Cash or Condition? Evidence from a Cash Transfer Experiment

Conditional cash transfers improve educational outcomes, while at the same time, unconditional cash transfers reduce teenage pregnancy and marriage.


Cash transfers have emerged as effective policy tools to fight poverty, however the debate between conditional and unconditional cash transfers persists. Conditional cash transfer programs (CCTs) are poverty alleviation schemes that provide payment to individuals in response to the completion of certain performance requirements, often relating to child school attendance or preventative medicine visits. A significant body of research shows that CCTs are effective in improving school outcomes among children in developing countries, and have now been introduced in more than 29 low-income countries worldwide. Evidence also shows, however, that unconditional cash transfer programs (UCTs), such as noncontributory pensions schemes, disability benefits, child allowance, and income support also have positive outcomes, such as reduced child labor, increased schooling, and improved childhood nutrition. There has been little research, however, comparing the marginal impact of attaching conditions to cash transfer programs when compared to UCTs. This paper examines the effects of a randomized experiment in Malawi that provided cash transfers that were either conditional on school attendance or unconditional to households with school-age girls.


CCTs are more effective than UCTs at affecting schooling outcomes, but not at delaying marriage and reducing pregnancy rates.

  • Improvements in attendance were observed, as the overall attendance rate was 8 percentage points higher in the CCT group than the control group. Additionally, while school dropout rates declined in both the UCT and CCT groups, the effect in the UCT was only 43% as large as in the CCT arm.
  • The English reading comprehension test scores were significantly better in the CCT group compared to the UCT group. Test scores on reading comprehension, mathematics, and cognitive ability improved in the CCT group but not the UCT group.  There were no statistically significant differences observed in the UCT group when compared to the control group.
  • Marriage and pregnancy rates dropped by 44% and 27% respectively in the UCT intervention group, but remained unchanged in the CCT group. These impacts were almost entirely observed among girls who dropped out of school after the start of the program, which suggests that UCTs alone are helpful in reducing pregnancy and marriage among girls who drop out of school.
  • CCTs are more cost-effective in raising school enrollment than UCTs: a $10 cash transfer would have been required per month to achieve the same gains in school enrollment as were demonstrated from the $5 per month transfer, in the CCT intervention.

In short, while CCTs may be more effective than UCTs in increasing school attendance, UCTs can offer an alternative and delay childbirth and marriage for girls in need of income support.


The study randomly assigns 176 enumeration areas, which are areas that contain an average of 250 households over several villages, in Malawi to receive cash transfers (treatment group) or not (control group). The treatment group was then broken down into a conditional (cash transfer conditional on attending school regularly) and unconditional group. The transfer amounts offered to parents were randomized at the village level, and varied across enumeration areas from $4 to $10. The amounts offered to girls were randomized at the individual level, and varied between $1 and $5. In addition, the CCT arm covered school fees, while the UCT arm adjusted transfers upward by the amount of school fees to make sure that the amounts offered in both arms were identical.

Baseline data was collected in October 2007 and January 2008 from 3,796 never-married females aged 13-22, before the cash transfer offers were made. Two follow-up rounds of data were collected one and two years after the first round. Data was collected from households, schools, school ledgers and achievement tests. The final sample focused on in this paper consisted of 2,284 school girls.

Related GAP Studies