Fixed a small bug in partnership targets

input/DoFilesTarget/03_calculate_partnership_target.do

@@ -0,0 +1,143 @@
+version 17.0
+clear all
+set more off
+
+* Paths (edit if needed)
+local work_dir "/Users/pineapple/Library/CloudStorage/OneDrive-UniversityofEssex/WorkCEMPA/SimPathsUK/validate_alignments"
+local input_dir  "`work_dir'/input"
+
+* ============================================================
+* 1) Load and append all initial population files 2011-2023
+* ============================================================
+tempfile appended
+clear
+save `appended', emptyok replace
+
+forval yr = 2011/2023 {
+    capture import delimited using "`input_dir'/population_initial_UK_`yr'.csv", ///
+        clear varnames(1) bindquote(strict)
+    if _rc != 0 {
+        di as error "WARNING: could not load population_initial_UK_`yr'.csv — skipping"
+        continue
+    }
+    destring demage idpers idbu idmother idfather wgthhcross, replace force
+    gen int file_year = `yr'
+    append using `appended'
+    save `appended', replace
+}
+
+use `appended', clear
+
+* ============================================================
+* 2) Define partnered share — BU membership logic
+*
+*  Eligible  : demage >= 18
+*  Partnered : there exists at least one OTHER adult (age >= 18)
+*              in the same benefit unit (file_year x idbu)
+*              who is NOT this person's mother or father
+*
+*  idpartner is NOT used — partnership is inferred purely from
+*  BU co-residency, consistent with getPartner() logic.
+* ============================================================
+keep if demage >= 18
+
+* Keep only the variables needed for the join
+keep file_year idbu idpers idmother idfather demage wgthhcross
+
+* --- Step A: build list of BU adult members to self-join against ---
+preserve
+    keep file_year idbu idpers
+    rename idpers other_idpers
+    tempfile bu_adults
+    save `bu_adults', replace
+restore
+
+* --- Step B: self-join on (file_year, idbu) ---
+* Each person is expanded against all adults in their BU
+joinby file_year idbu using `bu_adults'
+
+* Remove self-matches
+drop if other_idpers == idpers
+
+* Remove cases where the co-resident is this person's parent
+drop if other_idpers == idmother | other_idpers == idfather
+
+* Each remaining row means a qualifying co-resident exists
+gen byte has_partner = 1
+
+* Reduce to one row per person: partnered = 1 if any qualifying row
+collapse (max) partnered = has_partner, by(file_year idbu idpers)
+
+* --- Step C: save partnered flags, then merge into full eligible pop ---
+* joinby silently drops persons with NO qualifying co-resident, so we
+* must restore them from the full dataset and assign partnered = 0.
+
+tempfile partnered_flags
+save `partnered_flags', replace
+
+* Reload full eligible population as master
+use `appended', clear
+destring demage idpers idbu idmother idfather wgthhcross, replace force
+keep if demage >= 18
+keep file_year idbu idpers wgthhcross
+
+* Merge partnered flags in as using — unmatched master = no partner
+merge 1:1 file_year idbu idpers using `partnered_flags', ///
+    keep(master match) nogen
+replace partnered = 0 if missing(partnered)
+
+* ============================================================
+* 3) Collapse to annual weighted share
+* ============================================================
+collapse (mean) partnered_share = partnered ///
+         (sum)  n_eligible = partnered      ///   raw count only for reference
+         [pw = wgthhcross], by(file_year)
+
+* n_eligible above is sum of weights — replace with unweighted count if preferred
+* For unweighted N alongside weighted share, use a two-step approach:
+* Step 1: save weighted share
+tempfile weighted_share
+save `weighted_share', replace
+
+* Step 2: unweighted counts
+use `appended', clear
+destring demage idpers idbu wgthhcross, replace force
+keep if demage >= 18
+keep file_year idbu idpers
+
+merge 1:1 file_year idbu idpers using `partnered_flags', ///
+    keep(master match) nogen
+replace partnered = 0 if missing(partnered)
+
+collapse (count) n_eligible = idpers ///
+         (sum)   n_partnered = partnered, by(file_year)
+
+merge 1:1 file_year using `weighted_share', nogen
+
+rename file_year year
+
+format partnered_share %12.7f
+
+label var n_eligible      "Eligible persons (age >= 18, unweighted N)"
+label var n_partnered     "Partnered persons (unweighted N)"
+label var partnered_share "Partnered share (weighted by wgthhcross)"
+
+order year n_eligible n_partnered partnered_share
+
+* ============================================================
+* 4) Full comparison table
+* ============================================================
+//export excel using "`work_dir'/partnered_share_initialPop_BUlogic.xlsx", ///
+//    firstrow(variables) replace
+
+* ============================================================
+* 5) Slim target-format file: year + partnered_share only
+* ============================================================
+preserve
+    keep year partnered_share
+    format partnered_share %12.7f
+    export excel using "`work_dir'/partnered_share_targets_BUlogic.xlsx", ///
+        firstrow(variables) replace
+restore
+
+list, sep(0)