Series Creation by Specified Day(s)/Month(s)/Year(s) etc

[px41]

Hi Good Day everyone,

I have a question concerning data grouping.

If I have daily stacked data in a workfile ( 5 years of daily stock prices across 30 different companies), is there a way to create a series that will give a listing of daily stock prices for the months of March-December (across the 5 years for all companies)? I am also trying to create a series with the daily data across all companies for Monday-Thursday for a similar reason. I have done some searching on the forum and looked though the user guide but I came across frequency conversion, which isn't exactly what I want.

If it helps, my overall aim is to do a simple hypothesis test, to test the mean of a stock's price in a particular month(s) (e.g mean of Jan-Feb price) against the mean of a stock's price in the rest of the months of the year (eg mean of March-Dec price). Needless to say I will need a distribution of prices for the specific months involved so that's why I'm trying to create a series that will list the prices of certain month(s) (and also for weekdays, for my day of the week test) then I can just use that time series to carry out my test.

My alternative is to copy and paste so I was just wondering if there is a function to generate a series based on specified months or days. I know there is a function to generate means by day/month/year etc (thanks again for that Gareth, if you happen to be reading this!)

Thanking you in advance for your valuable help.

[EViews Gareth]

You can create a series that is equal to the sum over companies like this:

Code: Select all

series dum = @sumsby(X, @crossid)

where X is the variable you want to sum over.

Once you have the sums, you can restrict the sample to be just March-December:

Code: Select all

smpl if @month>2

or Mon-Thurs:

Code: Select all

smpl if @day>=1 and @day<=4

(where you'll have to check those numbers are correct).

Once you have the sample set, you can find means etc...

[px41]

Hi Gareth, thanks for the reply.

I'm a bit unclear as to the necessity of the sums. Will this sum my data given the crossid (@weekday etc) and variable(series name)? As an eg, are you saying "smpl if @day=1" interprets the data as if there were only mondays (assuming 1=monday) and thus will build graphs and histograms etc for only Monday dates across all firms? If yes, this will be fine!

If not, since I already have means I just need a population of prices to test it against. Thus, let me ask another question, is there a function that can reproduce the daily data for say, the months of Mar-Dec, across all firms and years, in a separate workfile? (ie this will be the population of Mar-Dec prices for all firms over the focal period and I can compare the Jan-Feb mean to this Mar-Dec mean). If this is possible it will also be of great help and will be the solution I'm looking for. Thanks again Gareth.

Regards.

[EViews Gareth]

Ah, I guess I wasn't sure what your original question was. I thought you wanted to sum across all companies.

If you don't want to do that - if you just want to work with certain dates, then yes, all you need to do is set the sample to be the dates you want.

[px41]

Oh ok, wonderful...! Sorry for being unclear.
You said I can set up the sample for the dates I want. What if I want a particular month? Say statistics for all months of January across all firms and years? The function are the same as you mentioned?

Here's what I mean:

(Assuming 1=mon, 2=tues, 3=wed, 4=thurs, 5= fri)

Entering the function:
1)"smpl if @day=1"................................will yield statistics for Mondays across all firms (such as means of all mondays, a histogram representing all mondays etc)?
2)"smpl if @day>=1 and @day<=4".............will yield similar statistics for Mon-Thurs across all firms?

Similarly, for months,
Assuming (1=Jan, 2=Feb....12=Dec)

Then entering
3)"smpl if @month=1 and 2"............................will yield statistics for Jan and Feb?
4)"smpl if @month>=1 and @day<=4"..................will yield statistics for Mar-Dec?

Is this correct? I hope yes...!

Thanks again for the advice Gareth.

[EViews Gareth]

Code: Select all

smpl if @month>=3

will lead to statistics for all companies for months March-December.

If you want March to July, it would be:

Code: Select all

smpl if @month>=3 and @month<=7

[px41]

Ooooooooooohhhh.....!

I see, I will give it a try as soon as possible.

Thanks very very much!!

[px41]

Hey Gareth, it worked!

Thanks again so much!

I was wondering though, i couldn't figure it out, what's the function to get statistics on the months of Jan-May AND Aug-Dec...given that i'll compare the statistics of those months to statistics of June and July (which should be "smpl if @month>=6 and @month<=7" right?).

Thanking you in advance.

[EViews Gareth]

Code: Select all

smpl if @month<6 or @month>7

[px41]

Thank you!