observation number problem

[ryan_DETROIT]

hi all,
i am trying to complete my master thesis. I am trying to manage empirical analysis by using panel data analysis. My thesis topic is the relationship between economic indicators and emerging stock market performance. i select 5 emerging stock markets and 4 economic indicators. dependent variable is stock index. independent variables are inflation, interest rate, money supply, GDP. Data are yearly between 1993 and 2008.
i would like to analyze each emerging stock market performance separately. in each panel data analysis i have 60 observations (4*15). you think that the observation number is problem? is it too small to conduct panel data analysis?

many thanks.

[random_access]

hi there
i think the number should be 75 ......five countries and 15 years......anyway i do not think it is a problem if u run pooled data.......i guess it sould be enough.....alternatively u can run time series analysis for each country.
the choice is urs.
all the best

[ryan_DETROIT]

hi there,
thanks for your message.
i think that, i have 15 annual data for each each independent variable when i run panel data analysis for each country. some people think that it is not enough for panel data analysis. is it true?
i would like to run panel data analysis for each country. it is okay ?
many thanks...

[random_access]

the idea of panel data is to have both cross sectional and time series together.....if u run each country individually then it is time series......and i believe it is better than panel data analysis.
anyway it depends on ur research questions and previous work
all the best

[ryan_DETROIT]

ok i got your reply. many thanks for that.
I just wanna know when i use panel data, i need to use dummy variables for emerging countries that i selected for my thesis. so, may i know which method i have to use ( Randon Effects vs. Fixed Effects) when i run dummy variables ?
BTW you have any idea how to run Hausman test in EViews 5 ?
best regards...

[random_access]

hi there
based on my experience (i'm not 100% sure) if u include a dummy variable u cannot run random or fixed effect u only can run pooled regression.
not sure about hausman test in eviews 5 but i will check and let u know.
all the best

[ryan_DETROIT]

ok. again thanks for your mail. let me clear it. if there are one dependent variable and four independent variable for each country. also all kind of data are yearly. does anybody reject that your data volume are so small ? in other words, can anybody object that data are short ? to remind , my data are yearly and between 1993 and 2008.

it presents problem?
many thanks...

[random_access]

if u have a good justification why u started from 1993 then i guess there will be no problems. if the data is available and it does create an issue for ur work then u might need to make it 20 years or more but if u need the data to start from 1993 to 2008 for a specific reason then i guess there is no problem.
all the best

[ryan_DETROIT]

hi random_access,
I think that i have a good justification :) i hope that I can handle with that.
BTW thanks for your support...