Sustaining existing social protection programmes during crises: What do we know? How can we know more?

Research on social assistance in crisis situations has focused predominantly on how social assistance can flex in response to rapid-onset emergencies such as floods or hurricanes and to slower-onset shocks such as drought. This paper identifies a substantial knowledge gap – namely, our understanding of the ways in which existing, government-led programmes can be sustained during crises to ensure that households that were already poor and vulnerable before a crisis continue to be supported. The limited literature available focuses on climate- and natural environment-related shocks – far less attention is paid to other crises. Conflict-affected situations are a major gap, although there is an emerging body of evidence of the ways in which focus on adapting delivery mechanisms has allowed social assistance and other social protection programmes to be sustained throughout the Covid-19 pandemic. The paper concludes that a better understanding of when, where and how existing programmes can be sustained during situations of violent conflict will help to ensure that poor and vulnerable households can be supported – either through government programmes or by enabling robust diagnosis of when efforts to sustaining existing programmes will be inadequate and an additional, external responses are required.


Introduction
Over the past decade, a combination of increasing use of cash and voucher assistance in emergencies, and an increased focus on how social protection systems and programmes can flex in response to shocks, has brought about sustained interest in adaptive and shock-responsive social protection (SRSP).In numerous countries, governments and their development partners are working to strengthen the capacity of social protection to reduce exposure of poor and vulnerable households to shocks and to support them when disasters occur.At global level, there are commitments among humanitarian and development agencies and the role of social protection in contributing to disaster risk management is recognised in the Sustainable Development Goals.
In this paper, we examine evidence about the extent to which social protection programmes can sustain existing delivery to existing caseloads during a crisis.In the Better Assistance in Crises (BASIC) Research proposal, we hypothesised that this was an evidence gap.This paper begins by exploring existing literature and evidence to assess whether this is the case and outlines a possible conceptual framing for future research.
Until early 2020, when the Covid-19 pandemic created seismic shifts in approaches to social protection and humanitarian cash and voucher programming, the SRSP and adaptive social protection (ASP) agendas had three main dimensions.First, debates on SRSP, ASP and links between humanitarian and social assistance had focused on expanding or adjusting coverage and on building new systems and there was less of a focus in the literature on how programmes were sustained and on the resilience of systems in the face of shocks.Second, agendas focused primarily on disasters related to climate and the natural environment -notably earthquakes, typhoons/hurricanes, floods, droughts and fires.Third, analyses primarily considered situations where non-state humanitarian and development actors engaged with national governments to address shocks.These agendas brought substantial gains, especially improvements in policy and legal frameworks and programme guidelines that, in turn, established mechanisms by which social protection could be scaled or adapted to tackle shocks and stresses.
Notwithstanding this progress, it seems that we have not come very far in addressing issues raised in 2007: 'How can partners build sustainability in [to] the design and delivery of programmes?'And what is the 'meaning of the term sustainability in relation to social protection in fragile state contexts ' (Harvey et al. 2007: 3 and 5).The Covid-19 pandemic has broadened the focus of our attention towards thinking about systems resilience (see below and Beazley et al. 2021) in crisis situations.But it is unclear how much learning from the Covid-19 pandemic is applicable to other types of crises.This paper seeks to explore if and how thinking beyond the current focus of SRSP and ASP has the potential to strengthen the delivery of social protection in Box 1.1: Defining adaptive and shock-responsive social protection 'Adaptive social protection (ASP) -social protection that helps build the resilience of poor and vulnerable households by investing in their capacity to prepare for, cope with and adapt to shocks, ensuring that they do not fall (deeper) into poverty' (World Bank 2020).
Shock-responsive social protection (SRSP) -'focuses on shocks that affect a large proportion of the population simultaneously (covariate shocks).It encompasses the adaptation of routine social protection programmes and systems to cope with changes in context and demand following large-scale shocks.This can be ex ante by building shock-responsive systems, plans and partnerships in advance of a shock to better prepare for emergency response; or ex post, to support households once the shock has occurred' (O'Brien et al. 2018: 7).The dominant tools of SRSP are vertical expansion (topping up what existing recipients get), horizontal expansion (including additional recipients), piggybacking on delivery systems, alignment (e.g. with existing benefit levels) and design tweaks (e.g.removing conditions on receiving benefits during periods of instability).crisis situations.The paper suggests that there is a missing piece in the current agenda: namely, to what extent can and do existing government social protection programmes sustain delivery of existing commitments -or possibly even restart delivery -to existing caseloads in crisis situations?
The paper argues that this could prove an important missing link, particularly in countries where maintaining their delivery of existing systems could prove effective and support poor and vulnerable households.The question posed here is not, therefore, about if and how governments and their partners are able to scale up or adapt programmes in the face of crises (whether through ex-ante or ex-post responses), but about how far existing provision of social protection continues and is sustained during a crisis.We refer to this in the paper as the 'sustain question' and have divided it into the following sub-questions: • Is the existing social protection caseload (i.e. the number of recipient households or individuals) maintained or does it reduce during a crisis?• Are the timing, size and frequency of transfer delivery sustained (e.g.where payments are made quarterly, does this continue and follow the planned timetable for disbursement)?• What features of social protection systems and programmes enable programme delivery to be sustained and the system itself to be resilient (or, conversely, disable the system)?• What revisions to programmes and delivery systems enable delivery to be sustained during and after crises?
The hypothesis of this paper is that this is an important yet poorly understood element of social protection in crisis situations but is the subject of far less applied empirical policy research compared to the preoccupation with elements of SRSP such as vertical and horizontal expansion.As Clarke and Dercon (2016) intimate, asking how existing programmes are sustained rather than how we might scale up is the 'dull' part of the disasters agenda.At the same time, the question has come onto the research agenda in the context of Covid-19.For many governments and international agencies, a focus on how to adapt existing programmes to ensure sustained support to existing recipients became a key part of the Covid-19 response through social protection.1This paper therefore investigates the extent of the gap and assesses what might be gained from a better understanding of when, where and how existing social protection systems and programmes can be sustained.
The paper begins by describing the limited results of two literature searches -the second of which was carried out following consultation and feedback on the first draft of this paper.In terms of literature, it notes that, apart from some evidence on sustaining social protection in situations of climate-related crisis, there is very little explicit and direct analysis.The evidence provides insights into the sorts of concepts and analytical frameworks that could prove useful in trying to better understand if and how existing programmes are sustained.The paper argues that improving our understanding could fill an important knowledge gap, particularly in countries with existing systems where maintaining delivery of those systems could be effective in supporting poor and vulnerable households, and where there are trade-offs between scaling up programmes to new beneficiaries and maintaining delivery to existing caseloads.
2. What research and evidence exist that focus on how existing social protection systems and programmes can be sustained in crisis situations?The Annexe describes the literature review process in more detail, highlighting the sources used for searches and combinations of search terms used; and indicates some of the problems found in identifying sources.Overall, there is not an established lexicon or systematically used language or terminology for this question.
In the initial phase of searching, search terms produced a very large number of results in Google and Google Scholar, but screening of the first 20 results for each combination of search terms found little relevant literature.
Even after a wider range of search terms was introduced following feedback and useful comments from peer reviews, orthodox literature searches produced limited results.Wider snowball techniques (e.g. for the research outputs found), tracing citations forwards (to later documents that cite the source) and backwards (reviewing the reference list in the source document) offered limited further perspectives and evidence on the question of sustaining delivery became increasingly opaque.
Overall, the literature search suggests that very few research outputs focus explicitly on the question covered in this paper.At the same time it is possible that there are findings in thousands of papers that are, implicitly and indirectly at least, faintly relevant.However, it was not the scope of this review to carry out a full, formal Systematic Review but rather to provide an exploratory analysis.As a result, the findings presented here focus on (1) research found in the literature searches that explicitly cover the question at hand; (2) research citing or cited by the literature found through searches; and (3) further research and analysis that provide material that is indirectly relevant to the sustain question.
The main findings from the literature are described in three case studies -East Africa, Nepal and Syria -and a wider summary of lessons and issues, including further evidence from Yemen and Iraq.

What other sources of data are available and were used?
Beyond the literature searches, monitoring and evaluation outputs of government programmes (where available) were assessed to identify evidence on the changing caseload and distribution of social protection transfers in conflict-affected settings.The primary source of these data was from the monitoring of donor Beyond programme monitoring data, a more exploratory approach was developed.This drew heavily on findings from previous research by the Secure Livelihoods Research Consortium to try to capture wider lessons about institutional arrangements and governance that successfully deliver social protection and basic services in fragile and conflict-affected settings (FCAS) and identify possible new ways for thinking about how programmes are sustained that go beyond technical elements and features.

What concepts and frameworks are used?
Overall, the terms used currently in social protection that could focus on how existing programmes can be sustained are, in fact, deployed in a rather different way.Across ASP, SRSP, shock-sensitive social protection and risk-informed social protection, the units of interest are individuals or households that receive support rather than the capacity of the system or programme to sustain delivery.They also all focus on providing additional, adapted or different support in response to current or future shocks.They do not focus on how to deliver existing commitments when the context becomes more difficult operationally.These frameworks provide a useful conceptual entry point for thinking more about how existing delivery is sustained; for example, through the emphasis in many of the frameworks on the systems that underpin effective delivery rather than individual programmes alone (Barca 2020).
SRSP focuses on ex-ante and ex-post actions that, together, allow social protection to be effective in providing support to households that are exposed to the deleterious impacts of various shocks.Until early 2020, it was almost exclusively applied to climate-related shocks but there is an emerging, substantial focus on SRSP for Covid-19 responses.SRSP has also tended to focus on flexing and scaling programmes and systems rather than on sustaining them.However, this perspective is starting to shift.
O'Brien (2020: 3) focuses on business continuity, flagging up the importance of contingency planning for delivery: Situations such as earthquakes, floods or conflict can disrupt services: the shock itself can pose problems for business continuity in social protection.Staff may not be able to get to work, payment infrastructure may be damaged, or funds may be diverted to other priorities.Recipients may be displaced internally or abroad, or be unable to reach the place where they usually get assistance, or may lose their identity documents.To maintain services in these difficult circumstances, implementers should be ready with contingency plans and may need to adjust their regular programmes.
(O'Brien 2020: 3) As part of its analysis of SRSP in response to Covid-19, the Maintains programme, an operational research programme led by Oxford Policy Management, focused on the sustain question.The Maintains framework split its SRSP study into: systems resilience, which was about business continuity; adaptation, which was about the more orthodox SRSP typology of vertical expansion, horizontal expansion, piggybacking, etc.'; and humanitarian assistance, which was institutionally independent of national government initiatives (Beazley et al. 2020(Beazley et al. , 2021)).
The Maintains project defined systems resilience as 'measures undertaken to enable the business continuity of social protection programmes and systems' (Rodolfo et al. 2020: 4).Systems resilience and business continuity overlap substantially with the 'sustain question' in this paper.It is similar in that it focuses on elements of the system that enable delivery to be sustained during crises, but different in that it is still oriented towards adapting/changing/making small tweaks to systems and programmes.Overall, it appears to focus more on how programmes might change, albeit in small ways and temporarily, than on programme and system features that are durable, robust and appropriate in both regular and crisis situations.
The Maintains system was developed specifically to analyse SRSP responses to the Covid-19 pandemic.As a result, many of the features of resilience and continuity relate to operational adaptations to ensure that social protection programmes 'do no harm'; in other words, they do not contribute, through their operations, to contagion or infection.Features include: hygiene protocols and social-distancing measures; consolidation of monthly payments into a single payment every three months; integrating mobile phone use for remote programming; and the suspension of conditionalities, especially public works requirements and school attendance.Some of these elements are predominantly about 'do no harm': how social protection can be delivered so as not to worsen the pandemic (e.g.hygiene and social distancing, consolidation of payments, suspension of conditionalities).But one feature, the use of mobile phones for remote programming, is more directly about continued delivery because of reduced physical access to areas where programmes are delivered.This latter feature is of greater interest than the others to BASIC Research because it is likely to be important for other crisis situations beyond Covid-19, such as conflict-and climate-related shocks where access to those receiving social assistance may be constrained.
SRSP is broader than shock response, but it still focuses on individuals and households rather than on the sensitivity of the system.In Malawi, the main location where it has been applied, it has four components: (1) reducing poverty and food insecurity; (2) meeting the chronic and seasonal needs of the poorest; (3) building resilience to shocks and climate change; and (4) supporting early, efficient and effective emergency action when needed.Shock-sensitive social protection systems prioritise core social protection objectives (elements 1-3), while also ensuring that any progress made is protected from predictable seasonal food insecurity and from frequent shocks (element 4) (Holmes et al. 2017).
The focus of risk-informed social protection (e.g.see FAO 2016) is on the risks that vulnerable households face and ensuring that social protection is designed to respond to them.The actions proposed by the Food and Agriculture Organization focus on scaling-up, in common with SRSP, rather than the risks of non-delivery by the social protection system itself in specific crisis situations.
ASP goes further still and generally focuses on the adaptive capacity of recipient individuals or households, rather than on the extent to which systems and programmes can adapt to maintain delivery of social protection when a crisis occurs.Where it discusses systems, it tends to focus on changes to programmes that allow longer-term transformations in livelihoods and asset portfolios rather than maintaining programme delivery.So, ASP programmes focus predominantly on if and how programmes contribute to households being better able to buffer themselves against the impact of climate change, rather than on whether, during a climate shock, social protection continues to be delivered or whether safety nets are present for distressed households to fall back on.
Several studies on social protection and climate change draw on the Building Resilience and Adaptation to Climate Extremes and Disasters (BRACED) programme's '3As' conceptual model of adaptive, anticipatory and absorptive capacity for responding to climate change (Bahadur et al. 2015).Absorptive capacity, with its focus on 'functional persistence', may provide a useful lens for thinking about whether social protection delivery systems continue to function in crisis situations.Although adaptive capacity has often been used in social protection research to consider how individuals and households might shift their activities to help avoid shocks and stresses, it is of interest here because its focus on systems for deliberate decision-making might be applied at an institutional or organisational level rather than an individual/household one.Elements of anticipatory capacity are already strongly represented in SRSP work (e.g.establishing social registries and risk-financing mechanisms to allow horizontal and vertical expansions of programmes following a shock) but are less relevant to the question of whether existing provision of social protection can be sustained.Although anticipatory capacity implies ex-ante actions, it remains focused on scaling and flexing rather than sustaining existing programmes.Although principally applied to climate change, the idea of absorptive capacity might equally be considered where a crisis is related to conflict, or Covid-19, or another form of shock entirely.
Elements of absorptive capacity are reflected in Hu et al. (2010), who differentiate between organisational disaster preparedness, institutional robustness and operational resilience.Whereas institutional robustness refers to the extent to which existing systems and procedures function normally during a crisis, operational resilience is about the ability of staff or actors to take on unfamiliar tasks, improvise and find ways to reuse and adapt remaining and surviving resources.
Resilience is a term that is commonly used in discussions about adaptive social protection, but again tends to be used to refer to social assistance recipients and the extent to which they can buffer themselves against the negative impacts of shocks and stresses.However, a systems view of resilience -for example, the United Nations (UN) Office for Disaster Risk Reduction definition of resilience as 'the ability of a system and its component parts to anticipate, absorb, accommodate, or recover from the effects of a hazardous event… including through the ensuring the preservation, restoration or improvement of its essential basic structures and functions' -could provide a useful framing for thinking about how social protection delivery is, or could be, sustained in crises and to move away from a very individualised focus on recipients to a broader view of social protection systems (Ulrichs, Slater and Costella 2019).It is tempting, therefore, to refer to the issues covered in this paper as shock-or crisis-resilient social protection.However, the term 'resilience' has been overwhelmingly deployed 2 with a focus on resilience to climate change within the social protection and humanitarian space in the past decade, so using it here risks confusing rather than illuminating our understanding.Talking about 'crisis-resilient social protection systems' might be one way forward.
Studies of governance systems in FCAS, mainly of donor programmes, provide a different set of insights, particularly those from adaptive programming: Adaptive programming responds to several key understandings about development: that development actors may not be able to fully grasp the circumstances on the ground until engaged; that these circumstances often change in rapid, complex and unpredictable ways; and finally that the complexity of development processes means actors rarely know at the outset how to achieve a given development outcome -even if there is agreement on the outcome of interest.(Valters, Cummings and Nixon 2016: 5) The definition itself points to challenges for those working on social protection.The acceptance and embrace of uncertainty, complexity and ignorance can be deeply uncomfortable for those designing and operating social protection programmes.It runs counter to the notably linear processes of social protection programme design: identify needs and beneficiaries; and provide certainty and reliability in support.There are further challenges.For example, following the Wenchuan earthquake in China in 2008, while the institutional robustness described by Hu et al. (2010) depended heavily on staff knowing and following existing protocols and procedures to maintain social protection delivery, adaptive programming depends on a shift from 'following the map' to 'using the compass' (Christie and Green 2019).For BASIC Research, the question translates to: 'What is the right balance between flexing and adapting, on the one hand, and maintaining routine and reliable delivery in the long term on the other?'Despite these challenges, adaptive programming might prove useful where the objective is to better understand how programming is sustained in crisis situations, not least because adaptive programming was 'born from a recognition that development programs need to adapt more effectively to the (frequently shifting) contexts in which they operate' (Nixon 2019).The distinction between adaptive delivery, adaptive programming and adaptive governance provides entry points for research to better understand where in the social protection supply chain the main features that enable (or disable) sustained programme delivery are found and what can be done to strengthen (or address) them.

Box 2.1: Adaptive, anticipatory and absorptive capacity
Adaptive capacity is the ability of social systems to adapt to multiple, long-term and future climate change risks, and to learn and adjust after a disaster.It is the capacity to take deliberate and planned decisions to achieve a desired state even when conditions have changed or are about to change because a system has the ability to 'react to evolving hazards and stresses so as to reduce the likelihood of the occurrence and/or the magnitude of harmful outcomes resulting from climate-related hazards' (Malone 2009: 6).
Anticipatory capacity is the ability of social systems to anticipate and reduce the impact of climate variability and extremes through preparedness and planning.Anticipatory capacity is seen in proactive action before a foreseen event to avoid upheaval, either by avoiding, reducing exposure to or minimising vulnerability to specific hazards (Kellett and Peters 2014).
Absorptive capacity is the ability of social systems to absorb and cope with the impacts of climate variability and extremes.It refers to the ability of social systems, using available skills and resources, to face and manage adverse conditions, emergencies or disasters (Hudner and Kurtz 2002).While anticipatory capacity comes into play before a shock or stress, absorptive capacity is exercised during and after a disturbance has occurred to reduce the immediate impact on people's livelihoods and basic needs.In conceptual terms, it is concerned principally with 'functional persistence' -that is, the ability of a system to buffer, bear and endure the impacts of climate extremes in the short term and avoid collapse (death, debilitation and destruction of livelihoods) (Blaikie et al. 2003;Folke 2006;Béné 2012).
Source: reproduced from Badahur et al. ( 2015) It may be particularly useful for understanding which combinations of stakeholders work best in funding, designing and delivering social protection programmes in FCAS.As Christie and Green (2019) note: the politics of uncertainty and risk in [FCAS] raise profound challenges to traditional donor approaches to partnership, which revolve around the 'holy trinity' of state, private sector and civil society organisations.To differing degrees, all three of these and the links between them may be weak or absent in [FCAS].This partial vacuum highlights other forms of 'public authority' (traditional chiefs, armed organisations, faith groups), with whom donors are much less accustomed to working, but who have always formed part of the socio-political landscape in [FCAS]'. 3pplying these various frameworks to the sustain question reveals several gaps.For resilience frameworks, the focus is predominantly on individuals and households, with a sharp tapering of attention on organisations and systems.For social protection frameworks, there is growing attention on systems, organisations and programmes, which in turn support individuals and households; but this focuses more on flexing and scaling than continuity and sustaining existing delivery.Putting resilience and social protection frameworks together leaves a substantial gap for the sustain question.Drawing on systems-based definitions of resilience and ideas of 'functional persistence', 'institutional robustness' and 'operational resilience' appear to offer the greatest potential for exploring when, where and how existing social protection programmes can be maintained in crises.The distinction between levels of decision-making and operations from adaptive programming -namely adaptive delivery, programming and governance -also provides an entry point for breaking down social protection systems into component parts and assessing which features enable sustained and effective provision.

Box 2.2: Adaptive delivery, programming and governance
Adaptive delivery is the daily work undertaken by a delivery team on the ground with their fingers on the social, political and economic pulse of the world in which they operate.Instead of implementing 'The Technical Plan,' they think politically, opportunistically and on their feet, continuously navigating through a fog of ever-changing conditions, many moving parts and players, ambiguity and uncertainty and towards political ends (shifts in power imbalances).
Adaptive programming is a slower and more structured process, usually in the hands of the senior team in the programme office and informed by frontline staff, and the patterns and players which they spot or emerge from delivery, as well as pressure from donors to deliver results.
Adaptive governance normally resides with officer(s) in the donor agency [or government department] responsible for funding the programme and following its progress.They must both manage upwards, coping with pressure for results, reporting and shifting strategic priorities; and downwards, ensuring that the programme accounts for how it is spending donor money, but that it also retains the freedom of strategic manoeuvre that lies at the heart of adaptive approaches.
The relationship between adaptive delivery, programming and governance is constantly evolving, and can sometimes be fraught.The basic currency of adaptive approaches is trust between the various players and tiers involved and confidence that the plan will remain realistic even as it changes.Equilibrium can be disrupted by any number of factors: a political or other exogenous shock; a change of leadership or policy; or a significant failure.Any of these events heightens the perception of risk and can trigger a reversion to more command-and-control approaches, which rapidly shrinks the space for innovation, improvisation and 'dancing with the system'.

Case studies
In the absence of a strong literature identifying examples of how existing social protection programmes are sustained in crisis situations, a small number of brainstorming sessions and consultations identified simple case studies that might help bring the conceptual discussion above to life.These examples have not been selected because they provide significant new insights or textbook examples or because they are particularly current; rather, they represent the author's reflections based on research and analysis in various countries.

Adapting programmes to sustain delivery of social protection during the Covid-19 pandemic in East Africa
A developmental evaluation of WFP's work on social protection in response to the Covid-19 pandemic highlighted the importance of sustaining delivery of existing programming alongside any efforts to scale up transfers or coverage.The evaluation noted that, while there is a justified focus on addressing the increased needs resulting from Covid-19, the challenges of meeting existing programming commitments for those who were already vulnerable and food insecure before Covid-19 are equally important.
Programme expansion and the development of new programmes dominated global attention on social protection for the Covid-19 response in 2020.However, despite substantial noise about vertical and horizontal expansions, both globally and regionally, much of the work that was done focused on sustaining existing delivery (Rodolof et al. 2021); but -initially, at least -this received far less attention.WFP's immediate focus in 2020 in East Africa was on adaptations required to existing operations to ensure all those currently receiving support could continue to do so, despite new access, monitoring and funding constraints.
This experience provides lessons about the importance and challenges of maintaining and sustaining existing social protection programmes.Households that were vulnerable, poor or food insecure before Covid-19 continued to struggle during the pandemic -and in many cases their vulnerability, poverty and food insecurity intensified.WFP in the region focused on maintaining and sustaining existing social protection support, working to mobilise additional funds for expansion only where existing caseloads were supported.This included adaptations to social protection and broader safety net programming and delivery in response to the unique circumstances and demands of Covid-19.

Box 3.1: Examples of programme adaptation
• 'Covid-proofing' operations to ensure that targeting, registration, training, asset creation activities and distributions did not become sites for infection to spread.This included: introducing new standard operating procedures (e.g.social distancing in queues); providing hygiene and sanitation equipment for beneficiaries; use of personal protective equipment by staff; temporarily relaxing levels of verification (especially the collection of biometric data during registration); and distributing larger transfers less regularly (e.g.monthly payments made bi-monthly).• Switching from providing meals at school to delivering 'school feeding at home' (WFP 2020) when schools were closed; and establishing new standard operating procedures and guidance for those delivering meals in countries such as Burundi where schools remained open.• Adapting asset creation and livelihoods activities to make them more appropriate in urban contexts.A good example is South Sudan, where WFP's urban safety net shifted support from livelihoods and income-generating activities focused on delivering personal services -such as hairdressing -to livelihoods that could take place in the open air and provide produce for market sale or own consumption, such as home gardening.
Source: Adapted from Slater et al. (2021) 3.2.Delivering on social security allowances in earthquake-and flood-affected Nepal Experiences in Nepal following the 2015 earthquake, and in subsequent years where parts of the Terai region experienced serious flooding, highlight some of the key challenges associated with sustaining social protection delivery in a crisis.
Social assistance in Nepal comprises various cash transfer programmes targeted at the elderly, people with disability, children under five in vulnerable households, endangered ethnic groups, and widows and single women.Collectively, they are called social security allowances (SSAs).Various sources suggest that there were delays in disbursing SSAs following the earthquake but the impact of these delays was mixed.For example, The Himalayan Times reported that 'as staff of local governments remained busy in collecting data of earthquake losses and relief distribution works to victims, VDCs [village development committees] and municipalities of Kavre district failed to distribute social security allowance[s]' (Parajuli 2015).In contrast, Holmes et al. (2018: 13) found that 'while there were delays in receiving the allowance, respondents had relatively little to say about the effect of earthquakes on the allowances'.
Arguably, in the hours, days and weeks immediately post-disaster, delays to social protection may not be especially serious, but in the months following the disaster they become increasingly so.Evidence from districts in Nepal affected by floods in 2017 highlight this challenge.In the period immediately following the floods, when affected households moved from their houses to higher level roads and received (untargeted) support in the form of water, blankets, tarpaulins for shelter and food, not receiving SSAs had little impact on the usual caseload of beneficiaries (Holmes et al. 2019).However, some days or weeks later, as the flood waters receded and people moved back to their houses, a more targeted distribution of emergency relief began.At this point, delays to SSAs began to bite as many SSA beneficiaries found that they were taken off the beneficiary list for relief (ibid.;Slater et al. 2018).The second phase of emergency relief used criteria for targeting based on household losses rather than impact and, among those targeted, households receiving SSAs were assumed to be less in need because of their SSA income.This was despite most SSA programmes supporting people who were labour-constrained; for them, the challenges during an environmental shock were even greater because of difficulties of travelling to collect SSAs or other support.
Both the earthquake and floods in Nepal highlight a causation challenge associated with assessing sustained delivery of social assistance in crisis situations.Even without disasters, SSAs in 2015-18 were often not delivered on time.2017).The result is that it is not always easy to tell whether poor disbursement performance is attributable to disasters or to a more generalised problem of weak administration.

Maintaining the National Social Aid Fund in Syria
Syria's Five Year Plan (2006-10) committed the government to several operational objectives related to social security enhancement including: • Ensuring the financial sustainability of social insurance; • Improving the effectiveness and efficiency of social insurance; • Extending the coverage of social insurance to larger groups of the population; and • Strengthening social safety nets.
Social safety nets in Syria in the 2000s were weak, made up predominantly of an essentially untargeted subsidy system that consumed some 13 per cent of GDP: Coupons for the purchase of rice, tea and sugar were distributed twice a month to everyone who had a family book, irrespective of economic status.This targeting strategy led to severe exclusion errors.For example, it excluded unmarried and divorced women from receiving social assistance.(Harsch 2019: 22) In 2007, the National Social Aid Fund (NSAF) was established with a budget made up of savings from the reform of subsidies.The NSAF had three objectives, to: reduce poverty through direct cash transfers for households below the poverty line; promote investment into human capital and prevent the intergenerational transmission of poverty; and empower the poor through providing opportunities to escape dependency.Substantial process was made in establishing and rolling out the NSAF.By 2011 it had: • Formally established the NSAF with decrees and bylaws covering the legislative, managerial and administrative structure and financial and accounting systems.• Strengthened capacities of 950 staff including social workers, supervisors and administration staff in headquarters and 63 district centres.• Identified, developed and tested targeting policies, mechanisms and tools for vulnerable ultra-poor households.• Implemented proxy means testing using survey data to estimate a caseload of 560,000 households, carried out field verification and registration of 120,000 households.• Extended targeting coverage to 279 poor villages.
• Conducted two disbursements of benefits to NSAF beneficiaries in 2011 to 439,000 households.
The progress made demonstrates substantial capacity in Syria to establish and deliver social assistance.However, in 2011 the Arab spring brought sweeping change, citizen-led uprisings and violent conflict to the region.At the same time, after several years of orientation and communication efforts, the government in early 2011 also began to reduce fuel subsidies, resulting in a rapid and steep increase in prices, with negative impacts on heating, agricultural production and transport (ibid.).Public protests ensued and the government lowered prices in March 2011.It cancelled the third planned NSAF disbursement to beneficiaries (Abdul Aziz and Hassani 2012).
The extent to which the NSAF has been sustained is difficult to unravel.Since 2011 and the outbreak of conflict in Syria, delivery of the NSAF has continued but data on caseload and disbursement are limited.Machado et al. (2018) state that the NSAF was discontinued after the 2011 uprisings but it has received a budget allocation annually.However, the budget is very low in USD terms.Notwithstanding difficulties interpreting between official and parallel currency exchange rates, it seems unlikely that NSAF is delivering transfers to anything like the numbers of 2011.Rampant inflation and market shortages will also be limiting the real value of transfers; in 2015, the price of goods rose by more than 20 per cent (ibid.).Factors behind the reduction of transfer value in real terms include losses in tax and oil revenues, the effects of sanctions on international trade, growth of the informal economy and weak administrative tax collection capacity (Harsch 2019: 44).At the same time, demand for support from the NSAF has increased exponentially, with the International Monetary Fund (IMF) estimating that poverty rates increased from around 12 per cent in 2007 to around 83 per cent in 2014.There are also insights from Syria about how far displaced households can continue to access forms of social insurance.Interestingly, Harsch (ibid.: 45) finds that 'overall pension coverage has actually increased over the last few years.This is explained by the fact that pensions are seen as a relatively reliable channel to provide income security in the instable times of conflict'.Demand for pensions, specifically survivors' pensions, is likely to increase, given the large numbers of casualties due to the conflict.In terms of access, the evidence is mixed.On paper, there are mechanisms to ensure accessibility -even for those in areas of conflict and those who have cross international border: Officially, pensions have continuously been paid out across the country, including areas outside the Government's control.In principle, they can be withdrawn with special cards from certain Government banks and automated teller machines (ATMs) in such regions.However, other sources have reported that this is not always the case and that pensions for some parts of the population have been withheld, with significant variations depending on the area.(ibid.: 46) Similarly, it is possible for Syrian refugees in other countries to stay enrolled in social insurance and receive pensions and other benefits.On paper they can do this by authorising a relative who remains in Syria to claim the pension on their behalf.Employees paying into the programme can take unpaid leave for up to four years but stay enrolled in the scheme.With such large numbers of Syrian refugees having now lived outside the country for more than four years, access to the scheme will remain protected for fewer and fewer people (ibid.).For existing pensioners receiving payments, inflation and currency devaluation substantially limit the purchasing power of pensions, despite several pension increases in the past decade.

Lessons from the literature
Beyond these specific case studies, several broader findings emerge from the literature reviewed.
There is very limited direct treatment of the question of if and how delivery of existing social protection programmes is sustained in crisis situations.Hu et al. (2010: 107) find that although '[t]here are a number of studies examining social protection in the context of disasters… [w]hat is missing in this body of research, however, are studies of the effects of disasters on the social protection system itself.'They describe disasters as a 'double blow' to social protection systems and programmes, noting both direct damage to human and physical resources, and increased demand for new or expanded social protection.
While the shock-responsive social protection agenda comprehensively addresses the second of these effects, the first receives far less attention.Work on shock-sensitive social protection is broader, focusing not only on responding to rapid-onset acute shocks but also on regular seasonal shocks (Holmes et al. 2017).However, it pays little attention to the sustained operations of social protection systems and programmes.A broader analytical framework than is provided for in ASP, SRSP and shock-sensitive social protection is required.
Until the Covid-19 pandemic, the existing literature focused predominantly on climate-and natural environment-related shocks.Even in these cases, there is a focus on relatively rapid-onset or seasonal shocks in which the immediate threat to people's lives recedes quickly.Examples include assessments of obstacles to delivering social protection where roads are flooded or people are displaced for several days, or because communications are down.In those cases, it is often assumed that barriers to delivery are (usually) addressed within a matter of hours, days or perhaps a few weeks.In the main, work on climate shocks also tends to focus on horizontal and vertical programme expansion (i.e.top-ups to existing beneficiaries or temporary expansion of the beneficiary list), with little or no attention paid to sustaining existing payments.
Even with drought-related shocks, where the resulting shortfalls in food consumption and income can be felt in the long term, scaled-up responses are temporary and rarely sustained for more than a few months.The 'early action' agenda highlights the importance of support to households that are vulnerable to climate change shocks and how this can prevent loss of life, assets and livelihoods; but this focuses on beneficiaries, rather than on preventing loss of capacity in delivery systems to operate (Costella et al. 2017).
Literature on conflict contexts is far more limited and focuses (in order of prevalence) on: (1) supporting new rather than existing caseloads, especially internally displaced persons and refugees; (2) how social protection can be used post-conflict (e.g. to support return and rehabilitation); and (3) social assistance to address conflict (e.g.demobilisation, disarmament and reintegration programmes).References to how social assistance can be maintained during conflict offer patchy rather than comprehensive coverage of issues and places, and appear not to have been the subject of substantial systematic or comprehensive research.One theory to explain this gap is the existence of enduring views that either government is largely absent; that government social assistance programmes are nascent, fledgling or weak; or, where governments are seen as predatory, that they depend heavily on donor funding, which is not renewed due to concerns about fiduciary risks and corruption.These perspectives began in 2004 ('The state itself is often weak in such circumstances and its capacity to provide services drastically curtailed or non-existent', Darcy 2004: 6) and have endured until at least 2018 ('even if government-led social protection programmes exist, for example national cash transfer programmes, fragility and conflict affects and can destroy their ability to function', Cooper 2018: 2).
The problem is not that these specific statements are inaccurate but that they are subsequently deployed as a substitute for objective and specific assessments at individual country level.More recently, these generalisations have been challenged.First, because countries where social protection systems are relatively strong and established still face the challenge of sustaining delivery; nascent government capacity is not necessarily the binding constraint in conflict-affected settings.In Ukraine, the UN estimates that in 2019 some 600,000 pensioners in non-government-controlled areas (NGCAs) had lost access to their pensions due to movement restrictions and complex administrative requirements.By 2020, a further 270,000 NGCA residents were unable to make pension withdrawals inside government-controlled areas due to restrictions on crossing the line of contact (United Nations Ukraine 2021).Second, at the Humanitarian Network and Partnerships Week 2021 Nupur Kukrety, a social protection policy specialist at UNCIEF, raised the question of how often there is serious assessment of government capacity before international agencies decide to go it alone.
There at least appears to be an appetite for a more nuanced perspective on when, where and how governments can maintain delivery of existing programmes during crises.At present, policy and programming guidance for FCAS is limited to generic statements about capacity; for example, Wallis and Buckle (2016) stress the importance in considerations of social protection and climate change of being aware of operational constraints, noting that the 'fragility and insecurity of a country may affect a programme's ability to implement' social protection and drawing on the example of northern Kenya where 'insecurity affected beneficiary coverage, the roll out of IDs, opening bank accounts and delivery in all project areas' (ibid.: 18).
The literature combines limited practical, technical lessons about sustaining existing programming in crisis situations, but pays only lip service to political drivers and incentives that enable (or disable) continuous and ongoing delivery of programmes.The main practical lessons, predictably, point to the importance of established systems for registries and targeting, payment systems and use of digital technologies.These are all covered in other BASIC Research theme papers, so are not discussed in detail here. 4essons about human resource capacities, particularly to deliver social assistance at local level, are the focus of the limited literature on sustaining existing programmes in crisis situations.In relation to humanitarian response, various studies explore human resource capacity among aid workers who are directly or indirectly exposed to trauma.Chemali et al. (2018) highlight high levels of burnout and turnover, as well as long-term poor mental health among humanitarian fieldworkers supporting Syrian refugees in Lebanon.They suggest that scarce training in this regard leaves staff ill-equipped to care for themselves and practise resilience while handling trauma in the field.Ager et al. (2012) find that national humanitarian staff members in Gulu, Uganda, have high exposure to chronic and traumatic stress, and high risk of a range of poor mental health outcomes that affect capacity to sustain delivery.Hu et al. (2010)'s study of institutional robustness and operational resilience following the Wenchuan earthquake in 2008 stresses that those delivering social protection (broadly defined to include all elements of emergency response) were directly affected by the shock.Of 3,441 government personnel in the hardest-hit county, Beichuan, 568 died and 172 were severely disabled.Among surviving personnel, 36.2 per cent had a close family member who died in the earthquake (ibid.: 109).As a result, staff faced increased workloads, responsibilities and working hours at a time of great personal difficulty.Six months after the earthquake, selfreported physical and mental health had deteriorated and negative coping strategies had increased.
The focus of the existing literature is very much located in the 'adaptive delivery' part of Christie and Green (2019)'s model (see above).Neither the wider governance of social protection systems nor the political economy drivers that influence social protection programming receive much attention in the literature on crisis situations.
A major knowledge gap is how gender and other vulnerabilities influence capacity to sustain delivery, especially among programme staff.The Covid-19 pandemic has highlighted the burden working women faced during the pandemic, particularly their disproportionate contribution to child care and reproduction of labour.Even in the case of Covid-19, there is little evidence on how this impacts business continuity in the social protection and disaster response sectors.The focus of analysis on the gender inequalities emerging from Covid-19 has been on women as actual or potential beneficiaries of social assistance, less so on the challenges staff faced in delivering social assistance.The latter remains a substantial knowledge gap beyond Covid-19.Are there differences in the capacities of men and women who work administering social assistance to deliver their work when children are at home rather than in school due to conflict?Following a climate shock, do staff with disabilities face greater obstacles performing tasks because of breakdowns in administrative infrastructure that directly affects them?
Differentiating between competence, capability and performance helps unlock how human resources can enable or disable sustained delivery of programmes in crisis situations.Beyond the evidence about the impacts on human resource capacities in humanitarian settings, our literature review found no research exploring human resource capacities among social workers delivering social protection that was related to working in conflict-affected contexts.However, work for the BASIC Research Programme exploring what is known about how capacity and coordination features affect the delivery of social assistance in crisis situations provides insights into how to assess human resource capacities.
Noting 'the importance of the context when measuring a person's daily activities', Holsbeeke et al. (2009) differentiate between capacity, capability and performance: Capacity describes what a person can do in a standardized, controlled environment.Capability describes what a person can do in his/her daily environment.Performance describes what a person actually does do in his/her daily environment.The person-environment interaction is the discriminating element between capacity, capability, and performance.(Holsbeeke et al. 2009: 849) This seems to be a useful distinction to make to understand why and how programme delivery is -or is notsustained in crisis situations.Staff may have the capacity and capability to deliver tasks, but in situations where there are power or internet outages, or it is not possible to travel to recipients' villages, staff find themselves unable to complete their tasks; other reasons might be because they have not been paid and lack motivation, because they are burnt out, or because they are suffering mental health impacts from their own traumatic experiences or bereavement.Understanding what the overall capacity constraints are to programme delivery and tackling them could go a long way to supporting the sustained delivery of programmes.Where tackling deficits is unrealistic, at least recognising these deficits and measuring them provides an evidence-rather than an assumption-based justification for working outside or in parallel with the existing system to ensure delivery.
International humanitarian and development actors can work both upstream and downstream from governments to sustain existing social protection programme delivery, but whether they do so depends on (their perceptions of) the capacity and legitimacy of governments.Exploring the potential for enhanced social protection in Africa, Beegle, Coudouel and Monsalve Montiel (2018) stress the roles that non-state actors can play in sustaining social protection.Upstream, donor agencies support around 46 per cent of social protection financing, while downstream, 'both NGOs and United Nations agencies provide social safety nets to hard-to-reach communities or in fragile or conflict-affected areas' (ibid.: 210).These dual roles can create problems, especially where an agency is both a funder and an implementing partner; or where humanitarian and social protection principles do not easily align in particular contexts.
In many countries, the form of social protection that is being sustained is a subsidy system rather than social transfers.Subsidy systems frequently face the ultimate 'catch-22' in crisis situations where demand for subsidies -particularly for consumer staples -increases, often exponentially, just as government revenues slide and energy prices increase.Examples are found particularly in oil-producing states, notably in Syria (see above), Iraq and Cameroon (Harsch 2019;Savage 2021;del Ninno and Tamiru 2012).The politics of subsidy reform, even where this is achieved by a transition from subsidies to targeted social assistance, are especially challenging in FCAS, where reforms are usually politically unpopular and so undermine the social contract.So, while subsidies are more easily sustained in terms of delivery systems, they are far more challenging to sustain in financial terms.This flags up the importance of unpicking various elements of the sustain question; in particular, differentiating between economic and political dynamics that encourage or discourage programmes from being sustained.
Maintaining social protection delivery in crisis situations may require cooperation beyond the usual 'formal' actors, but not enough is known about the resilience of local and community actors and the mechanisms by which they connect with government systems.There is little evidence on whether the engagement of local governance, non-governmental organisations and communities in social protection contributes to its maintenance.Countries with weak systems tend to depend on community structures to deliver social protection (Ovadiya et al. 2015) but we do not know whether these are more or less durable than more centralised/official/formal systems for delivery.Evidence from sub-Saharan Africa points to a breakdown in community-led mechanisms when they are placed under severe stress (e.g. during the worst periods of the HIV/AIDS pandemic when mutual support and non-government/local social protection systems broke down).
However, humanitarian agencies frequently work with and through local and community organisations, and elsewhere there is evidence of them being relatively durable.Ovadiya et al. (ibid.)describe how Togo has drawn on an existing community-level mechanism for its school feeding programme.They describe femmesmamans, village women who are 'a familiar feature in Togolese villages, preparing and selling food in the market or on the street' (ibid: 31).They suggest that incorporating femmes-mamans into school feeding has proven to be successful in a low-capacity setting and has reached many beneficiaries.The extent to which this should be interpreted as community engagement versus private sector engagement is a moot pointone repeated in other BASIC Research Programme outputs including the Lebanon rapid review and the politics paper -but reflects that various configurations of stakeholders and hybrid delivery systems are found in many crisis situations.There is also analysis from Yemen identifying configurations of stakeholders that have enabled existing programmes -and piggybacked emergency programmes -to sustain delivery despite conflict.Al-Ahmadi and de Silva (2018) find that: Prior to the current conflict, Yemen had relatively well-functioning social protection programmes, implemented through national institutions like the Social Fund for Development (SFD), the Public Works Program (PWP), and the Social Welfare Fund (SWF), in which the government, donors and development partners have invested over the past two decades.The operational autonomy granted to SFD and PWP, their emphasis on developing effective administrative and delivery systems, their community-based networks (CBOs and local councils), as well as their accountability to the government, to donors, and to local communities have all contributed to making these institutions resilient against the multifaceted implementation challenges that arise during conflict.(ibid.:23) Reliability of finance is a key feature of programme sustainability.While this is true of any social protection programme in any situation, it has particular resonance in crisis situations where donors can play a role in enabling reliability or undermining it.Reviewing disbursement data and monitoring reports highlights many programmes where donor agencies have postponed or withdrawn support, usually because of concerns about predatory governments, fiduciary risks and human rights abuses.There are good examples of how different financing streams can allow funding for social protection and other basic services to continue even where trust between government and donor agencies breaks down.For example, the Ethiopian Productive Safety Net (PSNP) was established in 2005 with a large share of funds routed directly through the treasury to states and woredas (districts), rather than via direct budget support (DBS), which could bring fiduciary or diversion risks.In 2005, when international agencies had concerns about the outcome of federal elections, this meant that PSNP disbursements could continue, even when donors chose to 'turn off the DBS tap'.
Further analysis of studies in other sectors -notably, the disaster risk management, security and safety sectors -provide additional insights.Several other sources of literature provide further insights and require further exploration.From the economics and finance literature, there are assessments of the impact of natural disasters on the banking sector (e.g from the Caribbean -see Brei, Mohan and Stobl 2019), with insights on the operations of private sector organisations contracted to deliver cash transfers in crisis situations.This is particularly the case for small-scale payment providers, which may have limited liquidity to pay transfers because many households draw down savings following a shock (ibid.;Smith and Bowen 2020).An example of where this knowledge might be useful is in Tigray in Ethiopia, where cash availability in shops and banks is a key constraint on paying PSNP transfers in larger towns and cities and has been disrupted due to conflict (Sabates-Wheeler and Lind 2021).Beyond social protection, preoccupations with both natural disasters and cybersecurity have led to calls for the implementation of more effective risk management systems that provide contingency planning for business continuity (Ali Torabi, Giahi and Sahebjamnia 2016).Davidovic et al. (2020: 23, citing Leonovich 2020) argue in an IMF working paper that stakeholders should develop business continuity plans 'to demonstrate shock-responsiveness and guard against cyberattacks related to remote work (Leonovich 2020).These plans should identify core business processes and provide alternatives to sustain operations during emergencies.'The literature searches carried out for this work found no examples in the literature above that focused explicitly on FCAS.

Conclusion -towards a research agenda for sustaining existing social protection in crisis situations
The review of literature and programme data suggests that analysis of the features of social protection that enable existing programme delivery to be sustained in crisis situations suggests that this is a significant knowledge gap and one to be explored in greater depth and more comprehensively than is possible in this short review.While there is no specific explicit literature on the theme, there is potential to go further in exploring particular contexts and the grey literature around them that was not possible within the scope of this paper.This presents a research gap and an opportunity for BASIC Research.
But how important is the sustain question?First, sustaining social protection can be important in upholding social contracts.Ovadiya et al. (2015) highlight the importance of pension provision in the social contract in East Timor.Mallet and Slater (2017: 5), focusing on conflict-affected countries in Africa and Asia, find that: when people start experiencing problems with their service provision, perceptions of government deteriorate.This reflects a more widespread finding that legitimation is a precarious, long-term process that can be easily undone: as our DRC survey team note, trust arrives by foot but disappears on horseback.(Ferf et al. 2016) Second, as noted above, the notion that government programmes collapse in crisis situations is an oversimplified view that is not objectively supported by empirical evidence.Understanding when, where, and how programmes can be sustained -particularly the technical and operational features and wider political and economic environment that support sustained delivery -could lead to more effective programme design and adaptation.
In terms of framing, various concepts are useful.From the Maintains framework, systems resilience and business continuity are useful, but could perhaps be adapted to allow greater emphasis on resilient systemic and programmatic features that enable continuity rather than making changes to programmes, small and temporary though they might be.As noted above, drawing on systems-based definitions of resilience and ideas of 'functional persistence', 'institutional robustness' and 'operational resilience' appear to offer the greatest potential for exploring when, where and how existing social protection programmes can be maintained in crises.The distinction between levels of decision-making and operations from adaptive programming -namely, adaptive delivery, programming and governance -also provides an entry point for breaking down social protection systems into component parts and assessing which features enable sustained and effective provision at delivery, programming and governance levels.Smith and Bowen (2020) stress the importance of assessing barriers and blockages in the supply chain.
In terms of contexts, it is suggested here that situations of violent conflict are the least well understood and should form the central focus of future research, with any additional research on climate-related crises and Covid-19 providing complementary analysis.
A comprehensive understanding of the features of social protection that enable them to be sustained in crisis situations will require the overarching research question to be unpacked as below.

Potential questions about specific programme operations
• Is the existing social protection caseload (i.e. the number of recipient households or individuals) maintained or does it reduce/has it reduced during a crisis?• Are the timing, size and frequency of transfer delivery sustained during a crisis (e.g.where payments are made quarterly, does this continue and follow the planned timetable for disbursement)?• What technical features of social protection systems and programmes enable (or disable) sustained programme delivery?Are some targeting, registration and verification approaches easier to maintain than others?Which delivery systems are most commonly sustained?• What revisions to programmes and delivery systems enable delivery to be sustained during a crisis?Potential questions about the features of the wider enabling (or disabling) environment and how these affect the sustained delivery of existing programmes • How do the political dimensions of a particular crisis influence whether social protection is sustained?
What are the political incentives to maintain delivery?• What are the economic dimensions to sustaining social protection?For example, are some programmes or systems better able to absorb currency devaluation or inflationary effects that have implications for fiscal space?• Are there gender and inclusion dimensions to the sustained impact of existing programmes?There are two elements to this: first, whether some programmes are better sustained than others (e.g. in Syria, supporting the elderly through pensions for formal sector workers has been sustained but for those outside the gender, equality and social inclusion system, it appears that social assistance from the NSAF has receded); and second, whether there are gender and vulnerability dimensions to staff capacity to sustain delivery of programmes (e.g.whether female staff have increased responsibilities that undermine their work administering programmes).• What are the governance dimensions and institutional arrangements that influence the capacity to sustain social protection during crises?For example, do long-standing parastatal agencies -such as those in Yemen and the Central African Republic -have greater capacity to maintain operations to reach transfer recipients than government ministries or departments?What programmatic, policy, organisational and institutional features make this possible?• How do local, non-governmental 'informal' social protection systems (i.e.actions to provide income and consumption support to vulnerable households that exist outside of governments or formal organisations) endure in crisis situations?What contextual features influence this? • What role do international agencies -either humanitarian or developmental -play in sustaining existing programme delivery by governments?

Annexe
Literature search approach and other sources of data Literature searches of SCOPUS, Google Scholar and Google were carried out using the combinations outlined in the table below.
There were two phases of literature searching, the second taking place following feedback on a first draft of this report, which introduced a wider range of Term 2 wordings, largely based around the concept of 'continuity'.
The SCOPUS searches were restricted to title, keywords and abstracts because so many of the terms in the second column are too generic and might be used in multiple ways, even in work specifically focused on social protection.For example, the search term 'sustain*' yielded results about whether households sustain food consumption or whether government budgets and external financing streams are sustainable irrespective of (crisis) context.
In Google Scholar and Google, screening was more challenging.Searches yielded very many results because both search full text rather than title, keywords and abstracts only.In each case, the first 20 items on each page were assessed, after -following each search -deleting cookies and search histories to avoid introducing bias based on browsing history.Where these 20 items yielded new results, further items were assessed.If a new search yielded no new results in the first 20 items in the list, no further screening was done.
In all cases, the screening resulted in cutting large numbers of results.Terms 3 (the BASIC long list of countries of particular interest) and 4 were input together in single searches, and also input separately.Term 5 was input along with 1 and 2 after initial results with more terms produced no relevant results.In the case of Term 5, while there are many sources of academic and policy research focused on Terms 1, 2 and 5, none of these explicitly consider the gender and inclusion dimensions of sustaining programming as is intended here.

2.1. What does the existing literature look like and what does it tell us?
Monitoring reports such as World Bank Implementation Status and Results Reports provide regularly updated data on the disbursement of funds to governments and programme recipients who have received support in any given reporting period.However, to clearly understand if programme delivery is disrupted requires more granular, geographically disaggregated and time series data than is publicly available.For example, in Ethiopia, understanding if or to what extent the Productive Safety Net Programme has operated in the Tigray region since the outbreak of conflict there in November 2020 requires access to woreda (district)-level data on disbursements and this in turn would help identify if and how the PSNP can be more sensitive to and resilient in situations of acute conflict (Sabates-Wheeler and Lind 2021).
supported programmes -particularly World Bank International Development Association and International Bank for Reconstruction and Development documentation and East Africa delivery data for the World Food Programme (WFP).The programmes reviewed are listed in the Annexe.