Building financial resilience in pastoral communities in Africa LESSONS LEARNED FROM IMPLEMENTING Francesco Fava THE KENYA LIVESTOCK INSURANCE PROGRAM (KLIP) Nathaniel Jensen James Sina Andrew Mude Barry Maher BUILDING FINANCIAL RESILIENCE IN PASTORAL COMMUNITIES IN AFRICA Contents Acknowledgement 3 Francesco Fava* 1 Introduction International Livestock Research Institute 4 Nathaniel Jensen 2 KLIP product design International Livestock Research Institute 7 James Sina 3 KLIP implementation The World Bank Group 9 Andrew Mude 4 KLIP performance and impacts African Development Bank 11 Barry Maher 5 Lessons learned and way The World Bank Group forward 15 6 Conclusions 19 References 20 Annexes 22 * Author for correspondence: f.fava@cgiar.org 2 BUILDING FINANCIAL RESILIENCE IN PASTORAL COMMUNITIES IN AFRICA Acknowledgement T his paper was authored by Francesco Fava (International Livestock Research Institute – ILRI), Nathan Jensen (ILRI), James Sina (World Bank), Barry Maher (World Bank) and Andrew Mude (African Development Bank) as part of the World Bank’s Disaster Risk Financing and Insurance Program (DRFIP) of the Finance, Competitiveness, and Innovation Global Practice. The team is grateful to Olivier Mahul (World Bank) and Niraj Verma (World Bank) for providing support and guidance and to John Plevin (World Bank), Evie Calcutt (World Bank), Cecil Nartey (AfDB), Rupsha Banerjee (ILRI) and Federica Carfagna (African Risk Capacity) for peer reviewing the paper. The authors acknowledge valuable content made publicly available from the KLIP implementing stakeholders that include the Ministry of Agriculture, Livestock, Fisheries and Irrigation of Kenya, International Livestock Research In- stitute and private sector insurance providers. The team gratefully acknowledges the support of the KLIP implemen- tation unit and contributions provided by government officials including Principal Secretary, State Department of Livestock, Harry Kimtai and KLIP Coordinator Dr Richard Kyuma. The work is supported through Disaster Risk Finance and Insurance Program (DRFIP), which is housed in the World Bank Group’s Finance, Competitiveness and Innovation global practice. The DRFIP helps developing countries to de- velop and implement comprehensive financial protection strategies. Support to the KLIP program and development of this paper was made possible through partnership with USAID, through the Agriculture Insurance Development Program, and FCDO, through the Disaster Protection Program. The work was conducted as part of the CGIAR Re- search Program on Livestock and is supported by contributors to the CGIAR Trust Fund. CGIAR is a global research partnership for a food-secure future. 3 BUILDING FINANCIAL RESILIENCE IN PASTORAL COMMUNITIES IN AFRICA 1 T Introduction he livestock sector generates a large portion of national economies in African dry- lands and is the principal source of livelihood for pastoralists living there. Pastoral- ism is the main livelihood for an estimated 268 million people and represents 10 to 44 percent of the gross domestic product (GDP) of African countries (FAO 2018). However, pastoralists are often among the poorest—for example, 41 percent of pastoralists across the Horn of Africa are estimated to live in extreme poverty, which is well above the na- tional averages of the region. 4 BUILDING FINANCIAL RESILIENCE IN PASTORAL COMMUNITIES IN AFRICA Frequent and severe drought is a tremendous risk for since 2008, implemented an ambitious research-for-de- pastoral communities; drought-related livestock losses velopment agenda that resulted in the design of a set of can push households into a poverty trap and chron- innovative drought index insurance solutions that could ic destitution (Jensen et al. 2015; Lybbert et al. 2004). cushion pastoralists against the impacts of drought, This has, in turn, severe consequences on the effica- named Index-Based Livestock Insurance (IBLI) program. cy of countries’ and international donors’ efforts to im- prove welfare and develop the economies in pastoral IBLI solutions are based on an index of the relative sea- regions. In spite of these challenges, donors' and gov- sonal forage availability in a given area, which is de- ernments' response to crises often arrives very late, on rived from indicators of vegetation conditions collected average six months after the drought has set in (Clarke by satellites (Vrieling et al. 2014). After years of imple- and Dercon 2016). The delayed response and access menting an index that estimated average livestock mor- to support lead to loss of livelihoods and sometimes tality, which made payouts at the end of the dry season lives, weakening recovery from the crisis even after the for ‘asset replacement’ (Chantarat et al. 2013), the prod- conditions have improved. Furthermore, the govern- uct was redesigned with a focus on ‘asset protection’; ments' ex post response to shocks is costly because payouts are provided at the onset of the drought during it results in budget reallocations that might affect basic the rainy season to facilitate pastoralists implement- government functions (Clarke and Dercon 2016). The ing early coping and mitigation strategies (such as pur- increased impacts of natural disasters due to climate chasing fodder/water/veterinary services, destocking and environmental changes call for adoption of new before emergencies, and migration planning) to protect and innovative ways of funding disasters. Crisis risk fi- their livestock against more severe impacts (Fava and nancing (CRF) instruments, defined as financing mech- Vrieling 2021). Thus, the sum insured is based on the anisms that target a reduction of adverse socioeco- estimated cost of feeding and keeping the animals alive nomic or ecological impacts of potential crises (Poole during the drought. et al. 2020), offer this opportunity. CRF programs can support a proactive and timely response to drought for The early trigger/early action approach pioneered in clients and beneficiaries and help governments make the IBLI ‘asset protection’ product design is, in principle, drought response plannable and more cost-efficient. particularly valuable for pastoralists (Jensen et al. 2019). More generally, this anticipatory approach is at the fore- Among the different CRF instruments, index-based risk front of an emerging paradigm in disaster risk financing, transfer products, such as index insurance, have gained highlighting the value of preemptive responses to an- considerable traction over the last two decades for initia- ticipated shocks. A recent study in Kenya, for example, tives targeting the impacts of drought shocks on African found that for every dollar invested in early response smallholder farming and pastoral systems. Insurance is and resilience measures, US$2.8 are saved in later hu- a financial protection tool that can be used to address manitarian response interventions (USAID 2018). While relatively low frequency but high impact shocks. Un- there is no question that the humanitarian responses like conventional insurance, which is based on a claim will continue to play a major role in supporting ex post verification process of the losses, index insurance uses disaster relief, the complementary use of anticipatory payout triggering mechanisms that rely on transparent financial instruments can make a relevant contribution and objectively measured indicators of drought (that is, to protecting the livelihoods of affected households the index). Payouts are made to all policyholders when during drought crises. For example, livestock deaths predetermined index thresholds are met, which are nor- due to forage scarcity can be minimized, such that ev- mally derived from historic realizations of the index val- ery dollar spent on purchasing insurance can protect ues. This mitigates some of the key issues that make US$251 worth of livestock assets. conventional insurance unlikely to work in African rural settings, such as the lack of historic ‘loss data’ required IBLI provision started as a fully commercial initiative with for assessing risk and profiling clients; the high imple- private insurance companies retailing it as microinsur- mentation costs in remote and sparsely populated areas ance product in northern Kenya and southern Ethiopia (that is, for verification, data collection, monitoring); and since 2010. However, the huge losses and damage that problems related to asymmetric information (for exam- occurred during the 2008–2011 droughts, estimated at ple, adverse selection and moral hazard). US$12 billion (70 percent from the livestock sector), act- ed as a trigger for the government of Kenya (GoK) to To address the challenge of extending formal insurance review its strategic planning on ending drought emer- to extensive pastoral systems, several international or- gencies. As a result, in 2013 the GoK recognized ag- ganizations, local institutions, and private partners have, ricultural insurance as an important tool for protecting 1 This is calculated by assuming a premium rate of 20 percent (approximately the rate applied in 2019-2020) and cost of one mature cow (TLU) expected at US$500. 5 BUILDING FINANCIAL RESILIENCE IN PASTORAL COMMUNITIES IN AFRICA farmers and herders against production crises during Recognizing the benefits of the subsidized coverage, drought. The Ministry of Agriculture, Livestock and Fish- but also its limitation for long-term financial sustain- eries (MALF) allocated funding under the Second Me- ability of the scheme, GoK has considered a voluntary dium-Term Plan Two, 2013–2017, for (a) the implemen- component with partial subsidy. The partial subsidy is tation of a National Livestock Insurance Scheme (NLIS) expected to be accessible by all pastoralists, but it is and (b) increasing producers’ access to credit and fi- capped at 10 TLUs and 50 percent of the commercial nancial services including agricultural insurance. These premium. The GoK's proposition to offer universal ac- efforts culminated in the design and launch of a pub- cess to partial subsidy rather than to select a subset of lic-private arrangement (PPA), called Kenyan Live- beneficiaries was informed by the high cost of targeting stock Insurance Program (KLIP), offering subsidized and a relatively low inclusion error of universal cover- IBLI coverage to selected beneficiaries. age because poverty rates in the region are high. KLIP started purchasing insurance coverage on be- This paper aims at summarizing the main components half of 5,000 vulnerable households from two coun- of KLIP and at discussing the lessons learned during ties (Turkana and Wajir) in October 2015. Each house- implementation. It targets policy makers, technical ex- hold received fully subsidized coverage for five tropical perts, practitioners, and researchers interested in de- livestock units (TLUs).2 KLIP rapidly scaled up to eight signing and supporting the design and implementa- counties in the arid and semi-arid lands (ASALs) and to tion of similar programs. Section 2 describes the KLIP about 18,000 pastoralists annually. In the first five years product design, Section 3 presents the KLIP operation- of KLIP implementation, over US$10 million was paid al implementation, Section 4 reports observed impacts, out to vulnerable pastoralists to protect their while Section 5 summarizes and critically discusses the livestock and livelihoods from severe drought events. key lessons learned. 2 Five TLUs were considered the least viable herd threshold size that if lost from drought shock could lead to irreversible livelihood damage, hence the choice of the gov- ernment to provide 100 percent subsidy for five TLUs to cushion vulnerable households. A TLU is a standardized measure for livestock, where 1 cattle = 1 TLU, 1 goats or sheep = 0.1 TLU, 1 camel = 1.4 TLU. 6 2 BUILDING FINANCIAL RESILIENCE IN PASTORAL COMMUNITIES IN AFRICA KLIP product design I ndex insurance products offered under KLIP relies on a low-cost, accessible, and well-established sat- ellite indicator of drought (that is, Normalized Differ- KLIP is currently operating in arid and semi-arid coun- ties of Kenya characterized by bimodal rainfall distribu- tion and provides an annual insurance cover for the two ence Vegetation Index - NDVI) (Annex 1), which is a risk periods, the short rains short dry (SRSD) season, proxy for vegetation condition. High NDVI values in- from October to February, and the long rains long dry dicate healthier vegetation and vice versa. NDVI val- (LRLD) season, from March to September. During the ues from coarse spatial resolution satellites (250 m two risk periods, potential payouts can be triggered at and coarser) are generally used because (a) time se- the end of the wet months based on temporally aver- ries are available to describe long-term variability in aged NDVI values (Figure 2). forage conditions, (b) their daily acquisition frequen- cy allows for more cloud-free observations to describe The State Department of Livestock (SDL) under the MALF vegetation changes throughout the season, and (c) provided guidance on the critical parameters for the KLIP documented evidence exist for a strong relationship insurance policy, such as the sum insured per TLU, the at- between rangeland biomass and NDVI for arid and tachment index thresholds, and the exit index thresholds. semi-arid rangelands (for example, Dingaan and Tsubo These parameters provide the basis for product pricing 2019; Schucknecht et al. 2017). and are a result of negotiation between the underwriter and the GoK, which needs to consider the trade-off be- NDVI time series are elaborated to obtain an area-ag- tween frequency of payouts, magnitude of payouts, and gregated index of relative seasonal forage availabili- premium rates. The SDL chose an attachment threshold ty (Figure 1 and Annex 2). The unit areas of insurance correspondent to one expected payout out of five sea- (UAIs) are determined by combining local knowledge sons, a fixed exit threshold (that is, currently, -1.61), and a from pastoral communities about their grazing and mi- minimum payout every 5 percent (Annex 2). In terms of gratory patterns, agro-ecological maps, and adminis- sum insured, the SDL calculated the cost of feeding 1 TLU trative borders (Chelanga et al. 2017). When the index to maintain it alive during a major drought, correspondent falls below a predefined threshold, a payout is triggered to US$140 per year. The annual KLIP policy allocates 42 and increases proportionally to the severity of estimat- percent and 58 percent of the sum insured to the two po- ed forage scarcity. This payout approach is based on tential payouts (Figure 2), respectively. The premium rate the assumption that forage scarcity is an indicator of is determined by the underwriting company(s) in consul- the early stages of drought progression toward more tation with their reinsurers. severe impacts (that is, livestock losses and food inse- curity). Pastoralists could then use insurance payouts to The current KLIP product design resulted from a contin- make production decisions that mitigate the upcoming uous process of refinements since the launch of the IBLI impacts, for example, by protecting their herds to pre- program in response to the feedback from stakeholders vent high mortality or emergency sales. and to the evolution of satellite technologies (Chantarat 7 BUILDING FINANCIAL RESILIENCE IN PASTORAL COMMUNITIES IN AFRICA Figure 1: KLIP index calculation steps NDVI imagery (10-day) Spatially aggregated NDVI Spatial averaging 0.10 - 0.15 0.15 - 0.20 0.20 - 0.25 0.25 - 0.30 0.30 - 0.40 0.40 - 0.60 ging vera Seasonal averaged NDVI al a Z-score NDVI (index) por Tem Standardization < -1.5 -1.5 - -1.0 0.10 - 0.15 -1.0 - -0.5 0.15 - 0.20 -0.5 - 0.5 0.20 - 0.25 0.6 - 1.0 0.25 - 0.30 1.0 - 1.5 0.30 - 0.40 >1.5 0.40 - 0.62 Source: Vrieling et al. 2016, with Authors’ modifications. Figure 2: KLIP contract cover period, index calculation period, and time of payouts 1 year contract coverage SRSD LRLD OCT NOV DEC JAN FEB MAR APR MAY JUN JUL AUG SEP NDVI averaging SR NDVI averaging LR SRSD Payout announcement LRLD Payout announcement Source: Authors’ elaboration et al. 2013; Vrieling et al. 2014; Vrieling et al. 2016; Fava or climatic trends (for example, shift in seasonality, and and Vrieling 2021), taking into close consideration the rainfall increase/decrease). importance of keeping the product simple, accessible, and transparent. KLIP is now planning to expand the In terms of basis risk, there is a large body of scientif- coverage from the current eight arid counties to addi- ic literature supporting a high correlation between NDVI tional six semi-arid counties. A feasibility study (Kahiu anomalies and biomass in African drylands (for example, and Fava 2018) emphasized that for some counties this Diouf et al. 2015; Garba et al. 2017; Mahyou et al. 2018; would require a refinement of the product design be- Schucknecht et al. 2017; Tian et al. 2016). However, the cause of differences in seasonality and heterogeneity lack of time series of ground observations of rangeland in land uses (including cropping). In this perspective, biomass in Kenya has prevented undertaking of a rigor- alternative approaches have also been proposed (De ous assessment of the KLIP index accuracy in detect- Oto et al. 2019). Similarly, under circumstances of climat- ing drought-related forage scarcity. However, the index ic changes, the product requires periodic revisions to has been shown to be relatively well correlated with live- account for possible increasing frequency of extremes stock mortality observations (Jensen et al. 2019). 8 3 BUILDING FINANCIAL RESILIENCE IN PASTORAL COMMUNITIES IN AFRICA KLIP implementation K LIP is implemented under a PPA where the public sector supports an enabling environment (that is, regulations and infrastructure), provides subsidies, and is shared with the insurance company providing under- writing services to KLIP. The roster includes relevant cli- ent data such as the name of the beneficiary, identity creates necessary awareness about the product, while number, mobile phone and/or bank account numbers, the private sector prices the product, provides under- and next of kin. The GoK also coordinates awareness writing services, and manages payouts when triggered. creation campaigns and broad communication efforts The successful launch of KLIP involved close collabora- for pastoral communities (Figure 3 upper half). tion between private and public sector actors with clear roles and responsibilities for key stakeholders. To facil- A calculating agent determines index performance and itate implementation, the SDL set up a Program Man- percentage payouts for each risk period, following the agement Unit (PMU) that coordinated the design and approach described in Section 2 (Annexes 1 and 2) and implementation of KLIP. The private sector formed a the parameters indicated by the policy. The calculating technical committee that drew membership from inter- agent is required to provide accurate, timely, transpar- ested private insurers, with reinsurers actively involved ent, and independent information to all the interested for structured interaction with the SDL. Private sector stakeholders. The calculating agent also manages and engagement from the design stage and during imple- maintains the data set used for loss adjustment, doc- mentation ensured that the roles between the private uments the data processing chain for full replicability, and public sectors were well articulated, facilitating the provides in-season updates and maintains a backup launch of the scheme. dataset (Figure 3 lower half). Figure 3 provides a graphic picture of institutional in- The index values and payout amounts are shared with teraction that has taken place during KLIP implementa- stakeholders and officially announced at the predefined tion, while in Annex 3 a detailed description of the roles dates (Figure 2). The contracted insurance company(s) of the public and private sector actors is provided. The makes payouts directly to the beneficiaries as per the GoK, working closely with other institutions and organi- list provided by SDL, using either mobile money or zations involved in natural crises response, developed bank transfers. Bank checks have been used to pay beneficiary criteria to identify those who should bene- those who lack mobile or bank accounts with the fit from subsidized insurance cover. The SDL, in close bankers’ checks distributed using county and collaboration with county governments, administrative provincial administration infrastructures. officials, and local communities, then used that criteria to select the vulnerable households that could benefit At the initial stages of KLIP, both the registration from free cover. The number of beneficiaries for each and payout processes were mostly done manually, UAI is pre-determined and forms important input to cal- causing significant delays and mistakes in the pay- culating total paid premium. The roster of beneficiaries ments. The utilization of existing financial service 9 BUILDING FINANCIAL RESILIENCE IN PASTORAL COMMUNITIES IN AFRICA Figure 3: Schematic of KLIP main stakeholders (upper) and working mechanisms (bottom) Government k Fee d bac dba Fee emium Insu ck nce Pr ranc e co Insura ver Insurance companies Knowledge Pastoralist (and re-insurers) & Support households Institutions Indemnity payouts Index and for asset protection Pastoralist payouts are (end of rain season) households communicated needs resources to insurance to protect their companies livestock Satellite monitoring of forage availability Calculating Agent: If forage availability falls determines the Index below a preset threshold, values and final payouts then a payout is triggered Source: courtesy of D.C. Khalai, with Authors’ modifications. infrastructure (that is, M-Pesa, agent banking) for pay- Knowledge institutions have played an important role ments of premiums and payouts, mobile-based ap- in (a) providing technical and policy guidance to the plications for sales and clients’ registration, and a GoK, (b) supporting the private sector and its agency blended face-to-face and mobile-learning approach networks to implement effective sales and distribution for insurance agents have helped increase efficien- channels for voluntary market, and (c) conducting ac- cy across the program. In addition, monthly updates tionable research and impact studies to inform product/ about the progress of the season are currently pro- process improvement. Support has also been offered vided to inform the government and the underwriter by developing and delivering training to extension of- about the probability of payouts, so that they can start ficers and agents who have in turn created awareness planning accordingly. among the pastoral communities. 10 4 BUILDING FINANCIAL RESILIENCE IN PASTORAL COMMUNITIES IN AFRICA KLIP performance and impacts K LIP impacts can be summarized at the macro level, including markets and private sector support and financial protection of the government budgets, and mi- microinsurance retail product to pastoralists in the dry- lands of northern Kenya and southern Ethiopia since 2010 (Figure 4). The initial launch and associated commercial- cro level, related to the livelihood and welfare of vul- ization and outreach were met with slow but steady de- nerable pastoralists. Figure 4 provides an overview of mand for the product. In response to these challenges, KLIPs impacts, which are discussed more in detail in the product has evolved and adapted over time, thus sup- the next subsections. Overall, evidence suggests signif- porting a sustained demand for the product which is still icant benefits from KLIP both in terms of establishing a commercially sold in Kenya and Ethiopia and increasingly positive feedback loop of mutual benefits (that is, a win- adopted (that is, about 7,000 policies sold commercially win) between the public and private sector, a key ele- in 2018) (Figures 5 and 6). The retail pilot has been in- ment for long-term sustainability, and in terms of deliv- strumental in incubating the innovation in the Kenyan and ering positive outcomes for the welfare and livelihoods Ethiopian context and in supporting the improvement of of pastoralists during crisis and noncrisis periods. How- the product over time. However, the level of uptake and ever, there is still a great need for better understanding profitability for the private sector remained a major chal- of the short- and long-term impacts of KLIP on individ- lenge (Jensen et al. 2018, Zewdie et al. 2020). ual, community, and environmental outcomes. Moving forward, investments in a broader and more robust At its launch in 2015, KLIP initially provided fully subsi- monitoring and evaluation infrastructure and a rigorous dized coverage to 5,000 pastoral households from Tur- impact assessment study should remain a key priority kana and Wajir counties. The program rapidly expand- to fully understand the value of initiative for resilience ed and, since 2017, 18,000 pastoral households have building of pastoral communities. been covered, representing over 80,000 beneficiaries, across eight counties of northern Kenya (Turkana, Wajir, Marsabit, Mandera, Garissa, Tana River, Samburu, and Isiolo) (Figure 6). Macro-level impacts on markets and private  Since the program's inception, the local insurance sec- tor in Kenya has largely benefited from the rapid ex- and public sectors pansion of KLIP not only because of the substantial increase of the premium volume but also in terms of IBLI products were adopted by private insurance com- the support received for increasing technical and op- panies in Kenya and Ethiopia and have been sold as a erational capacity to implement agricultural insurance 11 BUILDING FINANCIAL RESILIENCE IN PASTORAL COMMUNITIES IN AFRICA Figure 4: Overview of KLIP benefits and impacts 1 US$10 million Protec gov. Risk transferred to • Premium payment reduces public financial burden in case of drought payouts since budget the private sector • Predictable and budgeted expenditures allow better resource inception made by allocation and harmonization with complementary initiatives the private sector 2 Public investment • Public sector premiums guarantee regular profit IBLI coverage expanded from on subsidies and • Investment in infrastructure facilities and crowds-in additional 3 to 8 counties Expand infrastructure services markets • More awareness on the product increases the potential for retail Number of IBLI sales policies increased from 4k to over 20k 3 Intensification: Increased investments in higher-returns production Good Seasons Reduced strategies drought risk • Strategic livestock sales when prices are high Greater income • Increased investments in veterinary services • Reduced precautionary savings Protect Less reliance on detrimental coping strategies during drought vulnerable Reduced income • Less distress selling of productive assets Drought Seasons loss during drought • Less "skipping" meals during drought Improved • Maintained investments in human capital post-drought economic and Early action to mitigate the impact of drought welfare Payments in • Destocking in anticipation of price and resource shocks outcomes anticipation of drought • Early purchase of inputs to sustain remaining herd during the coming drought Source: Authors’ elaboration Figure 5: IBLI commercial sales from 2010 to beneficiaries and payouts is shown in Figure 7. Some 2019 including Kenya and Ethiopia beneficiaries have received cumulative payouts amount- ing to over US$1,500, for example, in Tana River (Figure 7, 60,000 right), a relatively high amount for Kenyan pastoral econo- my. The seasonal per capita payout maps for all KLIP UAIs since the launch of the program are provided in Annex 4. 40,000 In terms of premiums, government expenditure reached Policies US$2.4 million per year in 2017−2018 (Table 1) and it is still approximately the same as of 2020. The GoK has 20,000 also invested growing resources for capacity building, awareness creation, and monitoring activities (up to 15 percent of the total budget) and has committed to fur- ther increase the budget allocation for the next three 0 2010 2012 2014 2016 2018 years to support geographic expansion to 14 counties Seasonal Sales Cumulative Sales and to increase the number of beneficiaries to 100,000 households. Source: Authors’ calculation As of end of LRLD 2020 season, the GoK has paid about solutions. In addition, the deeper penetration of agent US$9.5 million in premiums and a total of about US$10 networks in pastoral areas and the need of establishing million in payouts have been made to KLIP beneficia- partnerships with telecom companies for digital finan- ries. Payouts were triggered in at least one index unit cial services delivery have also stimulated new busi- during six of ten seasons (Table 1 and Annex 4). The se- ness opportunities. vere drought occurring in 2016–2017 (Uhe et al. 2018) led to three consecutive widespread seasonal payouts All 18,000 beneficiaries have received at least one pay- for a total of over US$7 million. In 2016–2017, this led out since KLIP was launched. The distribution of KLIP to a major loss for insurance companies, followed by 12 BUILDING FINANCIAL RESILIENCE IN PASTORAL COMMUNITIES IN AFRICA Figure 6: Schematic of IBLI evolution in Kenya from the launch of the first policy in 2010 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2019 2020 IBLI 106 1057 4695 5318 7000 Policies KLIP 5000 14000 18000 Policies Severe First Sharia Severe drought drought compliant triggers KLIP triggers product payouts for over widespread payout in US$7 million IBLII payouts Wajir Launch of Launch of Launch of first asset asset Launch of Full transition Mombasa Addis Ababa Horn of Africa replacement replacement asset to asset Executive IGAD livestock contract in contract in protection protection KLIP Ministerial insurance Marsabit, Borena, contract contract Workshop Workshop Initiative Kenya Ethiopia Note: The map illustrates expansion of coverage with IBLI (green), IBLI and KLIP combined (yellow), and planned KLIP expansion (light blue) in Kenya. Source: Authors’ elaboration Figure 7: (a) Map of KLIP beneficiaries per UAI and (b) map of the total per capita payouts for all KLIP UAIs from the launch of the program in 2015 to the LRLD season 2020 Note: KLIP started in different counties/UAIs in different years (see Figure 5). This is also reflected in the total payouts received by beneficiaries. Source: Authors’ elaboration another significant payout in 2019 (Table  1). However, pastoralists during severe and protracted drought cri- while the substantial payouts made have raised con- ses. Similarly, it has increased the confidence of the cerns about the financial sustainability of the program, GoK to continue supporting and expand the program, particularly from the private sector, these outcomes showing that it is possible to use private sector capi- have been well received by the GoK and insured ben- tal to manage risk and reduce pressure to use public eficiaries. The payout has proven the reliability of the funds, thus giving the government fiscal space to con- product, demonstrating the ability to make payouts to tinue to implement high-yielding development projects 13 BUILDING FINANCIAL RESILIENCE IN PASTORAL COMMUNITIES IN AFRICA Table 1: Summary of KLIP premiums, seasonal payouts, loss ratios, insured households, and TLUs Year Season No. of No. of No. of TLUs Total sum Premiuma Payoutsa Loss ratio (%) counties households Insured 2015–16 Short rains 2 5,000 25,000 5.59 0.56 6.24 Long rains 2 5,000 25,000 0.035 2016–17 Short rains 6 14,010 70,050 8.92 1.64 2.150 320.95 Long rains 6 13,776 68,880 3.130 2017–18 Short rains 8 18,012 90,060 12.61 2.46 1.750 71.02 Long rains 8 18,012 90,060 2018–19 Short rains 8 18’,012 90,060 12.61 2.41 0.880 160.75 Long rains 8 18,012 90,060 2.990 2019–20 Short rains 8 18,012 90,060 12.61 2.41 0.000 0.00 Long rains 8 18,012 90,060 0.000 Total 52.34 9.5 10.940 115.25 Note: a. US$, million. which can be otherwise compromised in the event of risk reducing strategies, such as distress selling of live- severe drought shocks. To address the private sector’s stock or skipping meals (Janzen and Carter 2019; Jen- concerns, the exit threshold was modified through a sen et al. 2017; Matsuda et al. 2019). It should be noted joint agreement by KLIP stakeholders, as part of a more that the reported evidence is related to the impacts of comprehensive actuarial review of the KLIP product insurance on households that choose to purchase in- that is currently ongoing. surance, whereas KLIP targets a particular group and transfers insurance to them. However, they should be broadly generalizable as long as the KLIP clients were equally informed on the details of their coverage. Impacts on vulnerable pastoralists A study using data from a survey of over 1,000 KLIP beneficiaries in Marsabit and Isiolo after the 2016–2017 drought examined how KLIP beneficiaries changed While impacts for vulnerable pastoralists are well docu- their coping strategies in anticipation of the coming of mented under the IBLI program and can be broadly ex- payments and then how they spent those funds once pected to be valid also for KLIP, no dedicated effort for they were received. Nearly all respondents reported us- continuous monitoring and impact assessment at the ing some of the payouts for human food, but most also household level has been put in place for KLIP. As such, used payouts to buy forage/fodder, water, and veteri- only few circumstantial studies are available specifically nary services for their livestock (Taye et al. 2019). Such for KLIP, leaving an important gap in our understanding a pattern of expenditures provides strong evidence that of the program’s impacts. Notwithstanding this premise, many households do have access to these livestock in- some important considerations can be made that could put markets, which are critical to protecting livestock also inform a future, more comprehensive agenda. during drought, and can use KLIP payouts to this end. Robust multi-year impact evaluation surveys on the IBLI Similar results were found by an independent impact program in Kenya and Ethiopia have evidenced consid- study led by German Agency for International Cooper- erable social and welfare benefits for pastoralists who ation (Deutsche Gesellschaft fur Internationale Zusam- have insured their livestock (Figure 4: box 3 - protect menarbeit, GIZ) (CED 2018). Self-reported satisfaction vulnerable). During good years, insured households with the program was high and many beneficiaries re- respond to their insurance coverage by increasing in- ported using the payouts at least partly for expenses on vestments in livestock veterinary and vaccination ser- their livestock (maintenance, restocking, and produc- vices, selling more livestock, and reducing their herd tion equipment) and for household needs. In addition, size (Jensen et al. 2017; Matsuda et al. 2019). These there is qualitative evidence of positive spillovers, such changes to production strategies result in positive im- as sharing payouts with neighbors. Despite the short pacts on indicators of well-being in good and drought lifetime of the program, KLIP households experience seasons, including increased household income per slightly lower levels of food insecurity and higher level adult equivalent and reduced reliance on costly ex ante of general insurance awareness. 14 5 BUILDING FINANCIAL RESILIENCE IN PASTORAL COMMUNITIES IN AFRICA Lessons learned and way forward T he first five years of implementation of KLIP have generated a huge amount of information on how to scale up a drought risk financing solution from the pilot innovative index insurance solution to build pastoralists’ resilience to drought. The active participation and finan- cial involvement of the public sector has been funda- stage to a national program. This is gaining considerable mental to the rapid expansion of the geographic scope attention from other African countries, who are express- and the number of insured households and is critical for ing interest toward implementing similar solutions. The continuation of the program, which aligns with agricul- following sections summarize key lessons learned. tural insurance initiatives worldwide. The GoK is provid- ing and supporting direct critical investments in aware- ness creation and financial infrastructure development (for example, registration systems) and is developing an of a PPA model Design  enabling regulatory and policy environment. The nega- tive side of a strong government leadership is the po- The PPA model developed under KLIP, while still needing tential program vulnerability to changing political con- refinements, is one of the key innovations of the program. text, which also reduces the private sector confidence A PPA approach was preferred for KLIP because private to make long-term investments. A strong involvement sector-only implementation proved difficult to scale and of the private sector into the planning and development keep the private sector interested in offering IBLI retail of the program, together with a medium-term budget al- coverage, due to the costs of distribution and relatively location framework from the government, can mitigate low uptake. The PPA helped in developing a new model this challenge. for sustainable livestock insurance provision and social protection. Two main considerations informed GoK deci- Private sector centrality and clearly defined role. The sion to provide full subsidy to some households: (a) pro- private sector has been the engine of KLIP, relying on vide substantial subsidy to crowd-in private sector and the experience and capacity built over years of imple- put in place mechanisms to expand the voluntary com- mentation of IBLI. KLIP relies on the private sector for ponent and (b) reduce SDL response cost when drought the administration of the insurance policies and the as- strikes. The full subsidy offered by GoK to vulnerable sociated financial transactions, the risk transfer to the households could be considered as social protection, reinsurance industry, and the actuarial analysis of the in- while voluntary purchases as commercial insurance. surance product, among others. The private sector also plays a critical role for the long-term sustainability of Government leadership. KLIP originates from a clear the program by supporting awareness creation efforts commitment of the GoK to support and scale an and by stimulating the expansion of the retail insurance 15 BUILDING FINANCIAL RESILIENCE IN PASTORAL COMMUNITIES IN AFRICA market and associated financial and extension services. of loss indemnification to a logic of early detection for Operational challenges encountered during KLIP, dis- early response. This proved to be more accurate and, cussed more in detail below, suggest, however, that even more importantly, more cost-effective in mitigating the private sector’s role needs to be carefully defined the impacts on drought, thus financially sustainable (by and that procurement and underwriting services com- reducing the cost of the premiums). Similar programs missioned by governments should include mechanisms should consider this innovative logic from the design (for example, performance assessments) to incentiv- phase and explore the use of the best technologies to ize investments on the retail market and financial infra- achieve the goal of early assessment toward mitigation structures to ensure that firms do not merely compete of catastrophic impacts. for the government’s tender without investing in exten- sion and on voluntary sales. Basis risk and quality assessment. The IBLI product im- plemented in KLIP has been proved to be simple, trans- Subsidized coverage. KLIP has provided fully subsi- parent, and, at least up to now, accurate in detecting dized coverage to all selected beneficiaries. This has major drought events. However, the lack of long-term been instrumental in developing the PPA and in rapidly spatially explicit ground data sets of rangeland biomass expanding the program. However, fully subsidized in- in Kenya did not allow a direct assessment of the index surance does have some important drawbacks if used accuracy with respect to the selected indicator of the ‘per se’ rather than as part of a comprehensive strategy risk covered (that is, forage scarcity). Similarly, there are toward incentivizing more awareness about the product no reliable long-term and high frequency datasets link- and the expansion of the insurance market in the target ing drought impacts on forage availability to livestock regions. In addition, the provision of full subsidies to conditions and household welfare/food security. These the same target beneficiaries in KLIP has in some cases gaps prevent a more comprehensive evaluation of the created double dipping and confusion issues with safe- basis risk and a comparative analysis of alternative in- ty net or cash transfer programs. Possible solutions to dices and product designs. The need of robust, trans- overcome this challenge include, for example, asking parent, and actionable strategies and methodologies target beneficiaries to pay at least a minimal contribu- for quality assessment of index insurance products, tion (that is, a token) for the coverage and introducing backed by dedicated ground data collection efforts, is a graduation process from the subsidy. A balanced use thus a priority for geographic scaling and product de- of smart subsidy schemes targeting different types of sign improvement with emerging technologies. beneficiaries and bonded to good practices from both the insurer and the clients could be an important step for the next stages of the program. implementation Operational  Effective implementation is as important as technical design Product  design, as experience with inaccurate registration of beneficiaries and delayed payments in KLIP has shown Accurate product design has been a pillar for KLIP and that the benefits coming from a sound product design continues to be a necessary component of the program can be largely undermined by operational challenges. in its upscaling trajectory. The capacity of the program Accurate identification and registration of the beneficia- to promptly adapt the product in response to stake- ries and efficient payout delivery mechanisms are es- holders’ feedback and to the evolving climatic/geo- sential. However, difficulties persist due to the lack of a graphic context has been of paramount importance to clear mechanism and workflow for preparatory steps to creating confidence about the reliability of KLIP. In addi- be undertaken in the event of a drought. Ensuring that tion, while there is sound evidence that the KLIP index the design of infrastructure for premium collection and is overall robust (that is, the basis risk is low), the geo- payout distribution is robust before the launch of simi- graphic scaling of the program to other Kenyan regions lar schemes is crucial to ensure development impact is (especially to semi-arid or sub-humid drylands dominat- achieved, trust is built in that scheme, and the scheme ed by agropastoral livelihoods) or new countries will re- is sustainable. quire careful product design work. This highlights the need for product design to become part of the planned Payout management. Drought shocks can be persistent government investments to support the program and/ and might pose considerable claim servicing challenges or the private sector. to the underwriting company(s). KLIP has paid out more than what has been collected as premium in some years, Early drought detection for early action. A milestone resulting in large loss ratio as it was seen in the 2016– for the program has been the shift from a livestock 2017 season. KLIP has paid out in six of ten seasons. mortality–asset replacement index design to a forage This has raised a cash call challenge, putting serious scarcity–asset protection paradigm, thus from a logic strain on timely handling of payouts. While the product 16 BUILDING FINANCIAL RESILIENCE IN PASTORAL COMMUNITIES IN AFRICA is designed to immediately pay after the end of the rainy Making a nd monitoring season, there have been delays in the range of more than three months after the result declaration. Such de- impacts lays in payouts are detrimental to sustainability of the scheme because they erode government and benefi- KLIP largely benefited from the long-term market and ciary’s confidence in the scheme. More efforts or weight capacity development efforts made in pastoral areas should be attached to the ability to service large claims as well as from the robust household-level impact as- during the underwriter selection process. It is common sessment conducted during the implementation of the for the underwriting company to focus on substantial retail IBLI initiative. However, considering its size, the premium collected through financing such scheme; program has shown limited capacity in carrying on with however, if the underwriting capabilities are weak, it will similar activities, especially at the level of pastoral com- struggle to service claims. munities. This is a serious drawback that needs to be addressed in the continuation of the program. Subsidy administration. Several options for subsidy ad- ministration exist and the best one should be selected Awareness creation and capacity development. based on the country’s procurement laws and regula- Awareness creation of clients and beneficiaries and ca- tions governing procurement of goods and services by pacity development at all levels in the public and pri- the public sector. In situations where some households vate sectors (that is, policy makers, institutional and are benefiting from full subsidies, the government may executive stakeholders, extension agents, insurance adopt the normal goods and services procurement ap- agents, program partners) have been important in the proach that in most cases involves expression of inter- KLIP agenda from its inception. Awareness creation ef- est, selecting the most qualified firm and getting a com- forts have been led by the government with support prehensive quote where the most competitive provider from knowledge institutions and variable level of invest- is selected. Experience in KLIP has shown that such an ment. However, targeting pastoral communities proved approach has undesired consequences, which include to be difficult for the government due to the cost and minimal investment to build necessary infrastructure for extension infrastructure deficiencies. When design- promoting voluntary uptake. The uncertainty that re- ing similar schemes in the future, it is critical to intro- sults from a winner-take-it-all procurement approach duce smart subsidized coverage early in the rollout of has negative impact on growth and development of the product, coupled with adequate budget provision the insurance market, so alternative options should be to cover the costs of awareness creation and capaci- considered. In addition, the following considerations ty development, using innovative techniques (such as should be made when launching the subsidized insur- e-learning and m-learning), workshops, and education- ance scheme with government support: (a) the full sub- al initiatives. This need for strengthening capacity at all sidy should be launched at the same time with a vol- levels is foundational and requires sufficient resources untary component (whether it is partially subsidized or for such schemes to achieve sustainability. not) and (b) an incentive-based structure to subsidies should be put in place, for example, by allocating the Program monitoring and impact assessment. The few full subsidies proportionally to the number of voluntary impact studies being conducted specifically on KLIP policies sold, thus incentivizing the private sector to in- were of limited scope, and no monitoring and rigorous vest in developing infrastructure for voluntary purchase impact assessment strategy has been effectively imple- promotion. Such an approach could ensure that private mented to assess or track the program’s impacts over sector players actively participate in the expansion of time. Several attempts have been made during KLIP to the program as per business growth strategy. implement a simple monitoring and evaluation frame- work, but the lack of clarity on objectives and funding Harmonized drought risk management. KLIP is part of streams has hampered the process. As a result, there a complex multilayered drought risk management strat- is still a strong need to increase understanding on the egy framework in Kenya, including early warning sys- evolution of KLIP and its impact on a growing number of tems, safety net programs, sovereign risk transfer, and beneficiaries (including the environmental component). insurance. Although they are complementary in prin- Such assessments are essential to ensuring that the ciple, these tools often have overlapping areas which insurance is operating as intended and improving the can create confusion among stakeholders, cause se- program. It is therefore critical to include monitoring, im- rious inefficiencies, and, without proper coordination, pact assessment, and cost-benefit analysis frameworks can result in missed opportunities for synergies. The fu- since the inception of such types of programs to ensure ture upscaling of KLIP would require a stronger effort of lessons and evidence are gathered. coordination and harmonization of KLIP with the other drought risk financing instruments, with the goal of pro- Linking financial resilience to physical resilience. While moting synergies between their finance mechanisms, KLIP is designed to contribute to building financial resil- targeting approaches, and management infrastructures. ience of pastoral households, its long-term development 17 BUILDING FINANCIAL RESILIENCE IN PASTORAL COMMUNITIES IN AFRICA effectiveness also critically depends on complementary inputs might be a serious constraint for the use of pay- interventions supporting physical resilience (that is, tar- outs in some areas or during catastrophic drought. While geting key livestock value chains), offering pastoralists it is difficult to assess if the financial resilience intervention multiple options for protecting their livestock and stimu- supported by KLIP could have provided a stimulus for the lating stronger markets. KLIP payouts have been used to development of livestock input markets, it is essential that, buy forage/fodder, water, and veterinary services, provid- in the continuation of the program, links with other live- ing evidence that many pastoral households do have ac- stock value chain interventions would be established, cess to these livestock input markets, but questions re- with targeted investments to create positive feedbacks main on the extent to which the limitations in the livestock loops toward comprehensive resilience building. 18 6 BUILDING FINANCIAL RESILIENCE IN PASTORAL COMMUNITIES IN AFRICA Conclusions K LIP has provided evidence that it is possible to de- velop a PPA to scale up an index-based insurance scheme to provide critical insurance protection to vul- replacement) to a drought early detection and im- pact mitigation paradigm (forage scarcity–asset protection) has been a fundamental step to improve nerable pastoralists in Sub-Saharan African drylands, the value and cost-effectiveness of the scheme. previously a feat which was thought to be impossible. • Accurate product design is critical to create trust Furthermore, it demonstrates how a government, by and deliver impacts, but the data infrastructure for creating an enabling environment and targeted subsi- assessing product quality and inter-comparison is dies, can crowd-in private sector capacity and expertise weak if not absent. The need of robust, transpar- to support its achievement of policy priorities. Needless ent, and actionable strategies and methodologies to say, with over five years of supporting the GoK de- for quality assessment of index insurance prod- sign and then implementing KLIP, a significant amount ucts is thus a priority. of knowledge and practical experience about the op- • Engaging with local and international stakeholders portunities and challenges of designing and operation- and tailoring the product to the specific agroeco- alizing a PPA has been gathered. Key lessons learned, logical and socioeconomic context and evolving that should be generalizable to similar programs, in- environmental conditions is a fundamental neces- clude the following: sity not just during the program design phases, but along the whole program implementation cycle. • Government leadership and direct investment in • Effective implementation is as (if not more) import- index insurance initiatives are possible and can be ant as technical design. Ensuring that the design effective if associated with a strong partnership of the premium collection and payment infrastruc- with the private sector with clearly defined roles ture is robust before the launch of similar schemes, and incentive structures. A mechanism for long- also leveraging on existing financial service infra- term public commitment needs to be established structures, is crucial to ensure development im- to guarantee the stability of the scheme. pact is achieved, trust is built in the scheme, and • Subsidies for scaling and consolidating the the scheme is sustainable. scheme are important and instrumental, but they • The future upscaling of KLIP and similar initiatives also need to be associated with smart targeting would require a strong effort of coordination and mechanisms and with incentives to the private harmonization of the different drought risk man- sector to develop and expand the market. agement initiatives to optimize their finance mech- • Creating awareness and strengthening capacity at anisms, targeting approaches, and data and man- all levels is foundational and requires enough re- agement infrastructures. sources for such schemes to achieve sustainability. • Impact assessment requires investment, planning, Moving forward, given the vastity of drylands in Africa and preparation. It is therefore recommended that and the millions of households considerably affected a rigorous impact study and cost-benefit analysis by drought shocks, there is significant scope to scale of the program be included in the design phase, to up KLIP-like approaches in other countries. The authors ensure lessons and evidence is gathered. trust that the lessons documented in this note can sup- • The shift from a drought assessment and impact port the effective design and implementation of future response paradigm (livestock mortality–asset programs. 19 BUILDING FINANCIAL RESILIENCE IN PASTORAL COMMUNITIES IN AFRICA References CED (Center for Evaluation and Development). 2018. Linear Regression Method.”  International Journal "Awareness, Perception and Outcomes of the Kenya for Research in Applied Science & Engineer- Livestock Insurance Program (KLIP)." Draft Evaluation ing Technology 5 (5):1627–1639. Report. Jensen, N., C. Barrett, and A. Mude. 2015. “The Favour- Chantarat, S., A. G. Mude, C. B. Barrett, and M. R. Carter. able Impacts of Index-Based Livestock Insurance: 2013. “Designing Index-based Livestock Insurance Evaluation Results from Ethiopia and Kenya.” ILRI for Managing Asset Risk in Northern Kenya.” Research Brief 52. Journal of Risk and Insurance 80:205-237. Jensen, N. D., C. B. Barrett and A. G. Mude. 2017. “Cash Chelanga, P., D. C. Khalai, F. Fava, and A. Mude. 2017. “De- Transfers and Index Insurance: A comparative im- termining Insurable Units for Index-based Livestock pact analysis from northern Kenya.” Journal of De- Insurance in Northern Kenya and Southern Ethio- velopment Economics 129: 14-28. pia.” ILRI Policy Brief 83. Jensen, N. D., A. G. Mude, and C. B. Barrett. 2018. “How Clarke, D., and S. Dercon. 2016.  “Dull Disasters? How Basis Risk and Spatiotemporal Adverse Selection Planning Ahead Will Make a Difference.” Oxford Influence Demand for Index Insurance: Evidence University Press. from Northern Kenya.” Food Policy 74:1 72–198. De Oto, L., A. Vrieling, F. Fava, and K. C. de Bie. 2019. Jensen, N., Q. Stoeffler, F. Fava, A. Vrieling, C. Atzberg- “Exploring Improvements to the Design of an Oper- er, M. Meroni, A. Mude, and M. Carter. 2019. ational Seasonal Forage Scarcity Index from NDVI “Does the Design Matter? Comparing Satellite- Time Series for Livestock Insurance in East Africa.” based Indices for Insuring Pastoralists against International Journal of Applied Earth Observation Drought.” Ecological Economics 162:59–73. and Geoinformation 82:101885. Kahiu, N., and F. Fava. 2018. "Feasibility Analysis for ex- Dingaan, M. N. V., and M. Tsubo. 2019. “Improved As- pansion of Kenya Livestock Insurance Programme sessment of Pasture Availability in Semi-arid Grass- (KLIP) in Agrosilvopastoral Counties in Kenya." land of South Africa.” Environmental Monitoring KLIP Tech-nical Report. and Assessment 2019:733. Lybbert, T., C. B. Barrett, S. Desta, and D. L. Coppock. Diouf, A. A., M. Brandt, A. Verger, M. E. Jarroudi, B. Djaby, 2004. “Stochastic Wealth Dynamics and Risk Man- R. Fensholt, J. A. Ndion, and B. Tychon. 2015. “Fod- agement among a Poor Population.” Economic der Biomass Monitoring in Sahelian Rangelands Journal 114:750–777. Using Phenological Metrics from FAPAR Time Mahyou, H., B. Tychon, M. Lang, and R. Balaghi. 2018. Series.” Remote Sensing 7 (7):9122–9148. “Phytomass Estimation Using eMODIS NDVI and FAO (Food and Agriculture Organization of the United Na- Ground Data in Arid Rangelands of Morocco.” Af- tions). 2018. "Pastoralism in Africa’s Drylands." Rome. rican Journal of Range & Forage Science 35:1–12. Fava, F. and A. Vrieling. 2021. "Earth Observation for Matsuda, A., K. Takahashi, and M. Ikegami. 2019. "Direct and Drought Risk Financing in Pastoral Systems of sub- indirect impact of index-based livestock insurance in Saharan Africa." Current Opinion in Environmental Southern Ethiopia."  The Geneva Papers on Risk and Sustainability 48:44-52. Insurance-Issues and Practice 44(3): 481-502. Janzen, S. A. and M. R. Carter. 2019. "After the drought: Poole, L., D. Clarke, and S. Swithern. 2020. The future The Impact of Microinsurance on Consumption of Crisis Financing: A Call to Action. London, UK: Smoothing and Asset Protection."  American Jour- Centre for Disaster Protection. https:// nal of Agricultural Economics 101(3):651-671. www.disas-terprotection.org/crisisfinance. Garba, I., B. Djaby, I. Salifou, I. Toure, A. Samba, Y. Yapo, Schucknecht, A., M. Meroni, F. Kayitakire, and A. Bourei- A. Agoumo, S. Soumana, A. Oumarou, and B. Ty- ma. 2017. “Phenology-based Biomass Estimation to chon. 2017. “Improving Fodder Biomass Modeling Support Rangeland Management in Semi-Arid in the Sahelian Zone of Niger Using the Multiple Environments.” Remote Sensing 9 (5):463. 20 BUILDING FINANCIAL RESILIENCE IN PASTORAL COMMUNITIES IN AFRICA Taye, M., V. Alulu, W. Gobu, and N. Jensen. 2019. “Live- Vrieling, A., M. Meroni, A. Shee, A. G. Mude, J. Woodard, stock Insurance Payouts and Coping Strategies of K. de Bie, and F. Rembold. 2014. “Historical Exten- Pastoralists during Drought.” ILRI Research Brief 90. sion of Operational NDVI Products for Livestock In- Tian, F., M. Brandt, Y. Y. Liu, A. Verger, T. Tagesson, A. surance in Kenya.” International Journal of Applied A. Diouf, K. Rasmussen, C. Mbow, Y. Wang, and R. Earth Observation and Geoinformation 28:238-251. Fensholt. 2016. “Remote Sensing of Vegetation Dy- Vrieling, A., M. Meroni, A. G. Mude, S. Chantarat, C. C. namics in Drylands: Evaluating Vegetation Optical Ummenhofer, and K. de Bie. 2016. “Early Assess- Depth (VOD) Using AVHRR NDVI and In Situ Green ment of Seasonal Forage Availability for Mitigat- Biomass Data over West African Sahel.”  Remote ing the Impact of Drought on East African pastoral- Sensing of Environment 177:265–276. ists.” Remote Sensing of Environment 174:44–55. Uhe, P., S. Philip, S. Kew, K. Shah, J. Kimutai, E. Mwan- Zewdie, Y., M. Taye and F. Fava. 2020. “Livestock insur- gi, ... and H Cullen. 2018. “Attributing Drivers of the ance for pastoralists in Ethiopia: Exploring opportu- 2016 Kenyan Drought.” International Journal of nities for scaling.” ILRI Policy Brief 29. Climatology 38:e554–e568. USAID (United States Agency for International Develop- ment). 2018. Economics of Resilience to Drought. Washington, DC: USAID. 21 BUILDING FINANCIAL RESILIENCE IN PASTORAL COMMUNITIES IN AFRICA Annexes BUILDING FINANCIAL RESILIENCE IN PASTORAL COMMUNITIES IN AFRICA ANNEX 1 Remote sensing data processing Introduction This annex describes the tasks required to process the remotely sensed NDVI time series data to obtain the Input for the index calculation. KLIP uses satellite imagery for calculating a seasonal forage scarcity index for each UAI. By the end of a given season (that is end of June for long rains, end of December for short rains), the forage scarcity index for each UAI is com- pared to that unit’s historical index readings to determine if a payout should be made. Three main remote sensing data processing steps are required: • Downloading the dekadal NDVI images from the internet during the seasonal coverage period for all the KLIP covered area; • Detecting and removing pixels with limited temporal variability; • Calculating a dekadal spatial average of the NDVI for each UAI. The steps are detailed in the following sections. the NDVI data Downloading  • For KLIP, the eMODIS NDVI C6 product is the data source. This NDVI data set is obtained from the Moderate Resolution Imaging Spectroradiometer (MODIS) flown onboard NASA’s AQUA satellite. The Earth Resources Observation Systems (EROS) Data Center (EDC) of the US Geological Survey (USGS) transforms the daily imag- es into 10-day composites (10 days is often referred to as ‘dekad’), which are referred to as eMODIS. • To reduce the remaining atmospheric effects, such as clouds, a temporal filtering is applied on the data, which requires three dekads before and three dekads after the value (image) to be filtered. For this reason, final 23 BUILDING FINANCIAL RESILIENCE IN PASTORAL COMMUNITIES IN AFRICA filtered eMODIS images are only available about one month later. Filtered eMODIS data are currently provided for Africa in the framework of FEWS-NET activities. The eMODIS data for the target areas are available from January 2003 up to date. • The long archive of data containing months with only filtered data is obtained for the East Africa region from https:// earlywarning.usgs.gov/fews/datadownloads/East%20Africa/eMODIS%20NDVI%20C6 (last access 02/09/2020). • The most recent 10-day composite usually is provided about three days after the end date of the compositing window. For example, the January 1−10 image is provided by January 13. Given the mentioned issues with filter- ing, the final filtered version of that same image will be available one month later, that is, by February 13. and masking Spatial averaging  • For each filtered 10-day NDVI image (for the nonoverlapping dekads), a spatial averaging of the NDVI values within a UAI needs to be performed (the updated shapefile should be requested to the SDL). The eMODIS pix- els are assigned to a specific UAI if their pixel center falls within that unit. In principle, all pixels for which this is true are used for calculating the spatial average. The only exceptions are those pixels that have a very limited temporal variability (that is if the difference between 5th and 95th percentiles of NDVI values in the historical time series is below 0.05). 24 BUILDING FINANCIAL RESILIENCE IN PASTORAL COMMUNITIES IN AFRICA ANNEX 2 KLIP index calculations Introduction This annex describes the tasks required to convert the processed remotely sensed NDVI time series into a forage scarcity index and to calculate seasonal payouts according to the KLIP methodology. The following calculation steps are necessary to convert dekadal NDVI values for each UAI into the insurance index: • Monthly NDVI calculation. Dekadal NDVI time series in each UAI are converted into monthly NDVI time series by averaging the three dekadal NDVI values in each month. • Temporal aggregation. The monthly NDVI data are averaged over time into a seasonal index (seasonal NDVI). Two consecutive rainfall seasons are observed in Northern Kenya, followed by respective dry seasons. They are referred to as ‘long rains’ (March–June) and ‘short rains’ (October–December). NDVI values for each UAI are averaged for the long rains and short rains months. • Standardizing aggregated NDVI data (standard score). The standard score (Z-score) of each season’s (that is, long rains and short rains) seasonal NDVI value is calculated with reference to the historical average and stan- dard deviation (the benchmark period is from 2003 to the last available observation at the time the KLIP annual policy is issued). Index calculation The following steps show the mathematical formulae for the index calculation process. The NDVI data used for constructing the index and parameterizing the range of insurance contracts are depicted as NDVI (​ NDVI​ ​  i,d,m,y​ ​​ ​ NDVI​ ) for each UAI i in dekad d, month m, and year y and are obtained by preprocessing pixel-level dekadal ​  i,d,m,y​ NDVI​ ​ NDVI data of all pixels in each UAI. The following steps are required to calculate the index from NDVI (​  i,d,m,y​ ​​ ​ NDVI​ i,d,m,y​​ ​): 1. Monthly NDVI (​ NDVI​ ​   i,m,y ​​​) for each UAI i in month m and year y is obtained by averaging the three dekadal val- ​ ues in each month: 2. Seasonal NDVI (​​​Avg ​ NDVI​   i,m,y ) for each UAI i in month m and year y is obtained by averaging monthly NDVI from ​ the beginning to the end of the long rain season (March–June) and short rain season (October–December): 3. Standard score (Z-score) of the seasonal NDVI for each UAI i in each month m and each year y (​​​ZAvg ​ NDVI​  i,m,y ) is then calculated using the historical mean (E) and standard deviation (SD) of the particular UAI and month: NDVIi​ Avg ​ (​  − E​  ,m,y​ NDVIi​ Avg ​ )​  ,m​ ZAvgNDVIi​ ​   =  ______________________  ,m,y​     ​        ​ (​ SD​ Avg ​ NDVIi​ )  ,m​ The KLIP index (I) is numerically equal to the ​​​​ ZAvgNDVI​   i,m,y : ​ I​ ​  = ZAvgNDVI​  i,m,y​  i,m,y​ 25 BUILDING FINANCIAL RESILIENCE IN PASTORAL COMMUNITIES IN AFRICA Payout determination A payout of the insurance contract for each UAI i is triggered when the index I in the last month M of the covered period (that is, June for long rains and December for short rains) in year y falls below a predetermined value of I i,M,y for that month, called the attachment threshold. In case the index is lower than another predetermined value called exit threshold, the total sum insured allocated to that season is paid out.3 Between the attachment and exit values, the payouts are a linear function of the index. A minimum payout at 5 percent of the total sum insured allocated to that season is given when the index value falls below the attachment, but the calculated payout is below the minimum payout threshold. Attachment and exit determination For each UAI and season (that is, long rains and short rains), the attachment and exit values are set according to the • following procedure: • The attachment values are set to the 20th percentile of the index values' empirical distribution including all the seasonal index data from 2003 to the date the policy is issued. The exit value is fixed to −1.61. 3 The sum insured allocated to the short rains and long rains seasons are 42 percent and 58 percent of the total sum insured annually, respectively. 26 BUILDING FINANCIAL RESILIENCE IN PASTORAL COMMUNITIES IN AFRICA ANNEX 3 Roles and responsibilities of public and private sectors The GoK, motivated by willingness to end drought emergencies, initiated the process of designing and implement- ing KLIP with technical assistance from both the World Bank and the International Livestock Research Institute (ILRI). Since the early stages of KLIP design the important role of both public and private sector players was recognized and a joint implementation approach was recommended with clarity on roles and responsibilities of public and pri- vate actors. Roles undertakenby the public sector during design and implementation of KLIP • Program management. KLIP was initiated by the government; therefore, the SDL in the MALF played the lead- ership role. The SDL formed a dedicated PMU to guide implementation and to coordinate government insti- tutions with stake in the implementation process, for example, Insurance Regulatory Authority (IRA). The PMU comprised technical officers, administration officers, and a process management adviser and continued to re- ceive technical assistance from ILRI and the World Bank during design and implementation of KLIP. The PMU is responsible for the day-to-day operational running of KLIP, including awareness creation and coordination of beneficiary selection, and working closely with underwriting company(s) to ensure payouts are delivered as per service-level agreements. IRA provided advisory on matters relating to regulation and product approval. • Beneficiary selection. The government provides full subsidy to vulnerable households, selected using criteria developed through consultative approach led by the SDL and involving other organizations or institutions work- ing on drought risk management such as the National Drought Management Authority (NDMA) and develop- ment partners providing humanitarian response. The selection of KLIP beneficiaries considers the following: (a) livestock ownership; (b) whether the household is already a beneficiary from existing social safety net schemes, such as the NDMA-led Hunger Safety Net Program (HNSP) that operates in some counties; (c) whether house- holds have formal financial access or demonstrate willingness to open a bank account or any other recognized formal payment system. The PMU, working closely with representatives from county governments and commu- nity leaders, undertook beneficiary registration, capturing all the necessary information, including names of the beneficiary, national Identification numbers, next of kin, and other relevant information. The PMU determined the number of households to receive subsidy from each UAI. The data on beneficiaries are shared with selected underwriter(s) for pricing and payout administration. • Subsidy administration. The SDL pays premium on behalf of selected households to selected underwriters, chosen through tendering process. Premium budget is allocated within the MALF budget earmarked for live- stock insurance purposes. The procurement of an insurance company to provide cover under KLIP is undertak- en by the SDL with support from IRA. It involves preparing tender documents which outline the conditions for the cover, short-listing interested companies that should meet certain selection procedures as outlined in the tender documents. The insurance company offering the best quote and other support services is contracted to offer the IBLI product. Interested insurance companies can quote as individuals or as a consortium. The winning insurance company is encouraged to share the business with interested local underwriters and reinsure the 27 BUILDING FINANCIAL RESILIENCE IN PASTORAL COMMUNITIES IN AFRICA remaining portion. The winning underwriter receives premium from the SDL for the selected beneficiaries and makes payout directly to the beneficiaries if there is a trigger. • Awareness creation. The KLIP PMU is also coordinating awareness creation and capacity-building efforts for pastoralists and institutional stakeholders through targeted campaigns in pastoral areas and county-lev- el awareness creation events. Up to 15 percent of the GoK’s annual budget for KLIP is allocated to this effort. by the private sector Roles undertaken  • Underwriting. The underwriting function is undertaken by private insurance companies which might choose to underwrite as individual insurance or form a pool of interested companies. The SDL chose to apply existing pro- curement laws to procure the right insurance company to underwrite IBLI under KLIP. Kenya procurement laws require the procuring entity to develop tender documents and invite insurance companies to express interest. The companies meeting a minimum selection criterion are invited to quote as per the terms and conditions of tender documents. The initial call for expression of interest resulted in a consortium of six insurance companies and one individual company being short-listed to provide quotation. The first and second KLIP cycles were in- sured by the selected consortium annually while in the subsequent tenders the government extended procure- ment from one year to three years after the private sector raised concerns of one-year contracts. The tender for the third cycle was won by a single insurance company to provide insurance cover for three years. The tender conditions encouraged business sharing and the winning entity could share the business with other interested insurance companies locally. The insurance companies consider KLIP insurance a high risk business and take just a small portion of risk by transferring a big part of the risk to reinsurance companies. At the design stages of KLIP, consideration was given to set up a Technical Support Unit (TSU) to house technical expertise centrally, given the costs of technical tasks related to livestock insurance. The TSU was expected to perform a range of services for the private sector that included (a) demand assessment; (b) product design and rating, including basis risk analysis; (c) design of operating systems and procedures; and (d) training of stakeholders and coordi- nation of awareness creation. While the role to be played by the TSU was considered important for the private sector to play its role effectively, it was never set up. The World Bank and ILRI continued to support some of those tasks with the danger of leaving the private sector weak when the technical assistance comes to closure. • Payouts management. Under KLIP’s fully subsidized component, the SDL provides the underwriting insurance company with a list of beneficiaries. The underwriting company is expected to make direct payouts to the list of beneficiaries provided in case trigger conditions are met. The contractual agreement between the SDL and un- derwriting company stipulates the period required to deliver payout to the beneficiaries. The underwriting com- pany is free to choose the most effective and affordable methods to deliver payouts. Bank and mobile money accounts are the preferred methods by insurance companies; bankers’ checks are used when the beneficiaries lack bank or mobile money account. The underwriting insurance company is expected to organize payouts us- ing its preferred model as long as beneficiaries receive the money without delay. There is a strong partnership between insurance companies, banks, and mobile money operators, which has been developed to deliver pay- outs. In addition, there is consideration to use banks’ infrastructure (and other distribution channels) to promote the voluntary component, which is a requirement by the GoK. • Calculating agent. The calculating agent is another private sector role. The underwriting company agrees with the SDL on a suitable calculating agent, expected to be knowledgeable on remote sensing and insurance to de- termine how the season has performed and the payouts. The initial calculating agent work was initially procured to the private sector by the underwriter. However, since the question on who should pay for the services (SDL or underwriting company) has not been adequately addressed, since 2016 ILRI has been providing the service on an interim basis while identifying a suitable approach for service procurement and payment. 28 BUILDING FINANCIAL RESILIENCE IN PASTORAL COMMUNITIES IN AFRICA ANNEX 4 KLIP payouts maps The following maps illustrate the seasonal payout per beneficiary (in KES) from the launch of the KLIP in 2015 to the long rains of 2020. Areas in white are not covered by KLIP. Source: Authors’ elaboration 29 BUILDING FINANCIAL RESILIENCE IN PASTORAL COMMUNITIES IN AFRICA Source: Authors’ elaboration 30 BUILDING FINANCIAL RESILIENCE IN PASTORAL COMMUNITIES IN AFRICA