Versión de HTML Básico
Lorena Moreno
84
Analiti a, Revista de análisis estadístico, Vol. 13 (1), 2017
Abstract
Poverty eradication policies based on Conditional Cash Transfers (CCTs) are widely
used in developing countries, being chronic stunting one of the outcome indicators most
studied. However, the few empirical analyses for Ecuador do not agree on the effects
of CCTs on this variable. In this context, and based on a large-scale survey, in this
article I propose a two-stage strategy to evaluate the effect of the Human Develop-
ment Bonus programme on children chronic stunting. First, I replicate the eligibility
status index through Categorical Principal Component Analysis, and then implement
a Fuzzy Regression Discontinuity Design. Main results show that there is a statis-
tically significant effect of 18% to 24% of the eligibility status on the probability of
treatment; a non-significant estimate of the intention-to-treat effect going from -0,5
to -0,3 standard deviations; and, a local average treatment effect ranging from -2,1 to
-1,6 standard deviations not significantly different from 0, depending on the approach
and the specification. Although, falsification tests support compliance with the iden-
tification assumption of the evaluation design, conclusions should be interpreted with
caution given the surveying period, the changes implemented in the programme and,
finally, the sample analysed.
Key Words:
social development, Conditional Cash Transfers, stunting, principal
component analysis, regression discontinuity design.
JEL Codes:
I31, I38, I12, C38, C31
1 Introduction
A vast number of empirical questions about poverty tackling in developing countries depend
on causal effects of government programmes. One widely implemented version of this type
of intervention are the Conditional Cash Transfers (CCT). These transfers aim to achieve
generalised social well-being by allowing vulnerable groups be responsible for their own
development. In order to attain this objective, the strategy incentivises social investments
through beneficiaries’ co-responsibilities. Namely, school assistance, and health checks of the
children pertaining to the recipients’ households.
In Latin American these programmes became popular in the late nineties, and since then,
several studies focused their attention on measuring the expected benefits. In between the
most recognisable are the Mexican
PROGRESA
, The Brasilian
Bolsa Familia
, an the Colom-
bian
Familias en Acci´on
. These and others programmes have been evaluated in the light of
their aligned objectives, being the most commonly studied outcomes, school enrolment and
assistance, chronic stunting and vaccine preventable diseases prevalence. The educational
outcomes have evidenced stable and positive effects (Ponce and Bedi, 2010), while, conclu-
sions on children health are more heterogeneous, particularly when focusing in malnutrition
(Paxson and Schady, 2007). This, added to the fact that stunting (measured as the deviation
2