Brief 15: Is it the cash or condition in Malawi?

EGAP Researcher: Craig McIntosh

Other Authors: Sarah Baird, Berk Özler

Geographical Region: Africa

Research Question:

In the Conditional Cash Transfer programs that are sweeping the developing world, what changes the lives of the very poor: the cash, or the condition on which the transfers are typically predicated?

Preparer: Seth Ariel Green

February 13, 2015

Background:

While much research has been dedicated to evaluating the effectiveness of both CCTs and UCTs, no prior experiment has allowed for the separate measurement of the effects of money and a condition. This experiment is a rigorous, randomized evaluation of a representative development program that allows policy-makers to see how the condition in CCTs may or may not change behavior.

Conditional cash transfer	CCT	Payment of money conditional upon fulfillment of a requirement
Unconditional cash transfer	UCT	Money with no requirements

Malawi is an under-developed, sub-Saharan African country of about 16 million people. AIDS is the leading cause of death among adults, and the average life expectancy is 54.8 years. Researchers have generally found that increased education, delayed pregnancy and increased age at first marriage are all associated with positive outcomes for both individuals and societies.

Research Design:

The experiment took place in the Zomba District of Malawi. Treatments were assigned randomly by Enumeration Areas (EAs), which are groups of 250 households spanning several villages. EAs include both urban and rural areas. The first treatment group, consisting of a random subset of girls and their families in 46 EAs, was offered money for regularly attending school, while the second treatment group received cash without any stipulations. There was also a control group that did not receive any form of cash, conditional or otherwise. In both treatment groups, the girls, ages 13-22 and never-married, were randomly offered $1, $2, $3, $4, or $5, while their families were randomly assigned $4, $6, $8, or $10. During the experiment, researchers interviewed a random subset of the UCT subjects to make sure that they knew that they did not need to go to school to receive money.

School attendance was measured continuously for two years. At the end of the experimental period, subjects were given math, English, and cognitive tests and interviewed about marriage and pregnancy.

Results:

Although both groups had increased school enrollment compared to a control group, the effect on the UCT group was not as large as it was on the CCT group. The CCT group also scored higher on English, math and cognitive skills tests. However, the UCT group had much lower marriage and pregnancy rates than either the CCT or the control group. While marriage rates for those enrolled in school were very similar between the two treatment arms, marriage rates were much lower for UCT subjects not enrolled in school. The authors conclude that a UCT was more effective in changing the behavior of girls who would not enroll in school in either a CCT or UCT condition. Girls in the UCT group were 27% less likely to have ever been pregnant than either the CCT or the control group.

Age also matters. In the CCT group, those under 16 scored much higher on English tests, while there was no statistically significant increase for those older than 16. The UCT arm was substantially more effective at preventing marriage and pregnancy for those 16 or older.

Policy Implications:

• Many of the behavioral outcomes in the UCT arm were experienced largely among those who dropped out of school, while those who dropped out of school in the CCT arm did not receive the same effects. The study suggests that dropout may be thought of as an indicator of vulnerability for adolescent girls in this context, implying a strong perverse consequence of cutting off transfers at precisely the moment girls leave school.
• Policymakers and organizations working on delaying marriage and parenthood should consider instituting UCT programs instead of CCTs. UCTs delayed marriage (by 44%) and childbearing (by 27%) after two years for those who dropped out of school after the start of the two-year intervention.
• The smallest transfers given in this experiment ($1 or $4) in the CCT arm had the same average impacts for schooling across the range of transfer amounts. Policymakers and organizations considering instituting CCT programs should take this into account.
• This experiment found that poor parents will invest at least some of the additional funds from a positive income shock towards educating their daughters. This is reflected in the fact that UCT had some positive effect on school enrollment.