Hello everybody,

I am currently writing my bachelor thesis on the effect of the Euro-introduction on trade flows between different UK NUTS1 regions and Euro-member countries. I am therefore using the gravity model looking at trade data between 1997-2004 (RTS data) with dummies for the exchange rate fixing in 1999 and the introduction of the Euro as a physical currency in 2002. In my first analysis I include trade flows between the UK regions and the 11 Euro-area countries, while in my second model I include all EU15 countries. This means that in my first model both my currency dummies will be time-variant, while in the second model for some countries they will be time invariant. Running the Hausman-test in both of my models, shows that I have to use fixed effects, which is a problem in my second model as the currency dummies are time-invariant for some countries. My supervisor has suggested that I use the MUNDLAK approach in this case, however I have also heard about using poisson pseudo maximum likelihood with comments such as "xtpoisson" (even though I am not very familiar with it)

I therefore have a couple of questions:
1. Can I use "xtpoisson" for my concrete case ?
2. Which of the two methods would you recommend?
3. I couldn´t find a lot of papers using Mundlak in a gravity model context (maybe I was searching wrong). Do you know of some papers with similiar problems as mine using it?

Thanks a lot for the help!