Page 2 of 2
Re: serial correlation & heteroskedasticity
Posted: Mon Jun 10, 2013 10:49 am
by startz
Yes, I think so. But maybe someone who does this more often than I do should chime in.
Re: serial correlation & heteroskedasticity
Posted: Wed Jun 12, 2013 10:07 am
by Cabinho
Your variables probably have unit roots. This may account for the serial correlation. I don't see a way for you to eliminate the serial correlation.
Startz, if I have heteroskedasticity (White test through ols) and want to run a tobit and logistic regression with the same model and still have heteroskedasticity after logging my data, what can I do to solve this?
Re: serial correlation & heteroskedasticity
Posted: Wed Jun 12, 2013 10:17 am
by startz
Your variables probably have unit roots. This may account for the serial correlation. I don't see a way for you to eliminate the serial correlation.
Startz, if I have heteroskedasticity (White test through ols) and want to run a tobit and logistic regression with the same model and still have heteroskedasticity after logging my data, what can I do to solve this?
Look under the Estimate/Options tab in your equation window.
Re: serial correlation & heteroskedasticity
Posted: Thu Jun 13, 2013 9:16 am
by Cabinho
Your variables probably have unit roots. This may account for the serial correlation. I don't see a way for you to eliminate the serial correlation.
Startz, if I have heteroskedasticity (White test through ols) and want to run a tobit and logistic regression with the same model and still have heteroskedasticity after logging my data, what can I do to solve this?
Look under the Estimate/Options tab in your equation window.
Should I run tobit with robust covariance Hubert/White?
Re: serial correlation & heteroskedasticity
Posted: Thu Jun 13, 2013 9:18 am
by startz
That seems like the right thing to do.
Re: serial correlation & heteroskedasticity
Posted: Thu Jun 13, 2013 9:20 am
by Cabinho
Thank you so much Startz!
Re: serial correlation & heteroskedasticity
Posted: Thu Dec 20, 2018 2:45 am
by tejdevkota
Hi ! everybody!
could anyone help me please? I am completely new in eviews.
I conducted ARDL bound test and got confirmed that there is cointegration. the value of calculated F test statistics is higher than the pro. value. Am I right?( its done in eviews 10) after that checked error correction form where I got the result as copied here. I have conducted this with dep. variable rgdpr and indep . variables ms2r, mcr, tor ndx and ncl . the time series data for 24 yrs. the ecm result is as .
ECM Regression
Case 2: Restricted Constant and No Trend
Variable Coefficient Std. Error t-Statistic Prob.
D(LNTOR) -0.192141 0.090229 -2.129484 0.0621
D(LNTOR(-1)) 0.733329 0.124959 5.868556 0.0002
D(LNNDX) 0.916016 0.175514 5.219051 0.0005
D(LNNDX(-1)) 2.352864 0.191734 12.27149 0.0000
D(LNNLC) 11.48273 0.680427 16.87576 0.0000
D(LNNLC(-1)) 8.741721 1.073612 8.142349 0.0000
CointEq(-1)* -1.890049 0.101288 -18.66021 0.0000
R-squared 0.982988 Mean dependent var 0.034643
Adjusted R-squared 0.976183 S.D. dependent var 1.460566
S.E. of regression 0.225407 Akaike info criterion 0.111554
Sum squared resid 0.762126 Schwarz criterion 0.458704
Log likelihood 5.772901 Hannan-Quinn criter. 0.193333
Durbin-Watson stat 2.432171
why are my other variables missing in the above result? anyone please suggest me.