World Bank (2022) A New Look at Old Dilemmas: Revisiting Targeting in Social Assistance

Abstract 

Targeting is a commonly used, but much debated, policy within global social assistance practice. This book examines the well- known dilemmas in light of the growing body of experience, new implementation capacities, and the potential to bring new data and data science to bear. Chapter 1 presents a series of essays on the factors that shape choices around why or whether or how narrowly/broadly to target different parts of social assistance. Chapter 2 updates the global empirics around the outcomes and costs of focusing benefits on the poor or vulnerable. Chapter 3 illustrates the options and choices that must be made in moving from an abstract vision of focusing resources on the poor or vulnerable to more specific concepts and implementable definitions and procedures, and how the many choices should be informed by values, empirics and context. Chapter 4 provides a brief treatment of delivery systems and processes showing their importance to distributional outcomes and suggesting the many facets with room for improvement. Chapter 5 discusses the choice between targeting methods, how differences in purposes and contexts shape those. Chapter 6 summarizes and comprehensively updates the know-how with respect to the data and inference used by the different household-specific targeting methods. Chapter 7 contains a primer on measurement issues, going much deeper than usual and explaining how better measurement can lead to clearer understanding of targeting issues. Chapter 8 explores machine learning algorithms for household-specific mechanisms for eligibility determination.

See more at A New Look at Old Dilemmas: Revisiting Targeting in Social Assistance (worldbank.org).

Key findings:

Targeted social protection interventions can play a valuable role in helping achieve and deliver Universal Social Protection. Targeted programs and universal programs together support broader social policy.

Targeting is an effective tool used in social protection to make the most of constrained fiscal space. For a given budget, prioritizing poorer households can produce more progress on reducing poverty and inequality, smoothing income, and other dimensions of welfare such as human capital.

There is no single targeting method that fits every situation. Context and policy objectives drive choices. Whether to use methods such as self-targeting, geographic targeting, demographic targeting, or household welfare-based targeting methods must be based on context and capacities.

Regardless of the targeting method, robust social protection delivery systems can help:

• reduce transactions costs or stigma for beneficiaries,
• minimize inclusion errors,
• facilitate crisis response,
• improve access to social assistance, especially for the poorest and most vulnerable populations such as indigenous, migrants, people living with disability and others.

Advances in technology—ICT, big data, artificial intelligence, and machine learning—offer the promise of significant improvements in targeting accuracy but are not a panacea. Better data may matter more than greater sophistication in data use.

Social protection targeting methods are changing as new data and technology as well as other innovations emerge.

Citation

“Grosh, Margaret; Leite, Phillippe; Wai-Poi, Matthew; Tesliuc, Emil. 2022. Revisiting Targeting in Social Assistance : A New Look at Old Dilemmas. Human Development Perspectives;. Washington, DC: World Bank. © World Bank. https://openknowledge.worldbank.org/handle/10986/37228 License: CC BY 3.0 IGO.”