Re: st: how to get monthly values from Bimonthly data

[Nick Cox <njcoxstata@gmail.com>]

I would leave these data almost as they are. What you can do is

gen year = real(substr(dates, -4, 4))
gen period = substr(dates, 1, 2)
gen month = cond(per == "JF", 1, cond(per == "MA", 3, cond(per ==
"MJ", 5, cond(per == "JA", 7, cond(per == "SO", 9, cond(per == "ND",
11, .))))))

gen date = ym(year, month)
tsset id date , delta(2)

The key point is that you have bimonthly data and that's fine. No
doubling up or interpolation is needed; in fact that can only produce
a spurious increase in your sample size.

Some people would undoubtedly prefer -recode- for the mapping "JF" -> 1, etc.

Nick

On Fri, Jul 27, 2012 at 3:48 AM, sabbas gidarokostas
<sabbasgidarokostas@googlemail.com> wrote:
> Dear all,
>
> I have the following panel data set
> clear all
> input str8  (id        text             dates           values1      values2 )
>
> "1"    "ot"            " "             ""                " "
> "1"    "wt"          " "               ""              " "
> "1"    "rt"        "MA 2005"            "32.4"          "2.5"
> "1"    "gh"        "MJ 2005"            "82.3"          "12.5"
> "1"    "bg"        "JA 2005"            "2.5"          "24.5"
> "1"    "se"        "SO 2005"            "82.5"          " "
> "1"    "cv"        "ND 2005"            "62.3"          "26.3"
> "1"    "cv"        "JF 2006"           "12.2"           "22.3"
> "1"    "cv"        "MA 2006"            "22.1"          "2.4"
> "2"    "ot"          " "                " "               " "
> "2"    "wt"           " "                " "         " "
> "2"    "rt"        "MA 2005"
> "2"    "gh"         "MJ 2005"            "32.3"          "23.5"
> "2"    "bg"        "JA 2005"            "52.4"          "22.3"
> "2"    "se"        "SO 2005"            "82.9"          "22.5"
> "2"    "cv"        "ND 2005"            " "          "2.5"
> "2"    "cv"        "JF 2006"            "332.4"          " "
> "2"    "cv"        "MA 2006"            "312.4"          "21.5"
> end
> The above data set is a sample.
> Basically,I have 30000 individuals and  instead of the last 2 columns
> that contain the numbers I have 29 such columns.
>
> The data set refers to bimontly values.
>
> Is there a way to  convert the above data to montly data?
> Specifically, What I think is that first I have to split the dates as follows
>
> M/2005
> A/2005
> M/2005
> J/2005
>
> and so forth and then to convert them to
>
> 3/2005
> 4/2005
> 6/2005
> and so forth.
>
> So regarding the first 3 columns I should get
>
> "1"    "ot"            " "
> "1"    "wt"          " "
> "1"    "rt"        "3/2005"
> "1"    "rt"        "4 2005"
> "1"    "gh"        "5/2005"
> "1"    "gh"        "6/2005"
>
> and so forth.I am not exactly sure how to handle the last two columns,
> though(numerical values)
>
> I think that one way would be to divide the bimonthly values by 2. Then I get
>
> "1"    "ot"            " "
> "1"    "wt"          " "
> "1"    "rt"        "3/2005"            "32.4/2"          "2.5/2"
> "1"    "rt"        "4 2005"              "32.4/2"          "2.5/2"
> "1"    "gh"        "5/2005"             "82.3/5"          "12.5/5"
> "1"    "gh"        "6/2005"                  "82.3/5"          "12.5/5"
>
> I am not sure that this makes sense as I have repeated values for two
> consequtive months.
>
> Another way would be interpolation?
>
> Any code provided is greatly appreciated.
*
*   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/