Executive Summary

In a context of increasing frequency, size and duration of disasters and crises globally, the limitations of standard approaches to humanitarian response have come to the forefront, causing governments and international agencies to pledge to "use existing resources and capabilities better to shrink humanitarian needs over the long term" (Grand Bargain, 2016). The social protection sector can have an important role to play in this process, as recent research on "Shock Responsive Social Protection" has confirmed (O'Brien et al., 2018, Beazley et al. forthcoming).

This research focuses on the specific role of social assistance data and broader information systems and capabilities. In particular, it looks at their potential role in identifying beneficiaries and delivering benefits to them in the aftermath of a shock.

The research presented draws on recent international experiences in using social assistance data systems for shock response. It builds on an earlier briefing note on the "Factors affecting the usefulness of existing social protection databases in disaster preparedness and response" (O'Brien and Barca, 2017), co‑financed by the UK's Department for International Development (DFID) and Australia's Department of Foreign Affairs and Trade (DFAT).

The main findings include the following:

Depending on their set‑up, existing social assistance data systems can offer a range of potential uses for shock response. For instance, as a source of household and individual level data; comprehensive socio‑economic data; operational data (that is useful to identify, trace and deliver benefits); geo‑referenced or geographically‑disaggregated data; and (in an increasing number of countries) data that can help to capture shock vulnerability in advance of a shock. They also sometimes feature interoperability or data sharing arrangements with other government registries and are underpinned by established capacity to collect, store, and manage data.

The varied nature and quality of social assistance registries and broader information systems however, means that their role and use in emergencies can only be identified with reference to the particularities of the registries in the country and context under review. Social assistance registries and broader information systems vary widely across countries – if they are set up at all. Key variations include: their coverage; whose data and what data they collect and store; how data are collected and updated; who is responsible for data collection, storage, and management; whether/how information is integrated to other government databases, and; what processes and authorisation levels are in place for data sharing (see Section 2 and Box 1).

There are six complementary dimensions that can be used as a framework to assess the potential utility of social assistance registries and their broader information systems to be useful in response to shocks – which derive from the variations described above (see set of questions provided in Table 6 and Section 3):

Completeness. This refers to the level of data coverage and number of records compared with what would be perceived as a full set of records–for instance, 100 percent of the population in affected areas, or 100 percent of those in need. An existing social assistance registry may assist an emergency response if the data cover all of those affected by the shock, or a high enough proportion. Important distinctions need to be made between data on beneficiaries and registered non‑beneficiaries, acknowledging that neither are likely to offer full coverage of populations affected (see Figure 2)

Relevance. Data are relevant if they contain the variables required for the intended purpose. Data collected for the provision of long‑term social assistance (i.e. another purpose) may not always be relevant for shock response if they do not contain variables that comprehensively identify households in affected areas, and ideally that assess household needs and enable an immediate response.

Currency. Data currency is the degree to which data are current (up to date), and thus represent households' real circumstances at the required point in time. It is, of course, impossible for standard social protection data to reflect the reality after a disaster, meaning some form of post‑disaster revalidation is always required. The relevant factor is how up to date existing data are overall – often an issue for concern in many countries reviewed.

Accessibility. This refers to the ease with which potential users ‑ most likely national or local government agencies and departments, or their partners ‑ can obtain the data. Accessibility can vary widely depending on who the users are and what processes and authorisation levels are in place for data sharing; the underlying policy and legislation; whether or not data are maintained and stored digitally; existing provisions for data security and privacy; what type of data interfaces are provided, etc.

Accuracy. Data are considered to be accurate if they are free from errors and omission. Accuracy means that a high level of confidence can be placed in the data, affecting their wider credibility and ultimately their usability.

Data protection. Data are secure when they are protected against unauthorised access, misuse, or corruption. Data privacy is guaranteed where data are utilised while protecting an individual's privacy preferences and their personally identifiable information. In emergency contexts, concerns regarding misusing or losing such information – potentially exposing households to further vulnerability – are heightened.

• Depending on the six dimensions discussed above, social assistance data on households and individuals can inform decision making before, during, and after a shock – as a complement to other data sources and data collection efforts: Before a shock hits, data can inform risk analysis and vulnerability assessments, as well as planning and preparedness measures.
  • When a shock is about to occur, and immediately after it, early warning systems can enable timely responses by leveraging existing data.
  • After the shock, data can inform key decisions in relation to identifying who to support (targeting) and the type of support required. Table 5 usefully breaks the main options down.
  • In the long‑run, data and information can enable learning and inform policy changes – for example by incorporating shock‑affected caseloads into routine social protection provision.
• While there are recent international experiences successfully using social assistance data, they are not widespread – and have sometimes encountered challenges. For example, focusing on shock response: Vertical expansions and programmes "piggybacking" on beneficiary data require very little additional efforts (e.g. in terms of adapting processes) and can therefore enable timely responses, if adequately planned in advance. However, they present significant drawbacks in terms of the coverage of affected populations, which need explicit addressing (see Section 4.2.1 and Box 7). Moreover, with no preparedness in terms of financing and coordination, timeliness can be significantly compromised (Section 5.1).
• Horizontal expansions (via existing or new "piggybacked" programmes), on the other hand, inherently involve more complex processes and political decisions (Section 4.2.2). Moreover, few countries have developed social registries with the characteristics needed for these to be truly useful in response to shocks (e.g. sufficient coverage). This does not mean such a strategy is not possible, it simply means it requires significant planning (as for HSNP in Kenya) – including a careful assessment of the various options available for leveraging routine data and systems (Table 5).
• Overall, there are some important potential benefits of using pre‑positioned data and information systems versus "starting from scratch" with new data collection, however building on existing systems is not always achievable and also may come with risks, and trade‑offs. For example: Timeliness of responses can be increased by leveraging existing data, information systems and capacity, if financing is available for timely disbursement of funds and procedures have been planned in advance. This can be achieved via vertical expansions or piggybacking on beneficiary data. For horizontal expansions, timeliness can be more complex to guarantee if not via planned efforts to: use existing data to target or inform expansions (e.g. via pre‑enrolment as for HSNP); build on existing information systems, data collection approaches and capacity, and; leverage data collection technology. On‑demand systems for data collection can play a role in countries where they are already present, but can be labour intensive, difficult to maintain in the aftermath of a crisis and present excessive direct, indirect and opportunity costs for applicants (Section 5.1).
• Ensuring coverage of affected populations, and fully avoiding inclusion and exclusion errors, can be complex when relying solely on existing social assistance data. Data collected before a shock will never give an exact assessment of needs in the aftermath of a shock (even when the eligibility criteria are altered or where better data are collected beforehand), and beneficiaries of existing programmes are not necessarily those who are most in need. Strategies to reach affected households whose data are not held within existing registries will always be needed – for example in contexts of cross‑border displacements it is unlikely that existing registries will contain records on refugees/ non‑citizens.

The most pressing and important trade‑off that needs to be discussed and evaluated by decision makers in advance of a shock is therefore the one between inclusion/exclusion errors (coverage) and timeliness. When it comes to crisis response, timeliness is usually more important than full targeting accuracy, especially in the first phase of assistance. Specifically, inclusion errors can and should be tolerated in the short term – especially as they can contribute to controlling tensions within recipient communities. Exclusion errors, on the other hand, should be minimised by design, and promptly addressed through a sound grievance redress process and complementary approaches to swiftly reach all affected households. The true question for policymakers is whether leveraging existing social assistance systems and data is the best way to balance this trade‑off – as it may not be (Section 5.2).

There is a value in leveraging shared data for increased coordination amongst social protection, DRM, and humanitarian actors, leading to improved knowledge/learning, reduced duplication of efforts, and potentially saving costs (for example, administrative costs of data collection, recurring costs of data management, and private costs to citizens. Two experiments in Pakistan and Malawi were set up to explicitly assess the cost‑effectiveness of leveraging pre‑positioned data, versus new data collection efforts in the aftermath of a shock. Both found some clear advantages, but also several areas that would need addressing ex‑ante to make the pilots scalable and useful, as summarised in Box 8 (Section 5.3).

There is a further important trade‑off between making social assistance data more accessible to external partners for reduced duplication of efforts/costs and guaranteeing data security and privacy. Responses that build on existing data and systems should not compromise the data security and privacy of registered individuals and households, placing households at risk of increased vulnerability – and potentially compromising humanitarian principles. Better approaches to data collection, management, and sharing – and agreements developed in advance of a shock – can help to minimise risks while ensuring accessibility of valuable data.

To conclude, building on existing data, information systems and related capacities has the potential to (Section 6):

• enable better planning and preparedness for shocks by supporting the identification of vulnerable households and the estimation of potential caseloads. Efforts to collect operationally relevant data and expand registries in vulnerable areas can support this function further;
• enable a more timely response, by linking to systems for early warning via pre‑agreed triggers and/or leveraging existing data and capacities;
• support processes for registering, selecting, and enrolling beneficiaries ("targeting"), reducing the duplication of efforts and enhancing the cost‑effectiveness of responses; and
• ensure learning and new data generated via the shock response informs government capacity and programming decisions (e.g. expansion of caseloads), enhancing longer‑term sustainability.

Nevertheless, the extent to which these benefits can truly be reaped depend on:

• Factors that go far beyond the realm of data and information management – which need to be better factored in within broader "preparedness" measures. E.g. lack of funding, swift decision making and approvals, and robust M&E systems.
• The practical set‑up of the registry and information system in question which ultimately affects its coverage, relevance, currency, accessibility, accuracy/usability and level of protection.
• The type of approach to shock response. Each option has very different implications, and there are major variations within each option that need careful assessment (see Table 5).
• The type of shock. Table 6 summarises some core considerations along different variations in shock characteristics emerging from this research.

The core conclusion is that – before using existing data and information systems at any cost – it will be essential for every country to make a careful assessment of: a) existing data and systems based on the six criteria discussed above; b) the benefits, risks and trade‑offs of using existing data versus "starting from scratch". This can inform a decision on how/if existing data and systems should be used.

It will also be important to consider whether existing systems can be strengthened through adjustments to the way data are collected, stored, and managed. This research stresses the importance of investing in preparatory measures leading to better data quality, which are based on a strong policy vision. It is not only the data that can play an important role, but also the broader capacities to collect, store, manage, and share that data – as well as the underlying information systems and institutional agreements for data sharing. Building each of these ex‑ante with an eye to their potential role ex‑post will be an essential preparedness step for countries building more adaptive social protection systems. Box 15 provides several examples while Figure 4 provides a useful summary infographic of the key steps needed in this process.