I am focusing on the econometric model research consistent with small-size samples and endogenous problems. My research is a total of 30 observations with time-series data when I carry out research to FDI. I am wondering about VAR, VECM, and ARDL models. It stated that VAR/ VECM is suited for large-size samples and Var is performed when all variables are stationary at the same level, I(0) or I(1), and VECM is both and only allowed to perform when there is the cointegration in the research model. ARDL model could implement with the small-size samples, and fix the endogenous problem in the research model...etc. Please, who can explain this to me more clearly? Thanks for your kind help.
I suggest you use the ARDL model. ARDL model is the best because it is the most powerful model for analysing data and is robust. You need to check whether the variables are stationary at the level or first difference. It is straightforward to do. Then you use the ARDL model. When analysing your data using the ARDL model, do not select the lag criteria yourself. Let the function (if you use E-view) select the lag criteria. That is what I did in my research article. Good luck.
Thanks for your kind reply. I am seeking your research articles relevant to study more.
You welcome. It is there in ResearchGate.
The Relationship Between Financial Patterns and Exogenous Variables: Empirical Evidence from Symmetric and Asymmetric ADRL.
ARDL is more efficient in smaller sized samples but your variables must be a combination of I(0) and I(1)
You should go for ARDL model. ARDL works best with small sample sizes and its can be used when variables are integrated of order 0, 1 or a combination of both. How care must be taken to ensure that there exits only a single cointegrating equation.
- Badges
- Science topic
- Similar topics
- Psychology