SUR regression (sureg) for comparing variation accross countries not working - "no observations r(2000)"

[Laurence Jones]

Hi, I am trying to estimate a simply model for multiple countries using seemingly unrelated regressions.

I use the command:

sureg (y_country1 x_var) (y_country2 x_var)

Where y_country1 is y where the country = 1 and . otherwise, and so on for country 2.
However, when I run the sureg, I get the r(2000) error "no observations". But the data is there when I check and I can run the regressions separately without issue. Is this because there is no crossover between countries? is this a condition?

Am I able to run my model like this, I am loosely following the method from https://onlinelibrary.wiley.com/doi/....1002/ijfe.443, who use this approach.

Many thanks

[William Lisowski]

You are trying to use sureg to estimate two different equations - that's fine - on two different populations - not so fine.

Very often a simple description really isn't clear without more detail, or at a minimum it is too difficult to guess at a good answer from what has been shared. When that happens other members will decide not to answer the question, or ask for an improved presentation that could have been provided to begin with.

Please help us help you. Show example data. The Statalist FAQ provides advice on effectively posing your questions, posting data, and sharing Stata output.

You will find that the time spend preparing a well-formed question will often pay off in a quicker answer that requires less followup.

[Laurence Jones]

Many thanks for the response. Just to clarify, SUR cannot be used on a population split by time or entity? It requires observational crossover between the 2 or more specified equations?

Many thanks

[William Lisowski]

My first paragraph in post #2 was a comment on your organization of your data, not necessarily on the nature of the data itself. It is not the definitive statement you infer in post #3.

Focus on the final three paragraphs of post #2. Even the best descriptions of data – of which yours is not – are no substitute for an actual example of the data. There are many ways your data might be organized that are consistent with your description, and each would imply a somewhat different approach.