recommendations (Kuss, 2015;Tesliuc, Smith, & Sunkutu, 2013). 2 In October 2013, the PF Minister of Finance (Alexander Chikwanda) announced that 1 This point was also mentioned in interviews with a former Minister of Agriculture, a local consultant, and a former Deputy Executive Director of a local think tank.government spending on social cash transfers would increase sevenfold in order 'to make a significant impact on reducing extreme poverty'.A major implication of this announcement was that the government increased its share of funding the SCT budget from 24% in 2013 to 75% in 2014 (Siachiwena, 2020).The minister also committed to increasing spending on other social protection programmes, including the Food Security Pack (FSP) for households with the potential to start small-scale farming (Chikwanda, 2013, p. 8).In every budget speech between 2014 and 2021, PF finance ministers reaffirmed their government's supposed commitment to expanding the coverage of SCTs through steadily increasing the budget allocation.
In 2011, fewer than 25,000 households were benefitting from cash transfers across the country (see Figure 1).Sata's administration inherited a modest expansion plan under coverage rose to 61,000 households in 2013.Thereafter, nominal coverage exploded, reaching almost 900,000 households in 2021 (see Figure 1).Under the PF, nominal coverage thus rose from fewer than one in one hundred to more than one in five of the country's households.The pace of this nominal expansion came as a surprise to donors and civil society who had expected only modest reforms. 3nalyses of this apparently dramatic expansion precisely point to the kinds of factors emphasised by Lavers and Hickey (2021).The model of SCTs for poor households had been promoted by a transnational policy coalition comprising international organisations (including the ILO, UNICEF, and the German and British aid agencies) together with technocrats in the Ministries of Finance and Community Development.In one interpretation, reform proposals won the support of a social democratic faction within the PF, including Sata himself, his Vice President Guy Scott, PF Secretary General Wynter Kabimba, PF Treasurer and Cabinet Minister Emmanuel Chenda, and other cabinet ministers.This faction saw SCTs as both the right thing to do and as a strategy that would secure the votes of poor Zambians (Siachiwena, 2020).In an alternative interpretation, Sata was an opportunist concerned with reorganising the distribution of rents (and hence the 'political settlement') (Pruce & Hickey, 2019).

Deinstitutionalisation of social protection, 2015-2020
Sata's government initially struggled to spend the extra money it had budgeted for the expansion of SCTs, releasing only 50% of the promised funding.Officials in MCDSS and donor officials point to the unsurprising institutional challenges of identifying, registering, and paying over 80,000 new beneficiaries in a single year. 5Improved capacity enabled the MCDSS to spend 82 percent of the budget in 2015.Available data on the funding for SCTs that were budgeted and spent by year are shown in Figure 2.Where available, the division of budgeted and actual expenditure between government and donors is reported.Unfortunately, this disaggregation is not available prior to 2016 (nor is it available for the budget data for 2018 and 2019). 6ata's death in October 2014 triggered a presidential by-election that was won by Lungu for the PF in January 2015.Lungu was not part of the social democratic faction of the PF but was associated instead with a rival pro-business faction that included the PF's financial backers.These included Sata's uncle and Finance Minister Chikwanda, who owned a construction company that received government contracts after 2011 (Africa Intelligence, 2014), and Willie Nsanda, Sata's campaign manager in 2011, who Sata then appointed as head of the powerful Road Development Agency (RDA).Road construction and other infrastructure were central to the PF's development agenda, with the RDA emerging as a powerful agency that awarded million-dollar contracts to mostly Chinese construction companies (Hinfelaar & Sichone, 2019).The PF government moved the RDA into the Presidency, with Nsanda reporting directly to Sata.All this reflected the power of an 'infrastructure' policy coalition, involving the Presidency, RDA, the Ministry of Housing and Infrastructure, and foreign (mostly Chinese contractors) and local contractors including many who had close links to the PF (Siachiwena, 2021).After Sata's death, this coalition threw its support behind Lunguwith Nsanda running his 2015 election campaign.
Lungu also forged an alliance with former President Rupiah Banda of the MMD.Lungu was dependent on Banda's support in the east of the country (then an MMD stronghold) to stymie the The MoF's 2017 annual economic report shows that donors spent only 1% of the total budget costs.This contrasts with DFID's data that show that donors contributed about 24% of total spending.The MoF's report does not explain why donor funding was uncharacteristically low.Given the levels of donor funding both before and after the 2013 expansion, the DFID's figures for 2017 appear to be more realistic.It is possible that the MoF did not have complete data at the time of publishing the 2017 report.resurgent opposition UPND.It was reported that Banda's support for Lungu would "involve major policy compromises" in order to accommodate the former's business interests (Africa Confidential, 2015).Banda had close ties to the companies that were the main suppliers of fertiliser for FISP under a controversial single-sourcing contract (Pangea Risk, 2021).
Under Lungu, the uneasy balance of power within the PF under Sata shifted away from the social democratic faction to the more pro-business faction that had backed Lungu's presidential campaign.Most of the marginalised factions were pushed out of the PF.This shift also had implications for the social protection policy coalition which was unable to secure adequate funding for MCDSS programmes, whilst patronage-based coalitions in agriculture and roads infrastructure benefited from massive overspends in the two sectors.
Whilst the budget for cash transfers appeared to provide for expansion, actual spending was only a fraction of what had been budgeted (see Figure 2).In 2016, the government spent less than half of its share of budgeted expenditure.Donors spent what they had promised, but their share was smaller than the government's.Close to 110,000 households (about 45 percent of the caseload) went unpaid for the first eight months of the year (Siachiwena, 2020).In 2017, the government spent most but not all of what it had budgeted.Donors stepped in, allocating additional funding beyond what they had initially budgeted.In 2018, the budget went up, but actual spending fell.After a scandal involving the diversion of cash transfer funds worth about US$ 31 million by public sector workers, the large donors (including both the British and Swedish governments) suspended funding.In 2019, donors spent nothing, and the government spent very little.In district after district, social welfare officers had almost no funds to distribute.The only work to occupy them was to update the registers of supposed beneficiaries. 7hroughout this period, the repeated announcement of budget increases fuelled the appearance of expansionand institutionalisationbut the reality was retrenchment.This apparent paradox resulted from attempts to balance pressures from competing policy coalitions.On the one hand, there was pressure on the government from some Members of Parliament (MPs) as well as international organisations and donors.According to one former Cabinet minister in Lungu's administration, "There was a lot of pressure on the government to expand cash transfers.The pressure was so high.People on the ground began to feel the impact so suddenly, [and poor] people that were not receiving [cash transfer benefits] started asking to be added to the programme.That is what informed the government to try and address the pressure.Ordinarily, it should not have expanded that way.But what happened is that we [MPs] were all coming to Parliament and giving feedback from our constituencies, we were all coming to Cabinet and giving feedback.There was a fear that the government would become unpopular if we didn't do something because we realized that the programme was very popular". 8ormer and current PF MPs shared similar sentiments.One stated that 'increasing the numbers of beneficiaries was partly negotiated for to allow politicians to say we are going to increase the numbers [and fulfill] a campaign promise'.9Despite their own personal ambivalence, PF MPs sought to appear to favour the expansion of social protection.
Defunding social protection was in part the result of the government's deepening fiscal crisis.Zambia's external debt increased more than fivefold between 2011 and 2019.The debt crisis was accompanied by a slowdown of economic growth.The Treasury categorised programmes into four tiers.Programmes in the first tier including emoluments and debt repayments were the highest priority, followed by the second tier including agriculture support and infrastructure.Cash transferswhich were considered 'discretionary spending' without any statutory basiswere in the lowest, fourth. 10As one PF Finance Minister explained, retrenching cash transfers had 'everything to do with the available strength and spread of resources': "When the financing gap widens, like when there is a drop in revenue as did happen in 2021 because of COVID, there was a drop in every form of taxable revenue.At the same time, there was a huge bill in terms of expenses.So, when revenue drops, what do you support?Is social cash transfer high enough on the agenda?This is where ideological motivations and elections come in.Those two tend to weigh heavily". 11ust as the expansion of external debt reflected the PF's prioritisation of expensive infrastructure development beyond what was consistent with the economic reality, so attempted austerity exposed the PF's priorities and the relative power of competing claimants for public expenditure.Patronagebased coalitions in agriculture and roads retained significant influence to ensure that their sectors were funded generously.These sectors were massively overfunded even as the government grappled to implement austerity.This demonstrated high levels of fiscal indiscipline and showed that there was strong support within the government for programmes that had potential electoral benefits and provided opportunities for corrupt kickbacks.

The influence of patronage-based coalitions, 2015-2020
International organisations had shaped the social protection agenda and thus the adoption of SCTs in Zambia and remained an important part of the coalition that appeared to have succeeded in institutionalising SCTs not only during 2013-14 (when the PF accelerated the expansion of coverage and apparently assumed a major share of the financial responsibility) but also in subsequent years (amidst repeated announcements of the further expansion of SCT coverage).But this institutionalisation turned out to be ephemeral.Some PF MPs might have pressed for the expansion of coverage, but the policy coalition supporting SCTs proved much less influential within the government than the rival patronagebased coalitions that advocated other forms of 'development' through public spending on agriculture and infrastructure.
A former PF Minister of Agriculture informed us that vested interests in the supply of fertiliser had lobbied the president to dismiss him after his ministry procured cheaper fertiliser than his predecessors had done by cutting out the middlemen and buying directly from the source.12Indeed, the minister was transferred to another ministry after one year.The patronage-based agricultural coalition was more supportive of expensive fertiliser contracts that provided more opportunities for kickbacks.Yet, the inflated costs of procuring fertiliser required the MoF to spend more money than allocated to the sector which inevitably meant cuts to other programmes.
PF MPs were supportive of infrastructure development in their constituencies because of a belief that physical infrastructure was more electorally rewarding than other forms of development. 13Lungu's 2016 campaign slogan was Sontapo epowabomba (Bemba for 'point at your works'), emphasising very visible infrastructure development (Siachiwena, 2021).Whilst actual expenditure on SCTs was far below the budgeted amounts (at least until 2020-21), actual expenditure on FISP, FRA, and roads was often more than 100 percent of what had been budgeteddespite the fact that these latter three programmes had much larger budget allocations.On average, over the period 2013-21, the overall 'budget execution rate' (taking into account both government and donor spending compared to the announced budgets) was 66 percent for SCTs and 79 percent for the FSP, compared to 123 percent for road infrastructure and an even higher percentage for FISP and the FRA (see Table 1).
Even in the throes of fiscal crisis, the government could have funded fully SCTs and FSP if it had not spent more than it had budgeted on other programmes, especially farm subsidies and roads.This prioritisation reflected the power of the coalitions demanding expenditure on agriculture and infrastructure relative to that of the coalition behind social protection.
The patronage-based coalitions behind agricultural subsidies and infrastructural expenditure had more powerful champions at the highest level of government.In neither case did the coalitions enjoy strong support from international organisations.Indeed, the World Bank repeatedly encouraged the government to reduce its expenditure on fertiliser subsidies, in part to free up resources for other programmes such as the SCTs.The Ministry of Agriculture was fiercely defensive of FISP, which provided subsidised farming inputsprimarily fertiliserto one million maize farmers across the country.The parastatal FRA brought maize from farmersoften at subsidised pricesto build national strategic reserves and to sell to millers (IAPRI, 2016).
Agricultural subsidies and infrastructural projects also accorded with the norms and values of most of the political elite, including MPs.Drawing on interviews with twenty-five MPs from the governing PF and opposition parties, mostly in 2016, Kuss and Gerstenberg (2023) show that MPs are familiar withand readily usethe technical discourses diffused (to them) by international organisations in support of SCTs.But many MPs 'remained ambivalent towards SCTs and retained their core beliefs about social justice and redistribution'.Most MPs continued to believe that the government's responsibility was to promote economic growth and hence opportunities for people to raise themselves out of poverty through hard work.They were wary of programmes that might encourage people to be lazy.This accords with our research.In 2017, we listened to a senior minister first speaking in favour of SCTs at a public eventorganised by an international organisationand then sharing her misgivings with us in a private meeting.Senior PF leaders in the Lungu government believed that poverty reduction was linked to food security.A former PF cabinet minister told us that 'food is number one on the [poverty reduction] agenda and that's why food must be affordable and accessible': 'Poverty reduction means allowing for more people to eat because when you eat, you have the strength to work, then you can put food on the table'. 14Crucially, PF leaders believed, food security required producing more food within Zambia and then ensuring that it was affordable as well as accessible to consumers.This narrative is one reason why the government poured funds into subsidising fertiliser for farmers to grow maize and then into buying that maize.
At least as importantly, however, agricultural and infrastructural programmes enjoyed strong support from ruling party leaders at provincial as well as national levels and from the private sector interests that benefitted from these public expenditures (including contractors).Some government officials and even ministers were sympathetic to the criticisms of FISP made by the World Bank and othersbut were unable to implement any reforms in the face of opposition from vested interests in the sector.A former Minister of Finance told us that "FISP is now folklore and has taken a life of its own.It is like a myth of a haunted house that no one wants to confront.But how many people have benefited from it?The people who benefit from the programme are the ones interested in the sustenance of the programme.The providers of fertilizer and seed are the ones who maintain the myth that FISP can cost the government [an election]". 15imilar dynamics also informed the motivation for increased expenditure on roads.There is even more evidence linking PF elites to road contracts.In 2019, Lungu implored China's AVIC International to ensure that it adhered to a policy of subcontracting 20% of its projects to local contractors (Ncube, 2019).More than US$ 4.5 million was given to local contractors because of this policy (Matambo & Onwuegbuchulam, 2021).Among the beneficiaries of these contracts were companies owned by the PF's Eastern Province Chairman, the PF's Chief Whip, and several individuals who contested the 2021 parliamentary and local government elections as PF candidates (Malunga, 2022).According to a former Zambian government policy analyst, spending on roads had at least two main benefits: 'One, you can see the road and it can win you votes.Two, it was a conduit to channel funds to the party.In that sense, roads were more of a priority than spending money on the poor'. 16Another interviewee indicated that spending on roads was difficult to contain because infrastructure development was 'a PF trademark and there was no stop to it' but also because 'to some extent, some people helped themselves [to kickbacks].' 17Social protection was retrenched under Lungu because the regime reverted to the kind of patronage politics associated with a political settlement based on competitive clientelism and the distribution of rents (Pruce & Hickey, 2019).The retrenchment of financing social protection was not just a temporary gap in delivery but a process of deinstitutionalisation that was closely associated with a growing fiscal crisis, pressure on the government to meet debt repayments, a focus on development priorities that gobbled resources at the expense of other sectors, and cuts to discretionary spending.The limited 'institutionalisation' of social protection under Sata gave way to 'de-institutionalisation' under Lungu.

Reversion to expansion, 2020-2021
In 2020-21, donors rapidly increased their expenditure on SCTs and the government followed suit, allowing for both the resumption of regular payments to the expanded number of beneficiaries and (it appears) the payment of arrear benefits (see Figure 2 and Table 1).The data show that 77% of the total SCT budget was executed in 2020.The government only met 28% of the cost and only spent one-third of their supposed budgetary commitment.Donors helped to fill the shortfall, spending much more than the budget had envisaged.In 2021, the government budgeted for a massive increase in its own expenditure and largely delivered on its promise: Between 2020 and 2021, actual government expenditure increased sevenfold (whilst actual donor expenditure increased by only 30%).The government contributed 66% of the total expenditure.This was the first time since 2013 that the SCT budget had been executed in full (indeed exceeded).
International donors were crucial to the rebound in SCT spending.Donors' concern with both the effects of the pandemic on poverty and the government's fiscal crisis led them to consider renewing financial support for SCTs.The World Bank assumed a leading role, establishing a multi-donor trust fund that pooled resources from donors (including the British, German, Irish, Swedish, and Swiss governments) 18 and establishing accounting systems and employing accountants to manage SCT funds. 19In June 2020, the WFP and UNICEF coordinated donor efforts to implement an emergency social cash transfer programme, which provided additional support to beneficiaries of the regular SCT and households that became vulnerable because of the pandemic. 20Furthermore, in 2020, the Zambian government accessed US$ 1.3 billion through the International Monetary Fund's Special Drawing Rights, which included conditionalities for the government to support social sectors in response to the COVID-19 pandemic. 21Also in 2020, the World Bank, through the International Development Association (IDA), disbursed an IDA loan to Zambia (negotiated the previous year) to support social sectors, including cash transfers. 22 Zambia's SCT programme would have remained retrenched in the absence of renewed donor support.Both COVID-19 and the 2021 elections presented 'policy windows' for the donor-led social policy coalition to push for reforms that stymied the deinstitutionalisation of cash transfers.In the absence of the pandemic and in the event that the PF had retained power in 2021, it is possible that the retrenchment of social protection would have continued further, barring macroeconomic stabilisation.The fact that social protection remains a Tier 4 programme that falls under discretionary spending, shows that the social policy coalition will inevitably be in a weaker position to persuade the MoF to disburse funding than the patronage-based coalitions whose programmes are in Tier 2 of the national budget.It is notable that donors played an important role in persuading the government to adopt pilot programmes in the 2000s but were unable to guarantee the institutionalisation of programmes.This is largely because the policy coalition that promoted adoption and expansion had limited influence over a PF administration with vested interests in the more lucrative and electorally popular agriculture and infrastructure programmes that had the support of strong patronage coalitions.

Discussion
The combination of the enthusiasm for social protection that is hegemonic in the global arena with scholars' use of process tracing to understand policy reform perhaps encourages us to understand reform as a unidirectional process that moves through clear phases, from (possible) adoption to (possible) institutionalisation.This is in part because research has focused on cases where social protection has expanded and even been institutionalised (in the sense used by Lavers and Hickey), even when progress towards institutionalisation has been slow or has stalled.Certainly, critics of SCTs worry that once they have been introduced, they are very difficult to remove.The pathways to institutionalisation may be path-dependent because of the usual feedback loops in terms of vested interests among recipients, the politicians competing for their votes, and the bureaucrats tasked with implementation.
The case of Zambia in the late 2010s reminds us that the pathways of reform can be far from linear, sometimes involving even reversals or deinstitutionalisation.Governments can cut back on expenditure, typically because they prioritise other areas of public spending and defund SCTs, especially during fiscal crises.In the case of Zambia, reform proceeded through an initial phase of slow expansion when pilot programmes were run and funded by donors to a phase of anticipated rapid expansion, which Lavers and Hickey assess entailed institutionalisation.As we have shown before, this institutionalisation was more apparent than real.As Lavers and Hickey predict, this apparent institutionalisation was not the product of the transnational policy coalition, which was taken by surprise by the government's announcement of expansion.Instead, it resulted from factional and ideological politics within the ruling party.Almost immediately, this apparent institutionalisation began to unravel when a new president representing a different faction of the ruling party assumed state power with the financial backing of elites with vested interests in the agriculture and infrastructure sectors.The post-2015 government budgeted for expansion but failed to commit the required funding.This reflected the weakness of the SCT-supporting policy coalition in the face of competing demands for spending on infrastructure and agriculture, which were seen as offering more lucrative opportunities for both corrupt enrichment and electoral support.Whilst the ruling party was certainly worried about elections, the leadership appears to have assessed that the route to re-election proceeded through the deployment of patronage to hold together a supportive, multi-regional coalition, rather than through pro-poor SCT programmes.This deinstitutionalisation was partially reversed in 2020-21, when foreign aid and loansthat both came with pro-poor conditionalitiesmade possible a pre-election spending spree, including on social protection.Contrary to Lavers and Hickey's general interpretation, the transnational policy coalition was an important player in this phase.
Zambia is not unique in Africa.There have been other cases where governments or donors have defunded social grant programmes for periods of time.In 2006, recipients of the Old Age Pension in Eswatini protested against delays in disbursements (Devereux & Cipryk, 2009).Payments under Ghana's Livelihood Empowerment Against Poverty (LEAP) programme were 'highly erratic' for some time after 2008 (Ragno et al., 2016, p. 163).In Malawi, SCT disbursements were suspended from 2010 to 2012 after donors withdrew, the government was slow to disburse benefits even in the one district where it had assumed responsibility; when payments resumed, they remained 'erratic' (Angeles et al., 2016, pp. 295-296;Siachiwena, 2023).In Zimbabwe, SCT payments were irregular in the early 2010s and faltered altogether during 2016running up to eight months latein the face of donor withdrawal and the government's fiscal crisis (Angeles et al., 2018, p. 15).In Kenya also, payments to beneficiaries were often delayed (Bosworth et al., 2016).In some of these cases, they may have been capacity problems.In most, however, governments cut back on or limited SCT expenditure because they prioritised other categories of expenditure.
Almost every country in Africa has a history of temporary programmes in response to natural disasters or war.Several countries introduced large programmes in response to COVID-19 but terminated them once the crisis had abated.COVID-19 also led to the suspension of some programmes, including massive school and preschool feeding schemes in South Africa (Gronbach, Megannon, & Seekings, 2022) and Botswana's workfare programme (Gronbach, Seekings, & Arthur, 2023).Many African governments announced emergency COVID-19 programmes but failed to implement them (Beazley et al., 2021a(Beazley et al., , 2021b)).By late 2023, social protection had generally reverted back to pre-pandemic levels (Gentilini, 2023).The retrenchment of non-emergency programmes in cases such as Zambia suggests that governments might view non-emergency programmes as similar to emergency programmes in that they can be turned off or cut back if and when they cease to be an immediate priority.
Retrenchment in cases like Zambia rarely requires legislative change because few programmes in Africa have been rooted in legislation.When programmes are rooted in legislation (or a constitution), the supportive policy coalition has additional mechanisms to protect their favoured programmesthrough civil society and the courtsas is clear in the case of South Africa (Black Sash, 2022).But even existing legislation does not preclude legislative change, as shown in cases outside of Africa (see e.g.Greve, 2020).Nor does legislation necessarily prevent governments reducing expenditures programmes by eroding the real value of benefits, amending the means test or imposing new conditionsall of which happened at different times in South Africa.
In Zambia, as in the Global North (see Greve, 2020), the retrenchment of social protection reflects a combination of fiscal pressures and ideology.Across much of Africa, political elites rarely see social protection as a sufficiently high priority that it is institutionalised.As the Zambia case shows, apparent institutionalisation can be driven by a faction within the ruling party without buy-in from the political elite generally.Across much of Africa, both African political elites tend to believe that individuals have a responsibility to work (if possible) whilst the government should not usurp the roles of the family (that is kin) and community in supporting family or community members in times of need.This conservative perspective is not unique to Africa, of course.Even more widespread is the view that governments should prioritise economic growth and development so as to expand opportunities for people to support themselves.The ideology of developmentalism is much stronger in Africa than in the advanced capitalist societies of the Global North.Even conservatives in Africa may believe that governments should provide for the poorat least conditionallyas an expression of the reciprocity and solidarity that underpins harmonious if unequal communities.Theyand othersare less likely to believe that the funding of social protection should be prioritised over more developmental funding.Social protection is widely seen as only 'scratching the surface of poverty'23 without addressing the structural and economic transformation that is key for both long-term national economic development and elites' own political survival.Even in the short term, voters might prefer 'developmental' programmesincluding subsidised inputs or sale prices for small farmersover social cash transfers.Research among MPs in Zambia (including within the then opposition UPND) shows that many remain unconvinced of the merits of social protection (Hallink, 2019;Kuss & Gerstenberg, 2023).
As the Zambian case also illustrates, fiscal pressures can lead to the retrenchment of social protection if other government programmes are prioritised.Political pressures might also lead to governing elites prioritising other programmes, especially if the other programmes offer better prospects for kickbacks or other financial benefit to the elites or key supporters.In Zambia, the government deprived social protection of funding in part because it poured money into infrastructure and fertiliser subsidies.This prioritisation conformed with their developmental preferences, offered opportunities for personal financial gain or political patronage, and shored up electoral support in potentially swing constituencies.The prioritisation of other spending and the deinstitutionalisation of social protection further reflected the weakness of the social policy coalition that was unable to win political support to legislate social protection,24 or let alone ring-fence SCT budgets to ensure that funds allocated to the programme could not be diverted to other sectors.

Conclusion
Between 2015 and 2020, social protection was deinstitutionalised in terms of the criteria of spending and de facto coverage.By these criteria, there was a reversion to institutionalisation in 2020-21: The Zambian government's disbursements for and the payment of SCTs picked up (see Figure 2).In 2020, actual government spending rose to 21 percent of the budgeted amount.In 2021, it rose further, to 68 percent.This, combined with improved donor funding, meant that SCTs were paid out fully in late 2020 and 2021.The imminence of an election in which the PF faced a severe risk of defeat (as polls suggested: Seekings & Siachiwena, 2021) and the onset of the COVID-19 pandemic provided a policy window for international organisations to increase their pressure on the government whilst also strengthening the electoral incentives on the government to resume SCT payments.This might suggest that competitive elections did play a major part in reinstitutionalising SCTs in Zambia.
This view needs to be qualified.First, even in 2020-21, the government failed to spend the amounts it had announced in the budget.Secondly, higher overall expenditure and expanded coverage were primarily driven by donors, in part because it coincided with the COVID-19 pandemic.Indeed, the emergency COVID-19 grant rolled out in urban areas in 2020 was not only funded by donors but also implemented in large part by donors.The expansion of coverage coincided with reduced state responsibility for financing and implementationthat is two of Lavers and Hickey's other criteria for institutionalisation.
Following elections and the change of government in 2021, public expenditure on SCTs has continued to rise (alongside continued donor support), and the government has resumed full responsibility for implementation, albeit with substantial funding through World Bank loans.This is at least partial reinstitutionalisation, in terms of the Lavers and Hickey's criteria.This does not mean, however, that the expansion of social protection is likely to resume, as if the retrenchment of 2015-20 had not happened.Nor does it mean that social protection is irreversible.The return to power of a coalition prioritising other areas of public expenditure remains a possibility.
The concept of 'institutionalisation' implies that programmes will endure over timebecause they are politically entrenched such that governments are constrained to retain (or even expand further) the programmes.The basis of institutionalisation is thus the 'coalition' of political actors with vested interests in the programme and hence the political costs to the government should it retrench the programme.Enshrining social protection in legislation might make it more difficult for governments to retrench programmes because this would require support in the legislature.Expanded coverage might make it difficult for governments to retrench programmes because citizensas voters or as potential protestershave an interest in defending the programmes.Expanded expenditure also results in bureaucrats and local community leaders having a vested interest in the programmes.Governments might also be trapped by their own discourses.
None of these interest groups appeared strong in the Zambian case.Voters might have expected that politicians would profess a commitment to poverty reduction but appear to have viewed social protection as a low priority.Neither bureaucrats nor community leaders appear to have objected to defunding.Criticism was limited even among opposition MPs, who were themselves more supportive of directly developmental programmes (Hallink, 2019;Kuss & Gerstenberg, 2023).Lungu and the PF might have been right to assess that the political costs of deprioritising social protection were modest.
The more general point is that the 'institutionalisation' of social requiresat least in countries without authoritarian regimesthat supporters of social protection spending retain power within the ruling party and that they or others outside of the party (including in international organisations) ensure that social protection remains high on the government's agenda and integral to the election strategy of the ruling party (or the opposition, in contexts where the opposition has a significant chance of winning).
The Zambian case illustrates that advocacy coalitionsincluding the transnational policy or advocacy coalition that lobby publicly as well as privately for the adoption and expansion of social protection programmesare just one kind of coalition.Faced with choices over expanding further (or, in the Lavers and Hickey's model, 'institutionalising') social protection, governments face competing coalitions of actors pushing for the prioritisation of their preferred programmes.In the Zambian case, a set of international organisations, allies in Zambian civil society, and sympathetic bureaucrats in the relevant ministry worked together to promote social protection, to persuade sceptics, and to recruit additional actors into the coalition.This was a concerted effort to promote specific policy reforms, based on overlapping normative beliefs.Opponents of social protection in Zambia in the late 2010s did not form a comparable coalition.In other countries, public expenditure on social protection might well be opposed by competing policy coalitions that prioritise instead public expenditure on (for example) public education or health care.This was not the case in Zambia.In the Zambian case, actors with overlapping political or economic interests in the prioritisation of infrastructure or fertiliser subsidies, but without any clearly shared normative beliefs, pushed opportunistically and behind the scenes for their preferred expenditures.The fact that they did not resemble a classic advocacy coalition did not make them any less effective.

Figure 2 .
Figure 2. Spending on social cash transfers.Source: Ministry of Finance Annual Economic Reports.Data for 2017 obtained from DFID's 2019 Annual Review of Social Protection. 4

Table 1 .
Spending on social protection and other social sectors (% of budget costs).Ministry of Finance Annual Economic Plans.The budget execution rate is the ratio of actual expenditure to budgeted expenditure. Source: