The purpose of this study is to help AusAID staff, their counterparts in partner governments and others working in the field of social protection to better understand the strengths and weaknesses of a targeting methodology known as the proxy means test (PMT). As social protection practitioners search for effective ways to target poor people in developing countries, proxy means testing has become increasingly popular. The methodology estimates household income by associating indicators or ‘proxies’ with household expenditure or consumption. This study assesses its accuracy, objectivity, transparency and ease of implementation.

Despite the substantial literature on the PMT there is a dearth of comprehensive analysis. This means that developing country governments—and those advising on social protection—do not have an adequate basis upon which to consider the merits of proxy means testing or assess the methodology against alternatives. ‘Targeting the Poorest’ gives policymakers and those working in social protection information to help them decide whether to use the PMT. It attempts to explain the methodology’s considerable inaccuracy at low levels of coverage and sets out other challenges to its use. The theoretical basis of the methodology and its implementation are examined as are its associated costs.

Proxy means testing uses multivariate regression to correlate certain proxies, such as assets and household characteristics, with poverty and income. This study assesses regression accuracy in Bangladesh, Indonesia, Rwanda and Sri Lanka and finds that the PMT has high in-built errors, especially at relatively low levels of coverage (20% of the population and below). Exclusion and inclusion errors vary between 44% and 55% when 20% of the population is covered and between 57% and 71% when 10% is covered.

Part of the reason for this is the imperfect correlation between multiple proxies and household consumption. Additionally, the PMT methodology is based on national household survey data that represent ‘reality’ at one point in time and are inherently inaccurate to varying degrees. Other issues are sampling errors in household surveys and assumptions made in applying the PMT, which increase the arbitrary nature of the methodology yet affect whether individual households receive social protection benefits.

Implementing proxy means testing presents a number of challenges. Enumerators are not always objective when conducting surveys and do not always have time to verify proxies within households. Some proxies can also be difficult to verify—such as level of education, age and household assets—and interviewees can influence survey results, with children and men not as reliable as women.

Another challenge of proxy means testing relates to crises and shocks faced by households, including minor ones that are part of every day life. As a result, households that fall into poverty but do not suffer a related change in the household characteristics and assets used as proxies cannot receive social protection benefits.

The PMT is expensive to administer and has associated social and political costs. There is evidence that it can generate social conflict and stigmatise beneficiaries. Politically the methodology—as with all other forms of poverty targeting—is less likely to be popular because it excludes the middle class and those who are better-off.

This study’s findings show that the PMT is inherently inaccurate, especially at low levels of coverage, and it relatively arbitrarily selects beneficiaries. It therefore functions more like a simple rationing mechanism that selects some poor and non-poor but excludes large numbers of eligible poor from receiving benefits and support.

‘Targeting the Poorest’ does not provide an in-depth assessment of other targeting methods or formally compare them with the PMT methodology. It suggests, however, that other methods used to develop social protection schemes—which do not directly target poor people—may be better at including intended beneficiaries and avoiding the pitfalls of the PMT.