Dear specialists/ experts
I am very new for ARDL test and these days I am trying to analyse my data which are having different stationary levels. I have panel data. My all variables are in natural log transformation. My dependent variable and six independent variables are stationary at level and two IV are stationary at I(1). Since mixed stationary levels I thought ARDL test will be fitted with my data. Eviews does not support for my data panel and says "near singular matrix) when try to perform the test. Therefore I thought to go with STATA. I have few basic questions.
1. Since my data in natural log transformation there are negative values in my data series. Is it a problem to carry out ARDL with negative values?
2. I have time invariant proxy-( distance) is it a problem for my analysis?
3.Are there any limitation on number of variables that can be used in ARDL test ?
I have stucked with panel data ARDL in STATA and hope your assistance and guidance with appropriate STATA commands
Dear my friend,
I recomment you to read the principles of using STATA and then start to run ARDL by using internet resourses such as:
https://www.google.com/url?sa=t&source=web&rct=j&url=https://www.stata.com/meeting/chicago16/slides/chicago16_kripfganz.pdf&ved=2ahUKEwi0x-298aHjAhUIGewKHT1DCesQFjAEegQIAxAB&usg=AOvVaw3q5rfBgCF3Yo86GTiUOXiT
In regards to your question:
1-It is not a problem.
2-No
3-Yes.. it deoends on your time period and some preliminary tests.
1. Negative values have no issues but when you will transform into log it become unidentified or zero. So it will create problem as missing values.
2. time-invariant variable create problem as lag is taken in ARDL estimation and did not get lag values in time-invariant variable.
3. Depends on some tests.
like Zeenat pointed out, logarithm is defined only for positive numbers. If you apply log transformation to negative numbers it will return NA.
time variant proxies you can give as constants to solve that issue
Please clarify whether you are using a panel or time series data set
Zeenat Fatima Please explain your above point number one " Negative values have no issues but when you will transform into log it become unidentified or zero. So it will create problem as missing values. "
- Badges
- Science topic
- Similar topics
- Engineering
- Petroleum Engineering
- Oil