Pool estimation

[jabro]

can somebody please help

My sample is as follows:
36 countries
1999 - 2007

whenever im trying to estimate my equation, eviews respond by saying insufficient observations, what might be the problem?

[startz]

can somebody please help

My sample is as follows:
36 countries
1999 - 2007

whenever im trying to estimate my equation, eviews respond by saying insufficient observations, what might be the problem?

First check to be sure that your data isn't full of NAs.
Second, you might post your equation here so people can see what you're trying to estimate.

[jabro]

i got right, thanks

what is the meaning of the following when you are estimating "near singular matrix"

[startz]

i got right, thanks

what is the meaning of the following when you are estimating "near singular matrix"

It means there is near linear dependence among your independent variables, near perfect multicollinearity. In practice, it usually means you have accidentally specified a redundant variable.

[jabro]

thanks startz

what is causing the near singular matrix is my dummy variable. will you please be so kind and help me as to how i should generate my equation and dummy variable

what i am trying to estimate is the relationship between telecommunication perforemance (dependant variabl) and investment in the sector, competition in the sector with the value of 1 if the sector is open for competition (a value of 0 if the sector is not open for competition), and the regulation in the sector (value of 1 if the regulator is autonomous and value of 0 if the regulator is not autonomous). meaning that competition and regulator is represented by a dummy variable.

as i said my sample is 46 countries, period 1996 to 2006

Yit = q1 + q2Dit + q3Dit + B2X2it + uit
fixed telephone line per 100 subcribers = intercept + dummy variable for competition (varies in time and individual) + dummy variable for regulation (varies in time and individual) + Fixed investment in the sector

what is the best in this case fixed or random effect

[startz]

thanks startz

what is causing the near singular matrix is my dummy variable. will you please be so kind and help me as to how i should generate my equation and dummy variable

what i am trying to estimate is the relationship between telecommunication perforemance (dependant variabl) and investment in the sector, competition in the sector with the value of 1 if the sector is open for competition (a value of 0 if the sector is not open for competition), and the regulation in the sector (value of 1 if the regulator is autonomous and value of 0 if the regulator is not autonomous). meaning that competition and regulator is represented by a dummy variable.

as i said my sample is 46 countries, period 1996 to 2006

Yit = q1 + q2Dit + q3Dit + B2X2it + uit
fixed telephone line per 100 subcribers = intercept + dummy variable for competition (varies in time and individual) + dummy variable for regulation (varies in time and individual) + Fixed investment in the sector

what is the best in this case fixed or random effect

Is it true that one of the dummy variables is always 1 when the other is 0? If so, you are in the "dummy variable trap," which would explain your problem.

[jabro]

i am confused. to put it in perspective lets say i have three countries 2 of them are open for competition and the other not. does the latter not imply that the dummy variable for the two will be one and the other zero.

[startz]

i am confused. to put it in perspective lets say i have three countries 2 of them are open for competition and the other not. does the latter not imply that the dummy variable for the two will be one and the other zero.

Just look at your data. If for every observation either D1=1 and D2=0 or D1=0 and D2=1, then you have fallen into the dummy variable trap.

[jabro]

how do i insert time variant and non time variant variable in eviews

[jabro]

can someby please help me to interpret the following:

Dependent Variable: GDP?
Variable Coefficient Std. Error t-Statistic Prob.

C 5.92921 0.470135 12.61173 0
TIT? 0.009881 0.018972 0.520837 0.6035
TTS? 0.141849 0.022275 6.368211 0


Weighted Statistics

R-squared 0.671108 Mean dependent var 0.609641
Adjusted R-squared 0.665534 S.D. dependent var 0.222395
S.E. of regression 0.128618 S um squared resid 1.952029
F-statistic 120.3902 Durbin-Watson stat 0.77652
Prob(F-statistic) 0

[probettly]

Hi all,
I am also having a problem with near perfect multicolliniearity in my panel data regression.
To make a story short, I am estimating influace of labour migration on unemployment rates in local regions.
When I run both entity and time fixed effects regression model, I get message with: "near singular matrix".
It is obvious for me that minimal wage that is constant between regions but varies between time periods causes near perfect multicollinearity.
My question is: how can I avoid the problem, and run both entity and time fixed effects regression model with my minimal wage variable?
Is it possible at all?
Thanks in advance!

[startz]

You can't have fixed effects and a variable which is constant within a group. It doesn't make any sense, which is what EViews is telling you. You probably want to drop the minimum wage variable.

[probettly]

Thx for fast reply,
I understand that fixed effect regression will not work with minimal wage variable.
I wonder if it's possible to make time fixed effect for all variables except minimal wage and entity effect for all variables at the same time.
And if not (what I assume) than what would you suggest:
1) dropp minimal wage > get less significant variables but both entity and time fixed effects regression
or
2) keep minimal wage > get more significant variables but only cross-section fixed effects?
(in both cases R squered adjusted is preety much the same around 0.80)
Best,
Anna

[morad]

hi,
i have also same problem. I am working with pool data, it consists of 8 cross-sections(8 companies) over 738 days using SUR.
i see the following error"near singular matrix" when i use an only dummy variable of my model ( d1=1 in only 6 days and d1=0 for other days) as cross section spesific coefficients.
it is important for me to have their spesific coefficients through fixed effects.
i would be grateful if somebody helps me know how to solve it.
thanks