Gendered Analysis of Unincorporated Sector Enterprises — ASUSE 2023-24

Investigating the structural barriers facing women entrepreneurs in India's vast unincorporated enterprise sector, using the latest Annual Survey of Unincorporated Sector Enterprises (ASUSE 2023-24) from the Ministry of Statistics and Programme Implementation (MoSPI).

---

📊 Key Findings

Finding	Women	Men
Home-based work	75.2%	27.0%
Formally registered	~10%	~43%
Has hired workers	~3%	~16%
Internet access	~14%	~32%
Median annual income	₹47,040	₹1,57,080
Female hiring ratio (among employers)	96%	4%

The analysis reveals four structural patterns:

• The Invisible Workshop — three-quarters of women entrepreneurs work from home, outside any regulatory, credit or market linkage system.
• The Paper Ceiling — 90% of women-owned businesses are informal/unregistered vs 57% for men.
• The Solopreneur Trap — only ~3% of women are employers; 97% work entirely alone.
• The Sisterhood Effect — women employers hire almost exclusively other women (96%); male employers hire almost exclusively men.

---

🛠️ Methodology

• Software: R 4.1+
• Survey design: Stratified cluster sampling via srvyr / survey packages — all estimates are nationally representative population-weighted figures.
• Design variables: PSU (primary sampling unit = FSU), Strata (2nd-stage stratum), Weight (MLT multiplier).
• Lonely PSU handling: options(survey.lonely.psu = "adjust").
• Outlier treatment: Income capped at ₹1 cr, Assets at ₹5 cr, Loans at ₹1 cr before survey design creation.
• Transgender note: n = 16 sample observations; variance estimates = NaN. Excluded from all charts; included in tabular checks only with a reliability caveat.
• Weight scaling: Raw sum(Weight) ≈ 7.3 billion. Ratios between genders are unaffected by this scale. Standalone population-count claims (e.g. "19.9 crore women entrepreneurs") require sum(Weight) / 100.

---

📂 Project Structure

asuse-gender-analysis/
│
├── scripts/
│   ├── 01_data_pipeline.r       ← Load + clean raw SPSS blocks → RDS
│   ├── 02_analysis_main.r       ← Survey metrics + 7 publication charts
│   └── 03_charts_advanced.r     ← Ridge, violin, lollipop, dumbbell, correlogram
│
├── charts/
│   ├── Chart1_...png  …  Chart7_...png   ← from 02_analysis_main.r
│   └── advanced/                          ← from 03_charts_advanced.r
│       ├── A_dumbbell_manufacturing_gender_gap.png
│       ├── B_lollipop_disadvantage_profile.png
│       ├── C_ridge_income_by_gender_location.png
│       ├── D_violin_asset_distribution.png
│       ├── E_correlogram_financial_structural.png
│       ├── F_dodged_problem_constraints.png
│       └── G_panel_four_exclusion_dimensions.png
│
├── data/
│   └── processed/
│       └── chart_data_survey_adjusted.rds
│
└── docs/
    ├── 2026-02-19-gender-gaps-asuse.md   ← Article draft
    ├── block02_columns.txt               ← Column reference
    ├── block08_columns.txt               ← Column reference
    ├── Integrated Analysis.txt
    └── archive/
        └── 2022-23_column_discovery.r    ← Diagnostic script (old round)

Raw data (ASUSE202324sav/*.sav) lives outside this repo at C:/Users/ashwin/Documents/IndiaSpend Article/ASUSE202324sav/ and is not tracked by Git (see .gitignore).

---

🚀 How to Run

# Step 1 — Build the analysis-ready dataset (~5 min)
source("scripts/01_data_pipeline.r")

# Step 2 — Survey-weighted metrics + 7 charts
source("scripts/02_analysis_main.r")

# Step 3 — Advanced visualisations (ridge, violin, lollipop, correlogram)
source("scripts/03_charts_advanced.r")

---

📦 R Package Requirements

install.packages(c(
  "haven", "tidyverse", "labelled", "janitor", "skimr",
  "naniar", "assertr", "lubridate", "srvyr", "survey",
  "scales", "ggrepel", "ggridges", "ggdist", "corrplot", "patchwork"
))

---

📜 Data Source

Ministry of Statistics and Programme Implementation (MoSPI), Government of India. Annual Survey of Unincorporated Sector Enterprises (ASUSE) 2023-24. mospi.gov.in