Q1: Is there a relationship between the optimal (and maximum) lag lengths of a series and its differences (1st, 2nd etc.) in ADF test for finding stationarity?
To clarify further the problem:
In Eviews (7.2), when ADF test is done to find the (non-)stationarity of a series, (View-Unit Roots - ADF);
"Test for unit root in: LEVEL"; "Include in Test Eq: Trend and Intercept"; "Lag Length: Auto selection SIC max: 17"
Q2: In the beginning of ADF, Eviews offering DEFAULT max length of 17 according to which info?"
(By the way, as a side info, Eviews is offering the same max default length for the series and its 1st and 2nd difference.)
Q3: Assume that ADF done and optimal lag length for a series is found to be 3. Does this "3" have a say (or restricts) in optimal lag length of the 1st difference of the series?
Assume that ADF done and optimal lag length for a series is found to be 3. Does this "3" have a say (or restricts) in optimal lag length of the 2nd difference of the series?
Can one say the ADF test for 1st and 2nd difference of a series with max length "BIGGER THAN 3" (given opt lag length for series:3) has NO MEANING?
If anybody knows smt about the above stated troubles, I would be very glad if a reference is shown. THANKS IN ADVANCE.
Relation btw optimal lag lengths of a series in ADF test
Moderators: EViews Gareth, EViews Moderator
-
- Posts: 8
- Joined: Wed Jan 23, 2013 8:22 am
-
- Posts: 8
- Joined: Wed Jan 23, 2013 8:22 am
Re: Relation btw optimal lag lengths of a series in ADF test
To prevent the misunderstanding, I am NOT asking the relation btw OPTIMAL and MAX lag length of a series.
I am asking:
1. The relation btw OPTIMAL lag length of a series and OPTIMAL lag lengths of the differences (1st, 2nd etc.) of the series.
2. Also, I am asking: the relation btw DEFAULT MAX lag length of a series and DEFAULT MAX lag lengths of the differences (1st, 2nd etc.) of the series dictated by Eviews.
I am asking:
1. The relation btw OPTIMAL lag length of a series and OPTIMAL lag lengths of the differences (1st, 2nd etc.) of the series.
2. Also, I am asking: the relation btw DEFAULT MAX lag length of a series and DEFAULT MAX lag lengths of the differences (1st, 2nd etc.) of the series dictated by Eviews.
-
- Posts: 8
- Joined: Wed Jan 23, 2013 8:22 am
Re: Relation btw optimal lag lengths of a series in ADF test
In my example the # of obs: 423 (That is, famous arabica robusta coffee example for the application of Toda-Yamamoto Granger Causality)
Band Width: m≡⟦12(T/100)^(1⁄4) ⟧=⟦12(423/100)^(1⁄4) ⟧=⟦17,20⟧=17
ADF keeps this default max lag length THE SAME for the series and its 1st,2nd, etc differences.
One of the above troubles I think cleared away by myself. But the 3RD Question is IMPORTANT. Any idea?
Band Width: m≡⟦12(T/100)^(1⁄4) ⟧=⟦12(423/100)^(1⁄4) ⟧=⟦17,20⟧=17
ADF keeps this default max lag length THE SAME for the series and its 1st,2nd, etc differences.
One of the above troubles I think cleared away by myself. But the 3RD Question is IMPORTANT. Any idea?
Re: Relation btw optimal lag lengths of a series in ADF test
Hi, when i set leg structure ---- lag lenght criteria and put 6 lage it says insuffient no. of observations so i put 5 legs then 3 criteria (A/C, SC, HQ) suggest 5 lag lengh but when i find cointegration and put 5 lag it does not take and say "Near Singular Matrix". any body help?
Re: Relation btw optimal lag lengths of a series in ADF test
Hi, i have 4 variables. i select all 4 variables and open it as a group. when i set leg structure ---- lag length criteria and put 6 lags it says insufficient no. of observations so i put 5 legs then 3 criteria (A/C, SC, HQ) suggest 5 lag length but when in finding co-integration i put 5 lags then it says "Near Singular Matrix". any body help? to find a lag length criteria, should i use all 4 variables or just one variable?
Return to “Econometric Discussions”
Who is online
Users browsing this forum: No registered users and 30 guests