Unlanded: Distribution of Land in Bihar

Bihar, the third largest state in India, with over 100M people, is also its poorest---the GDP ppp is a mere $630. Agriculture constitues 23% of the economy which makes land ownership a vital source of income for many people.

Using the Bihar land records data---plot-level land records (N= 41.87 million plots or 12.13 individuals/accounts across 35,626 villages)---we describe the distribution of ownership of land, and how ownership varies by caste, religion, and gender. We also show ownership by last name and list top 10 land owning first and last names.

In particular, we investigate the distribution of a) the number of plots and b) total land area by account number. We also check how the ownership of land is distributed across castes and religion and using naampy, we infer what proportion of land is registered to women. We compare these to the base distribution.

Data

• data

Scripts

All scripts are in scripts/.

Summary & Comparisons

Script	Purpose
00_summary_basic_bihar_land_record.ipynb	Basic summary stats of Bihar land records
01_secc_gender_caste_proportions.ipynb	SECC gender/caste proportions comparison
02_plot_districts.ipynb	District-level visualizations

Data Processing Pipeline

Script	Purpose
10_land_distribution_bihar.ipynb	Core land distribution analysis
20_get_ryot_hindi_caste.ipynb	Extract unique names, assign caste codes
30_get_religion.ipynb	Infer religion using pranaam
40_translate_hindi_to_english.ipynb	Transliterate Hindi to English
50_get_gender.ipynb	Infer gender using naampy
get_caste_outkast.ipynb	Infer caste using outkast

Distribution Analysis

Script	Purpose
60_land_distribution_bihar_religion_coded.ipynb	Distribution by religion
61_land_distribution_bihar_gender.ipynb	Distribution by gender
61_land_distribution_bihar_religion_pranaam.ipynb	Distribution by religion (pranaam)
62_land_distribution_bihar_caste.ipynb	Distribution by caste
caste_outkast_dist.ipynb	Caste distribution via outkast
muslims_dist.ipynb	Muslim population land distribution
benchmark-ratio-viz.ipynb	Benchmark comparisons

Geographic & Admin

Script	Purpose
crosswalk_district_block_village.ipynb	Geographic hierarchy mapping
merge_district_block_village.ipynb	Merge administrative units
flagged_land_accounts.ipynb	Analysis of flagged accounts

LLM Annotation

Scripts in scripts/llm_annotation/ for batch name classification via OpenAI API:

• annotate_names_batch.py - Batch annotation via OpenAI Batch API
• annotate_names.py - Streaming annotation with retry logic
• schema.py - Pydantic models for annotation responses
• prompt.py - System prompt for name classification

Utilities

Scripts in scripts/utilities/:

• utils.py - Data loading and processing functions
• graph_utils.py - Visualization utilities

Workflow

make setup   # Create venv and install dependencies
make idata   # Build intermediate datasets (caste, religion, gender)
make build   # Create data/figure directories

Authors

Aaditya Dar, Lucas Shen, and Gaurav Sood