Hello,
I have a panel dataset of bilateral trade disaggregated at the HS4 sector level and I would like to estimate a theoretically consistent gravity model using this data. I am using the ppmlhdfe command.
I would like to include exporter-year-sector and importer-year-sector fixed effects to account for the multilateral resistance term. Using this simple dataset with only two-sector(categories), I am doing the following:


Code:
use "http://fmwww.bc.edu/RePEc/bocode/e/EXAMPLE_TRADE_FTA_DATA" if category!="TOTAL",clear

egen imp = group(isoimp)

egen exp = group(isoexp)

egen cate = group(category)

eststo: ppmlhdfe trade fta, absorb(imp#year#cate exp#year#cate imp#exp) cluster(imp exp)
I have a few questions:
1) Is this the correct way to specify exporter-year-sector and importer-year-sector fixed effects in the command?
2) I am including dyadic fixed (imp#exp) effects to account for the potential endogeneity of the policy variable (Baier and Bergstrand 2007). If my policy variable of interest varies at the sector level, would it be more appropriate to include dyadic-sector fixed effects (imp#exp#cate)?
3) I am currently clustering errors at the dyad level, should I cluster them also considering the sectoral level cluster(imp exp cate)?


Thanks a lot in advance for all the help