F-Stat or Chi Squared?

[michaelk]

(I am using e-views for this question).

When testing for serial correlation, heteroskedasticity and the wald test during a VECM or VAR or ARDL, I have the option to use either the F-stat or Chi-squared to test for significance.

My data is annual, with 36 observations (1980-2015).

Which test is more appropriate for my number of observations? The F-stat or Chi-squared?

Why?

Can you provide any literature to support your answer?

I have examples of using Chi-squared (with LM test) within the following videos with a small sample size (10)

https://www.youtube.com/watch?v=sZ-J_KM9TvQ

https://www.youtube.com/watch?v=LKgP6A9mO44


Thank you