string monthly date into Stata monthly date

[Nuno Soares]

Hi everyone,

I'm having a problem converting a string monthly date into Stata monthly date. The string date starts in 1999 and goes until 2013 and is shown as:
Jan-99
Feb-11
and so on.

I've tried using date(var,"MY") to no avail. The problem seem not to have the day in the originalvariable or that months are in text and not integers. I really don't need (because i don't have) the day, but need to have the data in a monthly format that Stata can interpret as such. Any ideas?

best wishes,

Nuno

[Nick Cox]

date() is a function for producing daily dates; it is not a generic function for producing any kind of date. This is multiply documented e.g. http://www.stata.com/help.cgi?date() but my own view is that date() should go undocumented in favour of daily()

Hence you need a different function or command. Experiment with monthly() to see that it is one solution:

. di monthly("Jan-11", "MY", 2050)
612

. di monthly("Jan-99", "MY", 2050)
468

With that solution you then need to specify a display format.

Alternatively, numdate (SSC) is a one-stop command for producing date variables.

. set obs 2
obs was 0, now 2

. gen sdate = cond(_n == 1, "Jan-99", "Jan-11")

. numdate monthly mdate = sdate, pattern(MY) topyear(2050)

. l

     +-----------------+
     |  sdate    mdate |
     |-----------------|
  1. | Jan-99   1999m1 |
  2. | Jan-11   2011m1 |
     +-----------------+

Note that date() or daily() is not useless here as you can follow with mofd():

 
. di mofd(daily("Jan-99", "MY", 2050))
468

Dates are a large and complicated topic unless you have just recent calendar years and everything is easy. There is no easy substitute for skimming http://www.stata.com/help.cgi?dates

[Nuno Soares]

date() is a function for producing daily dates; it is not a generic function for producing any kind of date. This is multiply documented e.g. http://www.stata.com/help.cgi?date() but my own view is that date() should go undocumented in favour of daily()

Hence you need a different function or command. Experiment with monthly() to see that it is one solution:

. di monthly("Jan-11", "MY", 2050)
612

. di monthly("Jan-99", "MY", 2050)
468

With that solution you then need to specify a display format.

Alternatively, numdate (SSC) is a one-stop command for producing date variables.

. set obs 2
obs was 0, now 2

. gen sdate = cond(_n == 1, "Jan-99", "Jan-11")

. numdate monthly mdate = sdate, pattern(MY) topyear(2050)

. l

+-----------------+
| sdate mdate |
|-----------------|
1. | Jan-99 1999m1 |
2. | Jan-11 2011m1 |
+-----------------+

Note that date() or daily() is not useless here as you can follow with mofd():

. di mofd(daily("Jan-99", "MY", 2050))
468

Dates are a large and complicated topic unless you have just recent calendar years and everything is easy. There is no easy substitute for skimming http://www.stata.com/help.cgi?dates

thanks Nick! the monthly command worked perfectly!

[Nick Cox]

Good, but monthly() is a function, not a command!