Productive Inclusion in Latin America: Policy and Operational Lessons A Synthesis Note 1 Table of Contents Acknowledgements ..................................................................................................................................... 3 1. Introduction ......................................................................................................................................... 4 2. Productive Inclusion: Definition and Operational Framework ...................................................... 7 2.1 Addressing Productive Constraints and Social Bottlenecks ................................................... 9 2.2 Productive Inclusion: An Operational Framework with Key Pillars................................... 11 3. Productive Inclusion Approaches in the Region- Key Operational Elements ............................. 14 3.1 Institutional and Policy Mandate ............................................................................................ 14 3.1.1 Target Population ................................................................................................................. 19 3.1.2 PI Components ...................................................................................................................... 22 3.2 Implementation Institutions for Program Delivery ............................................................... 28 3.3 Information Systems to Enhance Coordination ..................................................................... 29 3.4 Incentives: Financing and Budgetary Arrangements Supporting the PI Package ............. 31 4. Evaluating PI: Challenges and Results So Far ............................................................................... 33 5. Lessons Learned ................................................................................................................................ 35 References .................................................................................................................................................. 37 Annex 1 Comparative table: Sample of Impact Evaluations of Productive Inclusion Programs in LCR ............................................................................................................................................................ 41 2 Acknowledgements This report was produced by a core World Bank team co-led by Edmundo Murrugarra (Senior Social Protection Economist) and Aylin Isik-Dikmelik (Senior Economist), and including Tomas Ricardo Rosada Villamar (Senior Agriculture Economist), Luz Stella Rodriguez (Social Protection Specialist), Ingrid Mujica (consultant), and Jon Mallek (consultant). Liz Girón (consultant) undertook a review of the impact evaluations of productive inclusion programs in the region. The report synthesizes findings and lessons learned from 5 country case studies, produced as part of the broader LCR-Knowledge Exchange on Productive Inclusion task. In this context, the extended team included: Raul Asensio (consultant, author of Peru country case study); Claudio Castillo (consultant and author of the Chile case study); Arilson Favareto (consultant, author of Brazil country case study); Celeste Molina and Laura Delalande (consultants and authors of the Mexico country case study); and Florencia Wagmaister and Ines Reineke (consultants and authors of the Argentina case study). Jon Mallek also edited the report. The work was carried out under the direction of Pablo Gottret, Manager of the Social Protection and Jobs Global Practice for Latin America and the Caribbean. The team is deeply grateful to Colin Andrews (Program Manager for Partnership for Economic Inclusion), Thomas Bossuroy (Senior Economist), and Mirey Ovadiya (Senior Social Protection Specialist), who kindly peer reviewed the report, and whose valuable suggestions improved the report. In addition, feedback received at various stages of this work from Pablo Acosta, Hugo Brousset Chaman, Veronica Lucia Diaz Hinostroza, Thiago Falco, Junko Onishi, Truman Packard, Marcela Ines Salvador, Elisa Seguin, Veronica Silva Villalobos, Maria Concepcion Steta Gandara, Asha Williams, William Wiseman, and the entire Social Protection and Jobs – Latin America and the Caribbean team are gratefully acknowledged. The report was financed by a generous grant from the Rapid Social Response Fund (RSR) Multi- Donor Trust Fund. The team wishes to extend its gratitude to the RSR management team for their unwavering support in the undertaking of this work. 3 1. Introduction The dramatic transformation in Latin America and the Caribbean (LAC) between 2000 and the 2010s was driven by fast growth, together with higher employment rates and increased real incomes. Between 2000 and 2014, the share of extreme poor1 was more than halved from 25.5% to 10.8%, with a dramatic decrease in the share of poor people from 42.8% to 23.4% (World Bank, 2019). This transformation was largely due to sustained economic growth coupled with a relatively better distribution of income due to higher labor earnings, the prevalence of anti-poverty policies, and better integration of the disadvantaged to their domestic economies (World Bank 2019). In fact, redistributive factors alone accounted for about 35 percent of the fall in poverty during the Golden Decade and have played a key role ever since (World Bank, 2019). On the other hand, labor income was among the key drivers of the particularly strong poverty reduction between 2004 and 2012 (Vakis et al, 2016). During this period of sustained growth, increased fiscal revenues supported the expansion of social protection programs, mainly in the form of non-contributory social assistance. Since the early 2000s, social assistance expanded greatly, mainly in the form of conditional cash transfer (CCT) programs aimed at mitigating poverty and tackling the intergenerational transmission of poverty through investments in human capital (health, education, nutrition), or social assistance to the elderly in (extreme) poverty through different variations of non-contributory social pensions (Rofman, Apella and Vezza, 2014). Given that the CCTs target specifically those in greatest need, they have a higher redistributive impact than social spending in general. With their increasing prevalence and coverage, these programs further reduced poverty rates, increased some health and education access and utilization outcomes, and helped plug millions of people into the local and global economy (World Bank, 2019). Despite the increasing coverage of social assistance and positive impacts on consumption and other social dimensions, policy makers were aware of the limitations in ensuring the sustainability of those gains. It was becoming increasingly more difficult and expensive to maintain the pace in the reduction of poverty, and the achieved gains proved fragile as shocks could result in increased hardship for the vulnerable population. In fact, while the fraction of poor people declined, the fraction of economically vulnerable people actually increased, from 34 percent in 2002 to 38 percent in 2016 (World Bank, 2018b) raising questions about the suitability of solely social assistance for the vulnerable, or the need for a complementary support. Also, regardless of the efforts of social ministries to reach marginalized segments of their population with mechanisms like busca activa or through the recruitment of social workers and case managers, it was evident that for certain groups in extreme poverty social assistance alone was not enough and therefore it was necessary to identify complementary strategies for income generation. Brazil had already recognized in 2004 the importance of “productive inclusion” as part a broad set of instruments of the newly created Social Development Ministry where the emphasis was on expanding social assistance coverage and ensuring food security during a period of high growth and overall poverty reduction. The difficulties in further reductions of extreme poverty, especially in rural areas, led to the Plano Brasil Sem Miseria to eradicate extreme poverty where productive inclusion was one of the four policy pillars (Favaretto, 2019). Similarly, in Peru, the Juntos cash transfer program -launched in 2005- was seen as an innovative public policy for social and economic inclusion (Asensio, 2019). In Mexico, the National Development Plan 2013-18 (NDP 2013-18) 1 Less than US$2.50 a day. 4 included “democratization of productivity” as one cross-cutting strategies to expand access to opportunities to all socioeconomic groups (SHCP, 2013). In sum, it became clear that to ensure sustainable poverty reduction, complementary measures were needed in addition to social assistance. This will become evident when larger shocks exposed the vulnerability of the recent gains. The food, fuel, and financial crisis evidenced both the potential and the limitations of social assistance programs and led to the recognition of the role of other social and economic constraints for income generation. During the food, fuel and financial (FFF) crisis of 2008-10 countries more dependent on external trade and remittances, like Mexico and Central America, observed significant increases in both urban and rural poverty. In urban areas the main transmission mechanism was weak labor markets affecting both employment and real wages, while in rural areas the decline in remittances was significant, especially in countries like El Salvador with a large share of migrants from rural areas (Acosta et al, 2012; Grosh et al, 2014). The existing social assistance programs helped cope with the impacts of the crisis, like in Mexico where an incentive called Alimentar Vivir Mejor was included in 2008 through 2011, as part of the social assistance program to compensate for the rise in food and energy prices. Other countries, like Argentina, used large public works and social assistance programs targeted at vulnerable and unskilled families. During the FFF crisis, countries in LAC realized that while social assistance had played an important role in poverty reduction during the rapid growth period, there was a policy need to strengthen the income generation capacity of the poor and vulnerable through skilling, labor intermediation, business development services, financial inclusion, and other support to productive activities. Economic inclusion strategies were also motivated by the reduced but persistent intergenerational transmission of poverty. Another rationale for interventions addressing income generation constraints emerged from the evidence and rigorous impact evaluations of more mature social protection programs. Mexico and Colombia, for example, identified that the gains in human capital investments through better nutrition and higher educational attainment were not necessarily accompanied by equivalent gains in future labor markets opportunities among the graduated youth, though there was emerging evidence of positive impact (SEDESOL, 2016; Parker and Vogl, 2018; Baez and Camacho, 2011). As a result, Colombia reformed a program to facilitate the participation of youth from poor and vulnerable families in training and apprenticeship activities by connecting to social assistance family members. In Argentina, regular social assistance was accompanied by a large youth training program to ensure school completion and technical trainings. The coverage expansion of social assistance in the 2000s unveiled the need for further investments in human capital to further acquire productive skills and reduce poverty. It became clear that overcoming poverty requires simultaneous access to social and economic opportunities to access better jobs, productive assets, markets, and other income generation sources. Since poverty is an accumulation of multiple deprivations, a broader package of benefits would be needed to move beyond basic social assistance and modernize poverty confrontation. In fact, the move to a multidimensional measurement of poverty in the region, was an important driving factor in making this link even more apparent (World Bank, 2018a). Given the emphasis on income generation and asset building objectives, many countries would launch initiatives leveraging on their existing capacity shaping the content of “productive inclusion” (PI) interventions. These interventions were based on previous experiences of productive development programs whether rural livelihoods or urban youth training programs, among others. 5 This agenda has gained additional attention in post emergency contexts, like the current COVID-19 crisis, where shock mitigation measures need to be complemented with medium term economic resilience initiatives. The increased frequency of natural and man- made disasters positioned social protection programs at the front of disaster response after the emergency conditions have subsided (Williams and Gonzales, 2020). The unprecedented health emergency and the resulting economic crisis due to COVID-19 also bring to fore the potential role of social assistance and PI approaches in addressing the need to protect lives and livelihoods. The COVID- 19 emergency and the resulting downturn affect the poor and vulnerable disproportionately as they have less means to cope: e.g. more likely to be uninsured making it harder to cope with the related health expenses, less likely to stay at home (e.g. unable to afford not working and more likely to work in jobs that can’t be performed remotely) increasing their likelihood for infection; more likely to suffer income/job losses etc. As a result, the poor and vulnerable are resorting to negative coping mechanisms (such as skipping meals, borrowing with high interest rates, migration, etc) expanding and intensifying the constraints they face further. The focus of PI approaches on addressing multiple and simultaneous constraints faced by the poor and vulnerable, coupled with the territorial/community approach observed in many cases, indicate the potential role these programs can play in the recovery. At the same time, it highlights the importance of getting the PI approaches right and building on lessons learned so far, while adapting these programs for the recovery.2 This report aims to document different approaches of PI in LAC countries, particularly focusing on the policy and implementation aspects, as well as the lessons learned from their design and implementation. This report builds mainly on five emblematic case studies of PI approaches in Argentina, Brazil, Chile, Mexico, and Peru, as well as on a stocktaking of nearly 130 existing productive inclusion initiatives, using the ASPIRE database as a starting point. The countries studied are diverse among several dimensions such as, poverty, employment structure, size and quality of the workforce etc. which have a bearing on the diversity of the productive inclusion approaches employed. Indeed, each country case study highlights a particular aspect of the PI approaches observed in the region, with a view to provide a more complete picture. The report is therefore a synthesis note pulling common findings together across key aspects. This report does not examine in detail the underlying political economy factors but reflects on them in the description of the component on institutional and policy mandate. The note is organized as follows. First, it presents the definition and an operational framework for productive inclusion, to provide a conceptual grounding for the analysis of different approaches observed in the region, identifying the key elements, from policy mandate-design to implementation. Next, common themes for successful PI interventions for each key pillar, are presented in turn, drawing from the rich information of the country case studies. The third section presents the experience and the challenges on the evaluation of PI programs in the region to date as well as a brief summary of the results. The note concludes with (operational) lessons learned to inform the evolution of the productive inclusion strategies for the low and middle-income countries in both LAC and other regions; and presents a brief overview of the pending issues on the agenda for the region for future research. areas for further expansion and next steps of productive inclusion strategies, 2 It is important to note that the case studies were undertaken, and a significant portion of the synthesis note drafted before the onset of the COVID-19 pandemic, hence any further analysis on this issue falls outside of the scope of this work. 6 2. Productive Inclusion: Definition and Operational Framework The objective of supporting income generation capacity for the poor and vulnerable has been termed differently depending on the regional context. While in more urban European countries productive inclusion typically refers to ‘activation of individuals in the labor market’ (and more generally to labor market intermediation), in some East and South Asian countries, it may be understood as strategies to “graduate” ultra-poor households out of extreme poverty to a more stable state through a time bound ‘package’ of benefits and services. In countries like Ethiopia and Tanzania, productive inclusion strategies are also known as productive safety nets (or cash plus or public works plus). In these countries, PI interventions include the standard, predictable provision of food or cash transfers (allowing effective consumption smoothing and avoiding asset depletion), productive community assets, improved access to social services, and financial instruments to allow the program to scale up in times of transitory crisis. In programs like Samurdhi in Sri Lanka or India’s National Rural Livelihood mission for example, PI refers to ‘livelihoods’ programs that usually combine agricultural extension (technical assistance) with small individual and community services. The different types of productive inclusion strategies (and programs) reflect the constraints faced by the target population in enhancing their income generating capacity. The recent State of Economic Inclusion (World Bank, 2020) provides a global review of (productive) economic inclusion programs. LAC countries also developed different productive inclusion approaches with the common aspiration of enhancing income opportunities for the poor and vulnerable population. In the late 1990’s and early 2000’s, many interventions in LAC were aiming to improve employability and productive capacity of vulnerable groups like the unemployed, the youth, and those in rural areas, among other groups.3 In early 2000s many social development ministries were established and social assistance was expanded significantly during this growth decade. The large coverage of the poor in social assistance, increased their political visibility and stressed the need to facilitate economic opportunities beyond the social assistance benefits. It is in this context that the term “productive inclusion” was coined to reflect the goal of strengthening the capacity of the poor to be actively included in productive or employability programs.4 As social assistance alone was not sufficient to attain inclusion in productive activities, some countries established formal coordination mechanisms across different agencies to ensure complementary productive services for the poor and vulnerable.5 It was clear, then, that attaining productive inclusion required tackling multiple constraints on productive dimensions but also needed to address basic social needs. Economic opportunities, such as a training programs, could not be exploited without addressing some social needs such as income support to cover opportunity costs or child-care. 3 See Ibarran and Rosas (2008) for a review of youth employment programs in the 1990s, and Ocampo (2001) on the evolution of agricultural and rural development policies that recognized the increasing productivity gap among rural producers. as examples of the policy interest to link vulnerable population with economic opportunities. 4 It should be noted that this note follows the evolution of social protection systems in LCR where the incremental capacity reached a point where synergies were feasible in some countries on productive inclusion approaches, and thus have a decidedly Social protection perspective. While the existence of other interventions aimed at improving productivity in rural areas or SMEs that are under other sectors (e.g. agriculture) are recognized in the broader sphere, they fall outside of the scope of this note. 5 The Secretaria de Articulação de Inclusão Productiva (SAIP) in the Brazilian Ministry of Social Development in an example. The SAIP was formally established in 2008 (de Souza, 2013). 7 In this report, productive inclusion strategies are defined as public policies aimed at improving household income generation capacity and asset building by jointly addressing productive and social constraints for the poor and vulnerable population. This definition entails several elements. First, it recognizes that poverty is an outcome that reflects a number of constraints, both productive and social, at the individual, family and community levels, and that these constraints may even reinforce each other. This definition coincides with the framework presented by Barrett, Carter, and Chavas (2019) that argue that poverty traps are, in many cases, low-return low-risk equilibria adopted by households facing a number of market failures together with low endowments of human or physical capital. This definition stresses that consumption and productive dimensions of households are interdependent or non-separable, especially in contexts with missing or imperfect markets.6 Factors affecting social and consumption decisions (e.g. sending a teenage child to school) are, then, interdependent from production decisions (e.g. amount of labor used in the farm or family business) especially when there are no functioning labor market where to hire additional workers. Similarly, programs aiming to support household consumption and human capital investments, like cash transfers, are found to have productive uses in household livelihood activities as liquidity constraints are loosened in imperfect credit markets (Fizsbein and Schady, 2009). Social and economic exclusion dimensions are, then, two sides of the same coin. Second, the definition also recognizes the gains from jointly tackling social and productive constraints, especially among the poor and vulnerable. In particular, this necessitates that the corresponding PI programs consists of a bundled set of interventions to address these multiple and interdependent constraints. The interdependence between consumption and production decisions is especially evident when facing unexpected shocks on either consumption (social) or productive dimensions. As poor and vulnerable households are more likely to face market imperfections, such as access to risk insurance, the shocks they bear are transferred to both social and productive dimensions. Market failures, however, have been also addressed by non-market solutions through community and social networks and other informal/implicit insurance mechanisms (Mobarak and Rosenzweig, 2013; Carter, Chavas and Barret, 2019). This interdependence between social and economic constraints at the household and community levels, is precisely what many productive inclusion initiatives took into account to maximize the positive synergies among them. For instance, a technical training package for vulnerable urban youth (addressing a productive constraint) may not have the desired take up among the` poor because of the opportunity costs they face. If combined with a social assistance benefit (addressing a social constraint), take up for training would be higher and the impacts of social assistance would last longer. Similarly, an intervention to increase agricultural productivity among poor farmers may have limited effects if market connectivity is low, suggesting the need to accompany agricultural extension with basic community infrastructure. As such, this definition can apply to both urban and rural settings, to areas well connected to markets and to isolated ones. Implicitly, productive inclusion approaches leverage social interventions to reduce the overall household risk. Productive interventions could raise the livelihood risk level due to adoption of new technologies, entry into new markets, uncertain output and input prices, or input availability, among other factors. In these contexts, social assistance and other interventions addressing social constraints can play a critical role in ensuring minimum living conditions which reduces the underlying consumption risk and facilitates productive decisions. Social assistance 6 Unexpected shocks could also establish non-separation between consumption and production spheres as underlying markets (labor, care, financial) need time to adjust. 8 and the interventions addressing social constraints, then, not only represent a positive income effect but could have a behavioral impact on risk management and productive decisions. Third, the productive inclusion definition recognizes the heterogeneity of constraints and household characteristics demanding a rationale for public interventions (and the potential role of non-government sector). The social and productive constraints reflect market failures and other structural factors underpinning persistent poverty, establishing a merit for public intervention to address those market failures. There are two implications. First, as households and communities vary in their capacities, assets, or risk behavior, public interventions aimed at enhancing productive inclusion need to be flexible to provide a pertinent bundle of services that address the population needs. Second, and most importantly, evaluating the rationale for a public intervention implies evaluating the potential capacity of the private sector to find solutions to those constraints. While most strategies have relied almost only on public sector interventions, this paper stresses the need to carefully examine the potential role of the private and non-government sector in finding solutions to address the existing market failures. Finally, as productive inclusion implies addressing several constraints, it will require a coordinated action to solve those constraints. As the consumption and productive complementarities became evident, practitioners realized that social assistance programs (generally aimed at supporting minimum consumption levels) were insufficient to achieve a sustainable increase in consumption and reduction in poverty. Rather than seeing productive interventions as substitutes to social assistance, practitioners worked on exploiting the mutually reinforcing or synergistic effects among them and designed complementary activities between social assistance and productive economic interventions. Many productive inclusion strategies have used social protection programs as an ‘effective ladder’ or base platform to provide low income and vulnerable populations with access to a bundle supporting productive inclusion. This requires clear arrangements among participating actors to ensure the synergies are exploited without any unintended consequence. 2.1 Addressing Productive Constraints and Social Bottlenecks The definition of productive inclusion involves simultaneously tackling several constraints affecting income generation and asset building to enhance the participation of the poor in economic activities. The constraints to asset-building and income generation of an individual, household, or community hinge on both productive as well as social constraints, as discussed before. This report groups those constraints by their social or productive dimension and provides examples of the benefits and services to address those constraints. Table 1 shows the social and productive constraints and the components of integrated services in existing productive inclusion strategies. Skills, for instance, are a critical factor for income generation so productive inclusion strategies typically include programs to enhance skills accumulation like technical or occupational training, or programs to enhance employability through labor marker insertion or job search. In the definition used in this report, just any technical training or labor intermediation may not qualify as productive inclusive unless they have adaptations to ensure their pertinence among the poor (e.g. scheduling of sessions), or or ensure the adequate take up and coverage addressing other social dimensions (e.g. provision of social assistance or childcare). Other programs may support the building of productive assets like the distribution of sewing machines in urban microenterprises or livestock and modern fertilizers in rural farms. But these business innovations and the associated higher risk exposure may require basic social assistance that operates as a basic 9 insurance among the poor. Whether the productive inclusion package addressed both productive and social constraints or several productive constraints, their implementation requires the participation of specialized agencies. Moreover, these interventions will need to be adapted to the socioeconomic and territorial characteristics and local actors will need to be involved. Table 1 - Social and Productive Constraints and the Components of Integrated Services Source: Authors’ own elaboration The challenges for PI strategies are more acute in areas where the structural conditions do not favor the emergence of productive opportunities for low-income and vulnerable populations: rural areas, areas with precarious or unequal productive structures, areas vulnerable to environmental or political shocks. Constraints for productive inclusion are not only at the individual or household level but may arise at the local/territorial level (i.e. neighborhood or community level). For example, programs aiming to increase employability through occupational training combined with social assistance in urban slums may also require enhanced security and basic transportation support to facilitate labor mobility to and from jobs. Similarly, dispersed and poorly connected rural villages where agricultural technical extension and dwelling improvements are planned may also require small local infrastructure to ensure adequate access to water or connectivity to markets. Populations living in natural disaster or shock-prone areas (or conflict zones) will need measures to reduce those risks in order to engage into sustainable productive activities. The objective of productive inclusion, then, recognizes the community or territorial conditions and incorporates measures to address them. Policymakers have been working to design packages of benefits and services (B&S) aimed at loosening key productive constraints and identifying and addressing relevant social constraints. These combined packages aim to improve households’ access to income opportunities, skills and productive assets and these packages vary across institutional settings, 10 target population and implementation agencies, among others. Even when these packages are typically focused on the extreme poor and vulnerable populations, policymakers and social ministers are aware that this ‘group’ is not homogeneous, and therefore the interventions need to be (i) selective – in order to narrow down the key components of the package to simplify implementation, respond to the specific characteristics of individuals/ households, and facilitate scaling up, (ii) flexible to enable local actors and participants to choose the priority components to reflect territorial realities, and (iii) coordinated – to encourage synergies among programs and local and central actors on the ground. In many countries, PI programs also have shock responsive elements, to enable households to rapidly adapt in the face of climate, social, economic and political shocks. This new avenue of policies coincided with the emergence of emboldened social development and protection ministers that have developed more capacity for the delivery of social protection programs. This new generation of policies typically involved more than one program or institution to overcome structural poverty and transition from social assistance to sustainable independence for low income and vulnerable groups. Institutional strengthening emerged when social development ministries were: (i) gaining space within national governments, with a clear mandate to guarantee social and economic rights of citizens, with the prioritization for low income groups, (ii) working in the harmonization and coordination of social protection programs, (iii) exploring alternative and more efficient forms to deliver social protection - including systems to better identify potential users (social registries), to follow the trajectory of households within the different forms of social protection (beneficiary registries), and to better deliver social protection services (payments and case management)-, and (iv) actively committed to the SDG agenda of “not leaving anyone behind”. 2.2 Productive Inclusion: An Operational Framework with Key Pillars After analyzing PI initiatives across LAC, a common framework begins to emerge across many of the most successful cases in the region. In particular, countries with stronger results in terms of coverage are the ones that share four key elements. First, they had a strong institutional policy mandate capable of providing a common goal and aligning relevant actors. Second, implementation institutions align actors responsible for delivery of the package including local governments and non-government actors. Third, there were strong and effective information systems that enabled them to coincide on the target population, monitor partial results and adjust implementation plans. Fourth, there were clear assessments of the budget and financial implications setting an incentive structure that ensured compliance across different stakeholders. Figure 1 depicts the fours I’s for productive inclusion coordination: the broader institutional and policy mandate is supported by institutions for implementation and program delivery, information systems, and incentive structure. The territorial dimension is a cross-cutting element which could play a key role in certain pillars and depending on the context. These pieces are described next. 11 Figure 1: Operational Framework for Productive Inclusion Source: Authors’ own elaboration Institutional and Policy Mandate The challenge in tackling multiple constraints and reaching a significant scale is that it requires multiple specialized agencies. As those agencies have their regular activities, modifications of those activities require a higher-level policy mandate to ensure institutional participation. The review of cases in Latin America shows that countries had a strong political champion, typically with a broader policy mandate. In Brazil, the productive inclusion agenda was established under the broader Zero Hunger program that provided a shared goal across different Government actors. In Peru, the Include to Grow (Incluir para Crecer) strategy included a pillar on economic inclusion. Within the Institutional and Policy Mandate, there are some critical decisions. One is to define the management of implementation and the specific institutional responsibilities (who does what?). More successful strategies established a small committee typically involving the key implementing agency (in many cases the ministry of social protection), the political champion (some appointee under the president), and the ministry of finance. Another decision is about the target population where different stakeholders may have different operational approaches (e.g. community- or individual-based). This higher-level space should also define the broad components to be implemented (the package). The LAC experience, discussed next, shows that a reduced package of services and benefits facilitates the adaptation at the local level where certain components are prioritized. Implementation Institutions for Program Delivery Once the broader policy mandate is established and the broad contours of the package are defined, a productive inclusion strategy requires the operational capacity to deliver those 12 components of the package. The LAC experience suggests that institutions with delivery capacity are involved with specific targets in term of population to be covered, services and benefits to be provided and the timing of those deliverables. These definitions avoid the delivery modality. For instance, in Argentina it is the private sector trade unions at the sub-national level who deliver the technical training on a population whose eligibility is defined at the central level (policy) but operationalized by the local employment offices. In Brazil, the rural productive inclusion package was delivered by the Ministry of Family Agriculture based on the poverty and food insecurity indicators set by the high-level committee led by the social development ministry. This clarity in their roles and the relative autonomy in the implementation of delivery facilitated the scaling up of successful strategies in the reviewed countries. A key actor in the delivery of the productive inclusion package is the local or community actor (typically local governments). Operationalizing the national policy at the local level requires adaptations to local context and these actors bring this local knowledge and potentially establish links with territorial development plans. The evidence discussed also flag the limited links with private sector opportunities in the implementation of these strategies, with the exception of Argentina. Information Systems to Enhance Coordination Implementation activities under a common policy umbrella require operational communications. In particular, about target population, composition and volume of services to be delivered, timing of those services, and capacity to adjust operational plans during implementation. The experiences show that strong information systems supported the scale up by using, for example, Social Registries to establish entry points and target populations, or establishing interoperability among institutions to facilitate referral and counter-referrals. These information systems, however, were not a rigid framework but helped adapt based on existing results. In Peru the information system (SISFOH) helped set the target population as those beneficiaries of social assistance, but this was relaxed during implementation to include other community families that were not social assistance beneficiaries. Chile utilized its large coverage social registry to establish implementation plans across national institutions but delivered by local authorities. Information systems, then, are critical to ensure the target population has a voice in the decision-making process and that services and benefits are adequately delivered. Incentive and budgetary arrangements to ensure delivery In addition to the political mandate, implementation capacity, and information systems, successful productive inclusion strategies introduced budgetary decision incentives to facilitate the process. It is critical to identify the incremental resource constraints when institutions are required to modify their service content or delivery modalities. Technical training institutes requested to deliver trainings in urban slums face increasing mobility and security costs, as in El Salvador or Brazil, and can quickly slump implementation. Early assessments of those constraints and adequate budgeting will help address this. In Chile, the use of performance-based budgets helped ensure the delivery of specific services to the target population by conditioning part of the planned budget disbursement to compliance with verifiable delivery information. In Brazil, the presence of the ministry of finance in the high-level committee also ensured the delivery by different agencies. At the local level, Peru utilized a local budget execution unit to ensure community participation and ensure relevance of the productive inclusion package and local ownership. 13 3. Productive Inclusion Approaches in the Region- Key Operational Elements Productive Inclusion has become an increasingly key component of social protection in the region. A mapping of PI interventions undertaken as part of this note, indicates that more than 130 public interventions exist across 11 countries in the region.7 These interventions were all (co-) implemented by the national government and were active as of September 2019. The staggering number of interventions reflect the fragmentation in this space, stemming from the multi-sectoral nature of the problem and the required coordination resulting in the proliferation of small programs. While these interventions to a large extent aim to address multiple constraints, most of them were small in nature focusing on particular populations such as youth, indigenous etc. The countries in the region have taken distinct approaches to PI, key operational elements of which are distilled in this section. While most interventions were small in nature, there were distinct approaches across the region reflecting the differences in context, systems, and instruments, which are analyzed more in detail in the 5 emblematic country case studies (i.e. Argentina, Brazil, Chile, Mexico, and Peru). These country cases present an opportunity to distill the key operational elements and lessons learned to designing and implementing these interventions, that are summarized in this section. 3.1 Institutional and Policy Mandate The transition from implementing individual programs to a model that integrates a bundle of benefits requires a broader policy mandate and intra- and inter-institutional coordination. The objective is to mobilize existing benefits and services (B&S) and concur around a group of people that typically have not been beneficiaries of productive interventions. As discussed before, this bundle of services is context-specific and thus will vary by country, region, delivery capacity, and target population. PI Strategies, then, do not have an unlimited menu of B&S, but rather a compact set of options that are subject to local adaption. In the case of Brazil, the broader policy mandate came in the form of the PI umbrella program, Programa Brasil Sem Miseris (PBSM), and its ensuing framework, the Rural Productive Inclusion Strategy (RPIS). Through the years, many separate anti-poverty initiatives had existed in many forms and across multiple government ministries, but there was a lack of a cohesive overarching strategy. However, a new administration in 2011 ushered in a bold goal of eliminating extreme poverty in Brazil in just a few short years and to achieve this, policymakers relied on a whole-of-government approach specifically utilizing productive inclusion policies through PBSM. The massive effort coordinated more than 120 actions across 20 government ministries in an upfront recognition of the multidimensional nature of poverty. There was a good deal of unification and integration of previously existing policies and programs, and new instruments were created to complement what was already in place. This top-down approach built a cumulativeness with previous actions, coupled with strategic innovations in the PBSM and the RPIS framework. As Brazil’s approach shows, PI strategies demand higher levels of multi-sectoral coordination in order to achieve not only complementarity among programs but also synergies. This means in practice that it is not enough to have two or more programs targeting the 7These countries are Argentina, Brazil, Chile, Colombia, Dominican Republic, Honduras, Jamaica, Mexico, Nicaragua, Paraguay, and Peru. 14 same territory and same population, but to carefully design multisectoral interventions with the aim that the activities of one program, for instance monetary transfers, increase the potential of success of other program, i.e. food security interventions (RIMISP, 2018). Synergy in this context is therefore the interaction or cooperation of two or more organizations or programs that produces a combined effect greater than the sum of their separate effects. Synergy could be achieved if programs converge on the same territory and the same population, with appropriate/relevant benefits and services that complement each other. Under this model, the household/individual trajectory is supposed to be monitored to guarantee that it is receiving the correct combination of benefits and services and ensuring it is overcoming structural poverty. Coordination also requires the design and implementation of political and interagency agreements with a clear definition of what is the expected value added from this collaboration. Achieving interinstitutional cooperation is without a doubt a challenging task due to the existence of institutional rigidities and inertia, prioritization of individual over collective objectives, and lack of incentives (i.e. competing resources) to adequately encourage joint work (RIMISP, 2019). The operationalization of PI initiatives has taken different forms within the region. While certain countries have opted for fostering collaboration between different ministries or agencies, usually under the form of coordination bodies to facilitate interaction between different institutions and levels of governments (Argentina, Chile, Mexico); others have decided to centralize the different components under the same organization (Peru). In Mexico for instance, the government put in place a technical committee to promote dialogue and coordination for the productive inclusion of low income and vulnerable population. This was due to the fact that their initial PI strategy focused on linking beneficiaries of CCT programs with existing programs. The STEIA (Subcomite Tecnico de Empleo, Ingreso y Ahorro) was installed in 2014 to link beneficiaries with more than 15 programs implemented by more than seven institutions (Figure 2). This institutional model continued when other PI initiatives were deployed (Territorios Productivos, De la mano con Prospera and PROINPO), seeking to prioritize Prospera8 beneficiaries and creating incentives to facilitate their access to existing interventions. However, despite efforts, interinstitutional coordination has proven to be difficult to achieve. The programs that most complied with the coordination and attention agreements for Prospera beneficiary population were the programs of SEDESOL such as Opciones Productivas. In the case of Mexico, linkages were highly dependent of political circumstances and the will to collaborate among technical teams. In other words, the more technical/operational position the coordinating institution had the greater intersectoral and interinstitutional impact would be, highlighting that SEDESOL would have been better placed, as the head of the social policy, to coordinate the PI strategy. In Chile, the Intersectoral Social Protection System (SIPS) operates out of the Ministry of Social Development and Family (MDSF). MDSF, also the primary financing mechanism of SIPS, coordinates heavily with primarily three other government ministries, including the Ministry of Labor and Social Welfare (MINTRAB) for labor intermediation programs, the Solidarity and Social Investment Fund (FOSIS) for financing direction, and the Ministry of Education (MINEDUC) for childcare and education services. A well-functioning social protection system , such as SIPS in Chile, can implement performance-based budgetary transfers triggered across 8 The program has recently been transformed and no longer exists as a Conditional Cash Transfer Program nor has an inclusion (productive, social, economic) due to shifts in social policy in Mexico. 15 agencies and can be an effective mechanism for coordination. Noteworthy advantages of this setup include effective resource targeting; greater access to detailed information of the target households; total resource execution and a well-organized system based on differential roles and functions by node. Despite the strong intersectoral nature of SIPS, it also maintains effective interaction with local government entities, in particular municipalities who directly implement most of the programs, to the benefit of the targeting population. The big advantage of the intersectoral SIPS model is its flexibility particularly at the local level where conditions can change quickly; it becomes considerably more rigid however when budget modifications are required since any change often must traverse the various ministries involved. Figure 2: Mexico’s institutional horizontal coordination Source: Authors’ elaboration In Argentina, the implementation of Hacemos Futuro (HF) by the Ministry of Health and Social Development (MSyDS) has brought important changes in how productive inclusion strategies are being designed and coordinated in the country.9 Although Argentina has a long history of public works programs targeted to low income and vulnerable populations, the majority were implemented through community organizations (cooperativas) and local governments. In 2015, MSyDS put in place Programa de Ingreso Social con Trabajo, a productive inclusion strategy targeted to individuals (and not to community organizations). This new implementation mechanism was adopted in 2018 by the HF program that works to encourage individual productive inclusion of vulnerable groups. HF’s coverage far exceeds the coverage of the A ctive Labor Market programs (ALMPs) implemented by the Ministry of Production and Labor (MPyL). While HF centralized the interlocution with potential beneficiaries, it still coordinates with a wide network of public and private organizations, including with strategic partners. HF centralized the interlocution with potential beneficiaries that was previously done through local 9 It should be noted that the information refers to the program elements of HF as of July 2019, some of which may have changed since then. Furthermore, being a relatively newly established program, the information largely is on design elements, while implementation details and results remain to be evaluated/seen. 16 authorities and defined ANSES (Administracion Nacional de la Seguridad Social) as the unique point of contact between program beneficiaries and the program. However, for the provision of the bundle of services, MSyDS coordinates with a wide network of public and private organizations, including the Ministry of Education, provincial ministries, training units (UCAPs), technical education providers, universities, public employment offices, providers of soft skills training, local governments, and SCOs (Figure 3). In addition, HF also created a Coordination of Labor Inclusion (CIL) that seeks to strengthen education and training outcomes as well as generate concrete opportunities for the labor inclusion of program’s beneficiaries. CIL coordinates with strategic partners, identifying labor opportunities and monitoring beneficiaries’ trajectories within the program. CIL aims to channel beneficiaries through the Ministry of Labor’s Employment Offices but the agreement with this institution had not been formalized by June 2019. Figure 3: Argentina: Hacemos Futuro – Interinstitutional coordination. Source: Authors’ own elaboration On the other hand, Peru has a different approach to encourage coordination and integration of services. In 2011, the Ministry of Development and Social Inclusion (MIDIS) was created with the aim of harmonizing social protection, based on the assumption that poverty and exclusion are multidimensional problems that also require multisectoral strategies. The creation of MIDIS involved significant changes in the design and the provision of social protection, including reformulation of objectives, consolidation of programs, and setting higher standards of service provision. Similarly, the creation of MIDIS pushed the development of the National Strategy of Social and Economic development, Incluir para crecer, based on five core axes: actions to eradicate chronic malnutrition, early childhood development, integral development of children and youth, economic inclusion, and protection of old age population. HW was the only program associated with the economic inclusion axe and therefore the government’s tool to facilitate the smooth transition of people from social assistance to independence and the attempt to make social and productive programs to converge over the same population. Box 1 presents an overview of Peru’s experience in rural programs that it drew on the design of HW. Although Peru has a large set of rural development programs under the Ministry of Agriculture, MIDIS instead exploited the country’s social fund capacity to reduce the initial coordination costs The government integrated the social fund Fondo de Cooperacion para el 17 Desarrollo Social (FONCODES) team, previously accountable to the President’s Office and to the Ministry of Agriculture, into MIDIS structure. This team was then given the responsibility to design and implement the HW program. The integration of a ‘productive’ institution within a Ministry mainly focused in the delivery of social assistance benefits has proven to be challenging due to differences of approach and sometimes, different policy goals and objectives, a problem that is observed in other countries as well. While the policy mandate was to articulate social and productive interventions, social assistance and HW teams went through some learning and harmonization in terms of objectives (consumption vs productivity), target population (extreme poor vs. asset holders), and even location (remote vs. market connected). This dialogue has produced a more nuanced strategy where heterogeneity of households and communities are associated to variation in results. This exemplifies the need to reconcile two different institutional cultures and create common incentives to collaborate beyond operational or political will. Box. 1: Haku Wiñay: building on the experience of more than five decades of rural development programs Peru has a long history of programs aimed at improving the productive capacity of the rural poor that goes back to the 1960s when the government undertook an agrarian reform that implemented actions to increase rural productivity. FONCODES (Fondo de Cooperacion para el Desarrollo Social) created in the 1990s was aimed at improving the development of community infrastructure (roads, schools, irrigation systems) for low-income rural areas. This decentralized program gave small community groups (nucleos ejectores) the responsibility to decide, invest, and supervise the development of community projects. In parallel, the government transformed the model for the provision of technical assistance for rural workers by targeting it to poorer areas and outsourcing the service to private service providers. In practice the market for private provision of technical assistance did not take off but the international cooperation, in particular FIDA, introduced a new model of initially low-scale interventions, that later was expanded to a larger number of areas. This model of rural technical assistance was based on five main components: (i) income generation through the connection with local markets, ii) high engagement from community members in the decision making process, (iii) use of peer to peer training methodologies, (iv) creation of small organizations of rural producers who were the program counterparts, and (v) recognition of the importance of adapting traditional and local knowledge in rural practices. The interventions of FIDA were concentrated in the sierra sur. The third big influence was the demand-based watershed management and soil conservation program, PROMANACHS, that seeks to improve sustainable production through the introduction of irrigation systems and improvement of agricultural practices. Finally, Mi Chacra productiva, the immediate antecedent of HW, was implemented in 2009 by the Ministry of Gender and Social Development, building on the experience of the NGO-led Sierra Productiva and other programs that seek to boost the potential of rural family economies with the introduction of small technical innovations. 18 3.1.1 Target Population The analysis of PI programs suggests that they are commonly focused on groups that have not been able to sustain development gains. In particular, they focus on the extreme poor and vulnerable, or groups that face additional barriers for social and economic inclusion, including gender, age, ethnicity, household composition, or geographic location (Table 2). These groups are ultimately the motivating factors behind PI programs. The design, extent, timing, and method of delivery are all carried out purposefully based on the group of intended recipients. In Mexico for instance, PI interventions were intended to reach the same population who benefit from the conditional cash transfer (CCT) Program. In 2014, when the Mexican government decided to transition from Oportunidades to Prospera: Programa de Inclusion Social, it introduced the idea that on top of the human capital objectives, Prospera should work on productive, social, and financial inclusion of the program’s beneficiaries. Therefore, in addition to the nutrition, health, and education components, the government announced a fourth component called Linkages (Vinculacion) aimed at connecting program beneficiaries with productive, social, financial, and labor opportunities. Without a formal conceptual definition of PI and a clear diagnostic of what they would like to achieve,10 National Coordination of Prospera worked in the definition of a series of operational mechanisms to facilitate the Linkages of Prospera beneficiaries with existing Productive Promotion Programs (programas de fomento productivo), implemented by other federal ministries including Agriculture, Economic Development, Labor, Urban and Territorial Development, and the Commission for the Development of Indigenous Populations, among others. Table 2. An Overview of the Target Population of PI Programs in 5 countries COUNTRY TARGET POPULATION Argentina Vulnerable or at-risk population 18-65 years. Priority to head of households with social and economic vulnerabilities (disabled children, victims Hacemos Futuro of gender violence, victims of social exploitation, 3+ children). Not only for extreme poor but also people with low level of assets (1 housing and 2 movable property). Brasil Sem Extreme poor families with insufficient production and with food insecurity. Miseria Chile SIPS Poor and vulnerable (specific programs may be serving broader populations) Mexico Prospera Cash Transfers’ beneficiaries. Prospera 10 Triano et al (2019). 19 Peru: 1. Municipalities are chosen based on an assessment made by Foncodes (socio-economic conditions, potential participation of local governments and available budget). Haku Winay 2. Then, Centros poblados are identified based on socio economic conditions and inclusion in public policies (percentage of population on monetary poverty, high levels of child chronic malnutrition, high levels of unmet basic needs). 3. Finally, households are identified in selected municipalities, originally based on number of Juntos families, and since 2017 based on the number of households in subsistence economies. Households should have at least 3 out of the 4 characteristics: i) living in rural areas with less than 2,000 inhabitants, i) women head of household with less than primary school education level, iii) head of household with quechua, aimara or amazonic languages, iv) belong to the poorest quintile of income. Source: Authors’ own elaboration Conversely to Mexico, Argentina decided to implement PI projects targeted to a wider population. Based on the lessons learned from a decade of PI programs, the Ministry of Health and Social Development in 2018 put in place Hacemos Futuro, a program aimed at empowering vulnerable populations by means of offering them simultaneous access to cash transfers, education and training, social and labor intermediation services, with the ultimate goal of increasing their economic autonomy and social integration. As in other LAC countries, with this program the government of Argentina was willing to create incentives for social assistance programs’ beneficiaries to adequately exit (egreso acompanado) the programs. For this program the target population was not only the extreme poor but the working population (18-65 years), those vulnerable or at risk, and prioritized female or transgender heads of households in a situation of socio-economic vulnerability (as outlined in Table 2). The target group went beyond extreme poverty, based on the recognition that even people with low level of assets (vulnerable) need strategies to adequately access productive opportunities and avoid falling into poverty. The verification of vulnerability of potential benefits in this program was done through analysis of information in the National Tax and Social Identification System (SINTYS for its acronym in Spanish). Applied experience from various countries has provided lessons about the need for being flexible with PI policies. This becomes obvious with the sometimes misunderstood and often oversimplified trade-off between flexibility or resilience, and efficiency. Many societies maintain an insatiable appetite towards greater efficiency, whether in technology, logistics, public spending, or personal timekeeping and beyond. However, when various higher levels of efficiency are achieved, what is often lost is layers of slack or overlap that otherwise exist as sometimes intentional, other times unintentional, safeguards (Martin, 2019). PI policies seek to add back in some of those safeguards through a variety of programs that target the same and different groups of people while striving for individual program efficiency. Two separate programs that seek to provide benefits for working mothers but with different criteria for qualification or different areas of coverage is less wasteful duplication than mutually reinforcing and supporting programmatic planning to provide working mothers the best possible chance at success. While flexibility and the accompanying resilience is important, a balance with efficiency is still important as resources are often limited. In practice, this tilt back towards resiliency and flexibility through Productive Inclusion policies comes out in several important ways. First, low income and vulnerable groups are profoundly heterogenous and therefore, they need a bundle of services that is flexible enough to adapt to the specific conditions of individuals, households, and regions. Second, from an 20 operational perspective, PI interventions combine different types of services from various institutions that often cover diverse types of populations. This variety of target populations creates the need to expand the focus on the interventions and use a ‘territorial/geographic’ approach. Third, working only with low income and vulnerable could be difficult or non-strategic since, in practice, low income and vulnerable groups belong to communities and permanently interact with other members that may not be classified as poor. Segmenting populations only based on income can impose artificial divisions or create internal divisions within the communities which could become an issue in the effective implementation of PI strategies. A good example to illustrate the need of flexibility in the definition of the target population of PI interventions is the Program Haku Winay (HW) in Peru. HW was created in 2012 in response to the need to facilitate the exit of the beneficiaries of the Cash Transfer (CT) program Juntos, through the combination of continued CT with the provision of a series of strategies aimed at improving their economic productivity. The idea was to promote a ‘virtuous cycle’ where improvements in the productive capacity of low-income households encouraged them to be sustainable and, in the long term, exit social assistance programs. Responding to this logic, the geographic targeting of HW was concentrated in districts with large presence of Juntos recipients. However, in practice, this objective was partially achieved. There was a clear disconnect between the ‘household’ logic of Juntos and the territorial/geographic one of HW. Since the beneficiaries of Juntos were highly dispersed, it was challenging for HW to create the community organizations required by the model to provide technical assistance. Also, poverty targeting generated tensions within the communities since some members considered it unfair that the benefits of the program were offered only to Juntos beneficiaries. This targeting generated distrust between Juntos and HW teams which in some occasions translated into lack of incentives to cooperate. Facing the challenges of converging PI with Juntos beneficiaries, in 2017 Peru’s Ministry of Development and Social Inclusion (MIDIS) adjusted the target population. The targeting of HW was moved away from Juntos beneficiaries to households under ‘subsistence economies’ (expressed in extension of land they own, 1.3 ha). Although in practice this new targeting mechanism significantly overlaps with Juntos, it removed the technical tensions of having to work only with those classified as poor or extreme poor. Also, MIDIS made the decision to encourage collaboration between Juntos and HW teams but maintain independent targeting and delivery mechanisms for both programs. Brazil’s PI experience did not necessarily bring any new target population into focus compared to previously existing programs, but it did bring many programs under one PI umbrella, Programa Brasil Sem Miseria (PBSM). In practice this meant almost the same target population as before in Brazil, however, participants of one program within the newly formed PBSM now had a better chance of qualifying for other programs, that they previously would not have had when the programs operated as stand-alone initiatives. PBSM made it easier for participants to learn what was available and take advantage of a broader range of benefits. The basic qualifier for PBSM initiatives when it launched in 2011 was a monthly income less than R$70. The umbrella aspect of PBSM injected much needed coordination and flexibility to an otherwise patchwork approach that existed previously. The overlapping, yet not duplicative, bundle of services offered under PBSM meant beneficiaries did not have to meet singularly strict target requirements based solely on just one criterion such as age, income, or geography, although Bolsa Familia families were prioritized. 21 Chile’s Intersectoral Social Protection System (SIPS), primarily aims to provid e benefits to the most poor and vulnerable. SIPS is comprised of three main sub-components that primarily serve different target populations within the general mandate of SIPS. Chile Crece Contigo (Chile Grows With You) is an early childhood development program that targets children of low-income working parents; Seguridades y Oportunidades (Securities and Opportunities) targets vulnerable indigenous populations, as well as women more generally, and is a labor support program; finally, Cuidados (Care), which aims to help the elderly and the disabled. The main targeting instrument for SIPS is the Ficha de Proteción Social (FPS), a socio-economic information system established to obtain greater information from target families which is then used to match those beneficiaries to programs throughout the system. 3.1.2 PI Components Most programs’ design includes a menu of services and benefits to address social and productive constraints. Data from a sample of 138 programs in 11 LAC countries suggests that more than half of PI interventions have more than one component. (see Box 2 for a brief description of the inventory of PI interventions). Furthermore, more than half of all the programs include at least one social component. In particular, nearly 70% of programs with two components and more than 80% of programs with three components have at least one social component (Figure 4). An example is Brazil’s successful Bolsa Familia program, a conditional cash transfer program that targets the extreme poor in Brazil, especially women-led households. The benefits are conditional on several criteria, including children under 17 attending school and prenatal examinations for pregnant women. Through PBSM and the Ministry of Social Development and Fight Against Hunger, Bolsa Familia is connected to other necessary programs such as Pronatec (education) and Brasil Carinhoso (income, ECD, and nutrition/health) to help families, including mothers, meet these conditions to reach the income guarantee. Figure 4. PI programs across the region by component a. Distribution of PI programs by number of b. PI programs with social inclusion component components 160 160 140 138 140 120 120 100 68 100 80 80 60 57 18 60 46 40 70 45 5 40 29 20 39 24 0 1 0 3 0 2 0 1 20 3 2 1 0 With social components Without social components 22 c. Programs with only one component d. Programs with two components 1 4 1 1 4 1 component 20 2 components 5 6 24 14 27 0 10 20 30 0 10 20 30 F P L S+C S+P C+P L+P I+P S+L P+F I+L Source: Authors’ calculation based on the data of stocktaking exercise. On the other hand, one third of all programs in the inventory only have a single component with most focusing on labor intermediation. More than half of the single component PI programs are focused on labor intermediation, more than 40% in access to productive assets, and less than 10% in access to financial systems. Among those programs with two components, 60% combine labor intermediation and monetary transfers, and 23% combine access to assets with access to financial services. Overall most programs with more than one component combine issues of income support and access to the labor market, or support for productive development with financial services. Selectivity to choose the right approach for each beneficiary/household is key on PI interventions. In addition, the duration and size of the bundle of B&S varies according to the needs of beneficiaries, but also according to resource availability (i.e. in the leading/implementing agency) and to existing capacity to coordinate the provision of services with additional public (and private) service providers. 23 Box. 2: Brief Description of the Regional PI Intervention Inventory The program inventory provides an overview of productive inclusion programs in Latin America and the Caribbean. The review focused on programs managed at the central government. Data was collected on 11 countries including Argentina, Brazil, Chile, Colombia, Dominican Republic, Honduras, Jamaica, Mexico, Nicaragua, Paraguay and Peru. In all over 300 programs were reviewed, 150 of which were classified as productive inclusion. Data was collected based on a desk review of official government reports and official websites, other regional databases, as well as information provided by country-specific informants. Information was cross-checked using the ASPIRE administrative data database. Data collected includes information on program design characteristics and components, such as start-up year, institution in charge, description of each of its components and beneficiaries targeted, as well as annual programmatic data on the number of beneficiaries and the amounts spent for each program. Due to differences in ease of access, data gathered was not homogeneous across all countries and beneficiary and expenditure information was not available for all programs. However, because the analysis focused mostly on the composition of the programs this limitation did not have a great impact for the purpose of the study. For the purpose of the review productive inclusion was defined as the range of mechanisms that support the participation of the poor and vulnerable in labor, product, credit and other markets, by jointly tackling a range of household and community level constraints. Programs were classified according to the type of vulnerability they addressed. These included five types of vulnerabilities: 1) income and poverty, which refers to low income low levels of other easy-to-observe household characteristic correlated to the income level, such as location/quality of the household's dwelling, ownership of durable goods, etc, 2) employment, which refers to unemployment and underemployment, 3) geographical which refers to location based vulnerabilities, 4) Demographic/social, which refers to demographic and social characteristics of the group targeted (e.g. age, gender, or ethnic group, disability), 4) access to services/markets, which refers to low access to financial and production markets. Programs were also classified according nature of the benefits (also referred to as component or measures) provided. The database distinguishes between six types of components belonging to two main categories: social inclusion and productive development. Measures aimed at providing income support, social services for households and community support were considered as social inclusion components; while measures aimed at fostering labor market services, supporting productive development, and supporting financial services were considered as productive development components. More specifically the following interventions were classified for each of the categories: 1. Income support: CCTS, family/child allowance, emergency support in cash, scholarship benefits, public works and direct job creation, food stamps and vouchers, food distribution programs, School feedings, nutritional programs, in-kind emergency support, targeted subsidies (health, educational fee waiver, housing/utility), transportation benefits, other. 2. Social services for the HH: includes day care services for vulnerable children, adult, basic and continuing education, workforce development/ skills, education, health, housings. 3. Community support: community infrastructure, monetary aid 4. Labor market services: Training (vocational, life skills), internships, employment incentives, public works and direct job creation, entrepreneurship support/startup incentives 5. Productive development support: Endowment of inputs and productive assets, technical and non-technical assistance, training of personal and social skills, agricultural extension/advisory services, entrepreneurship promotion, seed capital, access to market information, improving market linkages in the value-chain, organization of producers, marketing support 6. Financial services: Access to accounts, access to affordable financial services and credit, digitalization of cash payments, promotion of mobile financial services, Financial education, Subsidized credits, formal credit promotion, financing of assets and promotion of savings, direct subsidies and soft loans for investments or equipment To be considered as productive inclusion a program had to target the poor and vulnerable and to contain at least one productive development component. Programs that included both productive and social components but were not targeted to the poor and vulnerable were excluded from the database. In all 150 programs were productive inclusion. 24 Rural and urban distinctions are evident. There is a clear preference in most programs to offer a menu of food security, technical assistance, asset transfer and micro-business development to rural poor and education or training and labor intermediation for urban beneficiaries. Compared to rural programs, urban programs tend to offer a more diverse portfolio of B&S, possibly due to the more limited supply of services at the rural level. Figure 5 – Most common components of PI strategies Source: Authors’ own elaboration In Peru HW is designed around four main components. As mentioned previously, at its creation the Haku Winay aimed to promote the economic inclusion of the extreme rural poor. MIDIS’ argument was that the poverty traps of rural households were due to three main factors: underused human capital, low productivity, and lack of access to markets. The objective was to promote income and food security transformations at the household level by means of improving agriculture productivity and access to markets. In order to address those constrains, HW was designed around four main components: (i) strengthening of rural family production systems (through the provision of technical assistance, training and productive assets), (ii) improvements of housing conditions (kitchens, water and waste disposal), (iii) rural inclusive business (business development services and access to capital through competitive funds), and (iv) financial capital (financial education and savings incentives). These components were delivered in a three-year period, with intense support from local extensionists, especially during the first two years. One of the distinct elements of the bundle of services provided by HW was that instead of bringing foreign or complex concepts, the program opted for promoting peer to peer learning incorporating idiosyncratic knowledge thorough local technical extensionists, known as yachachiqs (“the one who knows”) who provide technical advice on both productive and financial areas. Also, HW introduced small basic technical innovations to improve small rural workers productivity (irrigation systems, breeding of small animals, installation of cultivated pastures, installation of orchards in open fields or in greenhouses, technologies for breeding native animals, etc.) (see Figure 6). 25 Figure 6. Components for Haku Winay Source: Authors’ own elaboration In Argentina, Hacemos Futuro (HF) was designed to respond to a series of barriers that prevent families to overcome poverty or be financially independent.11 These barriers were related to the number of children, the presence of some form of disability in the household, economic barriers (lack or limited assets), lack of employment, conflictive family dynamics (i.e. gender violence or sexual exploitation), and gender discrimination, among others. To address these barriers the government designed a package of benefits targeted to individuals (not households or communities like in previous PI experiences). The benefits package includes conditional cash transfers paired with access to health services (regular medical checkups); scholarships to finalize high school education or job training programs; labor intermediation services - including labor profiling and coaching to access and retain job opportunities; and coverage of other services like the monotributo social12 and microinsurance. The program duration is set to be one year, which is relatively short compared to other countries, though some beneficiaries may have been beneficiaries of preceding programs with different names, considering that this is a new program. One important feature of this program is that it is compatible with other universal benefits like the Asignacion Universal por Hijo (AUH) and universal maternity benefit (AU por embarazo), which are universal monetary transfers aimed at protecting human capital. Transfers are conditioned to the regular update of information, participation in education and training services,13 and compliance with medical checkups. To encourage job retention, the program also implemented a series of innovative strategies that are aimed at facilitating the exit (egreso acompanado) from social assistance benefits (See Box 3). 11 It is important to note that this is a new program, with the implementation in its early stages. Thus, the information refers largely to the design features of the program the implementation of which remain to be evaluated. 12 Monotributo social definition to include. 13 As per training conditionalities, the program demands at least to comply with 120 hours per year for people attending primary or secondary education and 300 hours for people receiving job training. Beneficiaries at university level studies should demonstrate they are participating in job orientation or small business development programs. 26 In 2016, the Mexican government established the operations for PROINPO, a strategy aimed at connecting cash transfers’ beneficiaries with existing PI programs. The services provided under PROINPO aimed to be comprehensive consisting of business development services and technical assistance capacity building) from the stage of project formulation until after the group of beneficiaries received financing. However, despite the effort to link programs from different sectors, the government realized that there were important differences in objectives and program design that prevent institutions from effective cooperation, resulting in its early closure. Box 3: Argentina HF: Strategies to promote job retention Hacemos Futuro is implementing a series of strategies aimed at ensuring adequate and sustainable labor and productive inclusion of their beneficiaries. These strategies include: - Protection of employment income (ingreso protegido del empleo): Beneficiaries that manage to access the labor market with an income higher to the minimum salary, receive the cash transfer for one year. Those who receive up to 50% of the minimum salary receive the cash transfer until they increase their income. Finally, those who receive a salary between 50-100% of the minimum salary receive 50% of the cash transfer. - Valor.ar- an individual coaching service for beneficiaries who can get a formal job that includes training and soft skills. This service is paired with temporary cash transfers. The program includes economic incentives for beneficiaries who manage to keep a job for more than 4 and 12 months, respectively. The program also offers incentives for coaching and labor intermediation firms (NGOs and government organizations) that help beneficiaries to retain their jobs for more than 12 months. - PIL Empalme, an employment subsidy offered by the Ministry of Labor, that transfers beneficiaries up to 24 months of a subsidy (equivalent to the cash transfers offered by HF). This amount is deducted from the salary paid by the employer. - Nexo, a strategy to implement social internships at the community level, for instance, by means of developing social infrastructure. Education and training services are complemented with soft skills including better practices in health, gender, and access to services and rights. For those beneficiaries who were not willing or able to access formal jobs, the program also has a menu of additional services to promote self-employment including business development services, microcredits, subsidies to buy equipment, and technical assistance. To address this, the government moved away from only linking CT beneficiaries with existing productive programs, to directly assume the responsibility of designing and implementing PI programs. Different programs have been created to: i) try to address the challenges that are determined by geography (Programa Piloto de Territorios Productivos) by means of funding small community income generation projects, ii) provide individual and community productive coaching (Proyecto de inclusion productiva rural), and iii) providing a package of simultaneous benefits and services that respond to the beneficiaries’ needs (Proyecto De la Mano con Progresa, the Mexican adaptation of the BRAC model). These programs have ran in parallel with the traditional linkage model through PROINPRO, bringing the total number of different productive inclusion models implemented to four. 27 3.2 Implementation Institutions for Program Delivery Similarly, it is important to gain support at the local level and ensure local ownership. Local authorities may serve as connectors between beneficiaries and benefits and services that are available in the field, since subnational governments have first-hand knowledge of the needs and demands of the local population. Failure to adequately include local authorities may result in creating distrust between national and local authorities or suboptimal results in the program’s implementation. Cooperation between the coordinating agency at the national level and the implementing bodies at the local level also ensures a greater opportunity for programs to succeed because of the stability and greater buy-in from many players. This sort of vertical coordination works best when roles and responsibilities are clearly defined across the different levels of government involved. In addressing specific constraints, such as social and community infrastructure, community organizations are the key stakeholders in diagnostics, decision and implementation. Some of the social and productive constraints need to be addressed through community interventions, like small infrastructure (e.g. water cisterns or roads), local services (e.g. security), or even assets directly related to productive initiatives (e.g. water canals). Implementation of productive inclusion initiatives should incorporate these community-level dimensions as they may be the binding constraint for higer productivity and earnings. The role of communities in deciding the content of the project in Peru’s Haku Wiñay, or the content of the technical training in urban Argentina, are examples of this. Moreover, this raises the importance of agencies involved in supporting community building activities such as social funds or other agencies in direct interaction with local governments. The original design of the HW program placed emphasis in two key actors for the delivery of the program: community organizations (nucleos ejecutores) and local (subnational) governments. The nucleos are small groups of rural households living in the same centro poblado, who in practice are the counterparts of Foncodes. Thanks to the special legislation of Foncodes issued back in the 1990s, the nucleos are able to make financial, legal, and administrative decisions for the development of their business plans, as well as to receive resources directly from the national and subnational governments, without intermediaries. The existence of nucleos ejecutores was meant to empower beneficiaries to make their preferred choices. On the other hand, the subnational governments were meant to participate in the local committees for the distribution of resources, with the aim of keeping them informed about the activities, facilitate linkages for the inclusive business component, and eventually bring additional resources for the development of the projects. In practice the experience shows that the delivery model is complex, demanding, and highly dependent on the local social and political dynamics. For instance, it is common that the nucleos have low autonomy vis a vis other stakeholder, for example, the extensionists, powerful social or political actors, or local governments who ended up having more leverage in the decision-making process. Similarly, there seems that local governments do not have enough incentives to cooperate with HW as they were bypassed in the distribution of resources that went delivered directly to the nucleos. Also, the projects’ cycle does not always coincide with the cycle of local government programs and hence subnational authorities prefer to invest their own resources in other kinds of activities that could bring them political recognition. Given these results, Foncodes has lower 28 expectations, respect overall, and participation of local governments and therefore municipal engagement has also been removed as one of the conditions for geographical target. In Argentina, HF was moving away from relying on local actors for beneficiary enrollment and delivery of services.14 This was motivated by past negative experiences in productive inclusion programs. However, this concentration of responsibilities in the central level could create potential problems in the long run, especially in accessing uninformed, less connected, less digital beneficiaries. While there is no one way to implement PI initiatives, programs should be delivered in a way that is flexible enough to adapt to specific conditions at the subnational level. An early and adequate assessment of the local political economy is important in crafting the flexibility of operation rules. This flexibility in responding to regional dynamics can increase the chances that initiatives are taken up by policymakers at the local level who can then create incentives for low income producers to engage in PI activities. It is also crucial that program criteria and rules respect social and community rationales (i.e. when working with ethnic minorities). Adaptability includes the capacity to recognize that some elements of the menu of B&S are not appropriate to address the risks and vulnerabilities of certain groups (i.e. yachachiq technical assistance was not relevant in the Amazonian region in Peru15). Therefore, PI should be able to rapidly adapt and implement context relevant interventions. Adaptability means timely recognition of any possible design flaws and the ability to rapidly correct course to ensure that programs can attain their objectives. Relatedly, this also means countries being able to recognize that not every territory is adequately resourced to host PI initiatives: there is need of a minimum stock of climatic, economic, market and connectivity conditions that enable low income producers to adequately integrate in productive dynamics. In these cases, the key is to identify alternative policy options to effectively protect the most excluded groups. 3.3 Information Systems to Enhance Coordination The existence of a central social information system or registry integrated into the heart of PI programs can provide a strong platform that boosts efficiency of policies overall. Such information systems can act as a single access point for beneficiaries that help simplify what might otherwise appear as a complex, tangled system to someone unfamiliar and on the outside. Social information registries can also act as the sole component of productive inclusion policies that touch every other component. These systems properly document the needs of applicants and provide a menu of benefits and services available to them specifically, including a referral component, where beneficiaries can be referred from one program to another in a sequence or based on need. As applicants become beneficiaries and move through the PI programs, registries provide immeasurable advantages in tracking progress and delivery of services and benefits. While central 14 It should be noted that the information refers to the program elements as of July 2019, some of which may have changed since then. 15 Partially to overcome this issue, the program launched an adaptation of the HW scheme for amazon areas, called Noa Jayatai; although there is still room to improve the adaptation of technologies to the Amazonian region. 29 social information systems or registries exist in many countries in the region, their role in the implementation of productive inclusion approaches differs, as outlined below. HF in Argentina is undertaking important steps to implement direct channels of communication with beneficiaries to avoid relying on local intermediaries. The information platforms include Formate en Red, a unique channel with all the information about training options (face to face, virtual, soft skills); Portal de Empleo, a labor intermediation portal with information for both applicants and firms; App Mi Argentina where the beneficiaries can get information about their status in the program such as payment timing, news, and training options; and HF and ANSES websites and other regular channels of citizen services (Figure 7). Figure 7. The Information Ecosystem for HF in Argentina Despite these efforts, the program does not seem to have a strategy for the integration and use of this data. A series of data will be collected through these channels, which can be utilized and integrated for use, such as how to monitor the labor trajectory of program’s beneficiaries or how to map available productive or labor opportunities. However, currently the program does not seem to have a strategy in place for how to integrate and use the rich set of data they plan to collect, indicating the importance of going beyond data collection and focusing on its integration and use. In Brazil, the Unified Registry lies at the heart of the PBSM. As an umbrella program presiding over numerous, diverse programs, the Unified Registry acts as a common thread that connects all of PBSM. The The registry is effectively a gateway to access the entirety of PI policies in Brazil and simultaneously a facilitating mechanism that connects PI programs to people. Local municipalities play a large role in social assistance, largely because it is up to them to add households to and maintain the registry, underscoring the crucial aspect of cooperation between different levels of government in PI programs, especially in Brazil. The main targeting instrument for SIPS in Chile is the Registro Social de Hogares (RHS). RHS is a socio-economic information system established to obtain greater information from target families which is then used to match those beneficiaries to programs throughout the system. The system existed before PI programs but was simplified and streamlined to obtain only the information needed for applicants to qualify for SIPS-related programs. Previous versions of RHS were burdensome and required documentation of income as well as assets. 30 3.4 Incentives: Financing and Budgetary Arrangements Supporting the PI Package Productive Inclusion policies are most effective when many actors commit assistance and resources to the cause and many beneficiaries trust the process enough to join. Incentives can be an effective mechanism to achieve this by bringing more ministries and institutions on board and attracting target populations to enroll. Incentives can help bring necessary actors around to join the cause since PI programs are often new and unknown. Considering the multi-sectoral nature and various levels of government required to carry out PI programs effectively, incentives may be required to reach a broader target population, facilitate change in ministries in how things were done previously, and to deliver relevant services. In Chile’s case, the SIPS contains an incentive-based coordination mechanism across Ministry to ensure close coordination. This mechanism commits the Ministry of Social Development and Families (MDSF) and the Ministry of Labor and Social Welfare (MINTRAB) to working closely together towards common outcomes. The two ministries review their SIPS agreement annually, with technical teams from each updating the central aspects such as coverage to be met, standards, and services and benefits for the following year within the SIPS framework. Crucially, both ministries commit financial resources towards the SIPS, and are thus both financially invested in its outcomes. After mutual approval of the agreement, MDSF provides its SIPS database to MINTRAB, which in turn connects beneficiaries to private entities. They continue to collaborate throughout the length of the agreement by sharing and reviewing financial and technical reports, as well as communicating when discrepancies arise. This necessary coordination ensures both ministries are invested in the stated goals of SIPS and incentivizes them to work closely with each other to achieve the most efficient and far-reaching outcome for SIPS beneficiaries. Figure 8 illustrates the main elements of this inter-institutional coordination mechanism. One of the lessons learned from the implementation of the HW program in Peru has been the difficulties to articulate interventions between institutions. These difficulties are closely related to the prevalence of independent objectives, goals, and budgets that reward individual rather than integrated performance, even within the same Ministry, hence the lack of incentives to cooperate. Direct incentives are also needed at the local level. For instance, to reduce the barriers of low-income producers to access markets, governments may need to organize local fairs or provide technical assistance on how to add value to their products. However, this type of incentive could also have a negative impact or create market distortions that may harm the production of possibly less poor but more independent small producers. 31 Figure 8. Flow of the Process for the Interinstitutional Coordination mechanism- Chile 32 4. Evaluating PI: Challenges and Results So Far The evaluation of productive inclusion programs in LAC is still incipient . Thorough assessments of PI programs could not take place without efficient and innovative evaluation methods undertaken at a key stage in the program cycle to provide crucial information about the effectiveness of the programs to make informed decisions. These evaluations also aim to provide information about the operation of the multi-program activities in order to recommend adjustments for contextual designs and the effective implementation of such programs in the future (Rubalcava, 2015). On the other hand, while most programs that include some PI element also have impact evaluations, the evaluation of the integrated interventions is relatively scarce (Universidad de los Andes, 2016). An additional challenge is detailing the overall costs of PI programs, which are often obscured in budgets across agencies and levels of government, as well as sometimes aggregated within other non-PI program budget lines. While impact evaluations were the primary approach to evaluate productive inclusion in LAC, most had a non-experimental approach. Evaluations that do exist tend to be non- experimental or observational, which estimate the impact of initiatives through various econometric techniques.16 Most evaluations use national surveys, census data, or administrative records as main data sources. However, using such secondary data sources could pose a challenge as they are not designed with PI evaluation purposes in mind. Indeed, the heterogeneity of PI programs and the variety of contexts where they are implemented could require alternative approaches. PI initiatives have particular program designs which tend to evolve over time according to the characteristics of their beneficiaries. Some national PI initiatives such as PROSPERA in Mexico have gone through several changes since their first design and have involved changes in target population and outcome indicators (Rubalcava, 2016).17 In that light, no single standard approach could be suggested for the evaluation of productive inclusion initiatives. Lack of data sources and in particular for PI initiatives with local scope, is a challenge for evaluation in the region. For instance, assessment of Bolsa Familia (BF) and PRONAF in Brazil or Oportunidades and Procampo in Mexico used agricultural censuses and national surveys to elaborate impact evaluations. In these cases, the data did not include variables to estimate the impact on key outcomes of PI specifically. In addition, these secondary sources usually only provide small samples from the targeted regions which imply low statistical power for estimations. Indeed, PI initiatives with local scope, where groups that receive benefits are small in population, are usually not represented well in national surveys and are more likely to require primary information gathering. Therefore, the evaluation of this type of initiative requires a specific record of the progress of these populations, along with an early implementation of the monitoring and evaluation aspect of that program. However, these methods can have limitations to estimate accurately the impact of PI programs. 16 Across the 11 impact evaluations analyzed, only two Peru’s Haku Winay and Nicaragua’s Atención a Crisis (AC) and Programas Complementarios (PC) have a randomized controlled trial design, while the remaining 9 are observational studies. 17 Process evaluations for some of these approaches (i.e. De la Mano con Prospera, Territorios Productivos) were undertaken which provided some valuable lessons. 33 Most of the evaluations of PI initiatives aim to determine the progress of programs in solving causal problems and to identify the limiting factors at an operational level. Likewise, in the case of PI initiatives that involve strategies to connect two or more programs, it is also important to analyze how every program implemented (or did not implement) the activities as planned and if the beneficiaries received one or more of the interventions in the B&S package available to them in a timely manner. Consequently, evaluation strategies must be designed to take into account the complex dynamics of some of the most integrated and complimentary programs. An alternative approach to conventional impact evaluations and process evaluations that could be suitable for the iterative nature of PI initiatives are Problem Driven Iterative Adaptation (PDIA) evaluations. The PDAI approach is a methodological framework that involves organizations generating, evaluating, and refining solutions according to specific contexts. This approach proposes a type of evaluations that satisfy the constant and reflective learning of the experiences implemented in order to guarantee the effectiveness of the interventions. There are many reasons to consider a PDIA approach for the evaluation of PI initiatives. One reason relies on the fact that intervention packages or guidelines for PI initiatives are in constant design and redesign. They include simultaneous or consecutive implementation of multiple actions and not a single treatment. Therefore, the program expected outcomes and indicators usually vary over time. While this main characteristic of PI initiatives usually limits the implementation of impact evaluations, this could be a learning space for PDAI evaluations. Another aspect to consider is the cost of the evaluations; in the case of the PDAI, the evaluation instruments that accompany the implementation of the initiatives are low cost and could provide an effective gauge of the effectiveness of the program. In sum, the selected evaluation approach must respond to the magnitude of the intervention and its characteristics. An evaluation scheme may involve one or more approaches. For example, for a PI strategy at the country level, as in the case of Mexico, it may be relevant to have an evaluation framework that involves impact evaluations and PDAI evaluations. On the other hand, for PI initiatives with local scope or in low-middle income countries, a PDIA evaluation or a process evaluation could be more informative and cost-effective. Additionally, in line with Rubalcaba (2016), it is suggested that PI impact evaluations should be complemented by qualitative and in-depth studies that allow the documentation of the characteristics of the intervention providing information to contextualize results and report changes. Overall, PI programs and policies across the region demonstrate positive results, though in varying degrees (Annex 1). In Brazil for instance, 88 percent of families in extreme poverty were reached by at least one program within PBSM. Families that accessed a program though RPIS achieved the best progress with an increase in family income of 134 percent (Mello 2018). Argentina’s HF managed to meet its early goals of number of recipients reached, although it still has far to go in reaching all or even most who qualify. The component of HW in Peru that transferred productive technologies to beneficiaries led to significant increases in farm output, while home improvements enhanced household health and well-being. Annex 1 presents a comparison of the various elements (design, data, limitations, results, etc.) for a sample of impact evaluations of Productive Inclusion Programs in LCR. 34 While there are many successes to celebrate across the various PI programs, there are an equal or larger share of challenges yet to be overcome as well. In Brazil, a large share of beneficiaries was only reached by one policy or program. Argentina’s HF had a labor intermediation take-up of only 2.7 percent as of March 2019, a crucial aspect of PI that should be improved for greater impact, highlighting the importance of implementation. In Peru, success of PI programs varied greatly, with many programs geared toward Andean productive and social conditions, but less so elsewhere. 5. Lessons Learned The various successes and challenges from the early iterations of PI policies in the country cases examined here can provide a useful list of lessons moving forward. The first is to simplify when crafting initial programs since keeping it simple can help facilitate easy and quick implementation. Small, early successes can help build buy-in from more skeptical stakeholders and form the foundation for more complex, further-reaching policies later on. Less complicated policies can also help make it easier to scale up if programs prove successful in the early stages. Streamlined policies are simultaneously inclusive, due to the ease of understanding and thus taking part in, but also exclusive as they often cater well to homogenous populations only. While not perfect, simpler programs prove easier to work with early on and retain the possibility to increase in complexity and reach more people as they increase in scale over time. Balance in scope of programs and capacity increases in importance after the early stages of implementation. Another lesson learned is to allow for adaptation and remain flexible. Flexibility at all stages of a policy’s lifetime is important but it is especially critical during implementation. There will almost certainly be unforeseen challenges such as capacity constraints, misallocated resources, and confusion. These challenges are not inherent failures of policies but rather growing pains and adapting can help get over the initial difficulties. Crucial to early flexibility is a robust system of monitoring and evaluation. Adaptation hinges on information about implementation, bottlenecks, and preliminary outputs and results. Even the best planned programs may require revisions on the fly. These reassessments and revisions are much easier if early returns and results are detailed and readily available. No results will be perfect as even the best monitoring systems will return imperfect, varying results from which it can be difficult to draw easy or clear conclusions. This lesson points more to the heterogeneity of a population than any shortcoming of a program or evaluation system. Even if all PI program participants have something in common (poverty), there very likely are a host of different traits that affect their experience and ability to successfully partake in a productive inclusion program. Results for PI programs are typically better among the less poor, given their asset base and lower risk aversion. Social assistance may help reducing the risk aversion and asset base constraints overall. Avoiding the narrative of “graduation” from social assistance would help highlight the complementarity between social assistance and PI. Social assistance and productive inclusion programs are designed to be complements, not substitutes; PI programs are ultimately a broader platform for B&S package delivery than that of social protection and assistance programs. While the better-off are more able to leverage a productive inclusion package and are more likely to escape poverty, the PI approach may not be preferred by the poorest and most risk averse since PI 35 policies usually involve higher levels of income risk. For instance, the experience of HW demonstrated that households who have more opportunities to take advantage of the bundle of services provided by the program are not necessarily the poorest, in other words there seems to be a minimum household threshold under which is unlikely for households to be able to adequately integrate into sustainable productive activities. This threshold includes objective aspects like the number of household members and availability of productive assets, as well as more subjective things like the entrepreneurial capacity of household members. The complementarity of social assistance and productive inclusion ideally smooth the transition out of poverty that implies better risk management for a heterogenous population. Establishing output- and delivery-based agreements with partners will ensure better coordination and thus better results. By focusing on delivery outcomes rather than implementation differences, various implementing agencies can share a common goal while retaining their independence and playing to their strengths. The most successful policies often invested in aligning output goals among numerous stakeholders and left the implementation and delivery of programs to relevant agencies. Finally, PI approaches work better if nested within territorial development plans. This approach often aligns productive inclusion goals with those of existing local development plans in a region to address specific community and local constraints. This approach often strengthens community participation and ownership while connecting participants more easily to both economic and labor markets. Strengthening this local market connection can then help translate improved productivity into real income gains. Additionally, local development plans usually already have private sector participation which can provide information (specific products or labor skills), resources (leveraging financing), or management (especially in micro entrepreneurship programs) that supercharge PI programs. While these lessons are valuable in informing policy directions, some pending issues in this agenda warrant further research. The policies discussed above from the past decade have largely been successful, yet there is ample room for strengthening the PI approaches, even in countries where these policies have been established for a number of years. Some of the pending issues in this agenda include, strengthening the productive support (e.g. agriculture and livestock) with food security and nutritional components; use of productive inclusion programs for ecological reasons; strengthening of the labor dimension of productive inclusion programs especially in rural areas where the focus is more on agriculture; and strengthening the coordination across and further involvement of other actors, ministries of labor, agriculture, and commerce/economy. Blending nutritional programs into PI initiatives for instance could produce gains in both rural and urban settings while taking advantage of the effective distribution network most nutritional programs boast and contributing towards food security. Meanwhile, PI programs could also prove effective vehicles to carry out environmental goals for ecological reasons such as forestry preservation. These partnerships could leverage the high-functioning, robust agricultural ministries in many countries to complement and enhance the effectiveness of PI policies, as well as reach a wider population. Other possible actors for PI program expansion could include Labor and Commerce ministries and a whole host of private enterprises that focus on human capital. 36 References Acosta, P., J. Baez, R. Beazley and E. Murrugarra. 2012. “The Impact of the Financial Crisis on Remittance Flows: The Case of El Salvador” in Sirkeci, Cohen and Ratha (eds.) Migration and Remittances during the Global Financial Crisis and Beyond, The World Bank. Aldana, U., Vásquez, T., & Yancari, J. (2016). Sierra Sur y Juntos: Un análisis a partir de la implementación conjunta de ambos programas en Chumbivilcas, Cusco, Perú. In J. Maldonando, R. Moreno-Sánchez, J. Gómez, & V. Jurado, Protección, producción, promoción: explorando sinergias entre protección social y fomento productivo rural en América Latina. (pp. 389-462). Facultad de Economía, Universidad de los Andes, Bogotá: Ediciones Uniandes. Asensio, R. 2019. Haku Winay y los dilemas de la inclusión económica en el Perú. World Bank, May 2019. Baez, J., A. Camacho. 2011. Assessing the long-term effects of conditional cash transfers on human capital: evidence from Colombia (English). Policy Research working paper. WPS 5681. Washington, DC: World Bank. Banerjee, A., Duflo, E., Goldberg, N., Karlan, D., Osei, R., Parienté, W., Udry, C. (2015). A multifaceted program causes lasting progress for the very poor: Evidence from six countries. Science. Barrett, C., M. Carter, and J.P. Chavas. 2019. The Economics of Poverty Traps. National Bureau of Economic Research Conference Report. Castillo, V., Ohaco, M., & Schleser, D. (2014). Evaluación de impacto en la inserción laboral de los beneficiarios de los cursos sectoriales de formación profesional. International Labour Organization. Cirillo, C., M. Györi, and F. Veras Soares. 2017. Targeting social protection and agricultural interventions: The potential for synergies, in Global Food Security, Volume 12: pp 67-72. De Sanfeliú, M. B., Angel, A., & Shi, M. A. (2016). Transferencias monetarias condicionadas y desarrollo rural en El Salvador. In J. Maldonando, R. Moreno-Sánchez, J. Gómez, & V. Jurado, Protección, producción, promoción: explorando sinergias entre protección social y fomento productivo rural en América Latina. (pp. 313-388). Facultad de Economía, Universidad de los Andes, Bogotá: Ediciones Uniandes. De Souza, Fátima. 2013. “Assistência social e inclusão produtiva: algumas indagações,” in O Social em Questão - Ano XVII - nº 30 – 2013. Del Pozo, C. (2014). Impacto de la articulación entre las transferencias monetarias condicionadas y el crédito agropecuario en la acumulación de activos production de hogares rurales en el Peru. Consorcio de Investigación Económica y Social. Escobal, J., & Ponce, C. 2016a. Metodología Cuantitativa de la Evaluácion de Impacto. In J. Escobal, & C. Ponce, Combinando protección social con generación de oportunidades económicas: una evaluación de los avances del programa Haku Wiñay (pp. 35-66). Lima: GRADE. Escobal, J., Ponce, C., & Paz, C. (2016b). Impactos del Proyecto desde una Aproximación Cuantitativa. In J. Escobal, & C. Ponce, Combinando protección social con generación de oportunidades económicas: una evaluación de los avances del programa Haku Wiñay (pp. 67-114). Lima: GRADE. 37 Favaretto, A. 2019. A Estratégia de Inclusão Produtiva Rural do Programa Brasil Sem Miséria: arranjo institucional, implementação e aprendizados. Report prepared for the Banco Mundial. Fernández, M. I., Leiva, M. F., Ortega, J., & Weason, M. (2016). Efectos sinérgicos entre en Ingreso Ético Familiar (IEF) y los programas de apoyo al emprendimiento en Chile, 2012-2014. In J. Maldonando, R. Moreno-Sánchez, J. Gómez, & V. Jurado, Protección, producción, promoción: explorando sinergias entre protección social y fomento productivo rural en América Latina. (pp. 127-180). Facultad de Economía, Universidad de los Andes, Bogotá: Ediciones Uniandes. Fiszbein, A., N. Schady. 2009. Conditional Cash Transfers: Reducing Present and Future Poverty. World Bank Policy Research Report. Washington, DC: World Bank. Garcia, F., Helfand, S. M., & Souza, A. P. (2016). Transferencias Monetarias condicionadas y desarrollo rural en Brasil: Posibles sinergias entre Bolsa Familia y PRONAF. In J. Maldonando, R. Moreno-Sánchez, J. Gómez, & V. Jurado, Protección, producción, promoción: explorando sinergias entre protección social y fomento productivo rural en América Latina. (pp. 69-126). Facultad de Economía, Universidad de los Andes, Bogotá: Ediciones Uniandes. González-Velosa, C., L. Ripani, D. Rosas. 2012. How Can Job Opportunities for Young People in Latin America be Improved? Labor Markets and Social Security Unit. IDB-TN-345 May 2012. Grosh, M., M. Bussolo, S. Freije. 2014. Understanding the Poverty Impact of the Global Financial Crisis in Latin America and the Caribbean. Directions in Development--Human Development. Washington, DC: World Bank. Ibarrarán, P., and D. Rosas. 2008. Evaluating the Impact of Job Training Programs in Latin America: Evidence from IDB funded operations. IZA Conference Paper. Johnson, R. and C. Kirabo Jackson. 2019. “Reducing Inequality through Dynamic Complementarity: Evidence from Head Start and Public School Spending” in American Economic Journal: Economic Policy 2019, 11(4): 310–349. (Head Start and K-12 spending mutually reinforcing) Macours, K., Premand, P., & Vakis, R. (2012). Transfers, diversification and household risk strategies - Experimental evidence with lessons for climate change adaptation. World Bank Policy Research Working Paper 6053. Martin, R. L. (2019, January 1). The High Price of Efficiency. Harvard Business Review. https://hbr.org/2019/01/rethinking-efficiency Mobarak, A., and M. Rosenzweig. 2013. "Informal Risk Sharing, Index Insurance, and Risk Taking in Developing Countries." American Economic Review, 103 (3): 375-80. Moya, A. (2016). Programas para la reducción de la pobreza: Una evaluación de las sinergias entre Oportunidades Rurales y Familias en Acción. In J. Maldonando, R. Moreno-Sánchez, J. Gómez, & V. Jurado, Protección, producción, promoción: explorando sinergias entre protección social y fomento productivo rural en América Latina. (pp. 245-312). Facultad de Economía, Universidad de los Andes, Bogotá: Ediciones Uniandes. Naude, A., Leal, G. D., Ramirez, F. R., & Aranda, O. S. (2016). El programa de combate a la pobreza y las políticas de desarrollo rural en Mexico: evaluación de impactos y opciones de políticas alternativas. In J. Maldonando, R. Moreno-Sánchez, J. Gómez, & V. Jurado, Protección, producción, promoción: explorando 38 sinergias entre protección social y fomento productivo rural en América Latina. (pp. 181-244). Facultad de Economía, Universidad de los Andes, Bogotá: Ediciones Uniandes. Ocampo, J.A. 2001. “Agricultura y desarrollo rural en América Latina,” in Desarrollo rural en América Latina y el Caribe: ¿la construcción de un nuevo modelo? - Santafé de Bogotá : CEPAL/Alfaomega. Parker, S. and T. Vogl. 2018. Do Conditional Cash Transfers Improve Economic Outcomes in the Next Generation? Evidence from Mexico. NBER Working Paper No. 24303. Rofman, R., I. Apella, E. Vezza, Evelyn [eds]. 2014. Beyond contributory pensions: fourteen experiences with coverage expansion in Latin America (English). Directions in development; Human Development. Washington, DC: World Bank Group. Rubalcava, L. N. (2015). Estrategia de Seguimiento, monitoreo y evaluación de los componentes de inclusión productiva, laboral y financiera de PROSPERA Programa de Inclusión Social. Ciudad de México. Secretaría de Desarrollo Social. 2008. Evaluación externa del Programa Oportunidades 2008. A diez años de intervención en zonas rurales (1997-2007). Tomo I: Efectos de Oportunidades en áreas rurales a diez años de intervención. Coordinación Nacional del Programa de Desarrollo Humano Oportunidades. Secretaria de Hacienda y Crédito Público (2013). Plan nacional de Desarrollo 2013-2018: Una nueva etapa para México. Nota informativa. Triano Enríquez, Manuel, Steta Gándara, María Concepción, Isik-Dikmelik, Aylin y Ávila Parra, Clemente (2019). La experiencia de Prospera con el proceso de inclusión productiva, laboral, financiera y social de hogares pobres en México. Aprendizajes en el contexto de cambio de la administración federal 2018. In Hernández Licona, Gonzalo, De la Garza, Thania, Zamudio, Janet. y Yaschine, Iliana (coords.) (2019). El Progresa-Oportunidades-Prospera, a 20 años de su creación. Ciudad de México: CONEVAL.Universidad de los Andes. (2016). Protección, producción, promoción: explorando sinergias entre protección social y fomento productivo rural en América Latina. (E. Uniandes, Ed.) Bogotá. Vakis, R., J. Rigolini, and L. Lucchetti. 2016. Left Behind: Chronic Poverty in Latin America and the Caribbean. Latin American Development Forum. Washington, DC: World Bank. Wagmaister, F and Reineke, I. 2019. Caso de estudio Argentina: Hacemos Futuro como Programa de Inclusion Laboral y Productiva. World Bank, June 2019. Williams, A. and S. Gonzales. 2020. Towards Adaptive Social Protection Systems in Latin America and the Caribbean: A Synthesis Note on using Social Protection to Mitigate and Respond to Disasters and Climate Risk. The World Bank. World Bank. 2018a. Poverty and Shared Prosperity 2018: Piecing Together the Poverty Puzzle. Washington, DC: World Bank. World Bank. 2018b. Stagnant Poverty Reduction in Latin America. Poverty and Inequality Monitoring: Latin America and the Caribbean. World Bank Group. http://documents.worldbank.org/curated/en/750811537474872191/pdf/130046-PUBLIC-Stagnant- Poverty-Reduction-LAC-2016-final-June2018.pdf 39 World Bank 2019. Effects of The Business Cycle on Social Indicators in Latin America and the Caribbean: When Dreams Meet Reality. Latin America and the Caribbean. World Bank Group. https://openknowledge.worldbank.org/bitstream/handle/10986/31483/9781464814136.pdf?sequence=6&i sAllowed=y World Bank. 2020. State of Economic Inclusion Report – Moving to Scale. Platform for Economic Inclusion. The World Bank. 40 Annex 1 Comparative table: Sample of Impact Evaluations of Productive Inclusion Programs in LCR Type of Design Program Productive Impact Evaluation/ Limitations/ Country Programs Author Year Data Results Concept Inclusion Evaluation Study Identification Estimation Challenges Program Outcomes strategy strategy Data: Treatment Both programs group: did not include a PRONAF had no impact and Comparable Due to a proper baseline BF had negative impact on Minimum Panel data confidentiality survey. agricultural productivity. Areas (CMA), using issues, data The interaction of both BF: where HH Agricultural was grouped Due to programs has a positive impact Transferencias i) Land productivity: real Conditional receive both censuses of at municipal confidential on agricultural productivity. Observational Monetarias value of the agricultural Cash Transfer programs. Difference in Brazil, 1995- level. restriction, However, as only 2.6 percent Superposition study condicionadas y Garcia, F.; production per ha. – (CCT) differences 1996 and Authors analysis on of family farmers received Bolsa Familia (BF) No coordination desarrollo rural en Helfand, S. ii)Income: income per Brazil 2016 Control with municipal 2006. constructed censuses cannot both programs at the same and PRONAF (evaluation on Used official Brasil: Posibles M. & Souza, family worker PRONAF: groups: level fixed CMA which be done at HH time, the positive effect of the overlapping areas) secondary sinergias entre A. P. iii) Child labour: # of Rural credit i) CMA where effects Unit of are level. interaction was too small to data. Bolsa Familia y under 14 years old who with technical HH received analysis: CMA geographic compensate for the direct PRONAF. work in the farm. assistance one of the two units which Small number of negative impact of BF. programs. Sample size: did not beneficiaries of The authors recommend ii) CMA where 4251 CMA. changed over both integrating the two programs HH have not time. interventions’ in order to exploit positive received any limits sample synergies. program. size for evaluations. National Survey of Treatment Rural group: HH Households, which received 2002 and both programs Propensity Oportunidades: El programa de 2007. Score From 2002 to 2007, none of CCT combate a la Control Matching the two programs managed to pobreza y las Unit of Observational Naude, A.Y., groups: (generalized i)Net rural income, ii) Data: significantly reduce the PROCAMPO: Superposition políticas de analysis: Study Leal, G. D., i) HH which propensity vulnerability index, on Panel data Both programs incidence of poverty in the Oportunidades and Financial No coordination desarrollo rural en Communes Mexico Ramirez, F. 2016 only received score due to asset poverty, (attrition rate did not include a rural setting and that their cash PROCAMPO compensation (evaluation on México: evaluación with 500 to Used national R. & Aranda, Oportunidades, multiple iii) capability poverty and 13 percent). proper baseline transfers did not substantially for the North overlapping areas) de impactos y 2,499 surveys. O. S. ii) HH that treatment and food poverty. survey. promote the production of American Free opciones de inhabitants. only received overlapping of food in rural agricultural Trade políticas PROCAMPÔ, treatment). households. Agreement alternativas. Sample size: iii) HH which 3290 haven’t communes received any (merged 2002 program. and 2007 samples). JUNTOS: CCT Impacto de la Treatment Agricultural CCT increase the investments Observational Productive assets: Data: Crédito Rural: Superposition articulación entre group: HH that PSM used to census: 1994 in rent of land, the quantity of Study cultivated land (ha and Both programs JUNTOS and Credit for No coordination las transferencias a) never build pseudo and 2012. seed from own production, to Peru Del Pozo, C. 2014 %), irrigated cultivated Panel data. did not include a Crédito Rural extremely poor (evaluation on monetarias applied for panel and then raise minor livestock. Used Census land (ha and %), number proper baseline families in rural overlapping areas) condicionadas y el credit; b) difference-in- Unit of Additionally, increases the data. of cows, calves, survey. areas. crédito applied and did differences analysis: HH land use of seasonally crops, 41 Type of Design Program Productive Impact Evaluation/ Limitations/ Country Programs Author Year Data Results Concept Inclusion Evaluation Study Identification Estimation Challenges Program Outcomes strategy strategy agropecuario en la not get it; c) (controlling sheep/goats, guinea pigs natural pasture and selling of acumulación de applied and got for fixed and rabbits. Sample size: livestock production. activos productivo it. effects). 377,236 HH de hogares rurales Control: PSM (108,971 en el Perú. to match control – no treated and beneficiaries non-treated in non- districts. JUNTOS Robustness districts and check using 268,265 – instrumental beneficiaries variables. in JUNTOS districts) Treated group: JUNTOS beneficiaries Survey date: that end of 2013 participated in and beginning Sierra Sur. of 2014. Control group: i)Adoption of new Unit of JUNTOS: CCT JUNTOS productive practices, analysis: Sierra Sur: Sierra Sur y Juntos: beneficiaries in Sample was Data: social capital (number of Household. Positive impact on new productive Observational Un análisis a partir districts with informally Both programs Superposition Single memberships to technology adopted by non- technical Study de la Aldana, U., No Sierra Sur stratified with did not include a JUNTOS and Sierra No coordination difference organization) Sample size: treated in treated areas. Peru assistance for implementación Vásquez, T. 2016 intervention in high, median proper baseline Sur (evaluation on using PSM ii)Assets (value of total treated: 382; No impact was found on producer Used Survey conjunta de ambos & Yancari, J. 3 cat.: a) willing and low survey. overlapping areas assets, land, among control: a) participation in farmers’s associations data programas en to participate income. others, income (gross 402, b) 221, c) associations and Chumbivilcas. in Sierra Sur; b) External validity. income, net income, 125. Sample cooperatives median or high working capital). size after propensity to matching: pay for Sierra treated: 320; Sur-type of control: a) projects; c) 374; b) 209; c) high propensity 117. to pay for Sierra Sur projects. Inter-agency Experimental: Administrative Income (Total family JUNTOS: CCT articulation model The data Increase in income from on- income; income from Haku Wiñay: Metodología randomization farm and off-farm self- difference-in- agriculture), wage labour, Productive and Some level of Cuantitativa de la into treated Unit of employment and reducing differences food expenditure, ability entrepreneurial coordination, Evaluación de and control analysis: income from wage labour. JUNTOS and Haku with sample to increase income, resist skills, including particularly, for Impacto. Escobal, J. & 2016a groups took Household. Peru Wiñay RCT adjusted by to climate and non- - - adoption of component 4, Impactos del Ponce, C. 2016b place at the Increase in the number of propensity climatic shocks, assets, simple and financial literacy, Proyecto desde village level Sample size: varieties of green leaves, score production of new low-cost which is una Aproximación using pairwise 428 vegetables, fruits cultivated as matching vegetables, health and technological exclusively for Cuantitativa. randomization. households well as on the sale revenue nutrition, food security innovations. Juntos Pairs were (207 treated from green leaves and and financial literacy. beneficiaries. matched and 221 vegetables. 42 Type of Design Program Productive Impact Evaluation/ Limitations/ Country Programs Author Year Data Results Concept Inclusion Evaluation Study Identification Estimation Challenges Program Outcomes strategy strategy according to control) issued socioeconomic from 36 Positive impact on knowledge features. villages. and confidence in the financial Attrition rate system. of 4.3%. No effect on actual use of savings accounts or financial services. IEF only: Human capital: participation in training Panel survey courses, participation in 2012-2014: Treatment adult education courses, beneficiaries ID: As the group: school attendance 7-18 and non- sample contain No impact on labour force Beneficiaries of years, school attendance beneficiaries a small participation or per capita Inter-Agency Efectos both programs for 6-year old Fixed effect of the IEF. proportion of autonomous income, IEF: CCT articulation sinérgicos entre (IEF and Fernández, model Overlapping households of neither for the full sample model Observational en Ingreso FOSIS). FOSIS only: M. I., combined with between two IEF and an even nor for women, who are the FOSIS: Some Study Ético Familiar Income and occupation: Ingreso Etico Leiva, M. PSM. Logit for programs: Panel data smaller main beneficiaries of the Business coordination (IEF) y los Two control per capita autonomous Chile Familiar (IEF) F., Ortega, 2016 each program 2012- 3% and (balanced proportion with CCT program in Chile. support (credit (incentives to Used programas de groups: income (net of benefit), and FOSIS J. & effect and 2014 - 4.7% sample) the FOSIS, the and training) have priority administrative apoyo al i) beneficiaries total women's work force Weason, multinomial to statistical power There was no impact on the for extreme access) - low data. emprendimiento of one of the participation, M. measure the Unit of that underlies interaction term for poor families coverage en Chile, 2012- two programs participation in training synergies. analysis: the impact participation in both IEF though 2014 ii) other that courses, Human capital: Household. analysis of any and FOSIS productive haven’t participation in adult of the programs support program. received any education courses, school Sample size: is very limited. program. attendance at 6 years. 2,308 households For synergies: The outcomes of both groups Treatment Lifelong group: People Learning: that took Competency- vocational Administrative Based Training, training Evaluación de data Certification courses in Cross-section impacto en la Positive impact on the Lifelong Training and Observational labor insertion. difference inserción laboral Castillo, Percentage of program Unit of knowledge and skills of Learning and Promotion of Study between Data: de los V., Ohaco, participants who finished analysis: workers to access jobs PAD (Vocational Youth Fully Control treated and Cross- Programs did Argentina beneficiarios de M., & 2014 the program and Individual quality. Furthermore, the training courses Employment Coordinated Used groups: People control using sectional data not implement a los cursos Schleser, are employed in the impact is bigger for in labor administrative with common Propensity baseline survey sectoriales de D. formal market. Sample size: vulnerable groups. insertion) PAD de data. individual Score formación 41,000 jóvenes: characteristics Matching profesional beneficiary Provision of that do not people (2011). Orientation take vocational and Training training Services courses in labor insertion. 43 Type of Design Program Productive Impact Evaluation/ Limitations/ Country Programs Author Year Data Results Concept Inclusion Evaluation Study Identification Estimation Challenges Program Outcomes strategy strategy OR only: Production: amount invested, land cultivated (% over total land area), production increased over 2008 Occupation/employment: Stratification daily hours of work, daily using the labour, family workers. Administrative following Financial services: data variables: microinsurance (%), No evidence of impact entry year in Treatment access to formal financial Unit of from OR on food the program, Programas para group: HH that services (%), access to analysis: consumption, total FA: CCT activity, la reducción de participated in informal financial services Household. Data: Due to consumption and the food Inter-Agency region and an Observational la pobreza: Una both programs (%) lack of a clear security index. The OR: Financial articulation index for the Familias en Study evaluación de since 2008, Assets: value of Sample size: comparison interaction between OR support for (incentives to quality of the Acción (FA) and las sinergias 2009 and 2012. Single productive assets, value 729 group for OR. and FA actually led to a Colombia farmer's have priority Moya, A. 2016 organization. Oportunidades Used entre difference of total assets, asset index beneficiary Time of reduction in total organizations access) – low Rurales (OR) administrative Oportunidades Control Consumption: food households of exposure to the consumption for the earlier to purchase coverage A second data Rurales y groups: HH consumption, total OR from 59 program was cohort. No impact was extension though stage involved Familias en that entered consumption treated used. found for other indicators, services the random Acción. both programs Food security: food farmers’ neither for the earlier nor sampling of in 2013. security index. organizations for the later cohorts. 729 (FO) (between households Synergies between OR 2008 and from these and FA: 2013). farmers’ All the above plus: organization. Impact on children (12- 18 years): school attendance, education underachievement, hours of work, hours in school, hours of chores Several Simple cross- Production: new animals Administrative Receiving both programs Sampling Transferencias comparison section or new crops were data had positive impacts as frame: FISDL CSR: CCT monetarias groups are difference introduced? (%), opposed to just receiving baseline condicionadas y used to assess between productivity Unit of CSR. census in the DR: Extension desarrollo rural the impact of treated and (USD/manzana) analysis: 100 CSR services to en El Salvador. receiving both control using Income: agricultural Municipality Households with both types Superposition Observational De municipalities Comunidades improve In Protección, type of PSM and income (USD), yearly of support were between 10 No Study Sanfeliú, (between Data: Solidarias Rurales production, producción, programs, CCT Regression total income (USD). Out of 100 to 15 pp more likely to El coordination M. B., 2005-2009 Both programs (CSR) and homestead promoción: 2016 plus different Discontinuity Asset: asset index municipalities introduce a new crop or Salvador (evaluation on Used Angel, A. depending on did not include a Desarrollo Rural gardens, explorando modalities of Design (age of Food security: lack of in which the type of animal (livestock or overlapping Administrative & Shi, M. the year the baseline survey (DR) natural sinergias entre Rural the child/CSR dietary diversity (%), CCT program poultry) compared to areas data A. program resources protección Development eligibility). savings or asset depletion CSR was similar households that only started being management social y fomento (DR) Four to buy food (%) Financial implemented, received the EP. However, implemented and access to productivo rural interventions. comparisons inclusion: bank accounts 77 have some no statistically significant in the markets en América Comparison is of interest: a) (%), formal credit (%), rural impacts were found for municipality) Latina. always between A (CSR + DR) Social participation: % development indicators such as asset updated in just one versus C (DR) participate in producers’ project (18 index (durable and 44 Type of Design Program Productive Impact Evaluation/ Limitations/ Country Programs Author Year Data Results Concept Inclusion Evaluation Study Identification Estimation Challenges Program Outcomes strategy strategy program versus - extra effect association, % participate CSR plus 2014 (listing): productive), agricultural two programs. of CSR, b) AC in community support to 211 166 hh. income or productivity (ExCSR+DR) associations value chains (value of crop per area). Versus C (DR) Empowerment: general (EP), 28 CSR - isolate the empowerment, plus food and monetary productive empowerment nutritional effect of CSR (separately for men and security comparing to women) (SAN), and 31 1), c) A CSR plus EP (CSR+DR) plus SAN) versus AC and 23 CSR (ExCSR+DR) only. Most - isolate of beneficiaries monetary of SAN effect of CSR, projects (87 d) AC percent) were (ExCSR+DR) also in CSR versus BD municipalities, (ExCSR) - whereas only effect of DR 37 percent EP after receiving beneficiaries CSR. were from CSR municipalities. All Positive impacts of AC and communities AC combined with PC. Consumption and in the six However, the authors also income: per capita total municipalities found that the combination consumption, per capita of the pilot AC and investment grants total income, per capita programme had a much larger impact food consumption, per were grouped (12.6pp) as compared to the Randomized capita non-food in blocks combination of AC and AC: CCT (one- control trial consumption, non- Unit of based on training or to AC only, year pilot) Transfers, (experimental agricultural wage income, analysis: microclimates, which yielded an impact of diversification Similar evaluations) profit of non-agriculture Household crop mix, 4pp. The combination AC Atención a Crisis PC: Vocational and household Macours, Fully households business, expected similarity in and investment grants led (AC) and training and risk strategies - K., coordinated. selected from Intent-to-treat increase in profits in 12 Sample size: road access to an increase of almost Nicaragua Programas investment RCT Experimental Premand, 2012 - Complementary randomized- (ITT) months. 3002 and 11pp in services and 6.3pp Complementarios grants evidence with P. & programs. out household- Assets: value livestock households to infrastructure. in processing primary (PC) (different lessons for Vakis, R. communities. level sold or self-consumed, participate in Through a products, the combination groups climate change regressions value business assets. the lottery, 44 AC and training had received each adaptation. (difference-in- Occupation: non- programme. blocks were impacts on services (3.4pp), component) differences). agricultural wage selected and and AC only sample had employment, non- half of the impacts on services agricultural wage self- communities (3.25pp) and commerce employment (elaboration, in each block (3pp). manufacturing, trade and were No impact was found on services). randomly non-agriculture wage assigned to employment for any of the treatment and subjects of the evaluation. 45 Type of Design Program Productive Impact Evaluation/ Limitations/ Country Programs Author Year Data Results Concept Inclusion Evaluation Study Identification Estimation Challenges Program Outcomes strategy strategy the other half to control. Baseline data were then collected in the 56 treatment and 50 control communities. These data were used to define households’ eligibility for the programme based on a proxy means test. 46