NSM error and chrashing

[codru]

Hi all,
I am trying to estimate a structural VAR for the 50 states using this specification (SUR and about 3600 data points over 35 years):

xak-wsak(-1)=c(1)*(xak(-1)-c(6)-c(2)*yak(-1))
yak-yak(-1)=c(11)*(xak(-1)-c(6)-c(12)*yak(-1))

xal-wsal(-1)=c(1)*(xal(-1)-c(6)-c(2)*yal(-1))
yal-yal(-1)=c(11)*(xal(-1)-c(6)-c(12)*yal(-1))

xar-xar(-1)=c(1)*(xar(-1)-c(6)-c(2)*yar(-1))
yar-yar(-1)=c(11)*(xsar(-1)-c(6)-c(12)*yar(-1))
.
.
.
When I include all states Eviews crashes. If I run the system with only 48 states I get Near Singular Matrix error but Eviews does not crash. If I take 20 states out Eviews completes the estimations and provides estimates with impressive significance but the signs points out to potential instability. My question is: Do you think the errors/crashes are related to a potential instability issue or are these two separate issues I have to address. Any help would be much appreciated. thanks.

[EViews Gareth]

Can you provide the version and build date of your copy of EViews (Help->About EViews)

[codru]

Eviews 8 (64 bit) build date=Dec 19, 2013
thank you.

[EViews Gareth]

Could you provide your workfile and steps to replicate the crash/bad behaviour?

[EViews Gareth]

Crash is caused by a bug that crashes EViews for some singular matrix errors on multi-threaded computers in systems. We'll fix the crash, although you still won't be able to estimate, due to the singular matrix.

[codru]

Thank you for your response. Would the patch be available through automatic update or do we have to download it from the site?

[EViews Gareth]

Automatic.