Proxy Means Test Research Articles

Targeting Poor Students With Proxy Means Test

Proxy Means Test (PMT) is one of the most efficient ways to target the poor. The procedure of PMT is using household characteristic variables, which have a relationship with income, as a proxy for poverty. That is, PMT is a measurement of wealth, that is so to say poverty without using income, consumption or expenditure. In this study, we created the PMT poverty scorecard to be used as a tool for targeting poor student in 10 provinces of Thailand, including Mae Hong Son, Nan, Nakhon Ratchasimi, Udon Thani, Nakhon Phanom, Chiang Rai, Trung, Kanchanaburi, Chanthaburi, and Phuket. We estimated the relationship between these household characteristic variables and student’s average monthly household income per capita by using Ordinary Least Square (OLS) regression separately by province to capture any geographical characteristic in each province. We then turned 11 of household characteristic estimated coefficients into PMT poverty scorecard for each province. The result suggests that overall, PMT poverty targeting works well in 10 provinces of Thailand in terms of low undercoverage rate, high targeting accuracy rate in both poverty and total accuracy, except for Phuket which has a huge leakage rate since its geographical characteristics which are quite different comparing to others. This leakage problem, however, needs to be explored more as student maybe, in fact, actually poor as same as PMT poverty targeting suggested. For policy recommendation, we suggested using PMT as the main poverty targeting approach together with an identification survey of student who has been targeted as poor to increase the efficiency of poverty targeting. Also, we suggested the addition of household characteristic variables in survey questionnaire could lead to an increase in efficiency of PMT poverty targeting in statistical term.

Estimating Poverty for Refugee Populations: Can Cross-Survey Imputation Methods Substitute for Data Scarcity?

The increasing growth of forced displacement worldwide has led to the stronger interest of various stakeholders in measuring poverty among refugee populations. However, refugee data remain scarce, particularly in relation to the measurement of income, consumption, or expenditure. This paper offers a first attempt to measure poverty among refugees using cross-survey imputations and administrative and survey data collected by the United Nations High Commissioner for Refugees in Jordan. Employing a small number of predictors currently available in the United Nations High Commissioner for Refugees registration system, the proposed methodology offers out-of-sample predicted poverty rates. These estimates are not statistically different from the actual poverty rates. The estimates are robust to different poverty lines, they are more accurate than those based on asset indexes or proxy means tests, and they perform well according to targeting indicators. They can also be obtained with relatively small samples. Despite these preliminary encouraging results, it is essential to replicate this experiment across countries using different data sets and welfare aggregates before validating the proposed method.

Poverty Alleviation Strategies Under Informality: Evidence for Latin America

Strategies based on growth and inequality reduction require a long-run horizon, and this paper therefore argues that those strategies need to be complemented by poverty alleviation programs. With regards to such programs, informality in Latin America and the Caribbean is a primary obstacle to carry out means testing income-support programs, and countries in the region have therefore mostly relied on proxy means testing mechanisms. This paper studies the relative effectiveness of these and other mechanisms by way of a formal model in which workers choose between job opportunities in the formal and informal sectors. Although the means testing mechanism allows for a more pro-poor design of transfers, it distorts labor decisions made by workers. On the other hand, (exogenous) proxy means testing does not cause distortions, but its pro-poor quality is constrained by the power of observable characteristics to infer income levels. However, since taxation is necessary to fund programs, redistribution becomes less effective, especially for programs other than means testing. The paper concludes by discussing the implications of these results for the design of more efficient targeting programs.

Targeting Humanitarian Aid Using Administrative Data: Model Design and Validation

We develop and assess the performance of an econometric targeting model for a large scale humanitarian aid program providing unconditional cash and food assistance to refugees in Lebanon. We use regularized linear regression to derive a prediction model for household expenditure based on demographic and background characteristics from administrative data that are routinely collected by humanitarian agencies. Standard metrics of prediction accuracy suggest this approach compares favorably to the commonly used scorecard Proxy Means Test, which requires a survey of the entire target population. We confirm these results through a blind validation test performed on a random sample collected after the model derivation.

Poverty Alleviation Strategies under Informality: Evidence for Latin America

Strategies based on growth and inequality reduction require a long-run horizon, and this paper therefore argues that those strategies need to be complemented by poverty alleviation programs. With regards to such programs, informality in Latin America and the Caribbean is a primary obstacle to carry out means testing income-support programs, and countries in the region have therefore mostly relied on proxy means testing mechanisms. This paper studies the relative effectiveness of these and other mechanisms by way of a formal model in which workers choose between job opportunities in the formal and informal sectors. Although the means testing mechanism allows for a more pro-poor design of transfers, it distorts labor decisions made by workers. On the other hand, (exogenous) proxy means testing does not cause distortions, but its pro-poor quality is constrained by the power of observable characteristics to infer income levels. However, since taxation is necessary to fund programs, redistribution becomes less effective, especially for programs other than means testing. The paper concludes by discussing the implications of these results for the design of more efficient targeting programs. Institutional subscribers to the NBER working paper series, and residents of developing countries may download this paper without additional charge at www.nber.org.

How Can African Governments Reach Poor Farmers with Fertiliser Subsidies? Exploring a Targeting Approach in Ghana

African governments have been pursuing reforms to improve the targeting of fertiliser subsidy programmes, but recent experience suggests that these reforms have not ensured that subsidies reach intended beneficiaries. Using a targeting approach based on proxy means tests with carefully selected indicators, this paper suggests that Ghana’s fertiliser subsidy programmes can be targeted to the country’s poor and smallholder farmers more efficiently and more cost-effectively. While a universal subsidy in 2012 is estimated to have reached 11 per cent of poor farmers, the proposed targeting approach would have reached 70 per cent of the poor farmers in northern Ghana and 50 per cent of poor farmers in southern Ghana. Targeting reduces the costs of leakages by about 72 per cent, thus justifying the costs of administering targeted programmes using the poverty proxies. Furthermore, we show that once the initial models are constructed, the targeting approach can be used for nearly 20 years without any significant losses in accuracy. We propose that policy-makers should consider implementing this targeting approach on a pilot scale involving a few communities and, if found successful in practice, in a larger-scale programme.

Universal Basic Incomes vs. Targeted Transfers: Anti-Poverty Programs in Developing Countries

Developing country governments are increasingly implementing cash assistance programs to combat poverty and inequality. This paper examines the potential tradeoffs between targeting these transfers towards low income households versus providing universal cash transfers, also known as a Universal Basic Income. We start by discussing how the fact that most households in poor countries do not pay income taxes changes how we conceptually think about Universal Basic Incomes. We then analyze data from two countries, Indonesia and Peru, to document the tradeoffs involved. The results suggest that, despite the imperfections in targeting using proxy-means tests, targeted transfers may result in substantially higher welfare gains than universal programs, because for a given total budget they deliver much higher transfers to the poor. On the other hand, targeted transfers do lead to more horizontal equity violations, and do create an implied tax on consumption in the region where benefits are phased out. We discuss how alternative targeting approaches, such as community-targeting and self-targeting, can be used to further improve targeting in some situations.

A poor means test? Econometric targeting in Africa

Proxy-means tests (PMTs) are popular for poverty-targeting with imperfect information. In a widely-used version, a regression for log consumption calibrates a PMT score based on covariates, which is then implemented for targeting out-of-sample. The performance of various PMT methods is assessed using data for nine African countries. Standard PMTs help filter out the non-poor, but exclude many poor people, thus diminishing the impact on poverty. Poverty-focused econometric methods such as using quantile regression generally do better. We also characterize the optimal informationally-feasible solution for poverty targeting and compare it to econometric methods. Even with a budget sufficient to eliminate poverty with full information, none of the targeting methods studied bring the poverty rate below about three-quarters of its initial value. The prevailing methods are particularly deficient in reaching the poorest. A basic-income scheme or transfers using a simple demographic scorecard often do as well, or even better, in reducing poverty.

The (Lack of) Distortionary Effects of Proxy-Means Tests: Results from a Nationwide Experiment in Indonesia

Many developing country governments determine eligibility for anti-poverty programs using censuses of household assets. Does this distort subsequent reporting of, or actual purchases of, those assets? We ran a nationwide experiment in Indonesia where, in randomly selected provinces, the government added questions on flat-screen televisions and cell-phone SIM cards to the targeting census administered to 25 million households. In a separate survey six months later, households in treated provinces report fewer televisions, though the effect dissipates thereafter. We find no change in actual television sales, or actual SIM card ownership, suggesting that consumption distortions are likely to be small.

Proxy Means Tests to Identify the Income Poor: Application for the Case of Vietnam

Poverty targeting plays an important role in ensuring the success of support programs for the poor. We assess the governmental poverty targeting in Vietnam, which identifies poor households by collecting income data using a simple two-page questionnaire. We propose a proxy means tests (PMT) method to predict income and poverty status of households using data on basic household characteristics. Compared with the poverty targeting using the simple income questionnaire, the PMT method has higher coverage and lower leakage rates. In addition, the proposed approach also allows for more transparent and comparable poverty estimation across small areas in Vietnam.

The targeting effectiveness of social transfers

ABSTRACTMany methodologies exist for dividing a population into those who are classified as eligible for social transfers and those who are ineligible. Popular targeting mechanisms include means tests, proxy means tests, categorical, geographic, community-based and self-selection. This paper reviews empirical evidence from a range of social protection programmes on the accuracy of these mechanisms, in terms of minimising four targeting errors: inclusion and exclusion, by eligibility and by poverty. This paper also reviews available evidence on the various costs associated with targeting, not only administrative but also private, social, psycho-social, incentive-based and political costs. Comparisons are difficult, but all mechanisms generate targeting errors and costs. Given the inevitability of trade-offs, there is no ‘best’ mechanism for targeting social transfers. The key determinant of relative accuracy and cost-effectiveness in each case is how well the targeting mechanism is designed and implemented.

Targeting Poverty Under Complementarities: Evidence from Indonesia's Unified Targeting System

Combining nationally representative administrative and survey data with official proxy means testing models and coefficients, we evaluate Indonesia's three largest social programs. The setting for our evaluation is the launch of Indonesia's Unified Targeting system, an innovation developed to reduce targeting errors and increase program complementarities. Introducing a new method of evaluation under the condition of multiple programs, we show that households receiving all three programs are at least 30 percentage points better off than those receiving none. Importantly, the bias from failing to account for program complementarities is greater in magnitude than the benefits of receiving a single program.

Retooling Poverty Targeting Using Out-of-Sample Validation and Machine Learning

Proxy means test (PMT) poverty targeting tools have become common tools for beneficiary targeting and poverty assessment where full means tests are costly. Currently popular estimation procedures for generating these tools prioritize minimization of in-sample prediction errors; however, the objective in generating such tools is out-of-sample prediction. We present evidence that prioritizing minimal out-of-sample error, identified through cross-validation and stochastic ensemble methods, in PMT tool development can substantially improve the out-of-sample performance of these targeting tools. We take the United States Agency for International Development (USAID) poverty assessment tool and base data for demonstration of these methods; however, the methods applied in this paper should be considered for PMT and other poverty-targeting tool development more broadly.

Reaching the Poor: Cash Transfer Program Targeting in Cameroon

Identifying and selecting poor households with efficient targeting methods is essential for effective poverty alleviation programs. This paper assesses the ex-post performance of two popular targeting mechanisms, Proxy Means Testing (PMT) and Community-Based Targeting (CBT), in a pilot cash transfer program in Cameroon. Several indicators and metrics to measure each method’s performance in terms of inclusion of poor households and exclusion of non-poor households are employed. Results consistently show that CBT performs poorly in terms of selecting households with low per capita consumption when compared to PMT. CBT appears to select households with low physical and human capital, regardless of actual consumption level. However, CBT also shows more variability in the selection decision than PMT even when alternative poverty definitions are used as robustness tests. The PMT method used in the pilot slightly outperforms other targeting methods (hybrid, alternative PMT, and universal targeting with equal budget), but errors remain high when selecting 35% of the population for program participation. The results suggest caution is needed in employing CBT methods to select households with low per capita consumption in an environment where poverty levels are high and administrative capacities are limited. CBT performance may be improved through more uniform and consistent guidance on program selection criteria and process, including explicit criteria that enable CBT monitoring, as well as a better integration between PMT and CBT.

Targeting Social Transfer Programmes: Comparing Design and Implementation Errors Across Alternative Mechanisms

AbstractAn innovative cash transfer programme in northern Kenya is the first of its kind to trial three targeting mechanisms to learn about which approach is most effective at identifying the poorest households while minimising inclusion and exclusion errors. Analysing data collected through a randomised controlled trial, we conclude that community‐based targeting is the most accurate of the three approaches, followed by categorical targeting by age and household dependency ratio. However, targeting performance is strongly affected by implementation capacity and modalities. Through a simulation exercise, we show that a proxy means test would have performed better than single categorical indicators. © 2015 UNU‐WIDER. Journal of International Development published by John Wiley & Sons, Ltd.

Targeting vulnerable households in urban Burkina Faso: effectiveness of geographical criteria but not of proxy-means testing.

The 2007/2008 food prices hike has increased the interest in social safety nets programmes to fight food insecurity. Targeting the most in need is central to achieve effectiveness of such interventions. In 2009 in Ouagadougou, Burkina Faso, a food voucher (FV) programme targeted the 25 000 most vulnerable households (8.3% of the population). Targeting used a two-stage process: first geographical selection of poorest districts (∼90 000 households); then, in those districts, identification of the most vulnerable households according to a proxy-means test (PMT). Targeted households were entitled to receive FV for 1 year. A first survey was conducted at the beginning of the FV distribution on a representative sample of 2273 households drawn from the poorest districts. One year later a second survey, conducted on a subsample of same households (n = 901), identified those who actually received FV (beneficiary). The performance of the whole process was assessed against household food expenditure, used as the reference measure for vulnerability with a cut-off point of 1513 FCFA (corresponding to the 8.3th percentile of the distribution of expenditure). The 'normalized share of transfers going to vulnerable households' (NSTVH), i.e. proportion of FVs allocated to households below the cut-point, was the main criteria of judgement. Almost twice as many FV were allocated to vulnerable households as compared with a theoretical random distribution all over Ouagadougou (NSTVH = 1.85). When considering the sole targeted districts the NSTVH was only 0.84 (i.e. no more effective than a random distribution), meaning that the geographical stage was effective to select vulnerable districts while the PMT did not perform well to identify the most vulnerable households in those districts. Results could have been improved if only targeted households had received FV (NSTVH = 2.61 and 1.18 for the whole Ouagadougou and targeted districts, respectively). Improved targeting procedures or alternate targeting instruments are needed.

Decentralised Beneficiary Targeting in Large-Scale Development Programmes: Insights from the Malawi Farm Input Subsidy Programme

This paper contributes to the long-standing debate on the merits of decentralized beneficiary targeting in the administration of development programs, focusing on the large-scale Malawi Farm Input Subsidy Program. Nationally-representative household survey data are used to systematically analyze the decentralized targeting performance of the program during the 2009-2010 agricultural season. The analysis begins with a standard targeting assessment based on the rates of program participation and the benefit amounts among the eligible and non-eligible populations, and provides decompositions of the national targeting performance into the inter-district, intra-district inter-community, and intra-district intra-community components. This approach identifies the relative contributions of targeting at each level. The results show that the Farm Input Subsidy Program is not poverty targeted and that the national government, districts, and communities are nearly uniform in their failure to target the poor, with any minimal targeting (or mis-targeting) overwhelmingly materializing at the community level. The findings are robust to the choice of the eligibility indicator and the decomposition method. The multivariate analysis of household program participation reinforces these results and reveals that the relatively well-off, rather than the poor or the wealthiest, and the locally well-connected have a higher likelihood of program participation and, on average, receive a greater number of input coupons. Since a key program objective is to increase food security and income among resource-poor farmers, the lack of targeting is a concern and should underlie considerations of alternative targeting approaches that, in part or completely, rely on proxy means tests at the local level.

Poverty Dynamics and Programme Graduation from Social Protection. A Transitional Model for Mexico's Oportunidades Programme

Social protection programmes have emerged as one of the most important anti-poverty policy strategies in developing countries. Their effects on poverty and well-being have been widely studied. Yet, there is limited knowledge on how a transfer programme should respond to the dynamics of poverty. This paper contributes to the existing literature on social protection by providing an analysis of the implications of poverty dynamics for the graduation of beneficiaries of Mexico’s Oportunidades programme. To the best of our knowledge, this is the first study that provides a framework for a generic graduation condition, to the extent that it can be applied to any other transfer programme with means tests or proxy-means tests. By estimating a Markovian transition model that accounts for unobserved heterogeneity, state dependence, and attrition, and using three rounds of the longitudinal Mexican Family Life Survey, we find that Oportunidades could ‘graduate’ only 28.9 and 26.7 per cent of beneficiary households in urban and rural areas, respectively. Our results also show that the ‘recertification’ or eligibility assessment of Oportunidades—which takes place every three years—could be optimized by conducting it every 3.5 and 4.1 years in urban and rural areas, respectively

Poverty identification: practice and policy implications in Vietnam

To identify poor households, the government of Vietnam applies a combination of proxy means tests and quick collection of income data. This paper examines how well the government's poverty identification reaches the really poor in Vietnam. It is found that there is a large difference between the poverty rate for provinces and districts reported by the government and the rates estimated using expenditure and income data from independent household surveys. There is also a large difference between the poverty status of households identified by local authorities and the poverty status identified by income or expenditure data. More than 50 per cent of the poor households identified by local authorities are not poor in terms of income or expenditure measures. A better identification approach would be to use only proxy means tests and not income data collected using the simple questionnaire.

Finding the Best Indicators to Identify the Poor

Proxy-means testing (PMT) is a method used to assess household or individual welfare level based on a set of observable indicators. The accuracy, and therefore usefulness of PMT relies on the selection of indicators that produce accurate predictions of household welfare. In this paper I propose a method to identify indicators that are robustly and strongly correlated with household welfare, measured by per capita consumption. From an initial set of 340 candidate variables drawn from the Indonesian Family Life Survey, I identify the variables that contribute most significantly to model predictive performance and that are therefore desirable to be included in a PMT formula. These variables span the categories of household private asset holdings, access to basic domestic energy, education level, sanitation and housing. A comparison of the predictive performance of PMT formulas including 10, 20 and 30 of the best predictors of welfare shows that leads to recommending formulas with 20 predictors. Such parsimonious models have similar predictive performance as the PMT formulas currently used in Indonesia, although these latter are based on models of 32 variables on average.