International development organisations and NGOs are all looking for effective interventions to alleviate chronic poverty in Africa. Is cash transfer headache or breakthrough?
Beyond reducing poverty, finding the right methods to improve the scope of project implementation and bring satisfying outcome is the key element of deciding the success of charitable organisations and development institutions. This evaluation report(policy brief) on cash transfer by GiveDirectly conducted in 2013 is an example of seeking the right method, in my opinion.
The evaluation was conducted by Johannes Haushofer and Jeremy Shapiro. The report makes it clear that Jeremy Shapiro who participated in the evaluation is a co-founder and former director of GiveDirectly.
Overall, the result of the intervention is encouraging. There was a significant increase in asset value, consumption, business & livestock income, food security, psychological well-being, and female empowerment. However, long-term outcomes such as health and education didn’t show a significant increase in this result.
Here are the nuts and bolts of the plan (see table below).
- Households meeting a basic eligibility criterion(living in a thatched roof house) were randomly chosen as the recipient.
- Treatment and control groups were set for impact comparison.
- Recipients were told they can spend the money as freely as they wish.
- The cash transfer is unconditional. That is, recipients do not have to satisfy any condition in order to receive money.
- Three design features were added and randomised.
- woman vs. man recipient
- monthly vs. lump-sum transfer
- large vs. small amount(USD1,100 vs. USD300)
With the randomisation like a, b, and c, the evaluation provides more opportunities to improve the quality of the outcome. Table below is the result of the evaluation.
Control group is in column 1 and the treatment group, the recipient, is in column 2. There are significant increases in treatment effect with more than 5% significance(* denotes significance at 10 pct., ** at 5 pct., and *** at 1 pct. level). Put simply, three asterisks means the outcome is statistically significant.
Significant increase by USD279 in asset value, USD36 in consumption(food, medical, social expenses, in particular), and USD17 increase in business and livestock income.
Significant increase by 0.25SD(standard deviation) in food security, and 0.20SD in psychological well-being, 0.23SD in female empowerment, compared to control group.
Monthly payment has an influence in food security with 0.26SD.
There is little evidence that female/male recipient makes significant difference.
Spillover effect is small or not significant with the exception of women empowerment. The result implies that treatment group shows an effect from the intervention relative to control group.
Some common and critical questions on cash transfer are:
- won’t it be spent on gambling, alcohol and the like?
- won’t it be just one time impact? Does it bring sustainability to the recipient households?
- won’t the cash inflow affect the price increase in the village?
My thought on this evaluation is that it’s meticulously well designed evaluation. I’m not very familiar with the complication of having multitude of randomised variables as in this report. But it’s certain that various results from those variables can help guide how to design an improved project/programme afterwards. Somewhat surprising fact, however, is that this type of report is not as common as it should. It’ll be great if more and more NGOs and other agencies use an impact evaluation/randomized controlled trials like in this case.
How to design a survey for estimating returns to education? Do returns to education depend on how and who you ask? Yes. Turns out short questionnaires do lead to biased estimates of returns to education. Even thought who responds ,self or proxy, did not make significant difference, the level of biased estimates differed between education … Continue reading “Development links”
Malawi’s food problem is getting more serious as the harvest season is approaching.