Please read- How to proceed with unbalanced undated panel?

[catherinechui]

HI Eview master,

I tried to import panel data, and I get this error message when tried to restructure the panel data under dated panel option.

Illegal panel structure daystoexpiry nested within dateid.

Currently the cross section is daystoexpiry and it is in one column , and date MM/DD/YYYY in excel is in another column. Then a third column called 'Hour' further divides one day into 6 trading hour. So I am guessing that I will need to create a specific time id for each hour for each day. So I followed through a thread 'create date series?' that was posted on 26th Aug by Gareth, and did the same thing as suggested below.

Thread posted:

EViews 7 and 8 support intraday frequencies, including hourly.

Just use Proc->Structure/Resize Current Page, then choose Dated-Specified by Date Series, then enter the names of your series.

Then i get this message when I tried the above:
Indicies do not uniquely identify observations. 11108 duplicate date found. Create CELLID series to identify observations?

I chose yes, and I see this message:
Adding CELLID as the inner index will create a 2 dimension unbalanced undated panel.


Can you advise what I should do ? Is the Date Series mentioned above referring to Hour? Also, how can I restructure the panel data by 'dated by time series' after the new time format is created, what would i enter in the 'date series' in workfile structure panel ?

Attached excel file for your reference.

many thanks
Catherine

[catherinechui]

I am using Eviews 8 version.

[EViews Gareth]

I can't really see the panel nature of your data.

It is easy to build a datetime series:

Code: Select all

series datetime = date + hour/24

But there isn't really a panel to be made. There isn't a value of "30" and "29" and "28" etc... for 01-01-11.

[catherinechui]

Hi Gareth,

you are right that there isnt a nature for panel data. My goal is to regress how price, hour, daystoexpiry and ourvolumne affect PnL. And I want to spot any seasonality on effect on daystoexpiry. So that will be a multivariant regression? If that is the case, how or can I structure the time series because there are 6 hours in a day. Or will Hour also be dummy variable? Also Price has only high or low value, so i would think I need to create a dummy variable.

Many thanks for prompt reply.

Catherine

[EViews Gareth]

I cannot help with those questions.