ESG ETFs: index rules, ratings, returns and the limits of sustainable funds

In depth12 min read · 13 sources

ESG ETFs track indices that select or weight companies by environmental, social and governance criteria (ESG). Behind the same label lie very different rules: from excluding a few industries to climate indices with a fixed path for cutting emissions. Here you will learn how these indices are built, what the EU's fund categories mean, why ESG ratings rarely agree, how to calculate carbon intensity and what research says about returns.

What ESG means

ESG brings together three groups of characteristics. Environmental covers, for example, greenhouse gases, water use and biodiversity; social covers working conditions, supply chains and product safety; governance covers the supervisory board, remuneration and anti-corruption. An ESG ETF is usually a variant of a well-known index, the parent index: the index provider takes its stocks, filters or weights them using ESG data and publishes the result with suffixes such as ESG, SRI, Screened, Climate Transition or Paris-Aligned in the name.

What matters is the goal a fund pursues: avoiding risk (companies that could be hurt by climate rules, lawsuits or scandals) or achieving impact. Most ESG ETFs pursue the first goal; they buy shares on the stock exchange, not new power plants.

The EU's SFDR: Articles 6, 8 and 9

In the EU, the SFDR has applied since March 2021 (SFDR 2019). This is Regulation (EU) 2019/2088, the Sustainable Finance Disclosure Regulation. It is a disclosure obligation, not a quality label. Every financial product falls into one of three groups, named after the articles of the regulation:

Because there were no fixed minimum criteria, Articles 8 and 9 were used in the market like labels. In its proposal for the revision, the Commission itself speaks of the articles being misused as quasi-labels, of greenwashing risks and of divergent interpretations (European Commission 2025). An Article 8 fund can therefore exclude very little or a great deal.

Since 2025, the EU rules for fund names have been stricter. Under guidelines from the European securities regulator ESMA, a fund with ESG or sustainability terms in its name must align at least 80% of its investments accordingly. Depending on the term, the exclusions of the EU climate benchmarks apply: for environmental, impact and sustainability terms, the stricter ones of the Paris-aligned benchmarks; for terms such as transition, social or governance, those of the climate transition benchmarks (ESMA 2024). New funds had to apply the guidelines immediately, existing ones by May 2025.

How sustainable indices are built

The rules behind ESG ETFs fall into four groups. Many indices combine several of them.

• Exclusion criteria: companies in certain industries (controversial weapons, tobacco, coal) or with serious violations of principles such as the UN Global Compact are removed. The deviation from the parent index usually remains small.
• Best-in-class: within each sector, only the companies with the best ESG scores remain. Sector weights resemble those of the parent index, the number of stocks falls significantly and the deviation increases.
• EU climate benchmarks: Climate Transition Benchmarks (CTB) and Paris-Aligned Benchmarks (PAB) follow minimum standards set by EU Climate Benchmarks Regulation (2020) (table below).
• Themes and impact: funds focused on clean energy or water concentrate on a few industries and are thus a bet on a theme. Genuine impact investing with measurable impact tends to take place away from the stock exchange.

The regulation measures intensity as emissions per million euros of enterprise value including cash (EVIC). Scope 3 emissions were phased in gradually: first for energy and mining, after two years for transport, construction, buildings, materials and industrials, and after four years for all sectors.

Example: carbon intensity and deviation from the parent index

The most common metric for a fund's climate exposure is the weighted average carbon intensity (WACI): each company's emissions divided by its revenue, weighted by its share of the fund.

For this simple case, the deviation from the parent index can be stated directly. The variant holds the rest of the index in the same proportions, so its return differs only by the share of the exclusion times the return difference between the rest and the excluded part. The tracking error is accordingly the weight of the exclusion times the volatility of this difference.

Why ESG ratings differ so much

An ESG rating condenses hundreds of individual data points into a single score. Unlike credit ratings, which all answer the same question (will the borrower pay?), ESG ratings measure different things depending on the provider. Berg et al. (2022) mapped the ratings of six major agencies onto a common taxonomy of categories and decomposed the divergence.

Measurement means that the same category, such as working conditions, is assessed using different indicators. Scope means that one provider takes lobbying into account and another does not. Weighting plays the smallest role. The authors also found a rater effect: a provider's overall impression of a company colors its assessment of the individual categories. “ESG leader” is therefore a statement about a method, not about the company itself.

What research shows about return and risk

The most important distinction in the research is between expected and realized returns. Pástor et al. (2021) show in an equilibrium model that green assets have lower expected returns, because investors like holding them and because they hedge against climate risks. Nevertheless, they do better when the preference for sustainability rises unexpectedly. This is exactly what Pástor et al. (2022) see in the data: the high returns of green stocks and bonds in recent years reflect unexpectedly increased climate concerns, not high expected returns; by their estimates, the expected returns of green stocks are below those of brown ones.

On the other side are premiums for shunned stocks. According to Hong and Kacperczyk (2009), shares of alcohol, tobacco and gambling companies are held less often by norm-constrained investors such as pension funds, are followed by fewer analysts and had higher expected returns than comparable stocks. Bolton and Kacperczyk (2021) found higher returns in US data for companies with higher emissions, even after controlling for size, valuation and momentum. They interpret this as a premium for climate risk. Aswani et al. (2024) disagree: the relationship depends on estimated rather than reported emissions and on absolute tonnes rather than intensity; with reported data or with intensity, it disappears.

Pedersen et al. (2021) bring both sides together: their ESG-efficient frontier shows the highest attainable Sharpe ratio for each ESG level. Depending on whether ESG data contain information about future profits or merely express preferences, ESG lowers or raises the required return. Because the frontier is flat at its peak, a light ESG filter costs hardly any Sharpe ratio, while a very strict one costs considerably more. In ESMA's analysis, ESG funds lagged comparable funds without ESG features in 2024, and Article 9 funds lagged Article 6 funds (ESMA 2026). All this points to a different profile of industries and risks rather than a reliable excess return from ESG.

Costs and deviation from the parent index

ESG ETFs are hardly more expensive today. In the cost report of ESMA, the EU securities regulator, ESG funds were on average about 5 basis points cheaper than comparable funds without ESG features; for equity ETFs, ongoing costs with and without ESG features were at a similar level (ESMA 2026). The bigger difference is the deviation from the parent index. It depends almost entirely on how much is excluded or reweighted:

• Indices with few exclusions (weapons, tobacco, coal) stay close to the parent index.
• Best-in-class and SRI variants often hold only some of the stocks and deviate significantly.
• Climate indices are usually built through an optimization and shift weight away from energy, utilities and materials.

These sector bets determine the return: if oil and gas stocks rise sharply, an index without them falls behind; if they fall, it comes out ahead. What a portfolio of several funds actually contains is revealed by a look-through.

Limitations: greenwashing, data gaps, Scope 3

• Greenwashing: the name and classification promise more than the rules deliver. In the EU, the new naming rules are meant to curb this; beyond that, the only remedy is to look at the index rules.
• Estimated data: many emissions are not reported but estimated by data providers. Aswani et al. (2024) show that such estimates are closely related to financial metrics. In part, they therefore measure size and industry, not climate performance.
• Scope 3: emissions from the supply chain and from the use of products are often a company's largest source (GHG Protocol 2011), but the hardest to measure. They are also double-counted: a carmaker's Scope 3 emissions are partly the Scope 1 emissions of its steel supplier.
• Per revenue or per enterprise value: intensity per enterprise value falls arithmetically when share prices rise; the EU climate benchmarks have to adjust for this.
• Impact: if you sell a share on the stock exchange, you are only passing it on to someone else. The company gets nothing from it. According to Pástor et al. (2021), impact arises mainly through the cost of capital: if many investors prefer green companies, these find it easier to invest, and brown ones find it harder.

For advanced readers: ESG in the portfolio model

Formally, an ESG filter is a constraint on the optimization: it pushes the portfolio away from the efficient frontier, and the cost can be measured as a loss of Sharpe ratio. Pedersen et al. (2021) show that optimal portfolios can then be built from four fixed building blocks (four-fund separation) instead of two as in classical theory, and they derive an ESG-adjusted CAPM.

In Pástor et al. (2021), the market and an ESG factor jointly price all stocks; an ESG portfolio therefore carries its own non-diversifiable risk. You can check which known factors an ESG index brings with it using a regression on the Fama-French factors. According to Pástor et al. (2022), a green factor explains a large part of the weak performance of value stocks in the years before their study.

The path of the climate benchmarks is also a matter of compounding. If a Paris-aligned index starts 50% below the universe and cuts its intensity by 7% every year, after ten years it stands at 0.5 × 0.93¹⁰ = 24.2% of the universe's initial intensity. If companies' emissions do not fall as fast, the index has to reallocate more and more. Tracking error and turnover rise over time.

How Portfolio Lab calculates it

In the Look-through, the “Sustainability” tab shows what can be said with freely available data, without purchased ESG ratings and without a score of its own. The basis is estimated emissions from individual assets such as power plants, coal, oil and gas production sites, refineries, steel, cement, aluminum and chemical plants, paper mills and ore mines, for the last complete calendar year in CO₂ equivalents, attributed to companies in proportion to ownership. This corresponds to the direct emissions of these assets; the app does not calculate Scope 3 or an intensity per revenue.

Portfolio Lab matches the funds' holdings via ISIN and Legal Entity Identifier (LEI), otherwise via name and country. The results are shares of the portfolio: in companies with coal power plants or coal mining, with oil and gas facilities, with coal, gas or oil power plants and with steel, cement, aluminum or chemical plants (each from 1 million t CO₂e a year), plus the energy and utilities sectors, size classes of attributed emissions and the largest affected holdings. Financials and government bonds are not attributed. A separate table shows the coverage, because no entry does not mean no emissions. A fund's SFDR category appears where it is maintained; otherwise there is only a hint derived from the fund name, marked as an assumption. More in the ESG methodology.

This article explains a method; it is not investment or tax advice. Worked examples are simplified and rounded. All terms in the glossary.