Partially clustered data

[Ioannis Siskos]

Dear Stata listers.

Being unable to find an answer into the Forum, I beg for your help on the following.
I have a panel dataset of macro factors for 30 countries for 15 years. I regress those macro-factors against my dependent variable which is in count form. Some of these countries belong to Free Trade Blocs, e.g NAFTA=US+Mexico+Canada, but some of them don't, e.g China.
Among independent variables I use country market size (mkt_size_country), but I also use (or at least I would like to) the market size of the rest of the countries in the Free Trade Bloc ((mkt_size_RFTB). The problem is that the countries that belong to a Free Trade Bloc have positive values for mkt_size_RFTB, whereas countries that do not belong to any Free Trade Bloc do not have mkt_size_RFTB.
My question is whether it is right to set mkt_size_RFTB equal to zero for the respective countries. I expect that this variable affects my count, so I want to have it in the model. If I leave it as missing data (no observation), these countries will be excluded from the analysis, which is not what I want.

As a desperate solution I was thinking is to try would be to use two different models, which will differ in the unit of analysis. The first one would be on country level and have mkt_size_country only and the second one to be on a Free Trade Bloc level and have mkt_size_FTB, i.e the market size of the Free Trade Bloc (for countries like China it would be the market size of the country).

However, I would be glad if someone could answer my first question and provide suggestions as what I can do in case it is not right to set mkt_size_RFTB equal to zero for the respective countries.
Thank you in advance.

[Andrew Musau]

Among independent variables I use country market size (mkt_size_country), but I also use (or at least I would like to) the market size of the rest of the countries in the Free Trade Bloc ((mkt_size_RFTB). The problem is that the countries that belong to a Free Trade Bloc have positive values for mkt_size_RFTB, whereas countries that do not belong to any Free Trade Bloc do not have mkt_size_RFTB.

Your variable is a proxy of the size of a country's free trade area. In this regard, countries that do not engage in free trade have a value of zero, so this is a true zero and not a missing value. These would be missing if, for example, such countries engage in free trade but you do not have information about bilateral or multilateral arrangements that are themselves not part of a free trade bloc. In the same way, one may have a trade openness variable on the right-hand side and this would be equal to zero for closed economies.

[Ioannis Siskos]

Dear Andrew,
Many thanks for your answer. That was also the idea behind, I was just not 100% certain whether it was right.