Hi, I am trying to conduct a fixed effects difference in difference model to estimate the effect of Spanish labour reform on its FDI. I am using a dataset of 14 western European countries, including spain, from 2000 to 2017, with the reform taking place in 2012.
My stata model is as follows:

xtreg netfdiinflow reformspain reform_others tradegdp inflation humancapital xrate investmentratio realgdppercapita corptax i.year, fe vce(bootstrap, reps(500) bca seed(1))


My questions are:
1. As my dataset includes countries which have undergone labour reform at different points in time, I include a reform_others variable to control for their own labour reforms. I do this because labour reform is not a continuous variable and so i can only really work with dummies in this difference in difference model. Is this okay to do?
2. I saw some clustering when I plotted the squared residuals of my dependent variable so I undertook a Breusch-Pagen test and found there was heteroscedasticity. The vce(robust)/vce(cluster clustervar) options lead to missing F-values and so i decided to bootstrap - is my bootstrapped standard error command correct as I don't fully understand what is the sufficient number of reps to use and what seed value to use? I ask this because, i ran the above regression and i get missing chi2 values for the Wald test and I don't understand why.
3. I am wondering if i should put in a time trend for each country to account for the fact that the control and treated countries did not have parallel trends before 2012 (so goes against the identifying assumption of diff in diff), but I am aware that this may over parametrise the model but is it still a worthwhile endevour given i already have time fixed effects?