Paper: arXiv 2305.12364

Authors: Taeisha Nundlall, Terence L Van Zyl

Abstract

Socially responsible investors build investment portfolios intending to incite social and environmental advancement alongside a financial return. Although Mean-Variance (MV) models successfully generate the highest possible return based on an investor’s risk tolerance, MV models do not make provisions for additional constraints relevant to socially responsible (SR) investors. In response to this problem, the MV model must consider Environmental, Social, and Governance (ESG) scores in optimisation. Based on the prominent MV model, this study implements portfolio optimisation for socially responsible investors. The amended MV model allows SR investors to enter markets with competitive SR portfolios despite facing a trade-off between their investment Sharpe Ratio and the average ESG score of the portfolio.

Complexity vs Empirical Score

  • Math Complexity: 4.0/10
  • Empirical Rigor: 6.0/10 (capped from 7.0: transaction costs not modelled; see penalty flags)
  • Quadrant: Street Traders — empirical and applied, lower math complexity (effective empirical rigor capped from 7.0)

Why this score: The paper primarily relies on standard Mean-Variance optimization and basic machine learning algorithms like Random Forest, which are established and computationally dense rather than mathematically advanced. However, it demonstrates high empirical rigor through detailed data preprocessing, use of real-world datasets (Yahoo Finance, VettaFi), backtest-ready procedures, and evaluation of forecast errors.

Research Flowchart

  flowchart TD
  A["Research Goal: Extend Mean-Variance Model<br>with ESG Constraints"] --> B["Data & Inputs"]
  B --> C["Methodology"]
  C --> D["Computational Process"]
  D --> E["Key Findings/Outcomes"]

  subgraph B ["Data/Inputs"]
      B1["Multi-Asset Historical Data"]
      B2["ESG Scores per Asset"]
      B3["Investor Risk Tolerance"]
  end

  subgraph C ["Methodology"]
      C1["Classical Mean-Variance Optimization"]
      C2["Constrained Optimization<br>with ESG Score Weights"]
  end

  subgraph D ["Computational Process"]
      D1["Calculate Covariance Matrix"]
      D2["Solve for Efficient Frontier<br>max Return per Volatility"]
      D3["Iterate for ESG-Weighted Solutions"]
  end

  subgraph E ["Key Findings/Outcomes"]
      E1["Competitive SR Portfolios"]
      E2["Trade-off Analysis:<br>Sharpe Ratio vs. Avg ESG Score"]
      E3["Practical Framework for<br>Socially Responsible Investors"]
  end

  B1 --> C1
  B2 --> C2
  B3 --> C2
  C1 --> D1
  C2 --> D2
  D1 --> D2
  D2 --> D3
  D3 --> E1
  D3 --> E2
  D3 --> E3

Evidence Card

What the authors claim, on what data, and who has checked it. Read from the paper's own text by a language model: every quote is verbatim, nothing was re-run. How to read this card

Proposes

The paper extends the classic Mean‑Variance portfolio model to jointly optimise Sharpe ratio and average ESG rating (ESG‑MV model).

Source sentence
we propose to extend the MV Model to include ESG-rating optimisation.

Assumes 3

  • Historical price data are limited to trading days between 2011‑11‑30 and 2021‑11‑30 to avoid excessive back‑filling. data
    Source sentence
    Our study limits the use of historical data to trading days between 2011-11-30 and 2021-11-30 to avoid excessive back-filling for the 80% of the ETFs that were non-existent before 2011-11-30.
  • No historical ESG ratings are available, so ESG scores cannot be used to improve optimisation precision. data
    Source sentence
    Due to the available data on the ESG ratings, we cannot use historical ESG ratings to improve the precision of our portfolio optimisation, as was done in Vo et al. [16].
  • A favourable Sharpe ratio is greater than 1. model
    Source sentence
    Sharpe ratios are computed to relate the return of an investment to the risk of such an investment, where a favourable Sharpe ratio is greater than 1.

Data

  • VettaFi ESG dataset · VettaFi fundamentals public
    U.S. ETFs with ESG scores
    Source sentence
    This study uses the ESG scores from this dataset, ranging from 0 to 10.
  • Yahoo Finance ETF price data · Yahoo Finance market-prices public
    1993-01-29 to 2021-11-30 · U.S. ETFs
    Source sentence
    Yahoo Finance has released datasets containing U.S ETFs and their historical prices from 1993-01-29 to 2021-11-30.

Recognised sources: Yahoo Finance

Costs

Not modeled A trading or execution result is reported without costs.

Finds 2

  • Random forest outperformed the deep learning models (LSTM and CNN) in testing, and was used for forecasting all ETFs. predictive-accuracy
    Source sentence
    In evaluating model performance, the RF model outperformed the deep learning models during model testing. Consequently, the RF model was used to forecast the adjusted closing values of all ETFs.
  • ESG‑MV models yield competitive portfolios but with a slight trade‑off: lower Sharpe ratio for higher average ESG scores. performance
    Source sentence
    Using historical and forecasted data, this study has shown that ESG-MV models produce competitive portfolios with a slight trade-off between a lower Sharpe ratio for higher mean ESG scores of the portfolio.

Checked by others

No other paper in the archive has been found to support or contradict these findings (compared with 14 related papers). Coverage is partial, so this is neither replication nor refutation.

Code

No repository by the authors found in the paper.

Extracted 2026-11-14 · same card as JSON · A language model chose these items: check the source sentence before relying on one.