GREEN VS. NON-GREEN PORTFOLIO PERFORMANCE ACROSS BRICS NATIONS: A MULTI-COUNTRY ANALYSIS OF EFFICIENCY, SCALABILITY, RISK-ADJUSTED RETURNS, AND MACHINE LEARNING FORECASTING

Abstract

The current paper presents a cross-country comparison of the performance of green and non-green investment portfolios across the five BRICS countries: Brazil, Russia, India, China, and South Africa. Using portfolio optimization methods, including Maximum Sharpe Ratio, Minimum Variance, Naive Equal Weight, Black-Litterman, and Sortino Ratio, we compare risk-adjusted returns, volatility, and stability of sustainability-oriented (green) and traditional (non-green) asset portfolios. Examples of asset classes studied include domestic equities, green and non-green cryptocurrencies, commodities, and renewable energy stocks. Historical market data is also used to train and evaluate machine learning models to assess forecast strength and portfolio generalization in unseen market situations. Findings suggest that non-green portfolios tend to deliver higher absolute returns, especially in India and Brazil, at the expense of higher volatility. Despite lower returns across various optimization strategies, green portfolios show improved and more consistent risk-adjusted performance (measured using the Sharpe and Sortino ratios) in most countries, with notable exceptions in South Africa. These results are further corroborated by efficient frontier analysis, which reveals structural differences in the emerging market risk profiles across BRICS economies. The evaluation of machine learning demonstrates moderate predictive ability, and overfitting during the testing stages is observed in certain cases, especially with non-green portfolios. These results have important implications for institutional investors, policymakers, and practitioners in the field of sustainable finance who must balance two sets of demands in emerging markets: financial profit and environmental responsibility.