Data Frequency Conversion

For questions regarding the import, export and manipulation of data in EViews, including graphing and basic statistics.

Moderators: EViews Gareth, EViews Jason, EViews Steve, EViews Moderator

ercan
Posts: 7
Joined: Sun Dec 07, 2008 2:23 am

Data Frequency Conversion

Postby ercan » Mon Dec 22, 2008 4:53 am

I would like to generate new series using daily trading data. Specifically, I would like to calculate average weekly values using daily data and store the new series in weeks.

any suggestions?

trubador
Did you use forum search?
Posts: 1518
Joined: Thu Nov 20, 2008 12:04 pm

Re: Data Frequency Conversion

Postby trubador » Mon Dec 22, 2008 5:13 am

Create a "New Page" in your workfile in weekly frequency. Then copy your daily series and paste into this page with "paste special". In the dialog box you will see the "Frequency conversion options". Select the "Average observations" from "High to low frequency method" drop down menu. There are also many other useful options here that may interest you...

TheRocK
Posts: 20
Joined: Sun Jan 18, 2009 4:18 am

Re: Data Frequency Conversion

Postby TheRocK » Sun Jan 18, 2009 4:32 am

Hi !

I have a question which is rather similar but cannot find an answer.

I have daily stock returns. From this, i need to compile quarterly, two-quarterly, three-quarterly and yearly returns, which is not a big issue since I am using continuous returns and can just "add them up". However, I also need to compute the standard deviations for the above mentioned intervalls using daily returns. Here is an example:

Lets consider the following time horizon.

4Q1991
1Q1992
2Q1992
3Q1992
4Q1993
1Q1993
2Q1993
3Q1993
4Q1993

What I need to do is firstly compute the quarterly return and standard deviation from the end of 4Q1991 to end of 1Q1992 and so forth out of daily data.
Then I also need to compute the return and standard deviation from the end of 4Q1991 to end of 2Q1992, from the end of 1Q1992 to end of 3Q1992 and so forth. In Excel, I would compute the daily standard deviation over this period and multiply it by 125 (assuming 250 trading days). However, when I do this for all time periods and all cross sections, the excel sheets are skyrocketing in size. Is there any way to compute this in eviews ?

Sorry for my bad English and thank you !

Kai

Dusan
Posts: 5
Joined: Mon Jan 19, 2009 4:49 am

Re: Data Frequency Conversion

Postby Dusan » Mon Jan 19, 2009 5:27 am

Hi !
I have a similar problem with theRock.I have a time series with daily data, and I want the monthly standard deviation for it.
Of course I can convert my data from daily to monthly but I dont know how I can get the standar deviation for each month.
Thanx

trubador
Did you use forum search?
Posts: 1518
Joined: Thu Nov 20, 2008 12:04 pm

Re: Data Frequency Conversion

Postby trubador » Mon Jan 19, 2009 5:47 am

You should follow similar steps as I mentioned in my previous post. You should create a seperate worksheet (new page) in your workfile and specify its frequency as monthly. After that, you can simply copy your series in the daily worksheet and paste it into this new monthly page with "paste special". In the dialog box, under "Merge by" section select the "General match merge criteria". And then select the "Std. deviation (sample)" from "Contraction method" drop down menu. You can also specify your new series as link or value...

Dusan
Posts: 5
Joined: Mon Jan 19, 2009 4:49 am

Re: Data Frequency Conversion

Postby Dusan » Mon Jan 19, 2009 6:04 am

Thanx Trubador!U saved me......

TheRocK
Posts: 20
Joined: Sun Jan 18, 2009 4:18 am

Re: Data Frequency Conversion

Postby TheRocK » Mon Jan 19, 2009 1:04 pm

Hi Trubador,

thanks for your answer. When I do this, does eViews compute the standard deviation out of the daily frequency data or out of the newly computed monthly data? Background is that I need the standard deviation for every quarter of my data. Obviously that is only possible if the return data has a lower frequency. Additionally, can I then tell eViews to compute the standard deviation of every quarter and give me the output in a new pool ?

Although this question not belongs in this section of the forum, I hope the admins will forgive me :-) In my panel regression, I will need to regress the returns of period t+1 and t+2 on several variables in period t. obviously this will result in serial correlation due to overlapping time periods.

I already found out that eviews supports several methods of coef. covariance methods which correct this. Since the serial correlation according to the durbin watson stat comes from the "time factor" and not the cross-sections, the "white period", "white diagonal", "period sur pcse" and "period weights pcse" are possibilities. since either of them changes the significance of my variables extremely, I wanted to ask you how I can determine which method is the most appropriate for me? I could not find any answer in the eViews help.

Thank you so much for your help. I really appreciate it.

All the best

Kai

trubador
Did you use forum search?
Posts: 1518
Joined: Thu Nov 20, 2008 12:04 pm

Re: Data Frequency Conversion

Postby trubador » Tue Jan 20, 2009 5:40 am

There is no need to create an intermediate worksheet with monthly frequency, which, if you do, standard deviations in your quarterly worksheet will be calculated according to monthly series. You can copy and paste special your daily series directly into your quarterly worksheet and calculate standard errors with respect to your daily data.

Regarding to your next question, I really do not have enough expertise on panel/pool regressions to give advice. Therefore, it should be a good idea to move your question under the "econometric discussions" thread and ask for help from moderators and other users.


Return to “Data Manipulation”

Who is online

Users browsing this forum: No registered users and 16 guests