Number of Observations in ADF and PP Unit Root Test
Posted: Tue Aug 17, 2010 6:12 am
Hi! Let me tell you what I've been working on first. I'm currently running the UR test for MSCI INDEX for 7 countries (Indonesia, Malaysia, Thailand, Philippines, Singapore, Japan and USA), monthly data from 1997-1998 (2 years). I try to use the three criterions (AIC, SIC, and HQC). I notice that EViews will try to give the maximum lags value before I run the test for each series. This time, the system give "5 lags" as the value. I got the results from both tests, but most of the series will also have 5 as the lag lengths as well (which is the maximum). Then, when I start to change the maximum value of lags, the lag length for each series (on the results page) will keep on changing if I put 6, 7, 8.. 10, both with and without trend. And if I put more than 10, the system will say "insufficient number of observations".
What I'm trying to ask is: do 24 observations are considered "too few" to run UR test for both ADF and PP? how many observations amount should I have to at least run the unit root test? I wanted to do Johansen's cointegration test and granger causality after the UR and so I need to make sure that the series are stationary in I(1) for that period of time. If these UR tests are not the appropriate ones for my case, can you guys recommend other UR tests as well?
Thank you guys for your help. I really appreciate it.
What I'm trying to ask is: do 24 observations are considered "too few" to run UR test for both ADF and PP? how many observations amount should I have to at least run the unit root test? I wanted to do Johansen's cointegration test and granger causality after the UR and so I need to make sure that the series are stationary in I(1) for that period of time. If these UR tests are not the appropriate ones for my case, can you guys recommend other UR tests as well?
Thank you guys for your help. I really appreciate it.