Professor Dean Karlan of Northwestern University, the Global Poverty Research Lab, and Innovations for Poverty Action (IPA) delivered a presentation entitled “The Graduation Model in Social Protection: What 41 Studies Tell Us” during the Ethiopia Evidence Use Workshop held in Addis Ababa in 2026. Drawing on evidence from 41 studies, he examined the effectiveness of the Graduation Model as a social protection strategy, summarized the findings of a global meta-analysis of graduation and economic inclusion programmes, and discussed key policy implications and future research priorities.
Professor Karlan began by explaining the concept of the Graduation Model, describing it as a time-bound and sequenced package of interventions designed to support the poorest and most vulnerable households over a period of 12 to 36 months. He noted that, rather than relying on a single intervention, the model combines several complementary components, including food assistance, health services, livelihood support, social development interventions, and financial services. According to him, these integrated interventions are intended not only to meet immediate needs but also to enable households to build sustainable livelihoods and eventually "graduate" from extreme poverty. He acknowledged the Consultative Group to Assist the Poor (CGAP) for developing and promoting the graduation approach.
Turning to the evidence base, Professor Karlan presented findings from a meta-analysis of Graduation and Economic Inclusion Programmes, conducted in collaboration with Samantha Carter, Ed Jee, Patrick Premand, and Julius Ruschenpohler. He explained that the analysis synthesized evidence from 41 impact evaluations, providing one of the most comprehensive assessments of the Graduation Model to date.
Professor Karlan then described what he referred to as two major policy conversations surrounding the Graduation Model. The first concerns the humanitarian-development nexus, particularly the application of graduation programmes among refugees and other displaced populations. The second relates to social protection, where policymakers increasingly view graduation programmes as part of a continuum ranging from cash transfers, to cash-plus programmes, and finally to comprehensive graduation interventions. He explained that graduation programmes differ from conventional cash transfers because they seek to address multiple market failures that extend beyond simple liquidity constraints.
Discussing the theoretical foundations of the model, Professor Karlan argued that while cash transfers help households overcome immediate financial shortages, graduation programmes are designed to address a broader set of constraints that prevent poor households from escaping poverty. These include limited access to productive assets, financial services, knowledge, skills, and social support. However, he acknowledged that graduation programmes generally require greater investment per beneficiary and may be more challenging to customize effectively for different contexts.
Presenting the empirical findings, Professor Karlan reported that the meta-analysis demonstrated consistently positive impacts of graduation programmes across several key development outcomes. On average, participating households experienced an increase in monthly consumption of approximately US$36, representing a 12 percent improvement over comparison groups. Similarly, household income increased by an average of US$27 per month, equivalent to an 18 percent increase. He further reported that programme participants accumulated an average of US$193 in additional productive assets, indicating that graduation interventions successfully promoted asset-building alongside income growth.
Professor Karlan also highlighted important improvements in psychological well-being. He explained that the meta-analysis found a positive impact on mental health, with an average improvement of 0.12 standard deviations among programme participants. According to him, these findings demonstrate that graduation programmes not only improve material living conditions but also contribute to better psychosocial outcomes, which are increasingly recognized as important dimensions of poverty reduction.
Addressing the sustainability of programme impacts, Professor Karlan observed that the available evidence provides suggestive but not yet conclusive support for long-term effects. While improvements in household consumption appear to persist beyond programme completion, evidence regarding sustained income growth remains less consistent. He explained that relatively few long-term follow-up studies currently exist, making it difficult to draw definitive conclusions regarding the durability of programme impacts over extended periods.
Professor Karlan also discussed programme cost-effectiveness. He noted that although graduation programmes require greater upfront investment than simpler cash transfer programmes, they appear capable of generating both short-term welfare improvements and longer-term productive investments. According to him, the evidence suggests that graduation interventions can simultaneously address immediate consumption needs while building productive capacity for future income generation.
Summarizing the overall evidence, Professor Karlan emphasized that the central lesson emerging from the meta-analysis is that "less is not less." He explained that well-designed graduation programmes do not necessarily require an excessive number of components; rather, carefully selected and well-coordinated interventions can successfully address the multiple constraints facing extremely poor households. Nevertheless, he cautioned that there is no single globally optimal programme design, as effective graduation models must be adapted to local institutional, economic, and social conditions.
Professor Karlan then identified several important research gaps requiring further investigation. First, he noted the need for greater understanding of the macro-level and local economy-wide effects of graduation programmes, particularly whether they influence local labor markets, prices, and broader economic development. Second, he emphasized the importance of examining how rapid advances in artificial intelligence, digital technologies, and changing labor markets should influence livelihood selection and programme design. Third, he argued that additional research is needed to determine the optimal level of productive capital provided to beneficiaries and to assess whether loan-based approaches could complement or partially replace grant-based support. He also suggested further investigation into individual programme components, including whether additional mental health interventions or other complementary services could enhance programme effectiveness.
Turning to policy implications, Professor Karlan emphasized that governments should pursue the expansion of graduation programmes while simultaneously embedding evidence laboratories within implementation systems. According to him, integrating rigorous monitoring and impact evaluation into programme implementation would allow governments to continuously adapt programme design based on emerging evidence rather than relying solely on standardized models. He also highlighted the continued role of non-governmental organizations (NGOs) in developing innovative approaches, piloting new programme models, and providing technical assistance to governments during scale-up.
In concluding his presentation, Professor Karlan reiterated that the accumulated evidence from 41 studies demonstrates that the Graduation Model is an effective social protection strategy capable of improving household consumption, income, productive assets, and mental well-being. While important questions remain regarding long-term impacts and optimal programme design, he argued that graduation programmes offer considerable promise for helping extremely poor households achieve sustainable economic inclusion. He concluded by emphasizing the importance of continued evidence generation, adaptive programme implementation, and close collaboration between researchers, governments, and development partners to maximize the effectiveness of social protection policies.