Versión de HTML Básico
Assessing the Effect of Conditional Cash Transfers in Children Chronic Stunting: The Human Development Bonus in Ecuador
101
13
Analiti a, Revista de análisis estadístico, Vol. 13 (1), 2017
5 Results
5.1 Falsification test
The BDH has been reaching to a significant proportion of Ecuadorian households for thir-
teen years, and likewise, other social programmes, the possibility of the manipulation of the
rules configure a permanent concern. This, not only due to government reasons, e.g. budget
allocation; but also, because it represents a serious threat to evaluation purposes, e.g. the
exogeneity of the eligibility index. For the RD design, the manipulation of the forcing vari-
able increases the plausibility of the non-compliance of the assumed identification condition.
Therefore, it makes sense to first present the results of this falsification check, along with
others mentioned. Figure 4 illustrates the McCrary tests for the selected sample.
Figure 4:
McCrary manipulation test
Note:
with a binsize of 0,5 and a bandwidth of 3 the local density function
of the households according to their RSII eligibility index does not shows
significant differences around the threshold of 28,2.
This finely-gridded smoothed histogram shows that there is no apparent difference in
density around the threshold. In fact, with a log discontinuity estimator of 0,046 and a
standard error of 0,164, a t-test of the null hypothesis of continuity fails to reject. Likewise,
the Cattaneo et al. (2015) test fails to reject the null hypothesis of no systematic manipulation
(for all estimation methods) of the forcing variable on the base of t-statistics associated to
p-values
>
0,1, within the bandwidth (6,465; 8,895) (Table 3).
The underlying rationale in this case, where an index is reproduced, is that there is no
apparent reason for households to under-report their assets or lie about their socioeconomic
status to manipulate the RSII since the focus of the LSMS is not related to BDH reception
(this could represent more of an issue for the actual Registry). Complementarily, covariate
balance tests indicate no significant jump at the discontinuity. Figure 5 depicts Calonico
19