Stata code to reconstruct gravity equation with CEPII data

[Melanie Kuehnast]

Hi everybody,

I am new here and also not very good with stata. For a project, I have to reconstruct the gravity equation constructed by Gudgin et al. (2017) to evaluate the effect of Brexit on trade flows. To do this I have downloaded data from CEPII. As in the paper by Gudgin et al. I have trimmed the data down to Britain's top 60 export markets using the code:

 keep if iso3num_d==840| iso3num_d==276 |

. The code is not complete because I have done this 60 times (iso3num being the variable that identifies a given country) but just so you get an idea of what I am doing. Then in order to construct the following regression: trade flows = constant + log(GDP_home) + log(GDP_dest) + log(GDP_home/POP_home) + log(GDP_dest/POP_dest) + UK_EU + EU_both + EU_home + EU_dest + FTA_both + NOR_EU + SWZ_EU

gen lgdp_o=ln(gdp_o)

gen lgdp_d=ln(gdp_d)

gen lgpdpop_o=ln(gdp_o/pop_o)

gen lgpdpop_d=ln(gdp_d/pop_d)

gen uk_eu=1 if iso3num_o==826 & eu_d==1
replace uk_eu=0 if iso3num_o==826 & eu_d==0

gen eu_both=1 if eu_d==1 & eu_o==1
replace eu_both=0 if eu_d==0 | eu_o==0

gen nor_eu=1 if iso3num_o==578 & eu_d==1
replace nor_eu=0 if iso3num_o==578 & eu_d==0

gen swz_eu=1 if iso3num_o==756 & eu_d==1
replace swz_eu=0 if iso3num_o==756 & eu_d==0

gen ltrade=ln(tradeflow_baci)

egen countrypair=group(iso3num_o iso3num_d) 
duplicates drop countrypair year, force
xtset countrypair year 

xtreg ltrade lgdp_o lgdp_d lgpdpop_o lgpdpop_d eu_o eu_d eu_both fta_wto uk_eu nor_eu swz_eu, fe 

estimates store fixed

after the code xtreg ... stata tells me that there are no observations

I do not know what I did wrong and why I do not obtain a regression output. Could you help me find the error and tell me how I can reconstruct the equation? Thank you in advance.

I hope all of this makes sense