This document provides a comprehensive review of the design and impact of social protection programs in low- and middle-income countries. It outlines the key challenges in implementing these programs, particularly concerning targeting and the large informal sector. This paper is valuable for researchers, policymakers, and practitioners working in international development and social welfare. These insights are especially relevant for understanding the unique context of developing countries and adapting social protection strategies accordingly.
Key Insights
Targeting Challenges in Developing Countries
Unlike high-income countries where income is easily observable, developing countries face the challenge of informal work and self-employment that exclude many from the income tax system (Jensen, 2022). The standard public economics frameworks do not readily apply in this context, necessitating new targeting approaches.
Proxy-Means Tests (PMT)
Proxy-means tests, which use assets and other observable characteristics to predict people’s incomes, form the basis for targeting in many developing countries. These tests allow governments to target the poor without income data, although mistargeting due to imperfect statistical models remains a challenge.
Heterogeneity and Audit-Based Estimates
The optimal redistributive scheme in a baseline model relies entirely on self-reports. However, when there is heterogeneity in the cost of misrepresenting incomes, it can be optimal for the government to use noisy audit-based estimates as part of the targeting process. The weight placed on audit-based estimates increases with the heterogeneity in misreporting costs (Saez, 2002).
Self-Selection Mechanisms
Self-selection mechanisms involve pairing benefits with actions that are less costly for the targeted population. Workfare programs, which require beneficiaries to work in exchange for payments, are a common example. However, these mechanisms can impose large costs on beneficiaries.
Community-Information Based Targeting
An alternative approach leverages local information about people’s poverty status by allowing communities to participate in choosing beneficiaries (Alderman, 2002; Galasso and Ravallion, 2005). Studies suggest that the optimal approach may depend on the government’s objective function and whether the community is better at identifying poverty according to local notions of neediness.
Key Statistics & Data
- Safety net programs cover about 2.5 billion people globally.
- More than 120 low- and middle-income countries run cash transfer programs.
- More than 70 low- and middle-income countries run social pension programs (World Bank, 2018).
- 63 percent of the population in low-income countries is engaged in agriculture (ILOSTAT, 2019).
- The R² (predictive power) of proxy-means tests is between 0.53 and 0.66 (Hanna and Olken, 2018).
- Those who joined MGNREGA gave up income equivalent to about 30-35 percent of the workfare income received (Murgai, Ravallion and van de Walle 2016).
- Self-targeting led to dramatically poorer beneficiaries, driven in large part by those who are wealthier on dimensions that would be missed by the PMT being less likely to apply (Alatas et al. 2016b).
- There are roughly 200 studies related to UCTs and CCTs; Bastagli et al. (2016) provide a systematic review of 201 cash transfer studies (25 percent of which are UCTs).
Methodology
The paper uses a combination of literature review and theoretical frameworks to analyze the design and impact of social protection programs. It draws on existing research and develops a simple model to examine the trade-offs in targeting transfers in the presence of noisy data. The model accounts for the features of developing countries such as incomes that are hard to measure precisely since much work is in the informal sector.
Implications and Conclusions
The findings suggest that the design of social protection programs in developing countries requires careful consideration of the unique challenges, including the large informal sector, the difficulty of observing income, and the need to balance efficiency and equity. Given these constraints, the choice of targeting methods, the design of transfer programs, and the implementation strategies must be tailored to the specific context and objectives. Future research should focus on the optimal design of dynamic targeting systems, the use of new administrative and satellite data, and the reduction of the take-up problem.
Key Points
- Social protection programs have become increasingly widespread in low- and middle-income countries.
- Designing these programs for a developing country context entails challenges different from developed economies.
- Governments in the developing world face the challenge of observing earnings for informal workers.
- Proxy-means tests allow governments to target the poor without income data, but mistargeting is a challenge.
- The optimal redistributive scheme always has the maximum feasible slope with respect to earnings given the budget constraint and has no universal component.
- If there is heterogeneity in the cost of misrepresenting incomes, it can be optimal for the government to use noisy audit-based estimates as part of the targeting process.
- One successful development intervention is to give households a productive livestock asset and support them in taking care of it.