2003-01, Coady, David, Grosh, Margaret, Hoddinott, John

Of the commonly used methods for directing transfers to the poor, there is little consensus about which is best. Policymakers need to know how effective different targeting mechanisms are, how the effectiveness differs by method and type of program, and the implications. Targeting success can be partially captured by one outcome indicator, the share of benefits going to the bottom 40 per cent of the population. For example, if a program delivers 60 per cent of its benefits to this group, the outcome indicator is (60 divided by 40 =) 1.5. The higher the indicator - i.e., the greater the percentage of benefits going to the poor relative to their population share - the more progressive is the targeting. The authors calculate their indicator for 85 of the programs in the database. The full study provides information on the use of targeting techniques, summary statistics on comparative program performance, and regression analysis to examine the correlations between methods and outcomes. The study drew broad conclusions, subject to the limitations described beforehand, suggesting that "Targeting can work, but it doesn't always. There is no clearly preferred method for all types of programs, or all country contexts. A weak ranking of outcomes achieved by different mechanisms was possible. And, implementation matters tremendously to outcomes". Targeting performance improved with country income levels, the extent to which governments are held accountable for their actions, and the degree of inequality.