Dropping duplicated results with panel data

[Pascal Tohouri]

Hello Stata people,

I have a panel dataset where the cross-sectional variable is wage (for different individuals) and the longitudinal variable is time.
I only want a time-series of total(wage) (sum of all employees' wages in a given year). I'm not interested in the wages of individuals.

I have used this command: bysort year: egen tot_wage=total(total_income)

Now I have a time series of the tot_wage variable. Unfortunately, because there were i observations in each year, the tot_wage series is replicated i times in each year. These values appear sequentially e.g.

...
2012 | 5000
2012 | 5000
2012 | 5000
2013 | 5500
2013 | 5500
...

Not ideal.

The individuals participating in the wage survey are not the same from year to year so I can't drop all but one person's observations. How might I obtain the simple time series I am after?

Best,
Pascal

[Attaullah Shah]

I have not completely grasped your question, however, if you want to keep on observation for each person in each year, the following code should do the job. Note that I am assuming the there is a variable individual that tracks the identity of each cross-sectional unit.

bysort individual year: keep if _n == _N

And if you want to keep one observation per year, then

bysort year: keep if _n == _N

[Carlo Lazzaro]

Pascal:
welcome to this forum.
You may be interested is something along the following lines:

. set obs 5
number of observations (_N) was 0, now 5

. g id=1

. g year=2000+_n

. g wage=10000*runiform()

. collapse (sum) wage, by(id)

. list

     +---------------+
     | id       wage |
     |---------------|
  1. |  1   16498.94 |
     +---------------+

[Pascal Tohouri]

Dear All,

I found that the 'collapse' command gave the desired result.
Thank you both for your responses, you've saved me a lot of bother!

All the best,
Pascal