You are on page 1of 2

Exercise on Evaluation of Development Programs

Chapter 4 of Development Economics: Theory and Practice


Data available in the supplemental Excel file

1. Impact evaluation of a Conditional Cash Transfer program using a randomized


control trial approach: Progresa in Mexico
The Mexican CCT program, Progresa/Oportunidades, was piloted with an RCT to
test the validity of the approach. We give you in worksheet “Progresa” data on 100
treated children (Progresa beneficiaries) and 162 control children, all in the 11-12 years
old age group and all in school in 1997. The outcome variable is whether they were
enrolled or not in 1998, i.e., what is the continuation rate. We also give you several
observable characteristics of these children:
Gender (Male = 1, female = 0)
Head of household education (years)
Distance to school (in km)
Indigenous (Indigenous = 1, non-indigenous = 0)

The treatment variable is Progresa (In Progresa = 1, not in Progresa = 0)


The outcome variable is Enrollment (Enrolled = 1, Not enrolled = 0)

1. You need first verify that the RCT was correctly done in randomly selecting a
Treatment and a Control group.
a. Explain how you do this verification
b. Using the four characteristics available, does it look as though the RCT
was correctly done? Show your results.
2. Calculate the impact of Progresa on enrollment for the whole population.
3. Calculate the impact of Progresa for males and females. Who benefited most?
4. Calculate the impact of Progresa for indigenous and non-indigenous children.
Who benefited most?
Hint: For questions 3 and 4, you may want to use the "AVERAGEIF" function in Excel
Report your results in the following table:

1
2. Impact evaluation of a land certification program using a double difference
approach: Procede in Mexico

During the period running from 1993 to 2006, Mexico engaged in a major land
certification program called Procede that gave property rights to some 2.5 million farm
households over half of the country’s agricultural land. The program was rolled out
gradually across localities. Before certification, these farm households only had the right
of use of the land (not ownership) and could be expropriated if they did not farm the land
productively. One research question is whether having a land title giving households
security of land ownership creates opportunities for migration since they no longer risk
losing the land if they do not farm it productively. To test this, we use data from the
population censuses by localities in 1980, 1990, and 2000. Net migration would be seen
in the change in population in localities that were treated (certified between 1993 and
1999) and not treated (certified in 2000 or after). In worksheet “Procede”, we give you
data on the average population size (number of persons) in 10,924 localities that were
certified between 1993 and 1999, and in 6,828 localities that were certified in 2000 or
after.
1. Calculate the average change in population between the censuses for the treated
and control localities.
2. By double difference, measure what was the impact of land certification on
average change in population. Did granting property rights increase or decrease
net in-migration and by how much?
3. How would you verify that the diff-in-diffs methodology applies to measuring the
impact of certification on migration during the 1990-2000 period: explain what
the test is and what it means.
4. Show in the table below and on a graph the results of the test you proposed in
response to the previous question.

Report your results in the following table:

You might also like