Matrix to Series (Eviews 7)

[mike_d]

Hi, I am trying to look at some statistics of a common sample of a few variables. What I have been trying to do is to first group the two variables of interest, then create a matrix that contains only common observations and finally convert back to a group in order to run a couple tests. I have been using the 'stom' and 'mtos' command but get a sample error. See below.

group gg s3 s9
stom(gg, mat)
mtos(mat, test_group)


The common sample contains 1645 observations, I also tried to enter the "smpl" command before mtos however this returned errors as well. Any direction would be greatly appreciated.

Thanks

Mike

[EViews Gareth]

Nothing wrong with the code per se. Don't quite follow what you're trying to do though.

[mike_d]

What I am trying to do is to deal with a problem I am having when running a cointegration test on pairs of variables. I am looping through many files, I run all tests on the open page 'i', copy output tables to results page, close the page 'i', move to next page 'i'. I was trying to look at the frequency tables of the two variables to be tested to determine if there are only a couple different values for either series (this is highly possible with my dataset), which can cause the error 'matrix near singular.' I have decided to try and approach this differently. I wrote a loop that checks the frequency table ensuring that no variable accounts for more than 50% of the observations. I used a counter, if the counter is less then 1 then the test is run, otherwise a blank table is created (I need tables for coint_test9 through coint_test14 for later on). However the blank tables for some reason are only being created on the first loop. I can paste my entire program if it helps, but basically this procedure loops through 246 files and I obtain the desired result for the first one only. Here is the portion of code I am referring to. Thanks.


scalar counter = 0
group g1 s3
'''johansen cointegration test
for !j = 9 to 14
!old_count = @errorcount
for !check = 9 to 20 step 1
if freq_s!j(!check,1) <> "Total" and freq_s!j(!check,3) > "50" then
counter = counter + 1
endif
next !check
if counter < 1 then
g1.add s!j
freeze(coint_test!j) g1.coint(s)
g1.drop s!j
endif
!new_count = @errorcount
if !old_count < !new_count then
table(16,10) coint_test!j
endif
counter = 0


next !j

[mike_d]

Sorry for wasting your time, I realized a much easier way to do this simply using @errorcount function.




setmaxerrs 6

for !j = 9 to 14

!old_count = @errorcount
g1.add s!j
freeze(coint_test!j) g1.coint(s)
g1.drop s!j

!new_count = @errorcount

if !new_count > !old_count then
table(16,8) coint_test!j
endif

next !j

clearerrs

[banhong86]

Hi Gareth, I have just started using EViews. Below are the codes from me to gather two vectors data into a matrix. I wanted to add up both the r2s and covariance from the same row, thus I opted to use @rsum, how it failed. Could you please give me some advice on this? Is there a way to add up the row in matrix or column in matrix? Thank you for your time.

matrix(10,2) m5
colplace(m5,r2s,1)
colplace(m5,covariance,2)
group g1
mtos(m5,g1)
series sum = @rsum(g1)
show sum

Regards,
Teh

[EViews Gareth]

Don't declare the group first:

Code: Select all

matrix(10,2) m5 colplace(m5,r2s,1) colplace(m5,covariance,2) mtos(m5,g1) series sum = @rsum(g1) show sum