EViews.com

Hi all,

I just ran a VEC with 2 endogenous and 5 exogenous variables. Now I wonder how I should interpret the numbers for my exogenous variables (not the standard error or t-statistics)? What does the minus sign stand for? How big should the number be to be considered significant? Below are the results. Thank you in advance!

Vector Error Correction Estimates
Date: 06/03/09 Time: 10:27
Sample (adjusted): 1994 2007
Included observations: 14 after adjustments
Standard errors in ( ) & t-statistics in [ ]


Cointegrating Eq: CointEq1


GDPGROWTH(-1) 1.000000

DEXPGDP(-1) -1.682983
(0.22172)
[-7.59050]

C 0.174989


Error Correction: D(GDPGROWTH) D(DEXPGDP)


CointEq1 0.197145 1.053958
(0.28769) (0.19525)
[ 0.68526] [ 5.39793]

C -0.097634 0.160050
(0.39213) (0.26613)
[-0.24899] [ 0.60140]

EXRATE 0.006276 -0.005289
(0.00584) (0.00396)
[ 1.07489] [-1.33461]

HDI 0.302065 -0.085976
(0.56237) (0.38167)
[ 0.53713] [-0.22526]

INFLATION 0.084941 0.671529
(0.45613) (0.30957)
[ 0.18622] [ 2.16925]

INVESTMENTS -0.640612 -0.635130
(0.29657) (0.20128)
[-2.16008] [-3.15552]

SAVINGS 0.088283 0.185195
(0.08645) (0.05867)
[ 1.02122] [ 3.15649]


R-squared 0.536183 0.828988
Adj. R-squared 0.138626 0.682406
Sum sq. resids 0.002865 0.001320
S.E. equation 0.020230 0.013730
F-statistic 1.348696 5.655452
Log likelihood 39.59491 45.02132
Akaike AIC -4.656416 -5.431618
Schwarz SC -4.336887 -5.112089
Mean dependent 0.003207 -0.000614
S.D. dependent 0.021798 0.024363


Determinant resid covariance (dof adj.) 4.01E-08
Determinant resid covariance 1.00E-08
Log likelihood 89.20401
Akaike information criterion -10.45772
Schwarz criterion -9.727364

You have to consider "standard error or t-statistics" if you are really interested in the significance. Assuming that you have basic knowledge on VEC models, the following link might be helpful: viewtopic.php?f=18&t=855. Moreover, your sample is too small compared to size of your model.

Dear Jopo,

What trubador said is true. Your estimation sample size is small and not suitable for cases of 2 endogenous and 5 exogenous variables. Regarding the interpretation of exogenous variables, it is similar to the interpretation of normal regression result, but we only emphasise that this is the short run relationship. “How big should the number to be considered significant?” This is depends on the significant level you choose, you should refer to the student t-table Critical values to make a decision.

Thank you,

Regards,
CF Tang

trubador wrote:You have to consider "standard error or t-statistics" if you are really interested in the significance. Assuming that you have basic knowledge on VEC models, the following link might be helpful: viewtopic.php?f=18&t=855. Moreover, your sample is too small compared to size of your model.

hi trubador - if the t stat is less than the standard error does this imply the coeeficent in the VEC is significant or insignificant?