Eduardo Fe

Estimation and inference in regression discontinuity designs with asymmetric kernels

We study the behaviour of the Wald estimator of causal effects in regression discontinuity design when local linear regression (LLR) methods are combined with an asymmetric gamma kernel. We show that the resulting statistic is no more complex to implement than existing methods, remains consistent at the usual non-parametric rate, and maintains an asymptotic normal distribution but, crucially, has bias and variance that do not depend on kernel-related constants. As a result, the new estimator is more efficient and yields more reliable inference. A limited Monte Carlo experiment is used to illustrate the efficiency gains. As a by product of the main discussion, we extend previous published work by establishing the asymptotic normality of the LLR estimator with a gamma kernel. Finally, the new method is used in a substantive application.

Pension Eligibility Rules and the Local Causal Effect of Retirement on Cognitive Functioning

We present a nonparametric bounds analysis of the local average treatment effect (LATE) of retirement on cognitive functioning. Partial identification methods are used to obtain identification regions for LATE and empirically assess the extent to which the critical exclusion restriction is satisfied. We find that retirement can lead to, at most, a drop of around 0.8 correct answer in a 20-question memory task. Importantly, we find strong evidence against the universal validity of pension eligibility criteria as sources of valid instrumental variables for retirement.

Exposure, Contemporaneousness and the Effect of Droughts on Children's Cognitive Development

In this paper, we combine a comprehensive Indonesian household survey with detailed meteorological data to explore what can be learned about the effect of drought on the long term cognitive development of Indonesian children living in rural areas. We face a common problem of latent exposure with observable contemporaneousness. The problem is compounded by plausible endogeneity and likely confounding. To estimate the effects of drought on Indonesian childrens scores in a fluid intelligence test, we consider a battery of different identification assumptions which vary in credibility and power. Our most powerful assumptions point identify the effect of contemporaneousness, however they have debatable credibility. Our most credible assumptions, on the other hand, convey little information about the effect of contemporaneousness. In between these two extreme, we consider a range of middle-of-the-way assumptions which partially identify the effect of contemporaneousness. Specifically, we characterise some of the assumptions required to establish the sign of the effect of contemporaneousness. Our results reveal differential effects of drought across sexes, however we find at least two competing explanations which would explain these difference: natural selection, on the one hand, and family dis-investing on girls in the fact of hardship, on the other.