R&D Alpha: Investment Intensity and Long-Term Stock Returns

Empirical evidence on the relationship between corporate R&D investment intensity and subsequent long-term equity returns.

Live Platform: https://research.finsoeasy.com
Full Paper (PDF): https://research.finsoeasy.com/rnd-alpha-paper.pdf

Abstract

This research examines whether companies with high research and development (R&D) expenditures relative to revenue generate superior long-term stock returns. Using data from S&P 500 constituents over 30 years (Jul1995-Jun2025), we find that firms in the top quintile of R&D intensity outperform bottom-quintile firms by approximately 3.73 percentage points annually.

The core thesis: R&D spending is expensed rather than capitalized under GAAP and IFRS, systematically understating the true economic value of innovation-intensive companies. This accounting treatment creates a persistent mispricing opportunity.

Statistical significance is established via:

• Fama-MacBeth cross-sectional regressions (p = 0.0737)
• Monthly factor spanning tests (FF5 alpha = 4.37%, p < 0.01)

Key Findings

Metric	Value	Period
Annual HML_RD Premium	+3.73%	Jul1995-Jun2025 (30 years)
Win Rate (Annual)	57%	17 of 30 years positive
RD20 Net Premium vs SPY	+7.52%/yr	Jul2001-Jun2025 (24 years)
FF5 Alpha	+4.37%/yr	Monthly spanning (p < 0.01)

Repository Structure

rd-alpha-research/
├── src/
│   ├── scoring/           # R&D intensity scoring algorithm
│   ├── backtesting/       # Portfolio backtest methodology
│   ├── factors/           # Factor premium calculations
│   └── data/              # Data acquisition utilities
├── docs/                  # Methodology documentation
├── notebooks/             # Jupyter notebooks for replication
└── data/                  # Data README and samples

Replication

Prerequisites

python >= 3.10
pandas, numpy, scipy
sqlalchemy (optional, for database storage)

Quick Start

# Clone the repository
git clone https://github.com/vastdreams/rd-alpha-research.git
cd rd-alpha-research

# Install dependencies
pip install -r requirements.txt

# Run the scoring algorithm on sample data
python src/scoring/rd_alpha_scorer.py --sample

Data Acquisition

Financial data is sourced from Financial Modeling Prep (FMP) API. To replicate:

1. Obtain an API key from financialmodelingprep.com
2. Set environment variable: export FMP_API_KEY=your_key
3. Run data ingestion: python src/data/ingest_financials.py

See data/README.md for detailed data provenance and schema.

Methodology

Scoring Formula

R&D Alpha Score = (RD_Intensity × Sector_Adj × Momentum × Quality) / Volatility

Where:

• RD_Intensity: R&D Expense / Revenue, capped by sector
• Sector_Adj: S&P 500 sector weight / High-R&D sector weight
• Momentum: 1 + (Prior 3-year excess return × 0.1), bounded [0.5, 2.0]
• Quality: Data quality score (0 to 1)
• Volatility: 3-year historical standard deviation, floored at 0.10

Backtest Protocol

• Formation Date: July 1 (Fama-French convention)
• Holding Period: 12 months with annual rebalancing
• Data Lag: Uses FY(T-1) financials for T-year formation
• Universe: S&P 500 constituents (point-in-time)
• Transaction Costs: 18.3 bps round-trip (Novy-Marx & Velikov 2016)

Citation

@techreport{sehgal2026rdalpha,
  author = {Sehgal, Abhishek},
  title = {R&D Alpha: Investment Intensity and Long-Term Stock Returns},
  year = {2026},
  month = {January},
  institution = {FSE Research & Investments Pty Ltd},
  url = {https://research.finsoeasy.com},
  note = {Working paper, Version 1.0}
}

License

MIT License. See LICENSE for details.

Acknowledgments

This research builds upon foundational work in factor investing and R&D capitalization literature, including studies by Lev & Sougiannis (1996), Chan, Lakonishok & Sougiannis (2001), Eberhart, Maxwell & Siddique (2004), and Ahmed, Bu & Ye (2023).

Author

Abhishek Sehgal
ORCID: 0009-0000-9424-4695
FSE Research & Investments Pty Ltd