create series

[martingale]

Hi

I use EViews 7.

I have tried without success to write a little program which creates series for standardized unexpected earnings (SUE). I have quarterly data for earnings per share(EPS) for 25 companies put together to a group called OBXEPS12AAR.

I try to use the following equation:
SUE=(EPS-EPS(-4))/(MOVSTDEV(EPS,4))

I need to make a program somewhat like:
For %i
series SUE=(EPS-EPS(-4))/(MOVSTDEV(EPS,4))
next

I haven't programmed before and I can't work the details out.
I 'd appreciate very much if anybody could come up with an example or suggestion or reference?

[EViews Gareth]

I'm not entirely sure what you are asking - or rather what you're trying to achieve.

The equation you wrote looks like it will correctly calculate SUE. What is going wrong?

Your for loop doesn't appear to have any reference to !i inside it. What are you looping over?

[martingale]

Hi and thanx for the answer.

I have made a group which has data(earnings per share) for 25 companies(the data is in 25 columns, one column for each company). I want to use a program rutine to calculate SUE for the 25 companies but I don't know how to write such a program. What I would like to achieve is to automate the process in stead of doing the same calculation 25 times.

[EViews Gareth]

Code: Select all

For %i =1 to OBXEPS12AAR.@count %name = obxeps12aar.@seriesname(!i) %newname = "SUE_" + %name series {%newname}=({%name}-{%name}(-4))/(MOVSTDEV({%name},4)) next

[martingale]

hi and thanx a lot, when I run the program I get the message "!i is not defined" (I uploaded the file in case)

[EViews Glenn]

The first line should read

Code: Select all

For !i =1 to OBXEPS12AAR.@count

not "%i".

[EViews Gareth]

My bad.

[martingale]

Code: Select all

For !i =1 to OBXEPS12AAR.@count %name = obxeps12aar.@seriesname(!i) %newname = "SUE_" + %name series {%newname}=({%name}-{%name}(-4))/(@MOVSTDEV({%name},4)) next

Thanx very very much Gareth and Glenn, the above code works!

Is there also a way to substitute observations with value 0 to "NA"? this is because the program halts when it come across 0's.

[EViews Glenn]

It shouldn't stop, it should just produce a message saying that NAs have been generated and continue ("Division by zero. Missing data generated"). The final series should already contain what you want.

[martingale]

Program halts EViews.jpg (49.83 KiB) Viewed 14282 times

The program halts at the first series "sue_n_fro_eps_" where the error message is generated. A solution is to increase maximum errors before halting but it might ignore sources of other types of errors. A better solution is to program a subistitution of 0's with "NA". How could this be done?

[martingale]

Code: Select all

for !i =1 to OBXEPS12AAR.@count %name = obxeps12aar.@seriesname(!i) series {%name}=@recode({%name}=0, NA, -{%name}) next

this code seems to do the trick (substitute 0 with NA), but is it possible to nest it in the first code?

[EViews Glenn]

By stop, I meant within the one series generation. You are correct that after that series is generated, the program will stop.

I note that your proposed solution modifies the original series which generally makes me nervous.

Moreover, I don't think this modification is really what you want to do. You are replacing all values where the series equals zero with NAs, otherwise you are including the negative of the original series. Apart from the negative part, which I think is just a mistake, that substitution will solve the problem, but will also modify your @MOVSTDEV results everywhere there is a data 0 in the span, even if the original @MOVSTEDEV result is nonzero. What I think you want to do is to replace the 0's in the resulting moving standard deviation series:

Code: Select all

For !i =1 to OBXEPS12AAR.@count %name = obxeps12aar.@seriesname(!i) series stdevtmp = @MOVSTDEV({%name},4) stdevtmp = @recode(stdevtmp=0, NA, stdevtemp) %newname = "SUE_" + %name series {%newname}=({%name}-{%name}(-4))/stdevtemp delete stdevtemp next

Note that I don't mind overwriting the STDEVTMP since it's a temporary series.

[martingale]

Hi Glenn
Thanx very much for the ansewr. You are right, the negative part was a plain spelling error.

[EViews Glenn]

Glad it's working.

As I said, I don't like modifying original data unless I'm sure that's what I want. I almost always generate into a new series. The possibility of error is all the more reason not to overwrite the original series unless you are *absolutely* sure that's what you want to do. Once the data are modified, it may be impossible to "unmodify".

[martingale]

Code: Select all

For !i =1 to OBXEPS12AAR.@count %name = obxeps12aar.@seriesname(!i) series stdevtemp = @MOVSTDEV({%name},4) stdevtemp = @recode(stdevtemp=0, NA, stdevtemp) %newname = "SUE_" + %name series {%newname}=({%name}-{%name}(-4))/stdevtemp delete stdevtemp next

Hi Glenn, had to change "tmp" to "temp".
It works now fine like you said.