In this paper we study the robustness of the results found recently by Guzzini
and Palestrini (J Econ Interact Coord 11:35–55, 2016). Since the original analysis was
carried out in a static setting, we perform a dynamic panel analysis by using the same
dataset. The inclusion of the lagged value of the endogenous variable, missing in the
original paper, could be justified for several reasons. Firstly, the statistical relationship
may have itself a dynamical nature; secondly the inclusion of lagged-endogenous
variable is a way to mitigate the possibility of an omitted variable problem. We find that
the results are only qualitatively the same, and we discuss the quantitative differences.