Identifying Year of First Move

[Darman Bacha]

Hello Friends,

I have a dataset with the following variables: "year", "pair_id", "house_id", "moved_once". "year" is the year, "pair_id" is a pair-specific identifier, "house_id" is a house-specific identifier, and "moved_once" is a dummy indicating if a pair has moved at least once at some point in the data (their "house_id" changed at least once). "moved_once" = 1 for all observations for pairs who moved at least once, and 0 for pairs who haven not moved. Below is an illustration of the data:

I need to add a variable "year_first_move" indicating the year in which a pair first moved ("house_id" first changed). "year_first_move" should equal this year-of-first-change but for all observations of a given "pair_id", and naturally only for "pair_id"s where "moved_once" = 1. "year_first_move" can be blank for "pair_id"s where "moved_once" = 0. An illustration is below:

I tried to create "year_first_move" with the following code:

gen year_first_move = .
bysort pair_id (year): forvalues i = 2(1)_N {
    local year_first_move_l = 1
    replace `year_first_move_l' = year[`i'] if house_id[`i'] != house_id[`i' - 1]
    if `year_first_move_l' > 1 {
        replace year_first_move = `year_first_move_l'
        continue, break
    }
}

The inner workings of the loop are still rough (e.g., I haven't figured out how to replace all values of "year_first_move" with the year-of-first-change), but my urgent problem is that Stata does not even allow by and forvalues together. Is there any way I can work around this issue in creating my desired variable? Any help would be greatly appreciated.

Thank you,

Darman