IDA19 MID-TERM REVIEW OF THE CRISIS RESPONSE WINDOW EARLY RESPONSE FINANCING Development Finance, Corporate IDA & IBRD (DFCII) September 23, 2021 ACRONYMS AND ABBREVIATIONS Fiscal Year (FY) = July 1 to June 30 All dollar amounts are US dollars AFE Africa East IDA International Development Association CERC Contingency Emergency Response Component IPC Integrated Food Security Phase Classification CRW Crisis Response Window IPF Investment Project Financing ERF Early Response Financing MTR Mid-term Review ESF Environmental and Social Framework PRA Prevention and Resilience Allocation FAM Famine Action Mechanism RECA Remaining Engaged during FAO Food and Agriculture Conflict Allocation Organization TAR Turn Around Allocation FCS Fragile and Conflict-affected Situation TEGFS Technical Expert Group on Food Security FCV Fragility, Conflict and Violence UN United Nations FEWS NET Famine Early Warning Systems Network UNICEF United Nations Children’s Fund FTCF Fast Track COVID-19 Facility WBG World Bank Group GEMS Geo-Enabling Initiative for WFP World Food Programme Monitoring and Supervision TABLE OF CONTENTS EXECUTIVE SUMMARY ........................................................................................................... i I. INTRODUCTION................................................................................................................ 1 II. SUMMARY OF ERF UTILIZATION............................................................................... 2 A. Food Insecurity ............................................................................................................... 3 B. Disease Outbreaks .......................................................................................................... 4 III. ERF ACTIVATION............................................................................................................. 5 A. Food Insecurity ............................................................................................................... 5 i. Trigger-Based Activation........................................................................................... 6 ii. Local Activation......................................................................................................... 8 iii. Overall Assessment .................................................................................................... 9 B. Disease Outbreaks ........................................................................................................ 10 IV. SPEED OF RESPONSE .................................................................................................... 10 V. ERF CAPS .......................................................................................................................... 12 VI. CONFLICT CRITERION ................................................................................................ 15 VII. CONCLUSION AND QUESTIONS FOR GUIDANCE ................................................ 18 LIST OF ANNEXES Annex 1. CRW ERF Allocations to Date in IDA19 ..................................................................... 19 Annex 2. ERF Triggers on Disease Outbreaks ............................................................................. 20 TABLE OF FIGURES, TABLES AND BOXES Figures Figure 2.1. Commitments by Sector ............................................................................................... 3 Figure 2.2. Allocations by Region .................................................................................................. 3 Figure 2.3. Commitments by Instrument Type ............................................................................... 3 Figure 2.4. Commitments by Operation Type ................................................................................ 3 Figure 3.1. ERF Activation by Type (July 2020–August 2021) ..................................................... 5 Figure 3.2. Assessment of ERF Trigger Performance by Country ................................................. 7 Figure B3.1.1. ERF Food Insecurity Trigger-based Thresholds ..................................................... 6 Figure 4.1. Median Processing Time Elapsed (Number of Days)a ............................................... 11 Boxes Box 3.1. ERF Trigger-based Activation for Food Insecurity ......................................................... 6 Box 6.1. Risk Identification and Mitigation in FCS ..................................................................... 17 Tables Table 5.1. Illustrative Scenarios for Assessing the ERF Cap on Pre-allocated CERCsa .............. 15 Table A2.1. Pathogen-Specific Laboratory Confirmed Case Number Thresholds ...................... 20 -i- EXECUTIVE SUMMARY i. In the Nineteenth Replenishment of the International Development Association (IDA19), the Early Response Financing (ERF) framework was introduced under the Crisis Response Window (CRW) to address slower-onset crises, namely, food insecurity and disease outbreaks. While the devastating effects of sudden-onset crises like earthquakes are more quickly visible and generally result in faster resource mobilization, this did not always happen with slower- onset crises. The ERF was thus designed to respond to slower-onset events that are in the earlier stages of progression but could escalate into major crises. Given the ERF’s novel nature, Management committed to review “the ERF framework, especially the triggers and their adequateness” at the IDA19 Mid-term Review. i With a truncated IDA19, this review will instead be discussed at the third IDA20 replenishment meeting in October 2021. ii. This paper reviews the first 14 months of the ERF’s operation, and this short duration should be borne in mind when assessing potential policy changes. The review spans July 2020 when the ERF took effect to September 1, 2021, and its scope covers technical aspects such as the ERF activation framework and financing caps. All ERF requests so far have been for food insecurity and more would have to be learned on how it responds to disease outbreaks. Overall, this review identifies areas for improvement but does not recommend sweeping changes. iii. The ERF’s operation to date has yielded useful lessons, with areas for improvement. Its activation framework for food insecurity has generally performed well and no changes are recommended. The trigger-based thresholds have accurately identified crisis risks in most instances, and the local activation approach confers flexibility to capture cases which may be missed by the triggers, thereby avoiding reliance on a single set of thresholds. Given the limited experience with disease outbreaks, it is proposed to maintain the existing disease outbreak thresholds. Meanwhile, ERF responses have not been as fast as envisioned, and Management is embarking on ways to streamline processing. As for funding caps, the ‘ERF’s belts-and-braces’ approach remains relevant but there is scope to raise the limit on pre-allocated Contingency Emergency Response Components (CERCs). Finally, the ERF currently requires that eligible food insecurity events must not be primarily driven by conflict, but there are practical and policy challenges with meeting this criterion. iv. Drawing from the lessons above, Management proposes to: (a) raise the ERF cap on pre-allocated CERCs from $12.5 million to $25 million; and (b) adjust the treatment of conflict-affected situations to align with the World Bank Group (WBG) Fragility, Conflict and Violence (FCV) Strategy and IDA’s FCV Envelope. While a $25 million cap may tie up more ERF funds for contingent purposes, it is expected to still leave the bulk of ERF resources available for actual crises that materialize. A higher cap is also consistent with feedback from country teams and reinforces the overall push in IDA20 for stronger crisis preparedness. To address the challenges with meeting the current conflict-related requirement, the treatment of conflict-affected situations under the ERF would shift from binary causality (that is, requiring that an ERF-eligible food insecurity event must not be primarily driven by conflict) toward careful consideration of whether IDA can meaningfully engage in conflict-affected settings to support food security interventions. i World Bank Group, Additions to IDA Resources : Nineteenth Replenishment – Ten Years to 2030 : Growth, People, Resilience, p. 66. I. INTRODUCTION 1. The Crisis Response Window (CRW), piloted in the Fifteenth Replenishment of the International Development Association (IDA15) and officially established in IDA16, is the cornerstone of IDA’s crisis response toolkit. It provides additional resources for responding to crises and supporting recovery to restore countries to their long-term development paths. CRW funds are additional to IDA country allocations and help protect core development spending that could otherwise be jeopardized amid more pressing crisis needs. 2. The CRW has been adapted to evolving crisis demands. Started as a pilot in IDA15 during the 2008/09 global economic and financial crisis, the CRW was permanently established in IDA16, marking a paradigm shift in IDA’s crisis response from ad hoc interventions to a more systematic approach. Before this, IDA’s crisis support had entailed restructurings and reallocations that diverted resources from long-term development or trust funds that took time to set up. The CRW hence served as a more predictable and systematic response mechanism. In IDA17, its scope was expanded to cover public health emergencies in addition to natural hazards and economic crises—a change prompted by the 2014–15 West Africa Ebola crisis. In IDA18 during the initial months of COVID-19, the CRW was used to provide critical resources under the World Bank Group’s (WBG’s) Fast Track COVID-19 Facility (FTCF) to shore up readiness and response in IDA countries. 3. In IDA19, the Early Response Financing (ERF) framework was introduced to address slower-onset crises, namely, food insecurity and disease outbreaks. While the devastating effects of sudden-onset crises like earthquakes are more quickly visible and generally result in faster resource mobilization, this did not always happen with slower-onset crises. The ERF was thus designed to respond to slower-onset events that are in the earlier stages of progression but could escalate into major crises. It also complements existing sources of funds. Prior to the introduction of the ERF, and in the spirit of the CRW being established as a ‘last resort’ vehicle, the CRW only responded to slower-onset food insecurity and disease outbreaks after they had become severe, which has led to missing the window of opportunity for early response. 4. The ERF was originally sized at up to $500 million in IDA19; 1 this was increased to up to $1 billion given higher-than-expected demand.2 This adjustment was in light of rising food insecurity due to the COVID-19 crisis and other factors. An activation framework using both numerical thresholds and qualitative analyses was designed to identify potentially eligible events and inform a decision as to whether or not to deploy ERF resources. Box 3.1 and Annex 2 elaborate on the activation frameworks for food insecurity and disease outbreaks respectively. Complementing the above are country-level funding caps as well as a cap on pre-allocated Contingency Emergency Response Components (CERCs) that are financed by the ERF, as elaborated in paragraph 28. 1 This was part of an overall IDA19 CRW envelope of $2.5 billion. 2 World Bank. 2021. Adjustments to IDA19. June 8. -2- 5. The ERF does not directly fund crisis preparedness activities but contributes to this agenda in other ways. 3 While mainly used for crisis response, the ERF also bolsters preparedness by financing activities that help avert a worsening of the crisis. Examples of such activities include distributing fertilizers and seeds to farmers to prepare for the next planting season and creating job opportunities to strengthen rural productive infrastructure and raise incomes. Early response hence does not entail preventing a crisis but is about responding earlier to one that is already slowly unfolding. 6. This paper reviews the first 14 months of the ERF’s operation. Given the novel nature of the ERF, Management committed to review “the ERF framework, especially the triggers and their adequateness” at the IDA19 Mid-term Review (MTR). 4 With a truncated IDA19, this review will be discussed at the third IDA20 replenishment meeting in October 2021. It evaluates the technical aspects of the ERF such as its activation framework and financing caps. The review period covers July 2020 (when the ERF took effect) to September 1, 2021, and this short duration should be borne in mind when assessing potential policy changes. 7. The paper is structured as follows. This introduction is followed by Section II which summarizes the utilization of the ERF to date. Sections III to VI comprise the main body of the review and assess how the ERF has fared in terms of the appropriateness of its activation framework, the speed of response, and the relevance of its financing caps and conflict criterion. Section VII summarizes the conclusion of the review and outlines the questions for IDA Participant’s guidance. II. SUMMARY OF ERF UTILIZATION 8. Since its inception in July 2020, the ERF has allocated $400 million 5 to eight countries, comprising 36 percent of CRW allocations in IDA19—with $212.5 million already committed. All ERF allocations have been for food insecurity, with no requests on disease outbreaks so far. Annex 1 sets out the full list of ERF allocations, and this section presents key analytics on ERF usage. At this stage, it is not yet possible to assess the impact and results of ERF- funded operations as there have been no disbursements yet (see Section IV). 3 The CRW, including the ERF, is a response mechanism and generally does not fund upstream activities before a crisis occurs. Other sources of funds such as IDA country allocations and the Regional Window are available to support ex ante risk reduction and crisis preparedness. 4 World Bank Group, Additions to IDA Resources: Nineteenth Replenishment – Ten Years to 2030: Growth, People, Resilience, p. 66. 5 ERF allocations for Afghanistan (comprising $12.5 million for a pre-allocated CERC as well as a subsequent allocation of $37.5 million) were also considered earlier. The World Bank has temporarily paused disbursements in operations in Afghanistan and is closely monitoring and assessing the situation in line with internal policies and procedures. The ERF allocations related to Afghanistan are hence largely excluded from this paper, except in places such as Section III on the ERF activation framework as that section relates to whether the ERF triggers have been breached or not, which is different from whether or not an ERF allocation has been made. -3- A. Food Insecurity 9. ERF utilization has mainly been in agriculture and social protection (Figure 2.1). By sector, the share of ERF commitments in agriculture is 54 percent, five times higher than those under the CRW’s non-ERF commitments. This is due to all ERF allocations to date being requested for food insecurity, as compared to non-ERF responses which covered a wider range of crises. Meanwhile, social protection projects are a common vehicle across both the CRW’s ERF and non-ERF modalities, reflecting their shock-agnostic nature and utility during crises as such projects can be scaled up rapidly and often have pre-identified beneficiaries that target vulnerable groups. By region, Africa Eastern and Southern (AFE) region is the largest beneficiary with 63 percent of ERF allocations totaling $250 million (Figure 2.2). By instrument, all ERF-funded operations have been channeled through Investment Project Financing (IPF) (Figure 2.3). The bulk of ERF commitments have been channeled through new projects rather than via additional and supplemental financing to existing operations (Figure 2.4). Figure 2.1. Commitments by Sector Figure 2.2. Allocations by Region 100% 70% 63% 60% 80% 50% 60% 37% Share by Region 40% 37% 30% 40% 20% 13% 13% 13% 20% 4% 7% 10% 4% 11% 0% 0% CRW ERF CRW Non-ERF AFE AFW MNA LCR SAR EAP Agriculture Social Protection Other ERF Allocations Food Security Hotspots Industry & Trade/Ser Public Admin Transportation Note: The diamonds show the share of food security hotspots in each Water/Sanit/Waste region. Both the columns and the diamonds sum to 100%. Figure 2.3. Commitments by Figure 2.4. Commitments by Instrument Type Operation Type 1% 100% 100% 90% 12% 90% 24% 80% 80% Share of Commitments 51% Share of Commitments 70% 70% 60% 60% 50% 100% 50% 40% 87% 40% 76% 30% 30% 49% 20% 20% 10% 10% 0% 0% CRW ERF CRW Non-ERF CRW ERF CRW Non-ERF New Projects Additional/Supplemental Financing IPF DPF PforR Note: The non-ERF columns in Figures 2.1, 2.3 and 2.4 exclude CRW funding for the FTCF given the exceptional and specialized nature of that allocation. -4- 10. Partnerships have been central to the ERF and continue to inform its work. 6 For instance, the ERF benefitted from engagements under the Famine Action Mechanism (FAM), a World Bank-led partnership for scaling up anticipatory and early action to emerging food security crises. The FAM’s exploratory work on innovative models to predict food crises and simulate scenarios informed the development of the ERF’s food security triggers. It is also a valuable sounding board of food security experts across donor, development, and humanitarian partners to test approaches for scaling up early action. Such engagements have led to collaboration between IDA, World Food Programme (WFP) and the United Nations Children’s Fund (UNICEF), for example as part of Somalia’s Shock Responsive Safety Net for Human Capital Project (P171346) which has provided cash transfers to poor households vulnerable to drought and malnutrition risks. In addition, the Food and Agriculture Organization (FAO) has provided valuable support to the ERF applications for Cameroon and the Central African Republic (CAR). B. Disease Outbreaks 11. There have been no ERF allocations for disease outbreaks. In FY21 there was only one case—the Ebola outbreak in Guinea—discussed as potentially requiring ERF support. The outbreak was controlled at an early stage and the number of cases did not meet the ERF threshold, so the request was not put forward. 7 There have also been no COVID-19 related requests, in part because the ERF only came into effect at the start of IDA19 in July 2020 when COVID-19 had already reached pandemic proportions and was no longer at the early response stage. 8 With the now-familiar practices of mask-wearing and social distancing, COVID-19 has also changed the lifestyles of many, which temporarily reduced the incidence of some infectious diseases such as influenza. 9 12. The ERF nonetheless remains relevant for responding to disease outbreaks as IDA clients face continued risk in this area. 10 In recent decades, the world has experienced HIV/AIDS, Avian flu, SARS, MERS, Zika, several regional Ebola outbreaks, and of course, COVID-19. COVID-19 may have also disrupted critical services such as routine immunization, thus increasing susceptibility to outbreaks of vaccine-preventable diseases in the future. 11 The ERF hence remains a key part of IDA’s crisis toolkit that can help countries respond to outbreaks at an early stage. 6 For a broader discussion on the complementarities of IDA’s work on food insecurity in relation to that of other partners, see World Bank. 2020. “Responding to the Emerging Food Security Crisis.” November 26. 7 See https://www.who.int/emergencies/disease-outbreak-news/item/2021-DON328. 8 The World Health Organization (WHO) declared that COVID-19 reached pandemic proportions on March 11, 2020. 9 Sullivan, Sheena G., et al. 2020. “Where Has All the Influenza Gone? The Impact of COVID-19 on the Circulation of Influenza and Other Respiratory Viruses, Australia, March to September 2020.” 10 This includes higher risk of infectious pathogens spilling over from animals to humans, development of antimicrobial resistance, spread of infectious diseases via travel and trade, and weak public health infrastructure. 11 Mansour, Z., J. Arab, R. Said, A. Rady, R. Hamadeh, et al. 2021. “Impact of COVID-19 Pandemic on the Utilization of Routine Immunization Services in Lebanon.” PLOS ONE 16 (2): e0246951. -5- III. ERF ACTIVATION A. Food Insecurity 13. Since its inception, eight countries have pursued ERF support for food insecurity via trigger-based activation and six have used the local activation option (Figure 3.1). Countries can seek ERF support via trigger-based activation or local activation. Trigger-based activation applies standardized risk thresholds across clients, where available (Box 3.1). 12 It is monitored for 20 countries based on Integrated Food Security Phase Classification (IPC) compatible data provided every four months by the Famine Early Warning Systems Network (FEWS NET). 13 Local activation is based on country-specific evidence such as local-level data on food prices, climate-related indicators, seasonal outlook assessments for crop and livestock conditions, and benchmarks on malnutrition and mortality. Local activation is especially relevant in countries not covered by FEWS NET (that is, where the trigger-based approach is not feasible due to a lack of food security classification data). It could also apply in situations where a client is covered by the trigger-based approach, but the triggers are not breached, yet local indicators signal a significant cause for concern. Of the countries pursuing ERF support for food insecurity, 14 eight ERF allocations have been made (Annex 1) and the others are being processed. Figure 3.1. ERF Activation by Type (July 2020–August 2021) Local Activation Local Activation No Activationa 1. CARb 1. Burundi 2. Haiti Trigger-based 2. Chad 3. Madagascar Activation 3. Malawi 30% 30% 4. Nigerc No Activation 4. Mali 5. South Sudan 5. Nigeria 6. Yemen 6. Uganda Trigger-based Activation 1. Afghanistan 5. Kenya 40% 2. Cameroon 6. Mozambique 3. DRC 7. Somalia 4. Ethiopia 8. Sudan Notes: a. These countries were covered by the trigger-based framework but did not breach the triggers, except for Nigeria. Nigeria breached the triggers but did not pursue ERF support. b. CAR is not covered by the trigger-based framework and pursued ERF support using local activation. c. Niger breached the ERF triggers but pursued ERF support using local activation (paragraph 18). d. ERF allocations to Afghanistan were considered earlier; the WB has since temporarily paused disbursements in operations in Afghanistan and is closely monitoring and assessing the situation in line with internal policies and procedures. For the purposes of section III of this paper, Afghanistan is included as this section relates to whether ERF triggers are breached—which is different from whether or not an ERF allocation to a country is made. 12 During the design phase of the ERF, the World Bank informally consulted with FAM partners on the trigger parameters. Other consultations also took place, for instance, there was a mission to Somalia to explore potential acceptance of the ERF trigger thresholds as part of the work on Somalia’s Anticipatory Action Plan, and these were broadly endorsed by a working group of humanitarian and development food security partners. 13 Created in 1985, FEWS NET is a provider of information and analyses on food insecurity. The IPC-compatible metric produced by FEWS NET provides the longest running dataset across the widest available country coverage. 14 All except for the CAR are monitored by FEWS NET. -6- Box 3.1. ERF Trigger-based Activation for Food Insecurity ERF trigger-based activation is defined by two rules that capture a minimum severity level and a worsening trend (Figure B3.1.1). Both rules must be met, either using projections or actual outcomes.a Rule 1 sets a minimum severity level at IPC Phase 3. This is a critical juncture where the affected populations are experiencing, or are at risk of experiencing, high food consumption gaps and elevated rates of acute malnutrition, which are typically accompanied by accelerated depletion of livelihood assets. Rule 2 captures worsening trends by requiring that the share of the population that is in, or projected to be in, IPC3+ conditions must be rising by five percentage points or more. This helps differentiate ERF trigger events from ‘chronic’ food insecurity conditions that should be addressed by other sources of financing such as IDA country allocations. Figure B3.1.1. ERF Food Insecurity Trigger-based Thresholds The percentage of a country’s population in, or projected to be Rule 1 in, IPC3+ (crisis or worse) levels of food insecurity must be ≥ A CRW ERF 20% or 5 million people. food security AND trigger occurs when both The increase in the percentage of a country’s population in, or rules are met. Rule 2 projected to be in, IPC3+ (crisis or worse) levels of food insecurity must be ≥ 5% points. While the triggers provide a critical signal about the potential trajectory of food insecurity, they do not lead to an automatic decision to allocate ERF resources. Instead, if a country team assesses that the breached triggers reflect conditions on the ground and ERF support is needed, the World Bank’s Technical Expert Group on Food Security (TEGFS) would conduct an assessment to verify if local food insecurity conditions corroborate the worsening outlook. The TEGFS’ analyses hence help inform a decision on whether to deploy ERF funds.b The case for ERF eligibility is presented to the IDA Board of Executive Directors (IDA Board) via Eligibility Notes, and this is followed by submission of ERF- funded projects for Board approval. Note: a All food security classifications, whether projected or not, should be considered estimates. For simplicity, ‘current’ conditions in FEWS NET nomenclature are referred to in this paper as ‘actual’ conditions. b The TEGFS does not have a decision-making role on the use of ERF resources. ERF allocations are approved by the IDA Board. i. Trigger-Based Activation 14. The reliability of the ERF food security triggers depends on the input data used to calculate whether the thresholds were breached. The triggers use both projected food security conditions as well as actual conditions provided by FEWS NET every four months. When the triggers are breached accurately using FEWS NET’s projections, this provides the most value as the potentially worsening conditions are identified four to eight months before they are expected to materialize. 15 There is also value in mobilizing actions based on deteriorating actual conditions, so as to help avoid further escalation into a major crisis. Nonetheless, if the projections are 15 Such early action is shown to be more effective. See Cabot Venton. 2016. The Economic Case for Early Humanitarian Response to the Ethiopia 2015/2016 Drought; USAID. 2018. Economics of Resilience to Drought: Somalia Analysis. -7- inaccurate, the value of early response can be diminished especially when: (a) the projections fail to accurately capture a future food crisis hence resulting in a missed opportunity to intervene early (false negatives); or (b) the projections falsely project worsening food insecurity which leads to the deployment of resources that may be used for other purposes (false positives). Figure 3.2. Assessment of ERF Trigger Performance by Country Country \ Timestep 2020 June 2020 Oct 2021 Feb Afghanistan 0 1 Burundi 0 0 0 Cameroon 0 FN 0 Chad 0 0 0 Democratic Republic of Congo 1 0 0 Ethiopia 0 1 Haiti 0 0 0 Kenya 0 0 Madagascar 0 0 0 Malawi 0 0 0 Mali 0 0 0 Mozambique 0 FN 0 Niger FN 0 0 Nigeria 0 0 0 Somalia FP 1 1 Sudan 0 FP Uganda 0 0 0 Yemen 0 0 0 Non-trigger on forecast basis predicted accurately in one or two timesteps. Forecast trigger predicted accurately in one or two timesteps. False positive (triggered based on forecast when actual conditions of the following timestep indicates no trigger). False negative (not triggered based on forecast but triggered based on actual conditions at the following timestep). N/A - Data for the following one or two timesteps unavailable. Note: Not all countries in Figure 3.2 have pursued ERF support. 15. The trigger-based approach has generally performed well, enabling the earlier identification of most emerging food security crises in IDA19. 16 About 90 percent of cases show that a projected breach (or non-breach) corresponded with an actual breach (or non-breach) in one of the subsequent two timesteps, 17 that is, projected conditions materialized as expected. When considering only those cases where the triggers projected a crisis, 71 percent of such cases turned out accurate. 18 Additionally, 57 percent of the food crises were captured in advance by the triggers using projected conditions, 19 which gave a four- to eight-month head start for mobilizing responses. 16. Only two false positives were recorded out of 50 instances where data are available. One such case was Somalia which breached the ERF triggers in June 2020, October 2020, February 16 This analysis applies to 18 countries for which sufficient data are available from FEWS NET, namely, Afghanistan, Burundi, Cameroon, Chad, Democratic Republic of Congo, Ethiopia, Haiti, Kenya, Madagascar, Malawi, Mali, Mozambique, Niger, Nigeria, Somalia, Sudan, Uganda, and Republic of Yemen. 17 A timestep is defined as four months, as FEWS NET provides data on food security conditions every four months. 18 Calculated as the total number of cases where crises were projected and were followed by a corresponding actual crisis (four or eight months later), divided by the total number of cases where crises were projected. 19 Calculated as the total number of cases where crises were projected and were followed by a corresponding actual crisis (four or eight months later) divided by the total number of actual crises. -8- 2021 and June 2021. In hindsight, the projected deterioration in June 2020 was an overestimate, although food security conditions did worsen in each of the subsequent timesteps. The other false positive was Sudan, where projected conditions breached the ERF triggers in February 2021 but are likely to be an overestimate of subsequent actual conditions. 20 There was nonetheless a material, albeit less significant, increase in food insecurity which was close to the level required for ERF eligibility. Upon examination of additional data, the Technical Expert Group on Food Security (TEGFS)—which comprises food security experts from across the World Bank (Box 3.1)—assessed that both Somalia and Sudan met the ERF eligibility criteria under the local activation approach. 17. There were three cases of recorded false negatives. In these instances, the trigger-based approach still led to ERF eligibility based on actual conditions, but the lead time for response was reduced compared to cases which were triggered using projected conditions. In Cameroon, projections fell short of crossing the ERF triggers, but subsequent actual conditions showed a larger increase in food insecurity. In Mozambique, a sudden spike in food insecurity was driven by Tropical Cyclone Eloise which could not have been reasonably projected but was captured by subsequent actual conditions. In Niger, the actual conditions in October 2020 showed increased food insecurity which was not accurately captured in previous projections. This deterioration was largely driven by localized food insecurity in western Niger which was subject to particularly high levels of uncertainty. ii. Local Activation 18. The local activation approach has been a valuable complement to the trigger-based mechanism, as it enabled proactive identification of additional countries at risk of food security crises. Of the six countries that pursued local activation, five were covered by the trigger- based methodology but pursued ERF eligibility via local activation. The local activation approach hence allowed the TEGFS to review supplemental evidence and validate that these countries faced emerging food security crisis risks that were broadly in line with ERF eligibility criteria. a. Haiti: While Haiti did not breach the ERF triggers, it was very close to doing so based on projected conditions in November 2020. It has met Rule 1 and was just shy of meeting Rule 2 (with a 4.4 percentage point increase in its population living in districts categorized as IPC3+, versus the required five percentage points). Given how close Haiti was to breaching the triggers and in light of more recent deterioration which was expected to push Haiti beyond the trigger thresholds, the country pursued ERF support via local activation. b. Madagascar: Although Madagascar is monitored by FEWS NET, it did not breach the ERF triggers as the total population living in affected districts, particularly in the south, fell short of meeting Rule 1. That said, the TEGFS assessed that Madagascar’s southern regions were facing material food security crisis risks. This was corroborated by ground- truthing of food security conditions during a mission to the south by the country team. The mission revealed evidence that food insecurity was driven by the worst drought in the area since 1991, with 49 percent (1.3 million people) of the total population in affected districts 20 While the projection period (four to eight months from February 2021) is not yet complete, latest projections indicate that future food insecurity would not escalate to the extent predicted. -9- facing IPC3+ conditions as well as significant levels of malnutrition. There were also confirmed cases of IPC5 (famine)—an extremely rare classification. Madagascar was flagged as an extreme case that merited ERF support under the local activation approach. c. Niger: Niger breached the ERF triggers in October 2020 but did not pursue ERF support at that time as food security conditions were expected to improve due to favorable cropping and rangeland conditions. While FEWS NET data subsequently showed some improvement, the information published by the Cadre Harmonisé 21—and endorsed by the Government of Niger—signaled worsening food security conditions. These concerns were in line with Niger’s initial breach of the ERF triggers in October 2020, and the country eventually sought ERF resources via local activation. d. South Sudan and Republic of Yemen: While South Sudan and the Republic of Yemen are covered by the trigger-based approach, both were technically not able to meet the ERF thresholds. While they met Rule 1 (with nearly 100 percent of their populations living in districts categorized as IPC3+), it was nearly impossible for them to meet Rule 2 on worsening conditions as there was limited scope for food security conditions in both countries to deteriorate further. Given that South Sudan and the Republic of Yemen have some of the highest levels of food insecurity in the world, these were exceptional cases. iii. Overall Assessment 19. While there is inherent uncertainty in food security projections, the performance of the ERF triggers has been encouraging and no changes are recommended to the activation thresholds. The very few false positives—Somalia and Sudan—were balanced against the value of identifying potential crises earlier for response. For the three false negatives where projections did not accurately capture subsequent shocks, the trigger-based approach was still able to capture the crises using actual conditions. Moreover, of the countries assessed as eligible under trigger- based activation, over half were able to identify the need for ERF support four to eight months earlier than would otherwise have been possible if projections were not used. 20. Moreover, the cost of operating the trigger-based activation framework is modest. The triggers provide a regular, globally comparable mechanism which acts as a proactive prompt for ERF allocations, as new information becomes available. They help ensure that escalating food insecurity events are monitored and assessed against clear criteria. Given the publicly available nature of the data used from FEWS NET, and clear processes in place for processing and calculating whether the trigger thresholds were breached, the cost of this primary method of signaling potential ERF eligibility is low. 21. The architecture for ERF activation has also built-in flexibility to capture cases which may be missed by the triggers, thereby avoiding reliance on a single set of thresholds. The local activation approach has allowed the ERF to intervene in cases in which the triggers were not breached, but where local-level evidence signaled a material cause for concern. More broadly, ERF eligibility assessments have incorporated a fuller body of evidence that included reviewing 21 The Cadre Harmonisé is an early warning system that is used to share information on the risks of food insecurity in 17 countries in West and Central Africa. - 10 - on-the-ground information to complement the binary check of whether or not the triggers were breached. The TEGFS, which comprises food security experts from across the World Bank, has played a key role in conducting the technical assessments to validate ERF eligibility. B. Disease Outbreaks 22. Given the limited experience of the ERF in tackling disease outbreaks, no changes to the disease outbreak triggers are currently recommended. Management affirms that the existing thresholds remain relevant (Annex 2), 22 and will remain alert to whether refinements are needed as IDA accumulates more experience with using ERF resources for disease outbreak response. IV. SPEED OF RESPONSE 23. The ERF processing times from identifying a potentially eligible event to approval of ERF-funded operations are broadly comparable to those under the CRW’s non-ERF modality. Non-ERF cases under the CRW are used as a comparator as they likewise entail crisis responses. The time elapsed from identifying the onset of a crisis to the submission of ERF Eligibility Notes to the IDA Board was 118 days, compared to 87 days for non-ERF cases (Figure 4.1). For ERF-funded projects that have been approved to date, 23 the time elapsed between eligibility confirmation and project approval is 130 days—longer than the 53 days under the non- ERF modality. That said, ERF processing times are broadly similar to those under the CRW’s non- ERF modality for new projects (that is, those that are not additional financing to existing operations), as explained in paragraph 25. No ERF-funded projects have started disbursing yet as the projects were only approved in the past few months. 24. Notably, ERF eligibility assessments by design start at an earlier stage of a crisis than those under the CRW’s non-ERF modality, and hence are expected to provide support earlier than in the non-ERF cases. The comparisons in Figure 4.1 should hence be interpreted with care as the start dates are different. Unlike the non-ERF modality where the process of assessing CRW eligibility takes place after a crisis is already severe, ERF eligibility assessments start in earlier stages of slower-onset events based on analyses of emerging crisis risks, typically four to eight months in advance as explained in Section III. Therefore, while ERF processing times are broadly comparable to those under the non-ERF modality, the earlier engagement of the ERF is expected to provide earlier crisis responses compared to non-ERF cases. 22 The ERF thresholds for disease outbreaks were based on the Pandemic Emergency Financing Facility (PEF) Cash Window activation criteria. The Cash Window was the other arm of PEF, besides the PEF Insurance Window. The thresholds were determined by expert epidemiological assessments of historical disease outbreaks. 23 Six ERF-funded projects have been approved to date. One project—the Strengthen Ethiopia’s Adaptive Safety Net Project (SEASN, P172479)—included a pre-allocated CERC funded by the ERF. The other projects were: Yemen Food Security Response and Resilience Project (P176129); Additional Financing to the Somalia Crisis Recovery Project (P176343); Central African Republic Emergency Food Crisis Response Project (P176754); South Sudan Resilient Agricultural Livelihoods Project (P169120); and Emergency Locust Response Project (P174546). Only the projects which did not involve pre-allocated CERCs were included in the analysis in Section IV of this paper. - 11 - Figure 4.1. Median Processing Time Elapsed (Number of Days)a CRW ERF CRW Non-ERF (IDA18-19) CRW Non-ERF New IPFsᵇ (IDA18-19) 130 124 118 110 111 97 87 53 From Identifying Eligible Event to Board From Board Eligibility Note to First From First Operation Approved to First Eligibility Noteᶜ Operation Approved Disbursement Note: a. Pre-allocated CERCs that are funded by the ERF are excluded from this analysis as they have different processing steps. Pre- allocated CERCs are contingent financing tools which are put in place before a crisis strikes. b. New IPFs refer to CRW-financed IPF projects that are processed as new operations, not as additional financing to existing IPFs. c. For the CRW’s non-ERF modality, the leftmost stage excludes some cases of economic shocks as the starting dates of such events are difficult to determine. 25. ERF-funded operations have taken longer to be prepared and approved compared to those under the CRW’s non-ERF modality, partly reflecting the choice of instruments for ERF-funding operations. All ERF commitments were made through IPFs and 75 percent of these were channeled via new operations, which tend to entail longer preparation times. In contrast, 77 percent of the CRW’s non-ERF IPF commitments were made via additional financing to existing operations, which are typically faster to process. ERF processing times were broadly similar to those under the CRW’s non-ERF modality for new IPF projects, that is, those which were not additional financing to existing operations, as shown by the gray columns in Figure 4.1. The reasons for using new instead of existing operations in ERF cases were varied: (a) timing (as some ERF recipients had new projects under preparation that were aligned with ERF objectives); (b) country portfolio consolidation; 24 and (c) transition from safeguards policies to the Environmental and Social Framework (ESF), which restricted provision of additional financing in FY21 to existing projects prepared under the former. 25 26. Country-specific circumstances also affected ERF processing speeds, especially given the more challenging operating environments in ERF recipient countries. Seventy-five percent of ERF recipients are in Fragile and Conflict-affected Situations (FCS). More broadly, of the 27 countries flagged as being particularly susceptible to food security crises (paragraph 33), 66 percent are also FCS. The ERF Eligibility Notes in some instances took longer to complete due to political- or conflict-related developments that necessitated further discussion on how best IDA 24 This led to ERF funds being channeled through new operations that were already under preparation. 25 This limitation would apply to both ERF and non-ERF projects in FY21. Before that, earlier non-ERF projects prepared under the old safeguards policies were able to have scale up Additional Financing. The ERF only came into effect in FY21, and ERF-funded projects hence did not have this option available. - 12 - should provide support. This was the case in Ethiopia, where deterioration of the conflict in the Tigray region required changes to the Eligibility Note and the project design, resulting in the integration of ERF support into the WBG’s overall resilience-building programs in the country. Another example was Haiti, where the ERF application was temporarily put on hold following the assassination of President Jovenel Moïse. In the Democratic Republic of Congo, the lengthy negotiations with third-party implementing agencies led to delays. 27. Management is embarking on ways to streamline ERF processing and accelerate preparation of ERF-funded operations, drawing from the early lessons above. For instance, the eligibility assessment process could be simplified by using a single consolidated template for both the TEGFS’ technical assessments and the Eligibility Notes to the IDA Board. There are also plans to expand outreach to country and project teams to further familiarize them with the ERF and its processes. Further, the preparation times of ERF-funded operations could be improved through outreach and training to raise awareness of the World Bank’s emergency procedures for additional financing. Both new operations and additional financing that use emergency procedures have been prepared more quickly, at an average of three months. In addition, Management will step up efforts to encourage the use of CERCs pre-allocated with ERF funds 26 as a preparedness measure in advance of crises, particularly for clients that are more susceptible to food insecurity or disease outbreaks. V. ERF CAPS 28. Three types of ERF caps were introduced in IDA19 as part of a ‘belts-and-braces’ approach. First, an aggregate cap was set at $1 billion to preserve the bulk of CRW funds for already-severe crises 27 under the non-ERF modality, and for Management to gain more experience with the ERF. Second, country-level caps apply. For food insecurity, this is $50 million per country per IDA cycle or the cost of the country’s response plan, whichever is lower. For disease outbreaks, it is $25 million per country per outbreak or the cost of the response plan, whichever is lower. Third, pre-allocated CERCs funded by the ERF are capped at $12.5 million per country. 28 These are pre-funded contingent tools that can be deployed rapidly during a crisis. As pre-allocated CERCs entail opportunity costs of setting aside ERF funds for a contingent event which may not occur, the caps are kept modest. This cap only applies to pre-allocated CERCs financed by the ERF and not to other types of CERCs. 29. Experience so far shows that the foregoing architecture remains broadly relevant, but there is scope to raise the cap on pre-allocated CERCs. When food insecurity worsened amid COVID-19, the IDA Board leaned forward to increase the ERF aggregate cap from $500 million to $1 billion.29 This higher cap continues to be appropriate as ERF usage so far—with $400 million of allocations and several requests in the offing—has been broadly in line with the halfway mark of IDA19. The country caps also remain pertinent as they allow the ERF to support a larger number 26 Pre-allocated CERCs funded by the ERF are described further in Section V. 27 This refers to severe natural hazards, public health emergencies and economic shocks under the CRW’s non-ERF modality. In contrast, the ERF responds to slower-onset crises at an earlier stage of progression. 28 Pre-allocated CERCs funded by the ERF can be used for food insecurity or disease outbreaks. 29 World Bank. 2021. Adjustments to IDA19. June 8. - 13 - of clients. With food insecurity being widespread and demand for ERF resources materially higher than was expected before COVID-19, the country caps have prevented disadvantaging those who need ERF support at a later stage. The country caps would conceivably also be useful should another pandemic or epidemic that affects a wide swathe of IDA clients materialize in future. So far, only Ethiopia has taken up a pre-allocated CERC funded by the ERF. This subdued demand is partly due to clients already facing heightened food insecurity in IDA19, which required direct response rather than contingent pre-allocated CERCs. In addition, many teams also noted the $12.5 million cap is too low to generate client interest in such CERCs. 30. Management hence proposes the following for IDA20: a. Keep the aggregate cap at $1 billion. This aligns with the anticipated high levels of food insecurity expected in IDA20, 30 especially as countries are still contending with related COVID-19 impacts such as damaged food supply chains, reduced trade, and loss of incomes. b. Preserve country-level caps. These allow the ERF to support a larger number of clients and helps avoid first come, first served—especially with food insecurity being widespread across IDA countries. Should the food insecurity or disease outbreak worsen, clients may also apply for additional CRW resources via its non-ERF modality. c. Raise the ERF cap on pre-allocated CERCs to $25 million. Management proposes to raise the cap to $25 million. Feedback from country teams indicated that a $25 million cap would be more appropriate in galvanizing take-up of such CERCs. Paragraphs 31 to 34 elaborate on the choice of the new level. A higher cap is also consistent with the push in IDA20 for stronger crisis preparedness. It is not proposed to entirely remove the cap as pre-allocated CERCs entail opportunity costs of tying up scarce IDA resources, and doing so may leave the ERF with inadequate resources to respond to actual crises that materialize. 31. Assessing ERF exposure from pre-allocated CERCs requires estimating overall ERF demand—which ideally entails a risk-informed modeled approach, but this is not currently feasible. For instance, food security projections for all IDA clients would be modeled and the number of countries that breach the ERF triggers would constitute potential overall ERF demand for food insecurity. Despite recent innovations, 31 these models do not produce consistent results across countries and most IDA clients are not covered due to data limitations. Another option is to assess the number of times on average the ERF would have been triggered in previous IDA cycles. The past is nonetheless not necessarily a good predictor of the future, especially as food insecurity has risen every year since 2014 and is much more widespread today. 30 Modelled five-year projections of populations in food crises for 56 IDA countries point to sustained unprecedented levels of food insecurity until 2026 (see Modelled Projections of Food Crisis 2021–2026, World Bank, June 2021). Additionally, based on analysis of food security classifications provided by FEWS NET, the number of people currently living in IPC3+ conditions across 38 countries have more than quadrupled since 2014. The Food Security Information Network (FSIN) has also recorded increases in the global total of people experiencing IPC3+ in nearly every issue of its annual Global Report on Food Crises. 31 World Bank. 2020. “Predicting Food Crises.” Policy Research Working Paper no. WPS 9412; World Bank. 2020. “Stochastic Modeling of Food Insecurity.” Policy Research Working Paper no. WPS 9413. - 14 - 32. Given the limitations above, the following criteria are used to simulate a scenario that represents a conservative upper bound for ERF food security-related demand in IDA20. 32 a. Criterion 1: Any country under the ERF trigger-based approach which met, or would have met, the ERF thresholds had these been in place since IDA15 33 is assumed to breach the triggers and pursue ERF support in IDA20. This is very conservative as the likelihood that all clients which crossed the ERF triggers previously will do likewise in IDA20 is low. b. Criterion 2: Any country which is flagged in a major FAO or WFP report since 2020 as either: (a) a food security crisis ‘hot spot’; or (b) being at risk of famine, is assumed to pursue ERF support in IDA20. This criterion captures recently confirmed conditions of elevated food insecurity which may signal potential ERF exposure in the near future. 33. Using the above, 27 countries 34 are flagged as being an upper bound for ERF food security-related demand in IDA20; this is however very conservative and actual demand is expected to be much lower. Even during the COVID-19 crisis which has had major negative impacts on food security, only about 13 clients have requested ERF resources so far. A scenario of 27 countries pursuing ERF support in IDA20 would require a significant covariate global crisis with repercussions on food supply and access that are significantly higher than the combined impacts of COVID-19, natural hazards and conflicts observed during IDA19—a very unlikely event. 34. Raising the cap on pre-allocated CERCs to $25 million is expected to leave the bulk of ERF resources available for actual crises that materialize. Two levers are used to estimate ERF exposure from pre-allocated CERCs: (a) the level of the cap; and (b) the share of clients that pursue pre-allocated CERCs. By varying these levers, one can derive different scenarios of ERF exposure (Table 5.1). Keeping the cap at $12.5 million is likely over-conservative as most scenarios show that more clients would be covered by ERF support than are flagged as being at risk, that is, they are over-insured. 35 This suggests there is room to raise the cap, especially given the low likelihood of all 27 flagged countries needing ERF support in IDA20. A scenario where half of the flagged universe (14 countries) take up pre-allocated CERCs of $25 million each would still leave $650 million of ERF funds that are not tied up in CERCs and available for actual crises that materialize (Scenario B). Moreover, some of the demand for tackling food insecurity is likely 32 As a sense check, all countries that pursued ERF support in IDA19 were captured by at least one of the two criteria, with most captured by both. This is a useful reference as these clients faced food security crisis conditions recently. 33 This analysis covers IDA15–19, and data are available from FY2010 onwards with occasional gaps due to imperfect or inconsistent historical FEWS NET coverage. IDA19 coverage is available through August 2021. 34 These are Afghanistan, Bangladesh, Burkina Faso, Burundi, Cameroon, Central African Republic, Chad, Democratic Republic of Congo, Ethiopia, Haiti, Honduras, Kenya, Liberia, Madagascar, Malawi, Mali, Mauritania, Mozambique, Myanmar, Nicaragua, Niger, Nigeria, Sierra Leone, Somalia, South Sudan, Sudan, and Republic of Yemen. 35 If the pre-allocated CERC cap is $12.5 million (second column from the left in Table 5.1), Scenario B shows 30 countries would receive some form of ERF support—higher than the flagged universe of 27 clients, which points to over-insurance. The analysis also covered two other scenarios where 50 percent and 75 percent of the flagged countries take up ERF pre-allocated CERCs. The results showed that a total of 35 countries and 40 countries respectively would receive ERF support, also higher than the flagged universe of 27 clients. - 15 - to come from clients with pre-allocated CERCS, implying that available resources would likely exceed $650 million. Table 5.1. Illustrative Scenarios for Assessing the ERF Cap on Pre-allocated CERCsa ERF Cap on Pre-allocated CERCs ($, millions) 12.5 25.0 37.5 50.0 Scenario A: 25% of the 27 Flagged Countries Take Up ERF Pre-allocated CERCs I. No. of Countries with ERF Pre-allocated CERCs 7 7 7 7 II. Total ERF Resources in Pre-allocated CERCs ($, millions) 87.5 175.0 262.5 350.0 III. Remaining ERF Balance ($, millions) 912.5 825.0 737.5 650.0 IV. No. of Countries Without ERF Pre-allocated CERCs That 18 16 14 13 Could Receive up to the $50 million ERF Country Capb V. Total No. of Countries Receiving ERF Support (I + IV) 25 23 21 20 Scenario B: 50% of the 27 Flagged Countries Take Up ERF Pre-allocated CERCs I. No. of Countries with ERF Pre-allocated CERCs 14 14 14 14 II. Total ERF Resources in Pre-allocated CERCs ($, millions) 175.0 350.0 525.0 700.0 III. Remaining ERF Balance ($, millions) 825.0 650.0 475.0 300.0 IV. No. of Countries Without ERF Pre-allocated CERCs That 16 13 9 6 Could Receive up to the $50 million ERF Country Capb V. Total No. of Countries Receiving ERF Support (I + IV) 30 27 23 20 Note: a. For simplicity, these scenarios assume that countries receiving the pre-allocated amount do not receive additional ERF funds. b. This is derived by dividing the ‘Remaining ERF Balance’ by the ERF food insecurity country cap of $50 million. It gives the number of countries without pre-allocated CERCs that could still receive ERF support. 35. The above would be complemented by further streamlining of CERC processing and greater efforts to raise awareness of pre-allocated CERCs. Including a CERC in an IPF operation requires additional work for client countries and task teams. To incentivize the inclusion of pre-allocated CERCs and CERCs in general, IDA is updating the CERC Guidance Note for staff, assessing areas for further streamlining of CERC procedures, and increasing its training and outreach on contingent financing mechanisms. In addition, IDA countries which receive ERF allocations commit to develop crisis preparedness plans, as mandated by the ERF rules. Such plans are a useful entry point for country dialogue on crisis preparedness, which could increase the uptake of pre-allocated CERCs and other contingent financing instruments that strengthen clients’ readiness to respond to crises. See Annex 10 of the IDA20 Deputies Report for details on efforts to further incentivize use of CERCs and contingent financing across the IDA portfolio. VI. CONFLICT CRITERION 36. Currently, countries applying for ERF support need to demonstrate that the food security crisis is not primarily driven by conflict. 36 At its inception, it was deemed that opening the ERF to food insecurity events mainly caused by conflict could lead to moral hazard and the risks of ERF funds being misused or inadvertently causing harm to people. Of particular concern 36 World Bank Group, Additions to IDA Resources: Nineteenth Replenishment – Ten Years to 2030: Growth, People, Resilience, p. 135, fn. 232. - 16 - were the risks involved where famine is used as an instrument of war. 37 There were also concerns related to financial sustainability as unresolved conflict could lead to recurrent requests for ERF support, potentially over prolonged periods. As such, the ERF-eligible food insecurity events currently need to be driven primarily by natural hazards, economic shocks, or public health emergencies—consistent with the crises covered under the CRW’s non-ERF modality. 37. While this requirement is well-intentioned, the experience to date indicates that there are practical and policy challenges with meeting it. First, food insecurity is often a complex issue with multiple and interrelated causes, where conflict situations create an underlying vulnerability and impacts on food security could be quickly exacerbated by other shocks. 38 It is hence difficult to single out a primary driver. For instance, COVID-19 impacts are adding to other drivers of food insecurity such as climate change, conflict and locust infestations. 39 Second, even if conflict can be established as a primary or key driver, denying ERF support in such cases may not be consistent with the broader WBG institutional strategy for Fragility, Conflict and Violence- affected (FCV) contexts. IDA’s engagement in FCS and other countries affected by FCV—as reflected in the FCV Strategy and the IDA19 FCV Envelope—includes remaining engaged during situations of crisis and conflict to the extent possible. Notably, all but two of the countries pursuing ERF support in IDA19 are on the FY22 List of FCS, and of these, all but one is considered to be in medium- or high-intensity conflict. Six of the countries are also accessing the FCV Envelope, which entails a commitment by both the government and the WBG to proactively address the drivers of FCV. The ERF hence works in complementarity with these efforts, providing a comprehensive response to both drivers and consequences of instability and lack of capacity. 40 38. At the institutional level, the WBG approach to engaging in FCS focuses on providing support with strong risk identification and mitigation measures. Within the IDA19 FCV Envelope, the RECA for example targets high-intensity conflict situations. 41 In line with the FCV Strategy, IDA financing—including RECA and other FCV Envelope allocations 42—is underpinned by careful risk identification and mitigation measures, reflecting the belief that such measures make it possible to effectively engage in conflict-affected situations (Box 6.1). In these circumstances, IDA takes all possible measures to ensure such support does not fuel conflict or perpetuate socioeconomic and other exclusions. This World Bank-wide practice is also applicable 37 De Waal, War and Famine in Africa. Article first published in 1993, and republished in the Institute of Development Studies Bulletin, Archive Collection, Volume 49 (April 2018). 38 The World Bank’s Operational Policy on Development Cooperation and Fragility, Conflict and Violence (former OP2.30), recognizes these interlinkages and states that: “The drivers and causes of fragility, conflict, and violence differ with each FCV situation; they may stem from an array of political, economic, or social factors, and may be compounded by shocks such as those related to climate change, disease outbreaks, or food insecurity.” 39 World Bank. 2020. “Responding to the Emerging Food Security Crisis.” November 26. 40 In CAR for example, the country program—which includes support from the FCV Envelope and Turn Around Allocation (TAA)—has been recalibrated to primarily target the well-being of the population, and the ERF project is a key component in responding to a deteriorating security and humanitarian situation. In South Sudan, the ERF- financed projects which focus on resilient agriculture and locust response complement the broader Remain Engaged during Conflict Allocation (RECA)-supported program that seeks to preserve human capital and lay the foundations for institutional development. 41 Quantitative indicators of conflict intensity are the heart of the eligibility criteria. 42 The Prevention and Resilience Allocation (PRA) targets countries at risk of escalating violent conflict. The TAA supports countries emerging from conflict. - 17 - for operations receiving financing from the FCV Envelope, which are specifically geared towards contexts with FCV risks. Box 6.1. Risk Identification and Mitigation in FCS Context-specific analyses of FCV dynamics and use of innovative tools are important for risk identification and mitigation. In South Sudan for example, project targeting, design, and adaptation are informed by careful analysis of conflict and violence drivers in relevant geographies, and the FCV sensitivity of project design will be further enhanced through additional analytical work. In the Republic of Yemen, regular risk monitoring and frequent communications have allowed the World Bank and United Nations (UN) teams to identify risks early and take action when they materialize. The risk identification and mitigation framework under the Republic of Yemen Food Security Response and Resilience Project identified the threat of air strikes on project sites and threat of physical violence for individuals involved in the implementation of the project as the greatest security risks to the project. Key mitigation measures include (a) mechanisms to identify security threats to the project and communicating changes in threat levels to the parties involved in project implementation; (b) communication and facilitation arrangements that enlist the support of relevant political and community actors to promote safe and politically neutral project implementation; (c) awareness-raising and training of communities and project beneficiaries; (d) remote supervision arrangements (Geo-Enabling Initiative for Monitoring and Supervision, or GEMS); and (e) working with politically neutral implementing partners. Staying engaged in some contexts entails reliance on third parties for program implementation. In the Republic of Yemen, the World Bank is operating on the basis of a request from international community and relies on cooperative operational arrangements with the United Nations (UN, through which the World Bank contributes financing and technical expertise, while UN agencies provide in- country implementation. In South Sudan, the World Bank maintains robust dialogue with the government while relying on UN agencies and the International Committee of the Red Cross to implement its program. 39. Management hence proposes to adjust the treatment of conflict-affected situations under the ERF, in alignment with the FCV Strategy and the Operational Policy on Development Cooperation and Fragility, Conflict and Violence. Given the challenges above, it would be more pragmatic to shift from binary causality (that is, whether conflict is a primary driver of food security crises or not) to recognizing there may be several primary drivers that are intertwined and compounding. It is hence proposed to align the ERF’s approach to risks associated with engaging in conflict situations with the treatment under all IDA financing, including those funded under the FCV Envelope, which is to identify the risks and mitigation measures and make risk-informed decisions on that basis. 43 Specifically in the case of the ERF, instead of requiring that an ERF-eligible food insecurity event must not be primarily driven by conflict, ERF eligibility would require country teams to summarize their understanding of key conflict-related risks, including any moral hazard, 44 to IDA support in the areas affected by food insecurity and the 43 As per the World Bank’s Operational Policy on Development Cooperation and Fragility, Conflict and Violence (former OP 2.30), the World Bank’s engagements in FCV situations follow a context-specific, conflict-sensitive approach based on adequate due diligence, diagnostics, risk analysis and stakeholder engagement to avoid causing or contributing to adverse impacts. 44 Conflict-related risks to IDA support could include, for example, risk of limited access to areas with increasing acute food insecurity, risk of diversion and misuse of funds and food/assets for political or other non- developmental objectives, and risk of exploitation of disadvantaged or vulnerable groups and individuals. - 18 - strategy for managing these. The specific measures that an ERF-supported operation will take to manage such conflict-related risks will be spelled out within the design and documentation of the operation itself, as is the case for all IDA-financed operations. Overall, this change will help IDA to tailor its interventions more effectively to specific contexts. Moreover, the ERF’s country-level caps (paragraphs 28 to 30) will help prevent frequent and recurrent requests for ERF resources in situations of prolonged conflict. VII. CONCLUSION AND QUESTIONS FOR GUIDANCE 40. The ERF’s operation to date has yielded useful lessons, with areas for improvement. Its activation framework for food insecurity has generally performed well and no changes are recommended. The trigger-based thresholds have accurately identified crisis risks in most instances, and the local activation approach confers flexibility to capture cases which may be missed by the triggers, thereby avoiding reliance on a single set of thresholds. Given the limited experience with disease outbreaks, it is proposed to maintain the existing disease outbreak thresholds which remain relevant. Meanwhile, ERF responses have not been as fast as envisioned, and Management is embarking on ways to streamline processing. As for funding caps, the ERF’s ‘belts-and-braces’ approach remains relevant but there is scope to raise the limit on pre-allocated CERCs. Finally, the ERF’s conflict criterion could be reframed away from binary causality, which is hard to prove definitively, toward careful consideration of whether IDA can meaningfully engage in conflict-affected settings to support food security interventions. 41. The ERF has only been in effect for a year and two months, and this short duration should be borne in mind when assessing its performance or potential policy changes. Moreover, all ERF requests have been for food insecurity so far and more study is required on how it responds to disease outbreaks. Overall, this review finds that there are areas for improvement but does not recommend sweeping changes. 42. Management seeks guidance from IDA Participants on the following: a. Do Participants agree with the approach to the ERF caps? (paragraph 30) b. Do Participants agree with adjusting the treatment of conflict-affected situations under the ERF? (paragraph 39) - 19 - Annex 1. CRW ERF Allocations to Date in IDA19 1 No. Country Region Crisis Type On the FY22 Activation $, FCS list Type million 1 Central African AFW Food Medium- Local 50 Republic Insecurity intensity Activation Conflict 2 Democratic AFE Food Medium- Trigger-based 50 Republic of Insecurity intensity Congo Conflict 3 Ethiopia AFE Food Medium- Pre-allocated 12.5 Insecurity intensity CERC Conflict Trigger-based 37.5 4 Haiti LCR Food Medium- Local 50 Insecurity intensity Activation Conflict 5 Madagascar AFE Food Non-FCS Local 50 Insecurity Activation 6 Somalia AFE Food High-intensity Trigger-based 50 Insecurity Conflict 7 South Sudan AFE Food Medium- Local 50 Insecurity intensity Activation Conflict 8 Yemen, Rep. MNA Food High-intensity Local 50 Insecurity Conflict Activation Total ERF Allocations 400 1 ERF allocations for Afghanistan (comprising $12.5 million for a pre-allocated CERC as well as a subsequent allocation of $37.5 million) were also considered earlier. The World Bank has temporarily paused disbursements in operations in Afghanistan and is closely monitoring and assessing the situation in line with internal policies and procedures. The ERF allocations related to Afghanistan are hence largely excluded from this paper, except in places such as Section III on the ERF activation framework as that section relates to whether the ERF triggers have been breached or not, which is different from whether or not an ERF allocation has been made. - 20 - Annex 2. ERF Triggers on Disease Outbreaks 1. Table A2.1 sets out the Crisis Response Window Early Response Financing (CRW-ERF) triggers relating to pathogen coverage and severity thresholds. These are intended to guide CRW ERF allocation decisions and are not meant as hard triggers. A potentially eligible outbreak will be referred to subject matter experts for a technical assessment. In exceptional circumstances (for example, cases of extreme country fragility with very high risks of cross-border transmission that could result in a regional epidemic or global pandemic), the CRW ERF may intervene before the triggers are met. Table A2.1. Pathogen-Specific Laboratory Confirmed Case Number Thresholds (Higher thresholds apply where epidemiological or genetic evidence of sustained human-to- human transmission is not available) Pathogen Threshold where epidemiological Threshold where epidemiological or genetic evidence of sustained or genetic evidence of sustained transmission within the human transmission within the human population is available: number of population is not available: epidemiologically linked, number of laboratory-confirmed laboratory-confirmed cases arising cases arising in IDA countries from in IDA countries from a single a single outbreak and occurring outbreak and occurring over a over a four-week period within the four-week period within the six six weeks immediately prior to the weeks immediately prior to the application application Influenza (novel influenza subtype, 30 100 or an influenza A virus whose hemagglutinin gene is antigenically distinct, due to an antigenic shift, from those in seasonal influenza viruses) Filoviruses (for example, Ebola, 30 100 Marburg) Coronaviruses which are not already 30 100 endemically transmitted in human populations (for example, MERS, SARS) Lassa fever n.a. 100 Rift Valley Fever n.a. 100 Crimean Congo Hemorrhagic Fever 30 100 Nipah virus n.a. 100 Chikungunya n.a. 200 Zika n.a. 200 Smallpox (or, in future, any other n.a. 2 human pathogen after its elimination from the human population) Other known or novel zoonotic 30 100 viruses (including, but not limited to equine encephalitis viruses, Hanta viruses, West Nile, Monkeypox)