I have problem with the dummy variables.

[chun]

Hello, I'm doing my thesis to see weather political instabilities affect FDI and I have some questions. I was gathered data from 31 countries between 1980 and 2008. I set FDI as countrol variable and there are five political dummy variables.
polity2, 1 means a country is autocracy and 0 means democracy.
inttot, 1 means there are interstate conflicts and o means there are no interstate conflicts.
civtot,1 means there are intrastate conflicts and o means there are no intrastate conflicts.
rights,1 means there are no freedom in the electoral process and o means there are freedom in the electoral process.
civil, 1 means there are no freedom in expression and o means there are freedom in expression.
I type down @expand(polity2,civtot,inttot, rights, civil) in the common coefficients box and this is the result I got.

Dependent Variable: FDI?
Method: Pooled Least Squares
Date: 01/13/12 Time: 12:34
Sample: 1980 2008
Included observations: 29
Cross-sections included: 30
Total pool (unbalanced) observations: 541
Cross sections without valid observations dropped

Variable Coefficient Std. Error t-Statistic Prob.

DEPR? 0.000239 0.002096 0.114142 0.9092
RGDPPC? 0.000299 0.000467 0.641920 0.5212
GDPPCLAG1? 5.52E-05 0.000125 0.442889 0.6580
INFL? -0.003827 0.004771 -0.802247 0.4228
TRADE? -0.138881 0.042912 -3.236377 0.0013
AID? 2.90E-09 6.92E-10 4.195268 0.0000
EDEBT? 0.000529 3.95E-05 13.39470 0.0000
EXR? -1.62E-10 5.95E-10 -0.272157 0.7856
POLITY2?=0,INTTOT?=0,CIVTOT?=0,RIGHTS?=0,CIVIL?=0 5.585604 1.125078 4.964636 0.0000
POLITY2?=1,INTTOT?=0,CIVTOT?=0,RIGHTS?=0,CIVIL?=0 -0.580929 5.925297 -0.098042 0.9219
POLITY2?=1,INTTOT?=0,CIVTOT?=0,RIGHTS?=1,CIVIL?=1 4.950566 1.446125 3.423332 0.0007

R-squared 0.305343 Mean dependent var 5.744005
Adjusted R-squared 0.292236 S.D. dependent var 13.66466
S.E. of regression 11.49590 Akaike info criterion 7.741981
Sum squared resid 70042.55 Schwarz criterion 7.829278
Log likelihood -2083.206 Hannan-Quinn criter. 7.776120
Durbin-Watson stat 0.563631

My question are;
1. Why are there only three combinations of the dummy set? I think it would be 5! like
POLITY2?=0,INTTOT?=0,CIVTOT?=0,RIGHTS?=0,CIVIL?=0
POLITY2?=0,INTTOT?=1,CIVTOT?=0,RIGHTS?=0,CIVIL?=0
POLITY2?=0,INTTOT?=1,CIVTOT?=1,RIGHTS?=0,CIVIL?=0
POLITY2?=0,INTTOT?=1,CIVTOT?=1,RIGHTS?=1,CIVIL?=0
and so on. Where are the other combinations?
2. Which political dummy variables does the coefficient belong to? 5.585 belongs to polity2 or the others.
3. What does the coefficient of the dummy variables mean?
Thank you so much, Chun

[EViews Gareth]

You should enter each dummy variable inside its own @expand statement:

Code: Select all

@expand(polity2) @expand(civtot) @expand(inttot) ...

[chun]

I did it but the error message, " Near singular matrix", showed up.

[EViews Gareth]

Dummy variable trap

[purishiru]

my problem almost similar, although when i want to type @expand(mode), the box saying mode is not define appear.
what is wrong?

[EViews Gareth]

Do you have a variable called mode?

[purishiru]

yes, it suppose to be a dummy consisting of exporting and joint venture as mode

[startz]

Post the exact command you used, and maybe your workfile.

[purishiru]

i posted two of my workfile.
the thesis i am doing is using about 80 firms, divided further into 6 industry, of which sample is multinational firm. here in the workfile are mining (8 firm) and basic industry(20 firm). the purpose is i am doing research if the difference in entry mode impacts return.

I am using a pool (eviews 6) and whenever i put the independent variable in the common coefficient pool estimation, it wont work if either EXP (export) or JV (joint venture) is included, there goes "near singular matrix". so i resource to use @expand there. it shows JV not define.
If I change the excel into a sheet containing mode (where all export is 0 and all joint venture is 1) with the same procedure, is "Mode not define"

my first problem is the dummy. i need dummy for represent 3 modes: export, joint venture and FDI. as the dummy should be 0 and 1. e.g. the firm is export firm, than the dummy for export is 1; for joint venture is 0 ,- and vice versa, while FDI is represent by intercept. ( dummy variable trap is because c= d1+ d2 ? or am i wrong?)

i have tried basically almost every example posted in the forum. it is either i have done it wrong or maybe i have input it incorrectly. it shows either "variable (ROE, or JV, or MODE) not define", or "near singular matrix" or "Syntax error"

Hope all the info can help to find the best solution and thank you in advance!!

[startz]

The workfiles you posted do not have a series named "mode."

[purishiru]

Yes, those are two sample workfile where i try to separate the mode into two variable of jv and export
here is one of the workfile where the both joined as mode

[startz]

You need to show us exactly what you typed and/or include the equation in the workfile.

[purishiru]

i used pool. so when the i click on estimate, i write the dependent (ROA) and the independents in the common coefficient space( c der? size? cpi? gdp? mode?)
i choose on balance panel and fixed effect, it gives near singular matrix

if i put on the common coefficient "c der? size? cpi? gdp? @expand (mode)? there, mode not define

[startz]

i used pool. so when the i click on estimate, i write the dependent (ROA) and the independents in the common coefficient space( c der? size? cpi? gdp? mode?)
i choose on balance panel and fixed effect, it gives near singular matrix

if i put on the common coefficient "c der? size? cpi? gdp? @expand (mode)? there, mode not define

mode is a dummy variable. I suspect the problem is that you have fallen into the dummy variable trap. You can't have fixed effects and a variable that is a constant for each cross section.

[purishiru]

if i delete c from the estimation coefficient, that would not work too. what should be done?