Frequency conversion problems
Posted: Wed Mar 26, 2014 11:09 am
Hi,
I am having some trouble with the frequency conversion of my data. I would like to instruct eviews during a high to low (for example daily to weekly) frequency conversion to always take the first observation of a week even if it's a NA value. What eViews usually does: it takes the next available observation. If there is no observation for monday in a daily page, then eViews converts the data into a weekly page using the observation of the next day, i.e tuesday. I would like to prevent that, because if I have two series, and one of them contains an observation for a monday and the other one doesn't, eViews will mix up the dates during the frequency conversion. I don't know, if I made that clear enough. So I created an example workfile with a different explanation of my problem below.
I have a worksheet containing three pages (daily, weekly, and monthly). To clarify my problem I created two series ("lesscompletedata" and "morecompletedata") using @trend for each of the two series. I then randomly deleted some observations for both series. I have another series which is just frml diff=lesscompletedata-morecompletedata. So it just takes the difference of the two series which is always zero or NA, since both series were creating using @trend and some values were manually deleted. Now, once I paste the two series and the formula (diff) into the page "weekly" eViews converts the data to a lower frequency (i.e. weekly), and since some data are missing, eViews then takes the next available observations. That will however cause the difference of the two series to be no longer always equal to zero or NA. So my question is, how can I convert my series into a lower frequency, take the difference of the two series, and make sure, that the difference will still always be zero or NA. So I don't wanto eViews to do any "frontrunning", i.e. use the next available observation.
Thanks a lot for help.
I am having some trouble with the frequency conversion of my data. I would like to instruct eviews during a high to low (for example daily to weekly) frequency conversion to always take the first observation of a week even if it's a NA value. What eViews usually does: it takes the next available observation. If there is no observation for monday in a daily page, then eViews converts the data into a weekly page using the observation of the next day, i.e tuesday. I would like to prevent that, because if I have two series, and one of them contains an observation for a monday and the other one doesn't, eViews will mix up the dates during the frequency conversion. I don't know, if I made that clear enough. So I created an example workfile with a different explanation of my problem below.
I have a worksheet containing three pages (daily, weekly, and monthly). To clarify my problem I created two series ("lesscompletedata" and "morecompletedata") using @trend for each of the two series. I then randomly deleted some observations for both series. I have another series which is just frml diff=lesscompletedata-morecompletedata. So it just takes the difference of the two series which is always zero or NA, since both series were creating using @trend and some values were manually deleted. Now, once I paste the two series and the formula (diff) into the page "weekly" eViews converts the data to a lower frequency (i.e. weekly), and since some data are missing, eViews then takes the next available observations. That will however cause the difference of the two series to be no longer always equal to zero or NA. So my question is, how can I convert my series into a lower frequency, take the difference of the two series, and make sure, that the difference will still always be zero or NA. So I don't wanto eViews to do any "frontrunning", i.e. use the next available observation.
Thanks a lot for help.