# Sources & method — Vol 1, Issue 1

## gdp_growth.csv

**Indicator:** Real GDP growth by fiscal year, per cent. FY20–FY26 are
official outturns (FY26 provisional); FY27 is the federal budget's growth
target.

- FY20 (−0.9), FY21 (+5.8), FY22 (+6.2): computed from constant-price GVA
  levels (2015-16 base) in Economic Survey 2024-25, Table 1.1 —
  https://www.finance.gov.pk/survey/chapter_25/1_Growth_and_Investment.pdf
- FY23 (−0.2), FY24 (2.6), FY25 (3.2), FY26 (3.7): the series as revised in
  the FY26 accounts — https://www.dawn.com/news/2006877 and
  https://www.brecorder.com/news/40425051
- FY27 target (4.0%), Rs18.771tn outlay, Rs15.26tn FBR target, 8.2% average
  inflation, 3.6% fiscal deficit, 2% primary surplus: federal budget 2026-27 —
  https://www.dawn.com/news/2007283 and
  https://www.brecorder.com/news/40425242

**Vintage note:** FY20–FY22 are computed from ES 2024-25 levels (GVA basis);
FY23 onward reflect later revisions. Differences across vintages are within
±0.5pp and do not affect the saw-tooth shape the figure describes.

**Shock attribution in the lead:** FY18's boom and the 2018-19 BoP crisis —
SBP's real effective exchange rate index (2010=100) peaked at 123.7 in April
2017 (via CEIC, https://www.ceicdata.com/en/indicator/pakistan/real-effective-exchange-rate ;
see also Hamid & Mir, "Exchange Rate Management and Economic Growth", Lahore
Journal of Economics 22:SE, 2017,
https://lahoreschoolofeconomics.edu.pk/assets/uploads/lje/Volume22/04_Hamid_and_Mir2.pdf );
~$7bn of reserves spent defending the rupee —
https://tribune.com.pk/story/1691123/2-govt-injected-7b-keep-rupee-overvalued-recent-years
and https://www.theigc.org/blogs/forced-devaluation-rupee-recipe-disaster
FY23 collapse followed the 2022 oil/commodity surge and import compression
(Economic Survey narrative, links above).

## sector_employment.csv

**Indicator:** Share of total employment by major sector, 2020-21 and 2024-25,
per cent, on the comparable (13th ICLS) basis as published by PBS.

- Labour Force Survey 2024-25, PBS —
  https://www.pbs.gov.pk/wp-content/uploads/2020/07/LFS-2024-25-Annual-Report.pdf
  and Key Insights summary —
  https://www.pbs.gov.pk/wp-content/uploads/2020/07/Key-Insights-of-Labour-Force-Survey.pdf

**Comparability warning (the subject of piece 03):** LFS 2024-25 adopts the
19th ICLS employment definition, which excludes own-use producers. On that
basis agriculture's share is 33.1 per cent and informal employment is 80.8 per
cent of the total. Cross-round comparisons in this issue use the 13th ICLS
comparable series; single-year 2024-25 levels quoted alone use the 19th ICLS
headline. Do not mix the two.

## productivity.csv

**Indicator:** Sector labour productivity relative to the economy-wide average
(sector share of GDP divided by sector share of employment). GDP shares are
FY26 (services 58.4% per the Economic Survey; manufacturing 12.1% and
wholesale & retail trade 17.8% per PBS national accounts as reported;
agriculture ~23.4%, approximate). Employment shares are LFS 2024-25 headline
(19th ICLS).

- Sector GDP shares — https://propakistani.pk/2026/06/04/just-10-sectors-dominate-90-gdp-of-pakistan/
  and https://www.brecorder.com/news/40425051
- Employment shares — LFS 2024-25 Key Insights,
  https://www.pbs.gov.pk/wp-content/uploads/2020/07/Key-Insights-of-Labour-Force-Survey.pdf

**Output per worker (cited in the lead, not a column here):**
real GDP index FY21→FY25 = 1.0597 × 0.998 × 1.026 × 1.032 ≈ 1.12 (growth rates
of 5.97, −0.2, 2.6 and 3.2 per cent; FY22 per the rebased national accounts);
employment 67.25m (LFS 2020-21) → 77.2m (LFS 2024-25), +14.8%. Output per
worker ≈ 1.12 / 1.148 − 1 ≈ −2.5%. The 2020-21 employment count is 13th ICLS
and the 2024-25 count 19th ICLS (narrower), so −2.5% is a lower bound on the
decline. Investment ratio 14.38% of GDP in FY26 per the Economic Survey 2025-26;
the recent path (13.97% FY23, 13.14% FY24, 13.76% FY25) is Total Investment
[I] in Economic Survey 2024-25, Table 1.1 —
https://www.finance.gov.pk/survey/chapter_25/1_Growth_and_Investment.pdf —
supporting "roughly where the rate has sat for years".

- Employed 67.25m in 2020-21 — PBS LFS 2020-21 key findings,
  https://www.pbs.gov.pk/sites/default/files/labour_force/publications/lfs2020_21/Key_Findings_of_Labour_Force_Survey_2020-21.pdf
- Employed 77.2m in 2024-25 — PBS LFS 2024-25 Key Insights (link above)
- FY22 growth 5.97% — https://tribune.com.pk/story/2357283/with-6-growth-rate-pakistans-economic-size-jumps-to-383-billion
- Investment 14.38% of GDP — Economic Survey 2025-26 via
  https://www.brecorder.com/news/40425051

## Other figures cited in the issue

- Real wage calculation in the lead: nominal wages +62% (LFS rounds, below);
  CPI between the survey years compounds average annual inflation of 12.2%
  (FY22), 29.2% (FY23), 23.4% (FY24) and 4.5% (FY25) to ≈ +87%
  (PBS/Economic Survey inflation chapters,
  https://www.finance.gov.pk/survey/chapter_25/7_Inflation.pdf ), giving a
  real change of 1.62/1.87 − 1 ≈ −13%.
- Informal employment 80.8% (19th ICLS), wages Rs24,028 → Rs39,042 (+62%
  nominal, 2020-21 → 2024-25): LFS 2024-25 —
  https://profit.pakistantoday.com.pk/2026/02/19/over-80-of-pakistans-workforce-engaged-in-informal-employment-labour-force-survey-reveals/
- Goods exports −5.4% to $25.8bn, imports +8.5% to $52.8bn, trade deficit
  $23.5bn (Jul–Mar), remittances ≈$38bn in 11 months (+9%), poverty headcount
  28.9%: Economic Survey 2025-26 —
  https://www.brecorder.com/news/40425051 and
  https://profit.pakistantoday.com.pk/2026/06/12/pakistan-economic-survey-2025-26-gdp-at-four-year-high-poverty-climbs-to-289percent
- Textile exports $17.93bn in FY26 (+0.26%); readymade garments a record
  $4.18bn (+5.5%): PBS trade statistics —
  https://mmnews.tv/pakistans-textile-exports-hit-17-93-billion-in-fy26/ and
  https://propakistani.pk/2026/08/02/pakistans-readymade-garment-exports-hit-highest-level-in-fy26/
- Tariff reform phase two (ACD cut on 3,149 tariff lines; RD capped at 20% on
  1,900+ lines) under the National Tariff Policy 2025-30 —
  https://profit.pakistantoday.com.pk/2026/06/13/government-cuts-customs-duties-on-92-tariff-lines-for-industrial-inputs
  and https://www.dawn.com/news/2007905
- HIES 2024-25: first fully digital round, 32,814 households, Sep 2024–Jun
  2025; microdata published by PBS — https://www.pbs.gov.pk/hies/

## lights_tehsils.csv & mpi_districts.csv (the methods notebook)

**lights_tehsils.csv:** all 552 tehsils with 2024 VIIRS annual-composite
radiance (nW/cm²/sr), Meta Relative Wealth Index, and WorldPop population —
extracted from Data Darbar's poverty module (CC BY 4.0),
https://hibasameen.github.io/datadarbar/poverty.html , which sources
NOAA/Colorado School of Mines (VIIRS), Meta Data for Good (RWI) and WorldPop.

**mpi_districts.csv:** multidimensional poverty (Alkire-Foster MPI, headcount
H, intensity A, survey n) for 119 districts, computed from PSLM microdata via
the same module.

**Statistics quoted in the notebook** (all reproducible from these two files):
corr(pop-weighted district log-radiance, district H) = −0.70 (n=119);
tehsil corr(log nl, RWI) = 0.83 overall, 0.69 within districts (503 tehsils
in districts with ≥3); median |year-on-year log change| in tehsil radiance
≈ 0.24 (~27%) over 459 tehsils with ≥5 annual observations. National 28.9%
is consumption poverty (HIES 2024-25) and is not comparable to the
multidimensional headcounts. Caveat stated in the piece: RWI's inputs include
night lights, so nl–RWI agreement partly reflects shared inputs.

## flfp_edu.csv

Female labour force participation (ages 15-64) by highest education level
completed, rural and urban, comparable 13th ICLS basis. Adaad estimate
from the PBS public-use microdata of both rounds (LFS 2020-21 and 2024-25),
using the reconstruction documented in flfp_delta_method.py. Education is
Question 4.9 (identical category codes in both rounds), grouped: none or
below primary (codes 1-3), primary (4), middle (5), matric (6),
intermediate (7), degree or higher (8+). Needs no district recovery, so it
covers all four provinces including urban areas. Population shares are
survey-weighted shares of women 15-64 within each area, 2024-25.

## flfp_area.csv

Female labour force participation rate (ages 10+) by area, comparable 13th
ICLS basis. 2020-21 levels (rural 28.0, urban 10.0, all areas 21.4) as
published in the PBS LFS 2020-21 annual report (Figure 3.3, key indicators
by sex and area). 2024-25 rural (31.7) and urban (13.4) are Adaad
calculations from the PBS public-use microdata using the 13th-ICLS
reconstruction documented in flfp_delta_method.py, which replicates every
published aggregate rate exactly (47.7 total, 24.4 female, 69.8 male,
52.3 rural, 40.8 urban); the all-areas female rate (24.4) is as published.

## flfp.csv (piece 02)

Labour force participation rates, 2020-21 v 2024-25, comparable 13th ICLS
basis: overall 44.9→47.7, female 21.4→24.4, male 67.9→69.8, rural 48.6→52.3,
urban 38.8→40.8; labour force 71.8m→85.6m.

- Gallup Pakistan Big Data analysis of LFS 2024-25 (4 Feb 2026) —
  https://gallup.com.pk/wp/wp-content/uploads/2026/02/LFS-PR-1-Labor-Force-Participation-in-Pakistan.pdf
- Female contributing family workers 55.9% — LFS 2020-21 report narrative;
  unpaid family helpers 19.1% of all employment — PBS LFS 2024-25 Key Insights —
  https://www.pbs.gov.pk/wp-content/uploads/2020/07/Key-Insights-of-Labour-Force-Survey.pdf
- Wages by sex (female Rs20,117→Rs37,347; male Rs24,643→Rs39,302 between the
  two rounds) — PBS LFS 2024-25 annual report figures as reported —
  https://profit.pakistantoday.com.pk/2026/02/19/over-80-of-pakistans-workforce-engaged-in-informal-employment-labour-force-survey-reveals/
  Wage-gap percentages (5% in 2024-25, 18% in 2020-21) are computed ratios.
- Female LFPR on both bases (13th ICLS 24.4%, 19th ICLS 22.7%; overall 47.7%
  v 46.3%), female LFPR by region (rural 29.3%, urban 12.8%, 19th ICLS),
  2.48m own-use subsistence producers, and the questionnaire
  treatment (Q5.10 codes 3-4) — PBS LFS 2024-25 Annual Report —
  https://www.pbs.gov.pk/wp-content/uploads/2020/07/LFS-2024-25-Annual-Report.pdf
- 45pp participation gap "one of the largest in the region" — Gallup analysis
  (link above).

## Night-lights image (Method Notebook, figure 2)

Raw VIIRS radiance, June 2024 monthly composite (NOAA/Colorado School of
Mines, vcmslcfg tile 75N060E), clipped to 60–78°E / 23–38°N so India and
Afghanistan frame Pakistan; log scale, clipped at 30 nW; district boundaries
from Data Darbar. Rendered by Adaad from the raw grid. Worked examples
verified in the tehsil layer: Ghotki gas-belt tehsils Daharki 2.23 / Mirpur
Mathelo 2.92 / Ghotki 3.96 nW v rural-Sindh median 0.49 (district MPI
headcount 64.8%; Daharki RWI percentile 46.5); Aliabad (Hunza) 0.18 nW with
RWI percentile 91.2.


## flfp_prov.csv & lfpr_districts_2020_21.csv (piece 02, figures 2-3)

Provincial female LFPR 2020-21 v 2024-25 (ages 15+): PBS Employment Trend
Report 2025, Table 3.2 —
https://www.pbs.gov.pk/wp-content/uploads/2020/07/Employment-Trend-Report-2025-Final.pdf
(no ICLS label on the provincial table; the report states 2024-25 headline
figures follow the 19th ICLS with key variables back-cast to 13th).
Degree-holder female unemployment 24.1% — same report, Table 3.3.

District refined participation and literacy rates by sex, 131 districts:
LFS 2020-21 district-level report, Annexure-I (KILM by district), parsed from
the published PDF. The 2020-21 round is the only district-representative LFS;
no district-level change can be computed. The scatter's x-axis is census
2023 female literacy (Data Darbar census extract), name-matched for 129 of
131 districts (Duki and Sikandar Abad have no census counterpart);
corr(census female literacy, female LFPR) = −0.05; using the annex's own
literacy rates instead gives −0.10.

## flfp_delta_districts.csv (piece 02, figure 3 — ADAAD ESTIMATE)

Change in rural female labour force participation by district, 2020-21 to
2024-25, for 83 districts of Punjab, Sindh and KP. 2020-21 rates as published
(Annexure-I of the district report); 2024-25 rates computed by Adaad from the
PBS public-use microdata —
https://www.pbs.gov.pk/wp-content/uploads/2020/07/LFS-2024-25-STATA.zip —
with districts recovered from the processing-code strata
(https://www.pbs.gov.pk/wp-content/uploads/2020/07/CODING-SCHEME-LFS-2024-25.pdf)
and validated against census 2023 rural populations (median |log error| 0.08)
and the 2020-21 code list
(https://www.pbs.gov.pk/wp-content/uploads/2020/07/Coding_Scheeme_LFS_2020-21_Districts_Codes-1.pdf).
13th-ICLS-comparable employment reconstruction replicates published rates
exactly (47.7/69.8/24.4; rural 52.3, urban 40.8). Rural domain only;
Balochistan excluded (division-level strata). corr(census female literacy,
change) = −0.05 (−0.03 with n≥200); median change +6.0pp; 59 of 83 rising.
Method script: flfp_delta_method.py (ships with the issue).

## religious_vote.csv, religio_constituency_2024.csv, religio_growth_districts.csv & religio_seats_2024.csv (piece 03)

National Assembly general-seat vote shares of religio-political parties by
election (1977-2024) and by tradition (Deobandi, Barelvi, multi-sect
alliance, Jamaat-e-Islami, Shia, Ahl-e-Hadith), plus 2024 constituency-level
shares on the 2023 delimitation (266 seats). Aggregated from ECP returns by
Aiwan-e-Jamhoor (CC BY 4.0) — https://hibasameen.github.io/aiwan-e-jamhoor/islam.html ;
classification per its methodology page. Headlines: 2.6% (1988) → 11.9%
(2024); previous peak 12.4% (2002, MMA alliance); 2024 top parties TLP
(Barelvi) 4.9%, JUI-F (Deobandi) 3.8%, JI 2.3%; 7 NA seats won (6 Deobandi,
1 Shia); 49 of 266 constituencies at ≥20%. Map geometry: simplified 2023
delimitation boundaries from the same archive.

Corrections and additions for publication (13 Aug 2026):

- **NA-8 Bajaur**: the April 2024 by-election (general poll postponed after a
  candidate's assassination) included a Sunni Ittehad Council candidacy
  functioning as a PTI proxy (Gul Zafar Khan, 47,282 votes, 25.9%). SIC is
  doctrinally Barelvi but this vote is not religious mobilisation; it is
  excluded here. NA-8 corrected to 10.1% (JI), removing it from the >=30%
  list (20 -> 19 seats). Flagged for correction in the upstream archive.
  https://www.electionpakistani.com/ge2024/NA-8.htm
- **religio_growth_districts.csv**: district mean of constituency religious
  shares, 2013/2018/2024, with pp deltas. Constituencies matched to districts
  by name because delimitations are not comparable across years (2013 on the
  2002 delimitation); cum-district seats are assigned to every component
  district, so those districts share a value. 73 districts match 2013<->2024.
- **religio_seats_2024.csv**: per-seat mechanism classification from the
  archive's 2024 candidate-level file — held (7), corridor_target (13 seats
  >=30% with the religious candidate second), tlp_margin (TLP third or lower
  polling more than the winner's margin over the runner-up; 69 seats incl.
  NA-8 before its correction, 61 in Punjab), high_share_other. Winner,
  runner-up and margin as returned on election night. "Polled more than the
  margin" is an upper bound on leverage, not an attribution of outcomes.
- Piece headline facts: TLP mean share by region — Potohar/north Punjab 8.4%
  (32 seats), central Punjab 7.9%, south Punjab 4.6%, Sindh 3.7%, KP 2.4%.
  Since-2018 district deltas cited in the text: Shikarpur +16.3, Kashmore
  +10.4, Dera Ismail Khan +10.5, Battagram +24.5.

## plotline.csv (piece 04 — Plotline)

Workers' remittances v goods exports, US$bn by fiscal year, FY18–FY26.
FY18–FY24 for both series are from Economic Survey 2024-25, Trade & Payments
chapter (remittances table and PBS customs exports table) —
https://www.finance.gov.pk/survey/chapter_25/8_Trade_and_Payments.pdf
Remittances then: FY25 $38.3bn, FY26 record $41.6bn (SBP; FY23's −14% dip
reflected an exchange-rate gap diverting flows to informal channels).

- https://tribune.com.pk/story/2425599/remittances-decline-by-14-in-fy23
- https://www.outlookbusiness.com/news/pak-receives-record-416-billion-in-remittances-surpassing-total-exports

Goods exports FY25 $32.3bn, FY26 $30.8bn provisional (SBP BoP, FOB); export
bases differ ~1% where they overlap; the FY25 join is declared in the method
note.

- https://www.brecorder.com/news/40104313 (FY21/FY22 context)
- https://www.sbp.org.pk/assets/document/ExportsImports-Goods.pdf
- https://profit.pakistantoday.com.pk/2024/07/03/pakistan-exceeds-export-target-with-30-64-billion-in-fy24

## Method note (carried onto every figure)

National aggregates are official figures as released; no Adaad re-estimation
is involved in this issue's headline numbers. Where two definitions of the
same quantity exist (13th vs 19th ICLS employment), the text names the basis.

## District profile: Rajanpur (piece 06)

All figures sourced: population 2.38m, growth 3.0%, urban share 26.5%,
literacy 36.1% (female 28.2%) — Census 2023 via Data Darbar; female LFPR
37.1% (15th of 131 districts) — LFS 2020-21 district report, Annexure-I;
multidimensional poverty headcount 69.3% (rank 28 of 119, n=1,319) —
PSLM-based MPI via Data Darbar; rural female participation change +23.4pp —
Adaad estimate (see flfp_delta_districts.csv). Punjab and Pakistan
comparators are population-weighted census aggregates; female LFPR
comparators are the LFS 2020-21 refined rates (26.3 Punjab, 21.4 national).
The Sialkot, Kech and Swat prototype profiles were retired; their URLs
redirect to Rajanpur.

## lights_districts.csv

District night-lights against poverty, the Method Notebook's Figure 1.
Radiance is the 2024 VIIRS annual composite, population-weighted across
each district's tehsils (from lights_tehsils.csv; log_radiance = log of
the weighted mean, with Awaran's zero preserved via the log(1+x) scale
used in the figure and the correlation). Poverty headcount is the
district multidimensional headcount from mpi_districts.csv. 119 districts
match; r = -0.69 between log(1+radiance) and the headcount.

