Welcome to the Institute for Digital Reseach and Education

Stata FAQ:  How can I extract month and year component from a variable with %tm format?

A variable in %tm format is a numeric variable which is the number of months elapsed since January 1960. One may want to extract the month or the year component from this variable. In this example, we will show how to perform this task by using a function from a suite of date converting functions.

use http://www.ats.ucla.edu/stat/stata/faq/dow1_monthly, clear

list in 1/10

     +-------------------------------+
     |      dm   dowclose     ln_dow |
     |-------------------------------|
  1. |  1953m1   288.4524   5.664502 |
  2. |  1953m2   283.9611   5.648786 |
  3. |  1953m3   286.7909   5.658709 |
  4. |  1953m4   275.2857   5.617776 |
  5. |  1953m5   276.9381   5.623769 |
     |-------------------------------|
  6. |  1953m6   266.8864   5.586792 |
  7. |  1953m7   270.3261   5.599611 |
  8. |  1953m8   272.2048   5.606421 |
  9. |  1953m9   261.9048   5.567924 |
 10. | 1953m10   270.7191   5.600968 |
     +-------------------------------+

Notice that variable dm is in %tm format. If we reformat it as in a general numeric format as shown below, we will see that it is a numeric variable. For example, the number of months from January 1953 to January 1960 is 84. This is what we see as the first observation except that it is negative since it is before January 1960, going backwards.

format dm %10.0g

di (53 - 60)*12
-84

list in 1/10

     +---------------------------+
     |  dm   dowclose     ln_dow |
     |---------------------------|
  1. | -84   288.4524   5.664502 |
  2. | -83   283.9611   5.648786 |
  3. | -82   286.7909   5.658709 |
  4. | -81   275.2857   5.617776 |
  5. | -80   276.9381   5.623769 |
     |---------------------------|
  6. | -79   266.8864   5.586792 |
  7. | -78   270.3261   5.599611 |
  8. | -77   272.2048   5.606421 |
  9. | -76   261.9048   5.567924 |
 10. | -75   270.7191   5.600968 |
     +---------------------------+

Now how do we extract the month and year information from variable dm? We know that Stata has month() and year() functions available. But we cannot use them yet, since these two functions require that the input be in date format. That is the input variable should also carry the information on day. So we are not quite there yet. But what is also available is the Stata's suite of converting time format functions. To our end, we can use  function dofm().

gen date = dofm(dm)

format date %d

list in 1/10

     +---------------------------------------+
     |  dm   dowclose     ln_dow        date |
     |---------------------------------------|
  1. | -84   288.4524   5.664502   01jan1953 |
  2. | -83   283.9611   5.648786   01feb1953 |
  3. | -82   286.7909   5.658709   01mar1953 |
  4. | -81   275.2857   5.617776   01apr1953 |
  5. | -80   276.9381   5.623769   01may1953 |
     |---------------------------------------|
  6. | -79   266.8864   5.586792   01jun1953 |
  7. | -78   270.3261   5.599611   01jul1953 |
  8. | -77   272.2048   5.606421   01aug1953 |
  9. | -76   261.9048   5.567924   01sep1953 |
 10. | -75   270.7191   5.600968   01oct1953 |
     +---------------------------------------+

Now we are ready to extract month and year information from variable date.

gen month=month(date)

gen yr=year(date)

list in 1/10

     +------------------------------------------------------+
     |  dm   dowclose     ln_dow        date   month     yr |
     |------------------------------------------------------|
  1. | -84   288.4524   5.664502   01jan1953       1   1953 |
  2. | -83   283.9611   5.648786   01feb1953       2   1953 |
  3. | -82   286.7909   5.658709   01mar1953       3   1953 |
  4. | -81   275.2857   5.617776   01apr1953       4   1953 |
  5. | -80   276.9381   5.623769   01may1953       5   1953 |
     |------------------------------------------------------|
  6. | -79   266.8864   5.586792   01jun1953       6   1953 |
  7. | -78   270.3261   5.599611   01jul1953       7   1953 |
  8. | -77   272.2048   5.606421   01aug1953       8   1953 |
  9. | -76   261.9048   5.567924   01sep1953       9   1953 |
 10. | -75   270.7191   5.600968   01oct1953      10   1953 |
     +------------------------------------------------------+

For more information on date conversion functions, use "help dcfcns".