Versión de HTML Básico
Lorena Moreno
116
Analiti a, Revista de análisis estadístico, Vol. 13 (1), 2017
youngest children.
Furthermore, and in order to complement HAZ findings, empirical work in the area
frequently analysis other anthropometric measures, for instance, weight-for-height z-scores
(WAZ). This indicator named global stunting is not very stable, since it is highly dependent
on short-term factors, which are even harder to account for in survey data. Though, to
mimic other researches I estimated the effect of BDH over WAZ. The estimators are still
non-significant and negative ranging from -1,5 to -1 (Appendix 7).
These results might at first seem controversial, though it is important trying to build
upon the understanding of the paths that could be identified as mediators of this socio-
biological outcome in the Ecuadorian context. According, to the LSMS from respondents
in the vicinity
±
3, 95,82% know about the conditions and 99% of them report to follow
these requirements. Then, seemingly the problem is not due to the lack of commitment,
but perhaps the effectiveness of the services provided. Though, it is important to mention
that individuals are incentivised to distort reality if asked about fulfilment of a particular
settlement when their actions are not monitored. Therefore, I inspected expenditure made
with the transfer, which though still self-reported is less likely to include false information,
since the LSMS questionnaire gathers detailed household expenses. For the specific sample,
the largest percentage is dedicated to a category that includes food and household equipment
(53,73%), followed by lower values for education (32,54%) and health (9,55%) (Appendix 8).
Though, it would be ideal to have more disaggregated categories, we can see that not many
household resources are destined for health. Participants might adopt this attitude since
they expect government services to be enough, though, if the aim is to gradually reduce
the dependence of their development on government actions, they could and should change
their priorities. This might be a compelling reason for the non-significant estimate results
11 years after the first implementation.
To sum up, given the fuzziness of the sample, the timing of the evaluation and the
processes undertaken by the government during the surveying period, there is so much un-
certainty that it is difficult to ensure a non-significant effect of the programme. Though,
recent studies of the BDH effect, which also rely on survey data, have in fact found either
negative or non-significant effects. Specifically, Gonzalez-Rozada and Llerena (2011) con-
cluded that mothers receiving BDH experience longer periods of unemployment compared
to those who do not; and Carranza and M´endez (2015) reported a non-significant effect over
exclusive breastfeeding which is a determinant factor of stunting. In this sense, the research
findings might not only reflect the embedded challenges of the design, but also a possible
decrease in the intensity of the effects of the programme, and the need of revision of this
policy scheme.
Moreover, and aiming to get conclusive results, it could be interesting to evaluate mediat-
ing indicators or intermediate effects. For example, the impact on diet composition (calories
by type of food) aiming to search for pathways that determine final outcomes. Similarly,
with health variables related to mothers, as breastfeeding or their consumption patterns.
34