SOCIAL PROTECTION & JOBS | P  OLICY & TECHNICAL NOTE APRIL 2021 | No. 24 HIGHLIGHT Can the luck of the draw help social Lottery — a simple random draw — has been used in selecting safety nets?1 beneficiaries of public work programs in context as diverse as Argentina Paul Bance and and the Central African Republic. However, despite a burgeoning Pascale Schnitzer literature on targeting, this approach has hardly been studied. This policy With an increase in the frequency and severity of shocks, note discusses how lotteries compare from the COVID-19 pandemic to climate change and violent conflicts, the quest to end poverty has suffered against other targeting methods in its worst setback in decades (WB 2020). This situation terms of efficiency, legitimacy, and drives calls to rethink the welfare state2 and supports an readiness; and if lotteries could be unprecedented expansion of social safety nets (Gentilini expanded beyond their traditional et al. 2020). In most countries though, financial resources are not yet available for an adequate coverage use for public works to cash transfers. of the population (Beegle et al. 2018 and WB 2018). While more research is needed, there This mismatch between resources and needs is deepen- is no immediate reason why lotteries ing policy debates about targeting of social assistance should not be used for targeting social 1 A shorter version of this note was published on February safety net interventions, especially 11, 2021 in the Let’s Talk Development Blog: https://blogs. worldbank.org/developmenttalk/can-luck-draw-improve-social- when responding to emergencies safety-nets. The authors are grateful to Arthur Alik-Lagrange, in ultra-poor and fragile settings. Paolo Belli, Paul Bisca, Diana Cheung, Aline Coudouel, Chisako Fukuda, Ugo Gentilini, Margaret Grosh, Cem Mete, and Quentin Stoeffler for their comments and inputs. 2 For more information see https://www.economist.com/ leaders/2021/03/06/how-to-make-a-social-safety-net-for-the post-covid-world 1 SOCIAL PROTECTION & JOBS | P  OLICY & TECHNICAL NOTE APRIL 2021 | No. 24 recipients. As social protection turns to adaptive to reduce A primer on lottery in public vulnerability to shocks (Bowen et al, 2020), the question of the efficiency of targeting (i.e. are we selecting the works programs intended population?), is increasingly coupled with trade- There are many variants of lotteries to select beneficiaries offs around readiness. Is the proposed targeting method of public works programs. Here is a standard case. Assume also a practical one given the operational constraints? your budget allows to recruit 250 workers in a village. You Legitimacy of targeting also becomes important as social open registration for the program without any condition, safety nets are deployed more and more in areas where i.e., all adult villagers are eligible. Say 1,000 people enroll social cohesion is at stake. Is a targeting method accept- to participate. On the lottery day, 1,000 numbered cou- able to the public or does it generate extra tensions or pons are placed in a ballot box. Each of the 1,000 par- conflict? Considering all these three dimensions — effi- ticipating villagers pick a coupon by hand. Those who get ciency, readiness and legitimacy — is especially important a coupon with a number between #1 and #250 become when vetting targeting for crisis response in ultra-poor and program beneficiaries. The individual with coupon #251 is fragile settings. the first on the waiting list and so on. There is an abundant literature on targeting and the per- It is a simple random draw: chance is the only factor at play formance of a range of methods: categorial, communi- here! So, how can a lottery make for a targeting mecha- ty-based, proxy-means testing, etc. (Devereux et al. 2017). nism? Let’s look at efficiency first. The requirement of having However, in these discussions and reviews, an approach to work to access the program’s benefit — “cash-for-work” is barely mentioned and seldom compared with others: in other words — generates self-selection among eligible lotteries. Authoritative global reviews of targeting meth- people. The standard model goes that the poorer the people ods do not even mention it as an option (Coady, Grosh, are, the lower the opportunity cost of participating in the & Hoddinott 2004, Slater et. al 2009, Devereux et al. program is (Besley & Coate 1992). In the jargon of con- 2017). However, lottery is a frequent tool. It is used in tract theory, it is a signaling game: the agents (the targeted public works programs in context as diverse as Argentina, group) credibly convey (by accepting the requirement) the Central African Republic, Egypt, and Laos. While less some information about themselves (their poverty status) common, other social assistance programs have relied on to the principal (the program agency). In short, only the lotteries too. And beyond social assistance, governments poor accept to participate in public work programs. Poverty have also favored lotteries for public policies, from school profiles of participants in public works as in Cote d’Ivoire admissions to military service drafts, visa allocation and and India have shown that, indeed, self-selection can work vaccine distributions3. So, is there a possible disconnect (Alik-Lagrange & Ravallion 2018; Bertrand et al. 2016). between research and practice? But is it fair? When budgets are not enough to cover all What are the pros and cons of lotteries for targeting in terms the poor, a lottery can give equally-deserving people an of efficiency, legitimacy, and readiness? Should we extend equal chance to receive the benefit (Stone 2007). The lot- their use beyond public works to other social safety nets? tery system is also transparent for it is easy to explain to all. People understand the concept of chance (or “luck”) very well across cultures and continents. Also, lotteries are held in the open and the process is participatory — agency matters, with people actively (and playfully) engaging 3 For examples see: https://dcps.dc.gov/page/my-school-dc-lottery-how-apply; https://www.sss.gov/about/return-to-draft/lottery/; https://www.dvlottery.com/. 2 SOCIAL PROTECTION & JOBS | P  OLICY & TECHNICAL NOTE APRIL 2021 | No. 24 An Example: Public Works Lotteries in the Democratic Republic of Congo The Social Fund of the Democratic Republic of Congo has been implemented public works program for the last 10 years, using lotteries as a targeting mechanism. 26,643 persons registered to participate in its most recent operation in Kananga, the capital of the Kasai Central province. 27 lotteries — one per neighborhood — were organized in public spaces to recruit 6,750 workers in November 2020. The costs of organizing the lotteries was minimal with no additional staffing required and less than US$1000 in equipment (plastic boxes, tarpaulins, etc.). It is a good example of readiness with a large scale and fast selection process. Despite being a fragile region, no major security incident or allegations of fraud have been reported by the authorities, the media, the peacekeeping mission, and the public (through community surveys and project’s hotline). The legitimacy of the process is confirmed by the broad acceptance of the lotteries as a fair, transparent, and participatory system. A post-lottery survey shows that the beneficiaries, slightly more than 50 percent women, are poor and very vulnerable to shocks. 72 percent are unemployed; among those employed, all work in the informal sector and their self-reported monthly average income is about US$20. 94 percent have not saved anything in the last six months. 88 percent live on a meal per day and 64 percent have skipped a meal in the last 30 days. 66 percent believe it is unlikely they could find US$25 to respond to an emergency. 26 percent come from displaced households, 3 percent include demobilized combatants, and 6 percent care disabled family members. While most children of beneficiaries go to school, 36 percent of them rely on traditional medicine when sick and 76 percent do not have direct access to drinking water. 24 percent do not feel safe where they live. These data show that self-selection worked and that lotteries have been not been detrimental to the efficiency of the targeting process. All data are available from the authors. Free and voluntary Transparent and Draw in public with Immediate enrollment registration participatory process full access to observers and benefits Photos courtesy of Fonds Social de la RDC 3 SOCIAL PROTECTION & JOBS | P  OLICY & TECHNICAL NOTE APRIL 2021 | No. 24 during the drafting. These considerations are especially In a nutshell: There is not enough assistance for all. Nobody important in fragile and violent settings where communi- knows who is who. Efforts to know are vain. And it does ties may distrust officials and external actors. In such con- not matter anyway as all need assistance. Unfortunately, texts, the do-no-harm principle invites us to let legitimacy these are frequent circumstances in crisis response. prevail over efficiency when assessing targeting methods. The new frontier: extending Last — and this is often overlooked in the targeting debate — operational readiness is essential for crisis response lotteries to cash transfers? and humanitarian programs. In that respect, lotteries are Now, would it make sense to consider lotteries for uncon- remarkable: they are fast and inexpensive; they require no ditional cash transfers? These programs are increasingly technology or administrative capacities; and they are rep- popular in social and humanitarian assistance (CaLP licable in any environment, both rural and urban. Londo, 2020), their positive impact on beneficiaries is well doc- a government-led public work program has been able to umented (Bastagli et al, 2019), and one could argue that carry out lotteries without incident in all the 71 districts they are the new gold standard (Ivaschenko et al. 2018). of the war-torn Central African Republic4. In a nutshell, By definition, unconditional cash transfers have no work lotteries are ideal for these emergency operations in low requirement, so self-selection does not apply, and the con- capacity, fragile contexts. ventional wisdom goes that lotteries make no sense any- more. Indeed, no cash transfers programs use lotteries for When is “luck” enough of a targeting households. targeting mechanism for public works? Let’s look at this question again, but from a different angle: How It does not mean that lottery is the silver bullet of targeting for social assistance though. When is luck enough of a tar- would lotteries compare against geting mechanism in public works programs? Assuming a the most widespread targeting budget constraint, lotteries may be the best strategy when: methods? i. There is asymmetric information. In plain English, there is a lack of accurate data about the targeted popula- We are in luck: several studies have assessed and compared tion, i.e., no social registry, fiscal records, or ways to the efficiency of actual targeting methods by benchmark- get an accurate picture of the people in need. ing them against a “worse-case” no-targeting scenario defined as a random selection of beneficiaries… in other ii. Fixing that data gap — what others targeting meth- words, simulating a lottery (see for example Coady, Grosh ods do — would be prohibitively costly or lengthy and Hoddinott 2004). What is the main result of these (think of timely shock response), or it would increase comparative exercises? In many cases, a lottery would per- the risk of violence, which is often the case in contexts form as good as community based targeting (Premand and of low social cohesion and distrust of outsiders. Schnitzer 2020) or survey-based methods (Brown et al. iii. The targeted population is homogenous enough that 2018) in ultra-poor settings, i.e., the population is poor on the difference between two eligible people does not average with low standard deviations. It is especially true matter anymore. It would be the case when respond- for emergency programs responding to food insecurity, or ing to most covariate shocks in ultra-poor settings. after applying geographical targeting (Schnitzer 2019). 4 For information on the program see: https://www.facebook.com/londorca/ 4 SOCIAL PROTECTION & JOBS | P  OLICY & TECHNICAL NOTE APRIL 2021 | No. 24 Overall, in ultra-poor settings, the variations between tar- transfer programs aiming to address food insecurity, to 79 geting methods in terms of efficiency are arguably small, percent in Niger were a survey based method was applied so that the question of efficiency may not be decisive and to identify food insecure households to benefit from a cash therefore relevant in the first place5. What about legitimacy transfer program. and readiness? Figure 1: Share of individuals not selected by a targeting Knowledge about the legitimacy of lotteries builds largely scheme that found the selection process fair on anecdotal evidence. For instance, in Niger, lotteries 100 90 79 where introduced to the program to select beneficiary vil- 80 73 72 73 lages in the context of a randomized impact evaluation. 60 38 40 40 Since then, the program kept relying on lotteries. As men- 20 tioned in Gertler et al. (2016), “Its value as a transpar- 0 Senegal Senegal Burkina Niger Niger Niger DRC ent, fair, and widely accepted operational tool to allocate HEA CBT PMT PMT FCS CBT Lottery benefits among equally deserving populations justified its Figure is based on Diagne (2017) for Senegal, and Bank’s own calcula- continued use [after the impact evaluation was over] in tions for Burkina Faso, DRC, Niger. the eyes of program implementers and local authorities.” Similarly, in post-conflict northern Liberia a communi- While our knowledge on the legitimacy of lotteries is lim- ty-driven reconstruction program selected beneficiary vil- ited, we know much more about the counterfactuals: cur- lages through public lotteries (Fearon et al. 2008). They rent targeting methods have been questioned and may report that “'we monitored the lottery process for conflict have exacerbated tensions in some instances (Pavanello risks. Reports strongly indicated that not only were there at al. 2016; Kardan et al. 2010; Sumarto 2020). This is no conflicts resulting from the randomization but that especially the case in ultra-poor settings where budgets communities viewed the process favorably and appreci- are largely insufficient to cover needs, and communities ated the equity of the procedure.” feel that everyone is equally poor or deserving. The chal- lenges of categorical targeting are well documented with Based on a compilation of studies with information on tar- programs for refugees in poor hosting communities (ODI geting legitimacy, Figure 1 shows the share of non-bene- 2020; Samuels et al. 2020). The complexity of survey-based ficiaries who nonetheless found the selection process fair. poverty scoring can create challenges of trust as people do The seven programs operate in different contexts and have not understand why some are selected while others they different objectives, design, benefits, and coverage, so a believe to be as equally poor are not (Adato et al. 2004). proper comparative analysis cannot be carried out and Community-based targeting faces the risk of manipulation conclusions in difference in results cannot be attributed by local committees and authorities (Conning & Kevane to targeting methods exclusively. Variation in legitimacy is 2002), and decentralized decision-making has sometimes high. In the best performing case — a lottery done in DRC been refused by communities fearing disputes among them. for a cash for work program — 90 percent of non-bene- ficiary community members felt that the selection process On operational readiness, the evidence is again largely through lotteries was fair. For methods other than lotteries, anecdotal and there is no proper comparative research. this number ranges from 38 in Senegal, where a commu- Most would agree that, in general, organizing lotteries nity based approach (Household Economy Analysis6) was is low-cost, takes days, and requires no skills; communi- used to identify potential beneficiaries for seasonal cash ty-based targeting is affordable and a matter of weeks, 5 https://quentinstoeffler.weebly.com/uploads/4/0/2/6/40265181/sahel_targeting_spj_bbl_11052020.pdf 6 For more information on this method see Schnitzer (2019). 5 SOCIAL PROTECTION & JOBS | P  OLICY & TECHNICAL NOTE APRIL 2021 | No. 24 and it works with some basic administration; and surveying More systematic research is needed to make this case. household poverty is more expensive, should be planned However, there is no immediate reason, based on effi- over months, and needs advanced skills and higher admin- ciency, legitimacy, and readiness, why lotteries should not istration. It does not mean that a method is better than be used — in combination with geographic targeting — another. It is a question of objectives. Households surveys for cash transfers in ultra-poor and fragile settings, espe- are the right investment for building sustainable social cially when responding to emergencies. protection systems and there is a strong case for com- munity-based approaches in many contexts. But absent a An invitation for more pre-existing social registry or an efficient local governance, these targeting methods may not compare well with a lot- discussions, research, and tery when you are short of budget, time, and capacities, experimentation that is in most crisis situations in developing countries. In addition, lotteries can easily be repeated, so that people Universal social protection remains the goal. However, excluded once get a chance at the next round. Programs today, in a context of growing needs and constrained can actually be set up to ensure a regular rotation of ben- financial spaces, selecting beneficiaries is increasingly dif- eficiaries, which in return reinforces both effectiveness and ficult. Let’s not shy away from lotteries in the targeting legitimacy compared to more static approaches of target- debate anymore. Lotteries do not have the elegance and ing whose results cannot be updated frequently. sophistication of others targeting methods, but their sim- plicity may turn to have value and offer a solid alternative in the most challenging contexts. 6 SOCIAL PROTECTION & JOBS | P  OLICY & TECHNICAL NOTE APRIL 2021 | No. 24 References Adato, Michelle, Terry Roopnaraine, Fabiola Alvarado Álvarez, Leticia Böttel Peña, and Gladys Meléndez Castrillo. "A social analysis of the Red de Protección Social (RPS) in Nicaragua." (2015). Alik-Lagrange, A., & Ravallion, M. (2018). Workfare versus transfers in rural India. World Development, 112, 244-258. Bastagli, Francesca, Jessica Hagen-Zanker, Luke Harman, Valentina Barca, Georgina Sturge, and Tanja Schmidt. "The impact of cash transfers: a review of the evidence from low-and middle-income countries." Journal of Social Policy 48, no. 3 (2019): 569-594. Beegle, Kathleen, Aline Coudouel, and Emma Monsalve, eds. Realizing the full potential of social safety nets in Africa. World Bank Publications, 2018. Bertrand, M., Crépon, B., Marguerie, A., & Premand, P. (2016). Impacts à Court et Moyen Terme sur les Jeunes des Travaux à Haute Intensité de Main d’œuvre (THIMO): Résultats de l’évaluation d’impact de la composante THIMO du Projet Emploi Jeunes et Développement des compétences (PEJEDEC) en Côte d’Ivoire. Washington DC: Banque Mondiale et Abidjan: BCP-Emploi. Besley, T., & Coate, S. (1992). Workfare versus welfare: Incentive arguments for work requirements in poverty-alleviation programs. The American Economic Review, 82(1), 249-261. Bowen, Thomas, Carlo Del Ninno, Colin Andrews, Sarah Coll-Black, Kelly Johnson, Yasuhiro Kawasoe, Adea Kryeziu, Barry Maher, and Asha Williams. Adaptive Social Protection: Building Resilience to Shocks. World Bank Publications, 2020. Brown, C., Ravallion, M., & van de Walle, D. (2018). A poor means test? Econometric targeting in Africa. Journal of Development Economics, 134, 109-124. Coady, David, Margaret Grosh, and John Hoddinott. Targeting of transfers in developing countries: Review of lessons and experience. The World Bank, 2004. Conning, Jonathan, and Michael Kevane. "Community-based targeting mechanisms for social safety nets: A critical review." World development 30, no. 3 (2002): 375-394. Devereux, S., Masset, E., Sabates-Wheeler, R., Samson, M., Rivas, A.-M., & Te Lintelo, D. (2017). The targeting effectiveness of social transfers. Journal of Development Effectiveness, 9(2), 162-211. Diagnes Fatou, 2017. Etude des synergies possibles entre les méthodologies de ciblage HEA et le Registre National Unique (RNU). Fearon, James, Macartan Humphreys, and Jeremy Weinstein. "Community-driven reconstruction in Lofa County: impact assessment." Unpublished manuscript (2008). Fiona Samuels, Francesca Bastagli and Maria Stavropoulou with Nur Tukmani, Hiba Abbani and Georgia Plank. Social cohesion and stability between Syrian refugees and host communities. (2020). Gentilini, Ugo, Mohamed Almenfi, Ian Orton, and Pamela Dale. "Social protection and jobs responses to COVID-19." (2020). https://openknowledge. worldbank.org/handle/10986/33635 Gertler, Paul J., Sebastian Martinez, Patrick Premand, Laura B. Rawlings, and Christel MJ Vermeersch. Impact evaluation in practice. The World Bank, 2016. Institute for Economics & Peace (IEP). COVID-19 and Peace, Sydney, June 2020. Available from: http://visionofhumanity.org/reports (accessed 15 March 2021). Ivaschenko, Oleksiy, Claudia P. Rodriguez Alas, Marina Novikova, Carolina Romero Robayo, Thomas Vaughan Bowen, and Linghui Zhu. The state of social safety nets 2018. No. 124300. The World Bank, 2018. Kardan, Andrew, Ian MacAuslan, and Ngoni Marimo. "Evaluation of Zimbabwe’s emergency cash transfer (ZECT) programme: Final report." Concern Worldwide, World Food Programme, and Oxford Policy Management (2010). Pavanello, Sara, Carol Watson, W. Onyango-Ouma, and Paul Bukuluki. "Effects of cash transfers on community interactions: emerging evidence." The Journal of Development Studies 52, no. 8 (2016): 1147-1161. 7 SOCIAL PROTECTION & JOBS | P  OLICY & TECHNICAL NOTE APRIL 2021 | No. 24 Premand, P., and Schnitzer, P. (2020). Efficiency, Legitimacy, and Impacts of Targeting Methods: Evidence from an Experiment in Niger, The World Bank Economic Review;, lhaa019, https://doi.org/10.1093/wber/lhaa019 Schnitzer, P. (2019). How to target households in adaptive social protection systems? Evidence from humanitarian and development approaches in Niger. The Journal of Development Studies, 55(sup1), 75-90. Slater, R., Farrington, J., Vigneri, M., Samson, M., & Akter, S. (2009). Targeting of social transfers: A review for DFID. London: ODI. Stone, P. (2007). Why lotteries are just. Journal of political philosophy, 15(3), 276-295. Sumarto, Mulyadi. "Welfare and Conflict: Policy Failure in the Indonesian Cash Transfer." Journal of Social Policy: 1-19, 2020. The Cash Learning Partnership (CaLP) (2020). The state of the world’s cash 2020: cash and voucher assistance in humanitarian aid. July 2020. https://www.calpnetwork.org/wp-content/uploads/2020/07/SOWC2020-Full-Report.pdf World Bank. 2018. The State of Social Safety Nets 2018. Washington, DC: World Bank. © World Bank. https://openknowledge.worldbank.org/ handle/10986/29115 License: CC BY 3.0 IGO. World Bank. 2020. Poverty and Shared Prosperity 2020: Reversals of Fortune. Washington, DC: World Bank. https://www.worldbank.org/en/ publication/poverty-and-shared-prosperity © 2021 International Bank for Reconstruction and Development / The World Bank 1818 H Street NW Washington DC 20433 Telephone: +1 (202) 473 1000 Internet: www.worldbank.org This work is a product of the staff of The World Bank with external contributions. The findings, interpretations, and conclusions expressed in this work do not necessarily reflect the views of The World Bank, its Board of Executive Directors, or the governments they represent. The World Bank does not guarantee the accuracy of the data included in this work. The boundaries, colors, denominations, and other information shown on any map in this work do not imply any judgment on the part of The World Bank concerning the legal status of any territory or the endorsement or acceptance of such boundaries. RIGHTS AND PERMISSIONS The material in this work is subject to copyright. Because The World Bank encourages dissemination of its knowledge, this work may be reproduced, in whole or in part, for noncommercial purposes as long as full attribution to this work is given. Any queries on rights and licenses, including subsidiary rights, should be addressed to World Bank Publications: The World Bank Group 1818 H Street NW Washington, DC 20433, USA fax: +1 (202) 522 2625 e-mail: pubrights@worldbank.org. Project 76635 8