Dear Stata-Users,
for my bachelor thesis i am conducting a cointegration analysis (kao, pedroni and westelund).
I have the following problem: Looking at online examples, for example one with only three variables (Panel-data cointegration tests | Stata), the analysis has 700 panels. I do not understand how and following which logic panels should b created. I think that in that example, the weekly date variables might have been used for the panels.
In my thesis, i have the inflation (dependent variable), GDP, unemployment and consumer confidence indices for 6 different countries. I thought that i could run a separate cointegration analysis for every country, checking for long-term cointegration. However, i cant find a way to structure my panels in a way that this is possible. The only way i found it to work was by choosing the different countries as panels, however this is not what i want to analyze.
Data is quarterly and the command xtset ID Date, quarterly works for the cointegration with all six countries simultaenously. However, if i want to do the cointegration for one country only, If i try to ID the values in accordance with the time variable, e.g. commanding gen ID = _n, and then choose ID as the Panel ID and date as the time variable, the error "no observations / r(2000)" is reported.
So the question is: How can i get more panels without seemingly having panels at all (as seen in the example)?
Thanks a lot for the help!
Best
David
for my bachelor thesis i am conducting a cointegration analysis (kao, pedroni and westelund).
I have the following problem: Looking at online examples, for example one with only three variables (Panel-data cointegration tests | Stata), the analysis has 700 panels. I do not understand how and following which logic panels should b created. I think that in that example, the weekly date variables might have been used for the panels.
In my thesis, i have the inflation (dependent variable), GDP, unemployment and consumer confidence indices for 6 different countries. I thought that i could run a separate cointegration analysis for every country, checking for long-term cointegration. However, i cant find a way to structure my panels in a way that this is possible. The only way i found it to work was by choosing the different countries as panels, however this is not what i want to analyze.
Data is quarterly and the command xtset ID Date, quarterly works for the cointegration with all six countries simultaenously. However, if i want to do the cointegration for one country only, If i try to ID the values in accordance with the time variable, e.g. commanding gen ID = _n, and then choose ID as the Panel ID and date as the time variable, the error "no observations / r(2000)" is reported.
So the question is: How can i get more panels without seemingly having panels at all (as seen in the example)?
Thanks a lot for the help!
Best
David
