Page 1 of 1

ARDL/ECM Cointegration equation and graphs

Posted: Fri Feb 05, 2021 1:18 pm
by hkarun
Respected Eviews users,
I am new to EViews and ARDL model specifications. I read Eviews blog and followed the instructions to run a test model.
I have two questions-
a naïve Q1- many practitioners suggest atleast 4 lags specification usually for a quarterly data while 12 for monthly series.
I have 5 regressors with a dummy as fixed regressor (monthly data). When I run the ARDL model I get an error message "SINGULAR MATRIX". If I reduce the lags size to 6/8, it works. My apologies but I could not understand the underlying theory/eviews application behind it. Can you be kind enough to guide me on this?

Q2. I need some guidance in understanding the cointegration relationship especially how to understand if "the cointegrating relationship is either of the usual kind, or is valid but degenerate".I ran a test model following the instructions. Attached are the results.
I tried using replacing one of variables (rtb_exante = REAL TBILLS) with other variables. However, I am getting more or less similar relationships as shown in the graph attached here. I am just trying to understand if I am committing any silly error in estimation process itself or could it be that my variables have this tendency/relationship amongst each other.

My apologies for a longer post but as I noticed you have been patient in answering to many queries. I thought it might be useful to ask Eviews users here and understand better.
Any guidance will be deeply appreciated.

Kind regards,