I am new to econometrics and am trying to estimate the effect of receiving a conditional cash transfer (CCT) in a fuzzy RDD. The threshold for receiving the CCT changes over the years, which is why I normalized it at 0. The running variable is household income per capita. Treatment is assigned with a binary variable (1 if treated, 0 if not treated) I plotted the probability of treatment. There is a discontinuity, but it is rather small. plot

This is mostly due to the fact that people may receive the cash transfer for 2 consecutive years before they are being re-evaulated. Thus, they might exceed the threshold household income per capita and yet still receive the transfer. Also, they can receive the transfer if they remain under the minimum income. This makes me question the strength of using treatment as an instrumental variable, as foreseen in fuzzy RDD.

Since the instrumental variable will not consider non-compliers, I wonder if this is even an issue at all. At the same time, the first-stage regression seems to suggest that the instrument is suitable. firststage regression

What should I make of these findings? Is this example not suitable for RDD?



Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Browse other questions tagged or ask your own question.