Effectively Targeting Anti-Poverty Programs in IndonesiaPDF version
Targeted cash transfer programs have become an increasingly common tool for poverty reduction in the developing world, but identifying the poor can be challenging because governments often lack reliable information about incomes. One strategy to effectively target the poor is to conduct a census that focuses on an inventory of assets and demographics, called proxy-means testing (PMT). While this strategy identifies objective, visible measures of consumption, communities may have better information about recent economic shocks. For example, a family might have fallen into poverty because of illness, but still own a large house that classifies it as non-poor under the PMT.
While community targeting utilizes local information, it also allows for the possibility that targeting decisions would be based on factors beyond poverty as defined by the government. The community may not care as much about certain groups among the poor or they may prefer to use the money to achieve broader social goals. Elites may also capture the process and select only their friends and relatives.
Indonesia is home to one of the largest targeted cash transfer programs in the developing world, the Bantuan Langsung Tunai program, or BLT. The BLT program provides transfers of about US $10 per month to about 20 million households below and near the poverty line. Targeting in the BLT program is accomplished with a hybrid strategy of community-based methods and PMTs, where village leaders determine a list of households who could qualify for the program and enumerators collect asset data only for the households that were suggested by the community leaders.
According to the World Bank, and using the PPP$2 per day poverty threshold, this targeting method led to a misallocation of funds to households not living below the poverty line.
Approximately 45% of the funds were mis-targeted to non-poor households, and 47% of the poor were excluded from the program in 2005-2006. Citizens also voiced substantial dissatisfaction with the lists, causing some village leaders to resign from their posts.
Researchers, in collaboration with the Central Statics Bureau of Indonesia, designed a field experiment in 640 randomly selected Indonesian villages to investigate the PMTs and community-based methods of targeting. A special, one-time cash transfer program was implemented by the Indonesian government in these villages that distributed 30,000 Rupiah (about US$3) to households falling below location-specific poverty lines. These lines were established using PMT, community-based targeting, or a hybrid method. Each method produced a rank ordering of all households in the community, and each of the three treatments were randomly assigned to a third of the villages. After the cash disbursement was complete, researchers collected data on the community’s satisfaction level using suggestion boxes, sub-village head interviews, facilitator feedback and household interviews.
PMT Treatment: 49 indicators encompassing the household’s home attributes (wall type, roof type, etc), assets (TV, motorbike, etc), household composition and household head’s education and occupation were used to determine a PMT Score. A list of beneficiaries was generated by selecting the pre-determined number of households with the lowest PMT scores in each sub-village.
Community Treatment: The sub-village residents determined the list of beneficiaries through a poverty-ranking exercise led by trained community facilitators. At a community meeting, residents ranked all households in the neighborhood from richest to poorest, with the poorest households becoming the beneficiaries of the program.
Hybrid Method: This process combined the community ranking procedure with a subsequent PMT verification. After the community meetings were complete, government enumerators visited the lowest-ranked households to collect the data needed to calculate the PMT score and determined eligibility.
The PMT method outperformed both the community and hybrid treatment in terms of the overall mis-target rates. Both the community and hybrid methods increase the mis-targeting rate by about 10% relative to the PMT method. Much of the difference in the error rate occurs near the cutoff for inclusion, and the community methods may even include more of the very poor, even if they do worse on average.
Instead, communities appeared to be using a different concept of poverty: the results of community-based methods were more correlated with individual community members’ self-assessments of their own status, instead of a metric of poverty based purely on consumption. Consistent with this, communities were overwhelmingly more satisfied with the community treatment than the PMT or hybrid treatments. In the community treatment, respondents in the endline survey said they would make fewer changes to the beneficiary list than in the PMT, were more likely to report that they were satisfied with the program than in the PMT and were less likely to submit complaints than in the PMT. Thus, if the government aims to target the poor based solely on per-capita consumption, the PMT performed best. However, if the government is willing to accept the local community’s own definitions of poverty, the community targeting methods were superior in terms of producing a beneficiary list that more closely resembles local poverty concepts and in terms of greater satisfaction and legitimacy. The hybrid method delivered the worst of both worlds – poor targeting performance and low legitimacy.
