By John McCarthy
In September 2025, Indonesian media reported that 45 per cent of social assistance recipients may have been mistargeted.¹ Earlier assessments suggested that the flagship social assistance programme (PKH) reduced stunting, increased school enrolment, and improved maternal health outcomes.² Now, reports circulated that more than 1.9 million households had received benefits for which they were not eligible. In response, the Government pledged to digitise the system and strengthen oversight. Officials began verifying 12 million recipients and removing those found ineligible from the database. The mass removal of 1.9 million PKH recipients triggered waves of protests and complaints, especially from those who felt they were victims of system errors or unfounded accusations.³
Conditional Cash Transfer programmes, such as this one, operate in 64 low- and middle-income countries delivering targeted payments to low-income families. CCTs provide cash to low-income families who meet conditions like school attendance or health checkups. This approach promises efficient targeting while investing in nutrition, health, and education. Yet critics debate the effectiveness of these programmes.⁴
In recent years, Indonesia and the Philippines introduced CCTs. Since 2005, Indonesia has expanded its Program Keluarga Harapan (PKH), or “Family Hope Programme,” to cover an estimated 40 million people — 10 million beneficiary families — making it the second-largest CCT programme in the world. Yet, Mexico, which pioneered the first CCT programme in the world, eventually abolished it once a progressive government won power. After two decades of implementation, Mexico found that poverty did not improve⁵. The Government’s desire to reduce administrative burdens and the impact of hard conditionalities on recipients prompted a critical re-evaluation of the approach.
Evaluations have demonstrated the positive impacts of CCTs on school enrolment and child nutrition. Assessments in Indonesia praise the impacts of PKH. Yet, the reality on the ground is more complicated. Local media outlets report widespread mistargeting of benefits, the inclusion of non-poor households, the exclusion of impoverished families, community tensions, jealousy, and protests over unfair distribution.
A study undertaken by the author and colleagues aimed to explore the consequences of CCT practices of knowing and measuring poverty in rural Indonesia. The study combined interviews with officials, document analysis, and fieldwork in rural communities. Methods included community poverty ranking exercises, household surveys, and comparisons with econometric targeting to contrast state and community assessments of poverty.
The ‘politics of knowledge’ describes how particular ideas about poverty achieve dominance and thereby shape policy. Technical models determine which approaches policymakers consider reasonable, sidelining alternative understandings of poverty.
CCT programmes rely on a politics of knowledge embedded in complex data systems and practices to identify and target poor households. Proxy means tests use observable household characteristics as proxies for poverty. Social registries collect extensive data on millions of families. Econometric algorithms rank and score households to determine eligibility. This technocratic approach aims to identify people experiencing poverty and allocate benefits efficiently and objectively.
The field research reveals the reality behind “objective” poverty data. Enumerators take shortcuts under time pressure. Village leaders struggle with pre-selected beneficiary lists that exclude deserving families, eroding community trust in local Government. Village heads evade conflict by avoiding responsibility for beneficiary lists and by side-stepping the exclusion of people from them. Savvy respondents hide assets to appear poorer.
The system involves a significant simplification of complex social realities. Simple asset indicators miss nuances of household welfare. A convoluted, centralised social welfare system is complex, error-prone, and susceptible to data management problems. Fragmented coordination⁶ between government agencies leads to late disbursement of funds. As a result, CCT targeting often misses many impoverished households while including non-poor beneficiaries.
In one community, the CCT programme reached only a fraction of households (44 per cent and 23 per cent in different case studies), which a separate village ranking exercise considered poor. Many recipients fell just above the official poverty line, while most food-insecure families received no benefits. The village wealth ranking considered many receiving benefits (23 per cent and 44 per cent) as no longer poor. This confirms other studies⁷ suggesting exclusion rates of up to 60 per cent for programmes targeting the poorest 10 per cent, and that up to 52 per cent of poor households⁸ miss out on benefits.
As this system encounters local realities, it involves translation processes in which programme implementers and field officials reinterpret programme designs to fit the local context. Patterns of inclusion and exclusion emerge that often conflict with community perceptions of who is most in need or who is most deserving of help. Ongoing “repair” processes, adjustments and workarounds occur as the programme encounters local realities and faces challenges, as officials try to fix targeting problems. Village leaders engage in informal redistribution to ease social tensions caused by exclusion, particularly those created by the gap between technocratic targeting and social expectations around deservedness, sharing, and mutual assistance. The knowledge system provokes strategic behaviours as households try to gain benefits.
Typically, accounts of these problems suggest that implementation issues cause mistargeting. Therefore, the Government endeavours to improve targeting accuracy⁹ through improving digital platforms and management information systems. However, these problems largely stem from the knowledge systems and measurement approaches used—the politics of knowledge behind CCTs.
Poverty programmes pose social questions and involve making normative and political decisions about who to help and how to address social inequality. As CCT systems focus on technical or managerial solutions, they displace other conceptualisations of how to define and address poverty. Their highly sophisticated models mask the underlying political decisions about distribution behind a ‘firewall of technicalities.’ As only technical experts can understand how decisions are made, this cloaks discussions of poverty programmes.
The CCT model offers a solution to the symptoms of poverty (giving money to people experiencing poverty) without fully addressing the drivers of poverty in the rural context¹⁰ (why they are poor). While other interventions are required, the CCT knowledge system doesn’t just measure poverty; it reshapes how poverty is understood and addressed. It also shapes field-level outcomes.
CCT practices of knowing and measuring poverty have paradoxical effects. While PKH makes direct payments to millions of impoverished households, improving statistical indicators for health service use, child nutrition, school attendance, and household food security¹¹, it generates significant mistargeting and endless repair processes alongside jealousy and social problems. Hence, while CCTs create new opportunities for poor households to access state resources, they also provoke responses and contestations, a contentious local politics of distribution¹².
Advancing to a more equitable, accountable, and inclusive social assistance system will require transforming the governing knowledge systems. This could involve adopting alternative approaches to poverty reduction that are more responsive to local contexts and community logics of inclusion. Effective social assistance requires moving past narrow technocratic approaches to poverty measurement. Policymakers can incorporate community understandings of need and adopt more inclusive targeting methods. Policy makers can recognise the limitations of proxy indicators and rapid survey approaches. Studies suggest¹³ it is too difficult to identify accurately who is poor using econometric targeting, even when programmes add a community-level screening process. Then policy will have to shift to other, more inclusive approaches – such as a more universal scheme or simple lifecycle transfers – rather than narrow targeting of the extremely poor. Assistance needs to align with local notions of fairness and social reciprocity.
Anti-poverty programmes fall short when they overlook social, economic and political realities that shape poverty in favour of technical solutions. Indonesia’s experience shows that effective poverty reduction requires broadening our ways of knowing and measuring poverty while tackling the structural foundations of poverty.
John McCarthy is a Professor of rural development and the anthropology of policy at the Crawford School of Public Policy, Australian National University.
¹ 45 Persen Bansos Tidak Tepat Sasaran, Setara Rp 14-17 Triliun
² https://www.povertyactionlab.org/evaluation/medium-term-impact-conditional-cash-transfers-health-and-education-indonesia
³ https://www.kabarbantuan.com/2025/06/mengapa-bantuan-sosial-anda-dihapus-ini-Penjelasan-Lengkapnya.html
⁴ https://pubmed.ncbi.nlm.nih.gov/32538436/
⁵ Poverty rates remained practically unchanged.
⁶ https://www.kemenkopmk.go.id/transformasi-program-bantuan-sosial-di-indonesia-memastikan-mendukung-tujuan-pembangunan-dan-secara
⁷ https://www.sciencedirect.com/science/article/abs/pii/S0304387818305819
⁸ https://www.developmentpathways.co.uk/wp-content/uploads/2017/03/Exclusion-by-design-An-assessment-of-the-effectiveness-of-the-proxy-means-test-poverty-targeting-mechanism-1-1.pdf
⁹ https://kms.kemenkopm.go.id/
¹⁰ https://press.uchicago.edu/ucp/books/book/distributed/P/bo185856608.html
¹¹ https://www.povertyactionlab.org/evaluation/medium-term-impact-conditional-cash-transfers-health-and-education-indonesia
¹² The Paradox of Agrarian Change: Food Security and the Politics of Soci – NUS Press
¹³ https://documents1.worldbank.org/curated/en/484991481639919564/pdf/WPS7915.pdf