«DIREC TIONS IN DE VELOPMENT Human Development Public Disclosure Authorized The Cash Dividend The Rise of Cash Transfer Programs in Sub-Saharan Africa ...»
Fiszbein and Schady (2009) have summarized existing evidence, which includes economic theory, simulations, and empirical analysis from natural experiments, on the effect of conditions in CCTs. They suggest that conditions may increase positive impacts on children’s education and health. Until the recent evaluation of Malawi’s Zomba CT was completed, evidence on conditions was not based on any experimental evaluations.
Malawi’s Zomba CT provided transfers conditional on school enrollment to some female adolescent beneficiaries and unconditional transfers to others. The CCT has proved to be more effective than the UCT in improving schooling outcomes, including enrollment, attendance, and test scores (see figure 3.10). At first glance, the UCT, but not the CCT, appears to decrease the probability of teenage pregnancy and early marriage. However, those results are more nuanced than they first appear, and 116 The Cash Dividend Figure 3.10 Results of CTs versus UCTs in Malawi
0.15 * 0.10 0.05 0.00
Source: Baird, McIntosh, and Özler 2010, table III (panel a) and table XI (panel b).
Notes: Sample size is 852 for panel a and 2,057 for panel b. Dependent variable equals total number of school terms enrolled in during the program out of six possible terms for panel a. In panel b, test scores are standardized with a mean equal to zero and standard deviation of one in the control group. Math and English tests were created on the basis of Malawi school curriculum. The TIMSS score is based on five questions from the 2007 TIMSS. Both panels show results from ordinary least squares regressions using baseline values of age dummies, strata dummies, a household asset index, highest grade attended, a dummy for ever had sex, and a dummy for whether respondent participated in development of tests in pilot (learning outcomes regressions only). Results for both panels are based on regressions run with robust standard errors clustered at the local (enumeration area) level and are weighted so that results are representative of the target population in the enumeration areas.
* indicates the result is statistically different from that of the control group at the 90 percent level; ** indicates the result is statistically different from that of the control group at the 95 percent level; *** indicates the result is statistically different from that of the control group at the 99 percent level; † indicates that the CCT result is significantly different from the UCT result at the 90 percent level; †† indicates that the CCT result is significantly different from the UCT result at the 95 percent level.
Design and Implementation of Cash Transfers in Sub-Saharan Africa 117 they highlight some of the trade-offs between providing CCTs and UCTs.
An additional question raised by the experiment is whether such results could be replicated outside of the experimental setting. See box 3.8, later in this chapter, for an explanation of program results.
Prevalence of Conditional Programs Conditions are applied in a number of Sub-Saharan African countries, including, but not limited to, Burkina Faso, Ghana, Kenya, Malawi, Nigeria, and Tanzania. In many cases, programs take a significant amount of time to build sufficient capacity to regularly enforce program conditions. Enforcing conditions has often depended on the capacity and buy-in of relevant line ministries to work with program units to monitor fulfillment of conditions.
The majority of CTs started in Sub-Saharan Africa since 2000 have been unconditional (see figure 3.11). However, the percentage of CCTs is not insignificant: one in four CTs were conditional. Newer programs are more likely to be conditional than programs that were established earlier. Of the identified programs established in 2007 or later, 40 percent were designed as CCTs. If the experience of the past several years reflects an ongoing trend, the relative representation of CCTs in the region is expected to continue to grow.
In terms of the general typology described in chapter 2, the region’s wealthier countries rely on UCTs (see figure 3.12). This tendency partially Figure 3.11 Changes in Relative Numbers of CT Programs: UCTs versus CCTs
Source: Authors’ representation.
Note: Data are current as of June 2009. Panel a has a sample size of 72; panel b has a sample size of 30. Samples are limited to programs that began in 2000 or later and that provided enough information to be confidently classified.
118 The Cash Dividend Figure 3.12 Use of UCTs versus CCTs by Country Income Status
Source: Authors’ representation.
Note: Sample size is 25 for upper-middle-income countries,13 for lower-middle-income countries, 43 for lowincome countries, and 20 for fragile states. Total for low-income countries adds to greater than 100 because two programs are both UCTs and CCTs.
reflects the belief held in some wealthier countries that rights-based social assistance is not compatible with CCTs. The lower-middle-income countries have the highest relative representation of CCTs (approximately one in three programs), while just over one in four programs in low-income countries are CCTs. The greater representation of CCTs in lower-middleincome countries than in low-income countries probably reflects some of the greater capacity constraints involved with enforcing conditions in the poorest countries. Fragile states have primarily implemented UCTs, as would be expected given their weak institutions and frequent emergencyfocused CTs.
Although an increasing proportion of programs in the region are designed as CCTs, this shift does not necessarily indicate that all programs will head in this direction. Some relatively new programs have purposefully decided that conditioning transfers is not currently the best option.
For example, Senegal’s CF-SCT was designed without any conditions attached for several reasons. An unconditional transfer was justified, given that the program was dealing with an acute crisis caused by rising food prices. The program’s designers assumed that if targeting correctly identified vulnerable households, the transfer would be used appropriately and conditions would be unnecessary. A strong communications strategy supporting maternal and children’s nutrition was also expected to help beneficiaries make appropriate investments in nutrition and health without the need for conditions. That being said, the program’s designers remained Design and Implementation of Cash Transfers in Sub-Saharan Africa 119 open to the idea of imposing soft conditions, and they planned to examine this possibility (World Bank 2009a). (Soft conditions require that beneficiary households agree verbally or in writing that they will abide by listed conditions; however, no penalty is charged if they do not comply.) Ethiopia’s PSNP-DS, though unconditional, has also left itself open to conditioning future transfers on attendance at literacy classes or participation in relatively undemanding labor activities (World Bank 2010a).
Variation in Approach to Conditionality throughout Sub-Saharan Africa Conditions that have commonly been used in Sub-Saharan Africa are similar to those in CCTs around the world. The most common conditions are education- and health-related requirements for children (see figure 3.13).
Additional common conditions require household adults to attend educational or training sessions and pregnant women to fulfill health-related requirements.
Source: Authors’ representation.
Note: Sample size is 18. Total adds to greater than 100 percent because most programs have multiple conditions.
“Other” includes elderly members’ medical checkups, successful school progression, avoidance of child labor, and formation of community committees to improve local sanitation. CCTs were excluded if specific data on conditions did not provide enough information for classification.
120 The Cash Dividend Several CCT programs in Sub-Saharan Africa have adapted programmatic variations that reflect the constraints present in many low-capacity settings, as well as concerns that conditions are too heavy a burden to place on many beneficiaries. These programs try to strike a compromise between conditional and unconditional transfers by imposing soft conditions, with no penalty charged if beneficiaries do not comply with the requirements. In the reviewed CTs, it was not always clear whether beneficiaries believed they would be penalized for noncompliance with soft conditions.
Some programs use both soft and hard conditions. Such a mixture of conditions is used in Nigeria’s Kano CCT for Girls’ Education. If the hard condition of 80 percent school attendance is not met, benefits are forfeited. Soft conditions are as follows: girls are expected to pass their classes and obtain birth certificates, siblings under age five must receive immunizations and attend medical checkups, mothers must attend preand postnatal classes, and mothers and partners must attend awarenessraising seminars (Gerelle 2009). The program’s experimental evaluation will shed light on the extent to which both types of conditions help to achieve the program’s goals.
Other Sub-Saharan African programs approach conditions from a similarly less punitive perspective. Rather than setting transfer levels and conditions and then deducting from transfer levels for noncompliance, they set a base transfer level that is given regardless of the household’s behavior. If the household chooses to engage in conditioned activities, it then receives a bonus payment. This design is used in Zambia’s Chipata district, where a bonus is supposed to be awarded to those who fulfill nonrequired criteria (Ministry of Community Development and Social Services and German Agency for Technical Cooperation 2007). Using bonuses rather than conditions is thought to help create a positive attitude toward the conditioned behavior and increase the likelihood that the behavior will continue over the long term. The difference between this design and that of a traditional CCT appears to be mainly semantic and ideological.
Conditions have been flexibly applied in practice. In reality, the line between unconditional transfers, transfers with soft conditions, and transfers with hard conditions is often blurred. For example, Kenya’s CT for OVC has conditions outlined but recognizes that requiring households to fulfill conditions is not always feasible. In those cases, the CT is supposed to operate as a UCT. A similar plan is in place for Tanzania’s CB-CCT;
Design and Implementation of Cash Transfers in Sub-Saharan Africa 121 conditions will not be enforced when education and health centers are far from program communities. Nigeria’s Kano CCT for Girls’ Education focuses on girls’ completion of primary school or transition to junior secondary school, and conditions depend on the local availability of the schools (Bouchet 2009). Eligible schools also must be receiving support under the national and state education plans (Ayala 2009). The programs just mentioned conduct supply-side capacity inventories to understand where to apply conditions.
Even in areas with an adequate supply of schools and health clinics, conditions are sometimes only gradually enforced. Monitoring of conditions in Kenya’s CT for OVC has been incrementally phased in over time. Only education conditions were being enforced by mid-2009.
Even though some conditions are still not enforced, households are encouraged, through community awareness campaigns, to invest in their children’s education and health care whenever possible.
Penalties for noncompliance and frequency of monitoring reflect concerns for beneficiaries’ well-being. Although programs in Sub-Saharan Africa share human capital objectives with other programs around the world, it is crucial that the cost to households of complying with conditions is carefully considered before a decision regarding conditioning is made. In a study of a Zambian SCT, Schüring (2010b) found that the poorest households reported that they would be most likely to forgo receiving a conditional transfer, presumably because of the burdens the conditions would place on the household. This result is a common concern about conditional programs, and it must be examined.