dplyr lag function returns NAs

Question

Does anybody have an explanation for such result using dplyr package?

I have a data.frame df

    library(dplyr)
    df = d