st: Inaccurate date-time variable

[rosa <rosahh.rosa@gmail.com>]

Dear users,

I am having problems generating an accurate date-time variable (using
Stata V 12) using a database of intra-day bank transactions. My final
goal is to declare the dataset to be time series and be able to obtain
a 5minute rolling standard deviation.

The problem is that I have a string variable looking like "2005-01-03
8:34:03" and after transforming and fomating it as desired, some of
the observations of the new date-time variable do not reflect the
exact time as indicated in the original one. So, the date-time
detailed above is transformed to 03jan2005 08:33:37 which is 26
seconds behind the original time. This problem aggravates when
repeated observations are incorrectly generated.

I have transformed  the date-time variables using the following command:

gen clock_date2= Clock(clock_date, "YMDhms")
format clock_date2 %tc

And I've also tried to format to ' %tCDDmonCCYY_HH:MM:SS ', however it
gives me the same inaccurate result.

Does anyone have an idea of how can I obtain an accurate date-time
variable and why this error is be happening?


I detail below some of the original and transformed date-time observations:

string var: clock_date	 clock_date2 (inaccurately transformed variable)
2005-01-03 8:03:31	 03jan2005 08:03:02
2005-01-03 8:10:30	 03jan2005 08:11:46
2005-01-03 8:11:27	 03jan2005 08:11:46
2005-01-03 8:11:47	 03jan2005 08:11:46
2005-01-03 8:33:37	 03jan2005 08:33:37
2005-01-03 8:34:03	 03jan2005 08:33:37
2005-01-03 8:34:35	 03jan2005 08:35:48


Thanks a lot in advance,

Rosa
*
*   For searches and help try:
*   http://www.stata.com/help.cgi?search
*   http://www.stata.com/support/statalist/faq
*   http://www.ats.ucla.edu/stat/stata/