Abstract

Indonesia's transition to the Data Tunggal Sosial Ekonomi Nasional (DTSEN) and its decile-based welfare ranking system represents a fundamental shift in how social assistance is targeted. Contrary to popular belief, a household's decile position is not determined by salary or income alone. This article examines the multidimensional indicators that actually determine whether a family ascends or descends the decile ladder. Drawing on academic literature, government documentation, and empirical studies, it analyzes the Proxy Means Test (PMT) methodology, the 39 variables used in DTSEN classification, and the theoretical foundations of multidimensional poverty measurement. The article also critically evaluates the targeting errors—both exclusion and inclusion—that arise from centralized algorithmic ranking. The findings suggest that while the decile system offers a more comprehensive approach than income-based targeting, its accuracy depends on local context, data verification mechanisms, and sensitivity to the structural dimensions of poverty. The article concludes with policy implications for improving targeting accuracy and ensuring that social assistance reaches the most vulnerable.

Introduction

Poverty in Indonesia has long been understood as a multidimensional phenomenon, yet public discourse often reduces it to a single metric: income. When the Indonesian government introduced the decile system through the Data Tunggal Sosial Ekonomi Nasional (DTSEN) in 2025, many citizens were surprised to discover that their welfare ranking did not correspond to their monthly salary. A becak driver might find himself in decile 9, while a civil servant could be classified in decile 5. This apparent anomaly is not a bureaucratic error but a reflection of how modern poverty targeting actually works. The decile system ranks households based on a composite of observable characteristics—housing conditions, asset ownership, education, health status, family composition, and access to basic services—rather than income alone.

This article argues that understanding the real indicators behind decile classification is essential for both policymakers and the public. It proceeds in four parts. First, it provides an overview of DTSEN and the decile system. Second, it examines the Proxy Means Test (PMT) methodology and the specific variables used in classification. Third, it analyzes the theoretical foundations of multidimensional poverty measurement, drawing on the Alkire-Foster method. Fourth, it critically evaluates the targeting errors and social exclusion that emerge from centralized algorithmic ranking. The article concludes with policy recommendations for improving the accuracy and equity of social assistance targeting.

The DTSEN and the Decile System: An Overview

The DTSEN represents the culmination of Indonesia's long struggle to create a unified social registry. It integrates data from multiple sources—the Data Terpadu Kesejahteraan Sosial (DTKS), the Pensasaran Percepatan Penghapusan Kemiskinan Ekstrem (P3KE), and the Registrasi Sosial Ekonomi (Regsosek)—into a single national database covering approximately 290 million individuals and 96 million families. The system was mandated by Presidential Instruction No. 8 of 2025, which emphasized three strategic pillars: reducing household expenditure burdens, increasing income, and eliminating pockets of poverty.

At the heart of DTSEN is the decile system, which divides the population into ten equal groups based on relative welfare. Decile 1 represents the 10 percent of families with the lowest welfare ranking—the extreme poor—while decile 10 represents the 10 percent with the highest welfare. Deciles 1 through 4 are collectively classified as the poorest 40 percent and are prioritized for most social assistance programs, including Program Keluarga Harapan (PKH), Bantuan Pangan Non-Tunai (BPNT), and the Penerima Bantuan Iuran Jaminan Kesehatan (PBI-JK).

Crucially, decile classification is dynamic. The government has repeatedly emphasized that deciles can change as household circumstances evolve, and that citizens may request data updates if their classification no longer reflects their actual condition. This dynamism is both a strength and a weakness: it allows for responsiveness to changing circumstances, but it also creates uncertainty for beneficiaries who may lose access to assistance without any apparent change in their economic situation.

The Proxy Means Test: Methodology and Variables

The technical foundation of decile classification is the Proxy Means Test (PMT), a statistical method widely used in developing countries to estimate household welfare when reliable income data are unavailable or unreliable. PMT works by identifying observable, easily verifiable characteristics that are correlated with poverty, then using these characteristics to predict a household's welfare score. The method is popular in Indonesia and has been used since the 2015 Integrated Database Updating.

The DTSEN uses approximately 39 variables in its PMT model. These variables span multiple domains. At the individual level, they include age, education, employment status, disability conditions, and chronic illnesses. At the household level, they encompass housing conditions, drinking water sources, cooking fuel and energy, asset ownership, electricity capacity and consumption, family composition, and access to sanitation. The specific indicators that carry the most predictive power include household assets, access to energy, education, sanitation, and housing quality.

The selection of these variables is not arbitrary. Research using Indonesian Family Life Survey data has shown that from an initial set of 340 candidate variables, the most predictive indicators of poverty are those related to asset ownership, energy access, education, and housing quality. This finding aligns with international best practices in poverty targeting, which emphasize multidimensional indicators over income alone.

However, the PMT methodology has inherent limitations. Because it relies on observable proxies rather than direct measurement of income or consumption, it can only estimate relative welfare rankings. As economists have pointed out, "DTSEN dengan metode Proxy Means Test pada dasarnya hanya estimasi peringkat sosial ekonomi"—it is an estimation of socio-economic ranking, not a precise measurement of wealth. This distinction is crucial for understanding why decile classifications sometimes appear inconsistent with lived experience.

Beyond Salary: The Real Indicators of Decile Classification

Housing Conditions and Basic Amenities

Housing quality is one of the strongest predictors of household welfare in the PMT model. The DTSEN collects detailed information on dwelling characteristics, including floor area per capita, wall and roof materials, and the availability of private sanitation facilities. Households with larger living spaces, permanent walls, and adequate sanitation receive higher welfare scores. Conversely, crowded living conditions with inadequate sanitation contribute to lower decile rankings.

Access to clean water and adequate sanitation is particularly important. Research using the Multidimensional Poverty Index (MPI) in Bandung found that sanitation deprivation affected more than 50 percent of poor households, making it the most prevalent form of deprivation. This finding underscores why sanitation indicators carry significant weight in the PMT model.

Asset Ownership

Asset ownership is another critical determinant of decile classification. The DTSEN tracks ownership of various household assets, including motorcycles, cars, refrigerators, televisions, smartphones, and computers. The presence of these assets can substantially raise a household's welfare score, even if the household's income is low.

This creates a paradox that has generated considerable public confusion. A family may have purchased a motorcycle on credit or received a smartphone as a gift, yet the PMT model interprets asset ownership as evidence of higher welfare. The system does not distinguish between assets acquired through wealth and those acquired through debt or informal transfers. This limitation is well-documented in the PMT literature: proxy means tests can be "less predictive than it is in practice" precisely because they cannot capture the nuances of asset acquisition and ownership.

Education

Educational attainment is consistently identified as one of the most predictive indicators of household welfare. The DTSEN records the education level of each household member, with particular attention to whether any member has completed nine years of basic education. Households where the head has not completed primary education are more likely to be classified in lower deciles.

The MPI analysis in Bandung found that schooling duration was a significant source of deprivation, affecting 12 to 31 percent of poor households. This finding is consistent with the broader literature on human capital and poverty: education is both a cause and a consequence of economic well-being, and its effects ripple across generations.

Health Status and Disability

Health indicators also play a role in decile classification. The DTSEN collects information on disability conditions, chronic illnesses, and access to health facilities. Households with members who have disabilities or chronic health conditions may receive adjusted welfare scores that account for the additional costs and reduced earning capacity associated with these conditions.

The presence of elderly family members, young children, or persons with disabilities increases the dependency ratio within a household, which in turn affects its welfare ranking. The DTSEN explicitly considers "keberadaan anggota keluarga rentan (lansia, anak, disabilitas)" as a factor in decile determination.

Family Composition and Dependency Ratios

Household size and composition are important determinants of welfare. A household with many dependents relative to income earners will have a lower per capita welfare score, all else being equal. The DTSEN accounts for the number of dependents in a family, including children, the elderly, and persons with disabilities.

This dimension of the PMT model reflects the reality that welfare is not simply a function of total household income but also of how that income is distributed among household members. A large family with a moderate income may be worse off per capita than a small family with a lower total income.

Employment Sector and Informality

While income itself is not directly recorded in the PMT model, employment characteristics are used as proxies. The PMT model includes variables for the employment sector of the household head, with agricultural employment often associated with lower welfare scores. Informal sector employment—which characterizes a large portion of Indonesia's labor force—is also correlated with lower welfare rankings.

However, this creates challenges for accurate targeting. Seasonal farm laborers and small-scale fishermen, for example, may have irregular incomes that are not captured by employment sector indicators alone. Research in Takalar Regency found that centralized decile clustering "triggers a high exclusion error rate, where central algorithms fail to capture local socio-economic dynamics, such as seasonal farm laborers and small-scale fishermen".

Methodological Challenges and Targeting Errors

Exclusion and Inclusion Errors

The decile system is not immune to targeting errors. Exclusion errors occur when households that should receive assistance are not classified as eligible, while inclusion errors occur when ineligible households receive benefits. Both types of errors have been documented in Indonesia's social assistance programs.

BPS has acknowledged that inclusion-exclusion errors in DTSEN remain at approximately 23 percent, improving from around 37 percent during the system's early development. In practical terms, this means that nearly one in four households may be misclassified. The consequences are significant: households that are incorrectly placed in higher deciles may lose access to social assistance, educational subsidies, and health insurance, even if their actual economic condition has not changed.

The problem is particularly acute for vulnerable groups with irregular or informal employment. A study in Malang City found that "social exclusion in the utilization of DTSEN emerges in several forms, including administrative exclusion resulting from incomplete civil registration documents, system-based exclusion caused by rigid data verification mechanisms, and socio-structural exclusion experienced by the extremely poor with high mobility or unstable places of residence".

The Limitations of Centralized Algorithms

The centralized nature of decile determination—carried out by BPS rather than local officials—has both advantages and disadvantages. On one hand, it reduces the potential for local political manipulation and ensures consistency across regions. On the other hand, it may fail to capture local socio-economic dynamics that are not reflected in national statistical models.

A study in Takalar Regency found that "centralized Decile clustering triggers a high exclusion error rate, where central algorithms fail to capture local socio-economic dynamics, such as seasonal farm laborers and small-scale fishermen". This finding highlights the tension between national standardization and local sensitivity in poverty targeting.

Social Exclusion and Administrative Barriers

Beyond technical errors, the DTSEN system can produce social exclusion through administrative barriers. The extremely poor often lack the documentation—identity cards, family registration cards, birth certificates—required to be included in the database. They may also lack the digital literacy to navigate the online systems for data verification and updating.

These administrative barriers disproportionately affect the most vulnerable populations: the homeless, migrant workers, informal sector workers, and those living in remote areas. The irony is that those who need assistance most are often the least able to navigate the bureaucratic requirements for receiving it.

The Multidimensional Nature of Poverty: Theoretical Foundations

The decile system's reliance on multiple indicators reflects a broader theoretical consensus in development economics: poverty is multidimensional, and income alone is an inadequate measure of well-being. This insight is rooted in Amartya Sen's capability approach, which defines poverty as the deprivation of basic capabilities rather than merely low income.

The Alkire-Foster method, developed by Sabina Alkire and James Foster, operationalizes this multidimensional approach by measuring overlapping deprivations across health, education, and living standards. The Multidimensional Poverty Index (MPI) derived from this method has been applied extensively in Indonesia. A study of Bandung City found that the MPI headcount ratio ranged from 12 to 24 percent, significantly higher than the monetary poverty rate of 3.96 to 4.37 percent. This discrepancy illustrates how monetary measures alone can underestimate the true extent of poverty.

Research on multidimensional deprivation in Indonesia has found that "despite falling monetary poverty rates, Indonesia is facing persistent multidimensional deprivation, especially among the extreme poor". The Alkire-Foster method reveals deprivations in health, education, living standards, basic needs, and social participation that are not captured by income-based measures.

The DTSEN decile system, while not a formal MPI, shares the multidimensional philosophy. By incorporating indicators across housing, assets, education, health, and family composition, it attempts to capture the complexity of poverty in a single ranking. However, the PMT methodology differs from the Alkire-Foster approach in important ways: PMT uses statistical correlation to predict welfare, while Alkire-Foster uses normative thresholds to identify deprivations. This distinction matters for understanding what the decile system can and cannot measure.

Implications for Policy and Social Assistance

The decile system has significant implications for social assistance policy in Indonesia. By providing a unified ranking of household welfare, it enables more targeted allocation of resources. Programs like PKH, BPNT, and PBI-JK use decile rankings to prioritize beneficiaries, and the government has proposed using deciles 9 and 10 to restrict access to subsidized fuel.

However, the system's accuracy challenges must be addressed. Several policy recommendations emerge from the literature:

First, local verification mechanisms should be strengthened. The involvement of community members and local officials in validating decile classifications can help identify households that have been misclassified due to data limitations or changing circumstances.

Second, transparency in the PMT model should be improved. BPS has not published the weights assigned to different variables, which makes it difficult for citizens to understand why they are classified in a particular decile. Greater transparency would enhance public trust and allow for more informed appeals.

Third, the system should be complemented by community-based targeting approaches. Research has shown that combining PMT with community validation can reduce both exclusion and inclusion errors.

Fourth, the data updating process should be made more accessible. Currently, citizens who believe their decile classification is incorrect face significant administrative hurdles to request updates. Simplifying this process would improve the system's responsiveness and fairness.

Fifth, the government should invest in digital literacy and administrative support for vulnerable populations. The extremely poor are often the least able to navigate the bureaucratic requirements for data inclusion and updating.

Conclusion

The Indonesian decile system represents a significant advance in social assistance targeting. By moving beyond income-based measures to a multidimensional approach, it acknowledges that poverty is about much more than salary. Housing conditions, asset ownership, education, health status, family composition, and access to basic services all contribute to a household's welfare ranking. The Proxy Means Test methodology, while imperfect, provides a systematic way to estimate welfare using observable characteristics.

However, the system's accuracy challenges are real and consequential. Exclusion errors of approximately 23 percent mean that many vulnerable households are not receiving the assistance they need. Centralized algorithms may fail to capture local socio-economic dynamics. Administrative barriers exclude the most vulnerable from data inclusion. These challenges require ongoing attention and reform.

Ultimately, the decile system is a tool, not a solution. It can help target resources more efficiently, but it cannot replace the nuanced understanding of poverty that comes from local knowledge and community engagement. The real indicators that determine a household's decile are not just statistical variables—they are the lived realities of housing, health, education, and social inclusion. A just and effective social assistance system must address all of these dimensions, not just the numbers on a spreadsheet.

References

  • Aji, P., Syarifah, Z. A., & Wibisono, I. D. (2026). Unequal gains beyond the poverty line: Multidimensional deprivation in Indonesia's quest to eradicate extreme poverty. Journal of Southeast Asian Economies, 43, 25–48.
  • Alkire, S., & Foster, J. (2011). Counting and multidimensional poverty measurement. Journal of Public Economics, 95(7–8), 476–487.
  • BPS. (2023). Analisis kemiskinan multidimensi Indonesia. Badan Pusat Statistik.
  • Garuda. (2023). Is the wealth index better than the proxy means test in poverty targeting? A study in Brebes and East Jakarta. Proceedings of STIS.
  • Rompegading, A. M., Rahman, A., & Mallarangang, M. A. (2026). Evaluasi dampak kebijakan berbasis Data Tunggal Sosial Ekonomi Nasional (DTSEN) terhadap akurasi sasaran penerima bantuan iuran jaminan kesehatan (PBI JK) bagi masyarakat miskin di Kabupaten Takalar. Iqtishaduna, 7(4).
  • Sius, K. T., Peten, Y. Y. P., Raju, K. V., Kaha, H. L., & Nyong, F. (2024). Public trust: Dampak exclusion dan inclusion error data Program Keluarga Harapan. Jurnal Ilmu Administrasi Negara (AsIAN), 12(1).
  • Sukmasari, I. R. R. (2026). Eksklusi sosial dalam pemanfaatan Data Tunggal Sosial Ekonomi Nasional sebagai basis penanganan kemiskinan ekstrem di Kota Malang [Master's thesis, Universitas Muhammadiyah Malang].
  • UNICEF. (2023). Towards child-sensitive social protection in Indonesia: Challenges and opportunities. UNICEF Indonesia.
  • Yusuf, A., & Sumner, A. (2017). Multidimensional poverty in Indonesia: How inclusive has economic growth been? Journal of Southeast Asian Economies, 34(2), 261–282.