«DIREC TIONS IN DE VELOPMENT Human Development Public Disclosure Authorized The Cash Dividend The Rise of Cash Transfer Programs in Sub-Saharan Africa ...»
Despite these benefits, costs are involved in implementing poverty-related targeting (White and others 2009). Although measuring those costs is not always easy, Grosh and others (2008) found that means and proxy means targeting costs in a sample of programs studied in Central Asia, Eastern Europe, and Latin America averaged approximately 4 percent of program costs and between 25 percent and 75 percent of administrative costs. Targeting costs as a percentage of benefits transferred for a group of programs in Latin America averaged below 1.5 percent. Within Sub-Saharan Africa, targeting costs as a percentage of total costs will depend on transfer size, program coverage, and the extent to which the program is well established.
In addition, targeting of transfers may be difficult to administer correctly:
eligible households may be difficult to identify and data collection may be expensive (Jones 2009), particularly in some low-capacity Sub-Saharan African settings. Households may have difficulty understanding targeting criteria and may not believe those criteria are fair. When targeting criteria are well understood, households may adjust their composition or behaviors to meet those criteria. Programs with targeted transfers also have to be careful to avoid leapfrogging of beneficiary income over nonbeneficiary income, and retargeting will need to occur at certain intervals (White and others 2009). An additional concern about targeting is its potential to create social tensions or stigmatization (Jones 2009). This issue is important in countries with widespread poverty.
Whether a program identifies beneficiaries using categorical or near-universal methods or chooses beneficiaries through more selective methods should be (continued next page) Design and Implementation of Cash Transfers in Sub-Saharan Africa 83
Box 3.2 (continued)
based on a calculated decision regarding the expected costs and benefits of the methods chosen. The expected future size of targeted groups will also have implications for the CT’s long-term fiscal sustainability and must be considered. Analysis of household survey data can help outline the empirical trade-offs associated with the targeting method chosen.
Of course, sometimes targeting is not open to discussion; beneficiary groups may have already been decided before the analytical work began (Slater and Farrington 2010). Targeting choices must ultimately be pragmatic decisions that weigh the various administrative, private, social, and political costs involved with the potential targeting methods (Grosh and others 2008).
Recommended References on Targeting For further information on issues associated with appropriate targeting, see Grosh and others (2008) and Coady, Grosh, and Hoddinott (2004b) for a thorough treatment of the topic. See also Slater and Farrington (2009), who highlight important issues to consider when targeting in low-income, low-capacity countries, and Slater and Farrington (2010), who provide a quick reference guide on key targeting principles, particularly in very low-income settings.
a. Notably, Niño-Zarazúa and others (2010) point out that countries in Southern Africa with universal social pensions have Gini coefficients greater than 0.5. They argue that this inequality has generated a desire for social protection while helping to keep transfer leakage low and providing fiscal space for redistribution using transfers.
Groups Targeted Groups commonly targeted in Sub-Saharan Africa include OVC or other HIV-affected individuals, the elderly, and people with disabilities or those who are unable to participate in the labor market. Other vulnerable groups often targeted include the extremely poor, potentially malnourished preschool children, and pregnant or lactating mothers. Targeted groups are not mutually exclusive, and significant overlap across the groups may occur. Some programs target a combination of these groups, such as OVC and extremely poor elderly people.
Figure 3.1 depicts targeted groups, broken down between conditional and unconditional CTs.
Many CCTs target children or OVC. Several target the unemployed, mothers and young children, and young adults, and one in four CCTs targets a combination of these groups.
84 The Cash Dividend Figure 3.1 Groups Targeted in Conditional and Unconditional Cash Transfer Programs in Sub-Saharan Africa percent
Source: Authors’ representation.
Note: Sample size is 15 for CCTs, 96 for UCTs, and 111 for all programs. Samples are based on programs with known information about the targeted group.
Unconditional cash transfers (UCTs) target a wider variety of groups.
The elderly are the most frequently targeted group, followed by victims of disasters and people with insecure food sources. Households without members capable of participating in the labor force are targeted slightly less frequently, at approximately 10 percent of the time. OVC, refugees, ex-combatants, and people living with disabilities are targeted in 5 percent to 10 percent of the UCTs.
Targeting Methodology Most CTs in Sub-Saharan Africa combine several targeting methods to select beneficiaries. For instance, many select households using community or proxy means targeting within selected geographic regions. In some cases, targeting criteria could logically be classified into multiple categories.
A summary of the most commonly used targeting methods in the reviewed programs is found in figure 3.2. Multiple targeting methods are counted for programs that used more than one method. Panel a shows that categorical targeting is the most widely used method in both CCTs Design and Implementation of Cash Transfers in Sub-Saharan Africa 85 Figure 3.2 Multiple Targeting Methods Used by Programs in Sub-Saharan Africa
Source: Authors’ representation.
Note: In panel a, totals for CCTs and UCTs are greater than 100 percent because multiple targeting methods were used in many programs. Sample size is 14 for CCTs and 52 for UCTs. Samples are based on programs for which specific targeting information was available.
In panel b, totals add up to greater than 100 percent by income category because multiple targeting methods were allowed per program. Sample size of CTs is 15 for upper-middle-income countries, 11 for lower-middleincome countries (excluding fragile states), 30 for low-income countries (excluding fragile states), and 9 for fragile states. Samples are restricted to programs for which specific targeting information was available.
86 The Cash Dividend and UCTs. Geographic targeting and community targeting are also used frequently in most programs, especially CCTs. A little over one in five programs applies means tests, proxy means tests, or self-targeting. Proxy means tests are much more common in CCTs, a practice similar to CCTs outside of Sub-Saharan Africa, and means testing and self-targeting are more popular in UCTs.
Some social pensions in the region are universal or categorically targeted (as in Botswana, Cape Verde, Lesotho, and Mauritius), whereas others are means or proxy means tested, meaning that they have criteria in addition to the age requirement (as in Namibia, South Africa, and Swaziland). This mixture is similar to that found in pension programs around the world.
Examining targeting methodologies by countries’ income status reveals significant differences across the groups (figure 3.2, panel b). Uppermiddle-income countries practice categorical, means testing, and selftargeting methods most frequently. Lower-middle-income and low-income countries frequently use categorical and geographic criteria. More than half of CTs in lower-middle-income countries use community-based targeting, and almost 9 out of 10 do the same in low-income countries. One in two programs in lower-middle-income countries uses means or proxy means testing, and one in three programs in low-income countries uses proxy means testing. Very little means testing is used in low-income countries. CTs in fragile states use categorical and self-targeting methods in many programs; community targeting is a fairly common targeting method for these countries as well.
Widespread use of community-based targeting. One of the most salient features of targeting methods used in the reviewed CTs is the widespread use of community-based targeting systems, which are used to a more limited extent in comparable CTs around the world. Community targeting has obvious benefits. Those in the immediate community are easily able to identify vulnerable households that should receive benefits. They are familiar with households’ needs and recent shocks they have faced, and they are likely to know whether households will use cash in a manner they deem responsible. Community-based targeting is relatively inexpensive, and it has the additional benefit of informing community members about the CT and involving them in it. Conversely, community-based targeting may impose additional costs on communities in terms of opportunity costs or social and political costs of carrying out the targeting at the local level.
Design and Implementation of Cash Transfers in Sub-Saharan Africa 87 A commonly cited concern associated with community-based targeting is the danger of nepotism or other types of favoritism leading to inclusion of persons who should not be eligible (and therefore the potential exclusion of eligible individuals or households). Favoritism in community-based targeting has been identified as an issue affecting CT programs in Malawi and Zambia. Miller, Tsoka, and Reichert (2010) report that village heads were sometimes able to inappropriately influence community members involved in selecting beneficiaries for Malawi’s SCT program, perhaps because community members were not able or confident enough to navigate local political dynamics. Across Sub-Saharan Africa, the strength of traditions of extended family responsibilities and obligations creates an environment favorable to the emergence of such problems.
Even when communities attempt to implement fair and accurate targeting, inconsistent application of targeting rules may arise when eligibility criteria have room for interpretation. Clear targeting criteria and training of community committees or members involved in targeting should help ease this problem, at least to an extent. Finally, even when community members objectively target households, community-based targeting may still be perceived as unfair or inconsistent. This issue may be especially contentious in the poorest areas, where very small material differences separate beneficiaries and nonbeneficiaries.
Some Sub-Saharan African countries have taken steps to combat targeting errors when using community-based targeting. Malawi’s SCT does not allow village heads to be on community social protection committees, which are in charge of targeting (Miller, Tsoka, and Reichart 2010). The SCT has also recently added a verification round to its targeting process.
Extension workers are now involved in verifying targeting decisions, as they know community members well and may be more impartial than others, given that they are usually from outside the community and extended family system. Zambia allows local leaders to be involved in targeting but has taken steps to create a confidential appeals process to ensure that targeting is as fair as possible (Hamonga 2006).
In some cases, communities have already received training and have the capacity to implement targeting relatively easily. In Rwanda, for example, information gathered by existing Ubudehe2 committees about local households’ relative welfare is used for targeting purposes in the Direct Support program. In Ethiopia’s PSNP, communities already had relevant experience targeting households for food aid, which has facilitated the program’s community-based targeting (World Bank 88 The Cash Dividend 2010a). Other programs first sensitize communities and then provide training to enable them to participate more effectively in the targeting process.
Additional checks on community-based targeting should be clearly understood by community members. Kenya’s CT for OVC uses community-based targeting combined with geographic, categorical, and even proxy means testing. Although the process has reduced targeting errors, the complicated method has also generated confusion surrounding beneficiary selection, leaving some households feeling unfairly excluded or unsure of why they are beneficiaries. This confusion may easily become a source of tension among community members.