SUR Coefficient Estimates

For technical questions regarding estimation of single equations, systems, VARs, Factor analysis and State Space Models in EViews. General econometric questions and advice should go in the Econometric Discussions forum.

Moderators: EViews Gareth, EViews Moderator

jti
Posts: 3
Joined: Fri Dec 12, 2008 10:45 am

SUR Coefficient Estimates

Postby jti » Sun Dec 14, 2008 9:19 am

Good day to all,

I have a question regarding seemingly unrelated regressions (SUR).

I am trying to estitmate a SUR-Model for some financial indexes. For this reason, I already derived VAR residuals from some conventional macroeconomic time serieses (e.g. Term-Spread) as inputs (innovations) for the SUR-Model.

The System looks like the following example:

emerging_markets = c(1) + c(2)*world_market + c(3)*term
frontier_markets = c(4) + c(2)*world_market + c(3)*term

After creating a system and estimating it via the E-View's SUR fuctionality, I get the estimation output, of course, which works perfectly.

However my first question is, are the corresponding t-statistics robust estimates (autocorrelation and heteroscedasticity and which method does e-views use, Newey/West?), so that the inferences are statistically ok? Checking the residuals for serial correlation (Box-Pierce-Ljung Test and Durbin Watson), reveals significant correlations. Unfortunately testing for normality in the residuals, does reject the null.

My second question is, how I will be able to extract the coefficient estimates for the individual equations?
e.g. a table like the following
c c(2) c(3)
emerging markets 0,04 0,65 0,45
t-stats (2,01) (4,56) (1,72)

frontier_markets -0,01 0,32 -0,02
t-stats (-1,82) (2,03) (1,51)

Many thanks in advance.

Kind Regards,

Jan

houda
Posts: 1
Joined: Wed Mar 04, 2009 6:23 am

Re: SUR Coefficient Estimates

Postby houda » Wed Mar 04, 2009 6:54 am

hello

pleaaseeeeeeeeeeee i need your help pleaaase
i want to know how to use SURE estimetion in eviews

startz
Non-normality and collinearity are NOT problems!
Posts: 3775
Joined: Wed Sep 17, 2008 2:25 pm

Re: SUR Coefficient Estimates

Postby startz » Wed Mar 04, 2009 7:44 am

jti wrote:Good day to all,

I have a question regarding seemingly unrelated regressions (SUR).

I am trying to estitmate a SUR-Model for some financial indexes. For this reason, I already derived VAR residuals from some conventional macroeconomic time serieses (e.g. Term-Spread) as inputs (innovations) for the SUR-Model.

The System looks like the following example:

emerging_markets = c(1) + c(2)*world_market + c(3)*term
frontier_markets = c(4) + c(2)*world_market + c(3)*term

After creating a system and estimating it via the E-View's SUR fuctionality, I get the estimation output, of course, which works perfectly.

However my first question is, are the corresponding t-statistics robust estimates (autocorrelation and heteroscedasticity and which method does e-views use, Newey/West?), so that the inferences are statistically ok? Checking the residuals for serial correlation (Box-Pierce-Ljung Test and Durbin Watson), reveals significant correlations. Unfortunately testing for normality in the residuals, does reject the null.

My second question is, how I will be able to extract the coefficient estimates for the individual equations?
e.g. a table like the following
c c(2) c(3)
emerging markets 0,04 0,65 0,45
t-stats (2,01) (4,56) (1,72)

frontier_markets -0,01 0,32 -0,02
t-stats (-1,82) (2,03) (1,51)

Many thanks in advance.

Kind Regards,

Jan


The standard errors correct for cross-equation correlation. They do not correct for heteroskedasticity or serial correlation.

Notice that in your example the same coefficient applies to world_market in both equations, but you show different estimates in your example table.

I don't believe EViews has a facility to format tables as you describe.


Return to “Estimation”

Who is online

Users browsing this forum: No registered users and 22 guests