BRSR Sustainability Regime Analysis

Overview

This project analyses Business Responsibility and Sustainability Reporting (BRSR) disclosures from Indian companies to infer latent corporate sustainability regimes using Hidden Markov Models (HMMs) and Natural Language Processing (NLP) techniques.

Motivation

BRSR is mandated by SEBI for large listed companies to enhance transparency in sustainability practices. However, there is often a disconnect between the breadth of disclosure and actual operational integration. This study addresses whether we can distinguish genuine commitment from symbolic compliance by analysing temporal disclosure patterns using probabilistic frameworks.

This analysis is based on an earlier personal project of mine, which can be found here.

Key Features

• Automated Data Extraction: Web scraping of BRSR PDFs from NSE India portal
• NLP Feature Engineering: Six interpretable dimensions of sustainability disclosure
• Hidden Markov Models: Inference of latent sustainability regimes
• Temporal Analysis: Longitudinal firm-level trajectory tracking
• Regime Classification: Three distinct corporate postures identification

Sustainability Regimes

The model identifies three latent regimes based on empirical disclosure patterns:

1. Minimal Compliance: Checklist-driven reporting with low quantification
2. Lobbyist-Influenced Dilution: Procedural language suggesting regulatory resistance
3. Enforcement-Oriented: Explicit targets and capital commitment indicating genuine integration

Project Structure

brsr-regime-analysis/
├── txt.txt                     # Sample BRSR data from NSE India
├── src/
│   ├── __init__.py
│   ├── data_scraper.py          # NSE web scraping functionality
│   ├── feature_extractor.py     # NLP feature engineering
│   ├── hmm_model.py            # Hidden Markov Model implementation
│   ├── visualization.py        # Plotting and analysis utilities
│   └── utils.py                # Helper functions
├── notebooks/
│   └── notebook.ipynb      # Original comprehensive notebook
├── data/
│   ├── raw/                    # Raw PDF downloads
│   └── processed/              # Processed datasets
├── tests/
│   ├── test_data_scraper.py
│   ├── test_feature_extractor.py
│   └── test_hmm_model.py
├── docs/
│   └── technical_details.md
├── requirements.txt
├── setup.py
├── .gitignore
├── LICENSE
└── README.md

Quick Start

Prerequisites

• Python 3.8+
• Chrome browser (for web scraping)
• Git

Installation

1. Clone the repository:

git clone https://github.com/yourusername/brsr-regime-analysis.git
cd brsr-regime-analysis

2. Create a virtual environment:

python -m venv venv
source venv/bin/activate  # On Windows: venv\Scripts\activate

3. Install dependencies:

pip install -r requirements.txt

Usage

Quick Start

# Run complete analysis with included sample data
python run_analysis.py

# Or run with custom output directory
python run_analysis.py --output-dir my_analysis_results

Data Collection

from src.data_scraper import BRSRScraper

# Use included sample data (txt.txt)
scraper = BRSRScraper('txt.txt')
data = scraper.scrape_brsr_data()

# Or scrape from custom HTML file
scraper = BRSRScraper('path/to/brsr_data.html')
data = scraper.scrape_brsr_data()

Feature Extraction

from src.feature_extractor import FeatureExtractor

extractor = FeatureExtractor()
features = extractor.extract_features(pdf_text)

HMM Modeling

from src.hmm_model import SustainabilityHMM

model = SustainabilityHMM(n_states=3)
regimes = model.fit_predict(features)

Complete Pipeline

from src.main import run_analysis

results = run_analysis()

Methodology

Feature Engineering

Six theoretically motivated disclosure dimensions:

1. Commitment Language: Targets, pledges, aspirational statements
2. Metric Intensity: Quantified indicators, percentages, year-on-year comparisons
3. Governance Language: Board committees, accountability mechanisms
4. Capital Allocation: Investment and expenditure signaling
5. Enforcement Language: Audit, verification, penalty references
6. Supply Chain Responsibility: Scope-3 due diligence, supplier standards

HMM Framework

• Baum-Welch Algorithm: Parameter estimation via expectation-maximization
• Viterbi Algorithm: Most probable state sequence identification
• Gaussian Emissions: Multivariate normal distributions for feature vectors
• Z-score Normalization: Within-firm standardization preserving temporal dynamics

Results & Evaluation

Model Validation

• Log-likelihood: Model fit assessment
• Information Criteria: AIC/BIC for model comparison
• Viterbi Accuracy: Against simulated ground truth
• Linguistic Interpretability: Emission parameters analysis

Key Findings

• Clear separation of three distinct disclosure regimes
• Temporal stability within firms with occasional transitions
• Strong correlation between regime assignments and qualitative assessments

Contributing

1. Fork the repository
2. Create a feature branch (git checkout -b feature/amazing-feature)
3. Commit your changes (git commit -m 'Add amazing feature')
4. Push to the branch (git push origin feature/amazing-feature)
5. Open a Pull Request

License

This project is licensed under the MIT License - see the LICENSE file for details.

Acknowledgments

• NSE India for BRSR data accessibility
• SEBI for sustainability reporting framework
• Academic literature on corporate sustainability disclosure

References

• Securities and Exchange Board of India (SEBI) BRSR Framework
• Hidden Markov Models for sequential data analysis
• NLP applications in financial disclosure analysis

---

This research contributes to evidence-based assessment of corporate sustainability disclosure quality and regulatory effectiveness in emerging markets.