Can Better Diversification Improve Sustainable Investing?

Why ESG Portfolios Still Need Better Diversification

A portfolio can hold companies with strong environmental, social, and governance ratings while remaining vulnerable to a narrow set of market risks. Selecting attractive businesses and combining them into a resilient investment portfolio are related tasks, but they require different decisions.

A new study, “Sustainable portfolio optimization with four objectives: Assessing the role of diversification,” examines that distinction directly.¹ Researchers compared portfolios optimized for return, downside risk, and sustainability with portfolios that also treated diversification as an explicit objective.

Their findings offer a practical lesson: diversification deserves its own place in portfolio construction. Adding it generally produced broader allocations and lower volatility, although investors sometimes sacrificed returns and accepted modestly lower ESG scores. The opportunity lies in making these compromises deliberate and understandable.

Why ESG Scores Cannot Replace Portfolio Diversification

ESG scores summarize selected characteristics of individual companies. Diversification concerns how investments behave together. A collection of highly rated businesses can still share exposure to interest rates, technology spending, consumer demand, or the same geographic market.

Consider a hypothetical portfolio containing several software companies with strong ESG ratings. Their business practices may satisfy the investor’s sustainability preferences, yet their share prices could react similarly to rising borrowing costs. Adding another highly rated software business might improve the portfolio’s average ESG score without meaningfully changing that exposure.

This distinction also applies to funds. Owning several ETFs does not necessarily eliminate concentration if their largest holdings and economic sensitivities overlap. Securities.io’s recent examination of risk in digital economy ETFs provides relevant context for looking beyond the number of investments held.

The study’s central question follows naturally: if diversification matters independently, what happens when an investment model is explicitly instructed to pursue it?

How the Sustainable Portfolio Study Tested Diversification

The researchers constructed two related models. Both sought higher expected returns, lower Value-at-Risk, and higher portfolio ESG scores. The second added a fourth goal: increasing the Diversification Ratio.

In its volatility-based form, this ratio compares the weighted average volatility of individual holdings with the volatility of the combined portfolio. It captures how much combining investments reduces fluctuations relative to holding exposures that move together. Counting securities alone cannot provide that information.

The analysis used daily market data and ESG scores from LSEG Data & Analytics. Its four investment universes contained stocks drawn from major European, British, and American indices.

Portfolios were estimated using rolling 500-trading-day windows, then evaluated over the following 20 trading days, with rebalancing every 20 days. This out-of-sample approach tests allocations against subsequent observations rather than judging them solely on the data used to construct them.

An evolutionary optimization algorithm, ev-MOGA, searched for portfolios offering different combinations of the objectives. Investors could then express preferred directions, such as placing greater emphasis on sustainability or downside protection. The framework acknowledges that different investors can reasonably choose different portfolios from the same opportunity set.

Lower Portfolio Risk Came With Measurable Trade-Offs

Across the four datasets, explicitly targeting diversification consistently lowered out-of-sample volatility and Value-at-Risk while producing broader allocations. Those improvements generally came with lower returns and modest reductions in ESG scores.

Results were more varied for maximum drawdowns and risk-adjusted performance. In the Euro Stoxx 50 analysis, adding diversification improved Sharpe and Sortino ratios for all five selected model portfolios. Returns increased for some investor profiles and declined for others.

Conventional approaches also remained competitive. The global minimum-variance benchmark delivered particularly strong financial results in the European case, including the lowest volatility and strongest risk-adjusted performance, but had the weakest ESG score.

That comparison makes the research more useful. It shows the price of satisfying additional preferences rather than suggesting that a sophisticated model automatically beats simpler alternatives. Sustainable portfolio construction becomes a negotiation among objectives, supported by evidence.

The Real Innovation Is Making Investor Choices Explicit

A broader implication is that portfolio software should explain what investors receive for each compromise. An ESG score, volatility estimate, or return projection becomes more informative when presented alongside the alternatives available.

For example, an investor might accept a small reduction in a portfolio’s ESG score if it meaningfully reduces dependence on one industry. Another might prioritize sustainability even when doing so narrows the eligible investment universe. Neither choice can be evaluated properly through a single headline metric.

Securities.io’s coverage of ESG education and retail fund flows highlights the importance of helping investors translate preferences into investment decisions. This study addresses the next step: showing what those preferences imply for actual allocations.

A useful portfolio discussion therefore needs three questions:

• Which sustainability characteristics must the portfolio preserve?
• Which shared market exposures should it limit?
• What return or implementation trade-offs are acceptable?

These questions can improve investment decisions without requiring investors to reproduce the researchers’ algorithm.

A Higher ESG Score Does Not Establish Real-World Impact

Another important distinction concerns what the sustainability objective measures. This study optimizes portfolio ESG scores. That does not independently demonstrate lower real-world emissions, improved labor conditions, or additional financing for environmentally beneficial projects.

A portfolio can change its reported characteristics simply by changing its holdings. Whether that change influences corporate behavior is a separate question requiring different evidence.

The distinction is timely. On October 12, 2026, the European Securities and Markets Authority published additional guidance addressing impact and transition claims. Clear descriptions of investment objectives help investors understand what a strategy actually seeks to accomplish.

For portfolio designers, the implication is straightforward: specify the desired outcome first. Managing sustainability-related financial risks, excluding particular activities, and pursuing measurable impact are different goals. A model can only optimize the characteristics it has been given.

Where Portfolio Optimization Still Needs Improvement

The research also identifies limitations that matter for implementation. Its historical sample spans six years, and the findings depend on a particular ESG dataset. Different rating methodologies or market environments could produce different allocations.

Transaction costs were not explicitly included in the optimization. With rebalancing permitted every 20 trading days, trading expenses could affect realized results. For taxable accounts, realizing gains could introduce another cost that the reported portfolio comparisons do not resolve.

Value-at-Risk also describes a loss threshold rather than the severity of losses beyond it. A portfolio with an improved VaR estimate can still experience a substantial extreme loss.

A practical extension would therefore compare diversification benefits after trading costs and taxes, alongside stress scenarios and alternative downside measures. The objective would be to establish whether the improvements remain useful under the conditions investors actually face.

MSCI Offers Exposure to Portfolio Analytics Infrastructure

These implementation challenges create a commercial connection to companies building portfolio analytics. MSCI provides one relevant avenue for investors interested in the technology supporting increasingly customized investment strategies.

Its optimization services combine portfolio construction capabilities with inputs including risk models, ESG and climate information, and transaction-cost models. This places the company within the infrastructure used to translate investment preferences into implementable allocations.

MSCI’s July 2026 paper on portfolio construction and competing constraints describes a related industry challenge: connecting investment objectives, portfolio exposures, risk controls, and execution within a consistent workflow.

The investment connection is thematic. The attached study does not establish MSCI’s adoption of its algorithm or predict additional company revenue. Investors assessing MSCI would still need to examine valuation, competition, and the commercial performance of its analytics businesses.

Latest MSCI (MSCI) News and Developments

Building Sustainable Portfolios Around Clear Priorities

The study strengthens the case for treating diversification as an explicit design choice. Strong company ratings and historical risk controls do not automatically ensure that a portfolio spreads its exposures effectively.

For investors, the most useful advance is greater visibility into competing objectives. A well-designed process should reveal how sustainability preferences affect concentration, how diversification changes expected outcomes, and whether those benefits survive implementation costs. Better analytics can support that process by making the choices clearer.

References:

¹ Garcia-Bernabeu, A., Hilario-Caballero, A., & Luciani, F. (2026). Sustainable portfolio optimization with four objectives: Assessing the role of diversification. Operations Research Perspectives. Advance online publication. https://doi.org/10.1016/j.orp.2026.100420