Skill or sustainability? Deconstructing the stock-selection timing abilities of European ESG fund managers

Environmental, Social and Governance (ESG) investing has grown in popularity in recent years. This growth is driven by massive flows into sustainable funds, reflecting investors' rising awareness of sustainability concerns. The remarkable expansion of socially responsible investing (SRI) is evident in the enormous rise in assets under management, which surpassed $41 trillion by 2022 globally, up from $22.8 trillion in 2016. This substantial capital represents one third of the total assets worldwide according to the Global Sustainable Investment Association ^[1]. This trend is anticipated to continue, with analysts predicting that ESG assets will grow from $50 trillion in 2025 to $180 trillion by 2034, exhibiting a CAGR of 18.8% during this forecast period ^[2]. Europe is by far the leading region in ESG concentration with a market share of 44% in 2025 ^[3]; concomitantly, research conducted by PwC Luxembourg predicts a 50% increase in ESG-oriented assets in the European Union by 2027 ^[4]. Notwithstanding the ethical and environmental considerations central to their strategy, ESG investors remain focused on their primary investment objective of achieving competitive financial returns.

During the last decade, a series of exogenous shocks (e.g. COVID-19, the Russia-Ukrainian war) has impacted the global economic environment, triggering severe stock market crashes. Thus, it is imperative for a fund manager to be able to predict market fluctuations and appropriately adjust the composition of the managed portfolio. Managers achieve this by making the portfolio more aggressive during market upswings and more conservative during bearish markets—an ability referred to in the literature as market timing ability. The findings in the existing literature regarding the market timing abilities of conventional fund managers remain inconclusive, with some studies confirming the absence of such skills (Gao et al., 2020; Koutsokostas et al., 2018; Wagner and Margaritis, 2017; Cuthbertson et al., 2016; Babalos et al., 2015), while others detecting successful timers (El Ammari et al., 2023; Vidal et al., 2015; Benos and Jochec, 2011; Bollen and Busse, 2001). These outcomes depend significantly on the data frequencies and methodological approaches employed.

Studies on SRI fund timing abilities are limited and also produce mixed results. While Ang et al. (2014) imply that managers are capable of effectively timing the market, Das and Rao (2015) argue that they are generally unsuccessful in delivering positive return timing and often exhibit a detrimental trade-off where timing attempts undermine stock selection. When comparing the market timing skills of SRI funds with those of their traditional peers, empirical findings typically indicate that the differences are not statistically significant (Matallín-Sáez et al., 2019; Muñoz et al., 2015; Leite and Cortez, 2014; Ferruz et al., 2010). These market-timing discrepancies remain insignificant even when accounting for the post-2022 macroeconomic environment, marked by the global energy crisis and a growing political backlash against ESG (Alda, 2025; Baklaci et al., 2024). Some studies though (Ji et al., 2021; Naqvi et al., 2021), affirm that renewable energy funds perform worse in terms of market timing than their conventional counterparts, further fueling the ongoing controversy.

Even though sustainable finance is receiving increasingly more scholarly attention, a significant gap remains in the granular assessment of active managerial skills within the ESG industry. Given the inherent trade-offs imposed by ESG mandates, the current conflicting evidence regarding the timing abilities of ESG funds suggests the necessity for a refined theoretical framework. ESG screening constraints the investment universe, potentially hindering a manager's execution efficiency (Pástor et al., 2021), even if it aligns with investors' preferences for lower downside risk (Bauer and Smeets, 2015). In contrast to previous research, this study shifts the focus to the “micro-timing” of individual stock selections rather than macro-market timing. We argue that traditional timing metrics, which often penalize low turnover, are inappropriate to evaluate timing skill within this domain, as ESG managers prioritize long-term investment horizons. Consequently, we isolate stock-selection timing as a distinct dimension of performance, by utilizing the holdings-based approach of Jiang et al. (2021). This method allows us to ascertain whether a manager's ability to time trades is actually impaired by the restricted liquidity and the narrower investment universe inherent in ESG portfolios. Alternatively, we can determine if these structural limitations are successfully offset by the manager's specialized information set. Thus, this research offers a more accurate evaluation of ESG managerial skill by disentangling these components, providing more robust insights than the aggregate performance metrics employed in previous studies.

This study examines the stock-selection timing abilities of ESG-compliant equity mutual funds over the period from January 1, 2012, to December 31, 2024. Our sample comprises 309 equity mutual funds domiciled in the UK, France, and Germany, representing Europe's primary sustainable investment hubs. Unlike studies relying on top-weighted securities, we utilize monthly portfolio holdings that include all fund positions. We use the stock-selection timing test of Jiang et al. (2021) and conduct bootstrap analysis to determine whether the observed results are attributed to skill rather than pure luck. We also investigate whether this stock-selection timing ability is more pronounced within the ESG segment compared to their conventional counterparts. Moreover, given the profound market volatility induced by the COVID-19 and the Russian invasion of Ukraine, we explore how these exogenous shocks influenced managers' ability to time stock-selection opportunities. To document the economic value generated for investors, we contrast the subsequent performance of positive and negative stock-selection timers across various horizons. Finally, we analyze the determinants of stock-selection timing ability, with a particular focus on its relation with fund flows.

The contribution of our study lies in the following aspects. First, to the best of our knowledge, this is the first study to provide evidence on the stock-selection timing skills of European ESG fund managers. While the performance of sustainable mandates has been extensively discussed in the literature, the timely execution of individual trades has been largely underexplored. We fill this gap by investigating whether ESG managers' specific strategies result in a superior micro-level information advantage. This is especially pertinent given the rapid growth of ESG assets under management and the increasing demand from investors to comprehend the precise factors influencing alpha in sustainable portfolios. Second, we offer a novel approach for assessing how ESG managers strike a balance between the necessity of active trade execution and their long-term goals. In this way we can assess more precisely whether the restrictions of ESG portfolios constrain a manager's ability to time stock-selection opportunities or whether their specialized focus provides a distinct edge. Third, we link fund portfolio holdings with real-world market shocks. By considering the volatile period following the onset of the pandemic, we provide a comprehensive understanding of how ESG fund managers navigate systemic crises compared to their conventional peers.

The rest of the paper is organized as follows. Section 2 critically reviews the extant literature and develops the research hypotheses. Section 3 describes the methodology and the data used in the study. In Section 4 the empirical results are presented and Section 5 concludes.

We base our hypotheses on a constrained portfolio optimization framework in order to assess how sustainability mandates affect stock-selection timing ability. Under classical portfolio theory, an unconstrained manager maximizes timing alpha using private informational signals. By limiting the investable universe, explicit ESG screens alter this objective function. This mathematically shifts the efficient frontier inward and creates a structural “timing drag.” This investment boundary determines the relative stringency of the portfolio constraint, limits a manager's sector signal elasticity, compresses the signal-to-noise ratio during macro-volatility, and introduces unique principal-agent utility trade-offs across institutional regulatory environments.

According to Ang et al. (2014), from January 2001 to December 2011, North American and European SRI funds showed significant market timing abilities. On the other hand, Das and Rao (2015), demonstrate a lack of managerial skill in forecasting future market movements, which contradicts their results. Crucially, both studies rely on the classical methodological frameworks of Treynor and Mazuy (1966) and Henriksson and Merton (1981). Nevertheless, these traditional models fail to account for the investment constraints imposed by ESG criteria and the long-term investment objectives inherent in sustainable strategies. Therefore, they are fundamentally ill-suited for evaluating the timing abilities of ESG fund managers.

Beyond these conceptual misalignments, conventional returns-based approaches are incapable of effectively disentangling a lack of skill from a deliberate low-turnover strategy. Standard regressions may misinterpret the absence of frequent trading as a deficiency in timing ability, as ESG fund managers often maintain high-conviction positions over extended horizons. Even if improperly specified, the existence of positive stock-selection coefficients in the studies of Ang et al. (2014) and Das and Rao (2015) suggests that ESG managers might, in fact, possess latent informational advantages. Building on this, we apply the holdings-based measure of Jiang et al. (2021) to effectively capture the timely execution of managers' trades. Accordingly, our first research hypothesis is formulated as follows:

Theoretically, the “cost” of sustainability can be interpreted under Modern Portfolio Theory, which implies that any restriction on the investable universe inevitably causes the efficient frontier to move inward. Leite and Cortez (2014) offer empirical support for this, arguing that a wider investment universe correlates with better stock-selection abilities. They show that SRI funds investing in European equities are significantly worse stock pickers than funds investing globally. Other studies have produced conflicting findings regarding the comparison of the timing abilities of sustainable funds with those of conventional funds. The differences in timing skills between SRI and traditional managers are frequently reported as insignificant, according to studies that focus on developed markets (Muñoz et al., 2015; Ferruz et al., 2010). Matallín-Sáez et al. (2019) confirm that the performance gap between SRI and conventional funds disappears when business cycles are considered, though they find that SRI funds may lag during expansionary periods. Extending this to recent structural changes, research examining the post-2022 environment shows that the statistical differences in timing between sustainable and conventional structures continue to be insignificant despite the severe sector rotations brought on by the global energy crisis and the backlash against ESG (Alda, 2025; Baklaci et al., 2024).

However, investors may incur a “sustainability premium” in the form of reduced timing efficacy, as suggested by recent evidence from specialized green sectors. Ji et al. (2021) and Naqvi et al. (2021) show that traditional energy funds outperform renewable energy funds in the Eurozone and emerging markets, respectively. This implies that, as managers are restricted from investing in high-opportunity sectors that do not adhere to the ESG criteria, the ESG screening procedure creates a structural “timing drag”. Our second research hypothesis is built on the reasoning that ESG limitations reduce a manager's “elasticity” in responding to market opportunities:

The effectiveness of managerial abilities, though, is heavily influenced by the macroeconomic environment and the prevailing market conditions. According to Jitmaneeroj (2023), market factors, such as recessionary prospects, can impact fund managers' skills. Managers may favor market timing-wide returns and liquidity over individual stock-selection, denoting a shift in managerial focus. This indicates that, during crisis periods, ESG managers are redirected toward managing macro-volatility, which consequently diminishes the accuracy of their stock-selection timing. On the other hand, increased market volatility causes abrupt and frequent price changes that erode the reliability of fundamental ESG signals. In such environments, the “noise” of the market may mask the “latent informational advantages” typically associated with strong ESG positions. Thus, the structural constraints of ESG screening and the need for defensive macro-positioning during periods of turmoil lead us to state our third study hypothesis:

Following Jiang et al. (2021), who show that stock-selection timing is a key factor in mutual fund performance, we argue that these skills are economically significant. To overcome the structural “timing drag” imposed by the narrowed investable universe, the precise execution of trades is essential for ESG managers. If ESG managers have latent informational advantages, these should translate into superior returns (Baklaci et al., 2024). Our fourth research hypothesis is, therefore, formed as follows:

On the other hand, low-rated ESG funds outperform their high-rated counterparts, indicating a performance penalty for rigorous ESG adherence (Papathanasiou and Koutsokostas, 2024). We contend that since stricter ESG screening confines the investment options, a high ESG rating is associated with weaker stock-selection timing. This limitation restricts a manager's “elasticity” and prevents them from switching to high-potential stocks that do not satisfy high sustainability scores. Therefore, it is anticipated that the ability to precisely select stocks would decrease as ESG requirements become more stringent. Thus, our fifth research hypothesis is as follows:

Undoubtedly, investor behavior and managerial incentives are greatly influenced by the institutional and regulatory environment of a fund's home nation. France has established itself as a regulatory pioneer in sustainable finance and has developed an ESG framework that is notably stricter than those in use in Germany or the UK. We posit that investors derive significant utility from non-performance characteristics in such a high-oversight setting, which leads them to prefer funds with better ESG credentials. Indeed, prior empirical evidence indicates that French socially responsible investors may have to pay a price for ethics (Leite and Cortez, 2015). Consequently, even if this comes at the expense of economic performance, we anticipate large capital inflows toward funds with high ESG scores. On the other hand, as we expect these abilities to be negatively correlated with the most rigorous ESG screens, funds that demonstrate strong stock-selection timing skills may experience outflows. Thus, in order to secure assets under management, French managers may be under pressure from investors to prioritize “sustainability compliance” over active market timing. To empirically explore this potential association between ESG compliance and fund flows, we build the following exploratory hypothesis:

In conclusion, although earlier studies have examined sustainable market timing, there is a significant gap regarding the impact of ESG-specific mandates and differing regulatory stringency on the stock-selection timing abilities of fund managers. By shifting the focus from traditional returns-based models to the tactical execution of ESG trades under different market and institutional conditions, our research hypotheses address this gap in the literature.

The dataset of our empirical research is retrieved from LSEG Datastream, encompassing equity mutual funds that adhere to ESG criteria within the most concentrated European markets; i.e. the UK, France, and Germany. Our sample consists of funds that have a domestic investment focus and do not hold extensive ^[5] foreign stocks in their portfolios. The rationale behind this approach is to ensure the homogeneity of our results. When a fund holds foreign investments, comparing the stock-selection timing abilities of fund managers becomes difficult because they are confronted with different cycles, market trends and available information. This could lead to inaccurate assessment of a manager's stock-selection timing ability. Additionally, by focusing on these specific domestic universes, we can account for institutional variations, regulatory differences, and currency risks, even though ESG funds often operate under global or Europe-wide mandates.

We require the FTSE 100, CAC 40, and DAX 30 ^[6] to be the primary investment benchmarks for ESG funds domiciled in the UK, France, and Germany, respectively. By purposefully isolating relatively large-cap, developed-market stocks, the use of these major benchmarks ensures a consistent and highly standardized reporting environment across our sample. We specify that at least twenty-four monthly portfolio holdings be available for each fund in our sample, as it often takes several months for a manager to formulate a comprehensive investment strategy. In order to mitigate potential incubation bias ^[7], we do not include funds with an average TNA of less than €5 million in the sample, since incubated funds tend to be small, as highlighted by Kacperczyk et al. (2005). We also exclude funds holding fewer than ten stocks from the sample to assess potential stock-selection timing abilities more accurately. Following these restrictions, our final sample consists of 309 equity mutual funds classified as ESG over the sample period; 198 are domiciled in the UK, 62 in France, and 49 in Germany. The timespan covers the period from January 1, 2012, to December 31, 2024. Fund holdings are matched strictly by their actual portfolio effective dates rather than their publication dates to account for reporting lags. The sample period is considered highly volatile, as several exogenous shocks provoked intense fluctuations in the markets. Therefore, it is imperative to analyze a period where market conditions did not provide consistent opportunities for active managers to select outperforming stocks.

Our sample is designed to isolate established funds that have a long-term ESG mandate. To ascertain that each fund maintained a consistent ESG mandate throughout the sample period, a two-stage verification process was implemented. First, we confirmed each fund's current ESG status by cross-referencing the Lipper Responsible Investment attributes—which require a prospectus-driven sustainable mandate—with fund-level overall ESG scores. However, as our sample begins in 2012, we do not filter by SFDR (Article 8/9) classifications to prevent look-back bias. Additionally, we do not apply an arbitrary numerical score threshold, since providers frequently make ex-post modifications to historical ESG values. Second, to verify the sustainability mandate was established at the start of our sample period, we scrutinize fund inception dates and historical investment objectives within LSEG workspace. Any fund that switched from a conventional to an ESG strategy after 2012—a process known as “green-rebranding”—was not considered. This ensures that our findings represent the constraints of a long-term sustainable framework rather than a mid-period change in investment approach, enabling us to retain a uniform sample of full-cycle ESG managers. We acknowledge that the reliance on active funds introduces survivorship bias in our study; however, this is a purposeful methodological trade-off. We specifically focus on funds that have shown operational and strategic persistence, since ESG investing is intrinsically defined by a long-term investment horizon. In this way, we deliberately avoid volatile, transient funds and instead concentrate on stable, institutional-grade funds.

Table 1 presents the time series averages of the cross-sectional mean and median for several fund characteristics, such as fund TNA, expense ratio, fund age, cash holdings, number of stocks included in the portfolio, normalized fund flow, fund return, ESG score, and active share. All fund characteristics are winsorized at the 1st and 99th percentiles to reduce the impact of extreme outliers.

As shown, ESG funds domiciled in Germany are larger in size, older in age and hold more cash in their portfolios. French mutual funds are rated higher in terms of ESG score and charge investors higher expense ratios. Funds operating in the UK hold more stocks in their portfolios, attract greater flows and deliver relatively better raw returns. Finally, in most cases, the means of the funds' characteristics are greater than the corresponding medians, indicating a high possibility that the distributions are right-skewed.

To examine whether ESG fund managers have the ability to time stock-selection opportunities, we employ the methodology proposed by Jiang et al. (2021). The basis for this strategy stems from the premise that fund managers should trade more actively when the market provides greater stock-selection opportunities, as they face transaction costs that can erode performance. Thus, if fund managers are able to forecast a stock-selection opportunity at t + 1, they will adjust their active trading appropriately at t. Therefore, the empirical framework is defined by the following equation:

where ACT_i,t denotes active trading at time t, E_t[SSO_t+1] represents the expected stock-selection opportunity at time t+1 and g_i is a measure of the fund manager's stock-selection timing ability.

By using realized stock-selection opportunity measure SSO_t+1 as a proxy for E_t[SSO_t+1], the aforementioned regression model takes the following form:

This specification is based on the standard rational expectations framework, as described by Jiang et al. (2021), where SSO_t+1 = E_t[SSO_t+1] + η_t+1, and η_t+1 represents an unforecastable white-noise error term. The model's composite error term ε_i,t+1 = u_t − g_iη_t+1 naturally absorbs the expectation error under this assumption, producing unbiased estimates of g_i. A positive and statistically significant g_i coefficient indicates that the fund manager possesses stock-selection timing ability. Conversely, a negative and statistically significant coefficient g_i suggests that the fund manager incorrectly times stock-selection opportunities.

We compute the stock-selection opportunity (SSO) measure using the full constituent lists of the relevant benchmarks, even though ESG funds are subject to specific investment limitations and exclusion screens. By utilizing the unconstrained index as a baseline, we can determine more precisely whether the observed results are the product of managerial skill or the structural “opportunity cost” imposed by ESG constraints. This approach is essential for evaluating whether the timing abilities of ESG managers differ from the broader market baseline. Such a benchmark is necessary to capture the absolute opportunity cost of missing unconstrained alpha, even though the screened universe represents the immediate target set for sustainable funds. Furthermore, the unrestricted and ESG-screened universes share a very high correlation in their underlying risk factor exposures since our sample is restricted to major large-cap indices (FTSE 100, CAC 40, DAX 30). This ensures that the full index provides a robust and unbiased baseline for cross-sectional opportunities.

We measure fund activeness by using active share (AS), following Cremers and Petajisto (2009). Active share is defined by the following equation:

where ω_s,i,t and ω_s,indexi,t are the portfolio weights of stock s in fund i and in the relevant benchmark index in period t.

As shown in the last row of Table 1, fund activeness follows a similar pattern across the sampled countries. Funds domiciled in the UK are relatively more active (35%) compared to funds based in France (33%) and Germany (32%). The time series of active share for each market is plotted in Panel A of Figure 1, which shows significant variations in fund activeness over time. Over the sample period, fund activeness was lower during 2017–2018 and reached its highest level preceding the outbreak of the COVID-19 pandemic.

We compute the alpha value of each stock by using the five-factor Fama and French (2015) model with the inclusion of Carhart (1997) momentum:

where R_s,t is the return of stock s, R_f,t is the risk-free rate, R_m,t is the market return, and SMB_t, HML_t, RMW_t, CMA_t, and MOM_t denote the returns on factor-mimicking portfolios for size, book-to-market, profitability, investment, and momentum pattern, respectively. The risk factors are calculated based on the conceptual framework developed by Otten and Bams (2002). All stocks listed on the respective benchmark indices are included. Initially, stocks are categorized according to their market capitalization at the end of the preceding month. The top 30% of stocks by market capitalization are assigned to the “big” portfolio, while the bottom 30% constitute the “small” portfolio. The SMB factor is calculated by subtracting the return of the big portfolio from that of the small portfolio. A similar approach is employed to construct the High minus Low (HML), Robust minus Weak (RMW), Conservative minus Aggressive (CMA), and Momentum (MOM) factors, with stocks being ranked based on their book-to-market ratio, operating profitability, investment intensity, and prior performance, respectively.

To measure stock-selection opportunities, we estimate the average positive six-factor Fama and French (FF6) alpha value using daily returns over the monthly period t+1. Stocks with positive alphas, indicating outperformance relative to their expected returns, represent attractive investment opportunities for actively managed mutual funds. We address look-ahead and overlapping estimation biases to ensure robust SSO construction. First, to avoid look-ahead bias, we use rolling historical windows ending immediately prior to month t+1 to estimate factor loadings in Eq. (4), ensuring that no future data is utilized. Second, we compute SSO across distinct calendar months rather than rolling windows to eliminate overlapping estimation bias and secure statistical independence between consecutive periods. The time series of the stock-selection opportunity based on the average positive FF6 alpha of individual stocks is plotted in Panel B of Figure 1. The figure illustrates significant variation in stock-selection opportunities over time, with the German market providing more chances for picking winning stocks, especially during 2018.

We apply the stock-selection timing test described in Eq. (2) for each fund to acquire the timing estimates (g_i), along with their corresponding t-statistics. Table 2 presents the cross-sectional distribution of the t-statistics for the timing coefficients (g_i), computed using the stock-selection opportunity measure described in Eq. (4). The table shows the percentage of funds with t-statistics exceeding or falling below specific cutoffs.

The empirical results reported in Table 2 indicate that the percentage of ESG funds with stock-selection timing ability is higher than the percentage of funds unable to time stock-selection opportunities promptly. For the full sample, approximately 21.68% of the funds have t-statistics for the g_i coefficients exceeding 1.96 under the Fama and French (2015) six-factor model ^[8]. On the other hand, approximately 12.62% of the funds have t-statistics below −1.96, exhibiting negative stock-selection timing. The kernel density of the cross-sectional distribution of the Newey-West t-statistics for all funds is plotted in Figure 2. The plot illustrates that the right tail of the t-statistics is thicker than the left tail, which is consistent with the findings in Table 2.

Further analysis reveals that the superior stock-selection timing in these funds is largely attributable to UK-domiciled entities. More specifically, 25.76% of the ESG funds based in the UK display positive and statistically significant timing coefficients, compared to 9.60% of funds with negative and statistically significant g_i estimates. Conversely, French (14.52%) and German (14.29%) ESG funds displaying such ability falls short of the negative-timer group; we find 17.74% and 18.37% of the French and German funds miscalculating stock-selection opportunities, respectively. Thus, we observe cross-country differences in stock-selection skill.

The most pronounced difference is found in the sustainability disclosure regimes that apply to each market. The EU Sustainable Finance Disclosure Regulation (SFDR) imposes rigorous requirements on funds operating in France and Germany. This regulation reduces information asymmetry by mandating standardized, public disclosure regarding ESG alignment. Consequently, ESG data is incorporated into security prices more quickly due to this increased transparency. Thus, the lower positive stock-selection timing in these markets can be explained by the diminished opportunity for fund managers to produce timing-based alpha from early interpretations of underlying ESG signals. In contrast, the UK's Sustainability Disclosure Requirements (SDR) were only introduced toward the end of our sample period in 2024. UK funds were operating in a pre-SDR context marked by a less harmonized and more fragmented disclosure system. Therefore, this lack of standardization may have created significant opportunities for informational arbitrage.

Table 3 provides a comparative summary of the regulatory environments in the UK, France, and Germany, to further elucidate these cross-national disparities. The information environment in which fund managers operate is probably determined by the structural variations in these regimes.

Long before the broader EU SFDR took effect, France established a highly efficient ESG information environment through the early application of Article 173 (now Article 29). Since the “early-mover” advantage on ESG signals is neutralized when such data is already incorporated into baseline prices, this institutionalized transparency likely explains why French managers exhibit lower stock-selection timing ability. Similarly, the SFDR's strong transparency regulations and BaFin's sustainable finance strategy likewise restrict the potential for timing-based alpha in German funds. Conversely, the UK's delayed transition to a standardized SDR framework created opportunities for active managers to exploit fragmented data to achieve superior stock-selection timing coefficients.

We recognize that other localized mechanisms may potentially contribute to these cross-country differences, even if these regulatory divergences offer a convincing institutional explanation. Timing alpha generation may be impacted by variables such as variations in sector concentrations across local benchmarks, disparities in domestic equity market liquidity, or different cultural preferences across regional investor bases. Thus, our regulatory framing should be interpreted as a primary explanatory channel rather than an exclusive causal driver.

Overall, our findings partially corroborate the evidence reported by Jiang et al. (2021), as only UK-based ESG funds demonstrate strong stock-selection timing ability, which also provides partial support for research Hypothesis 1 (H1). Nevertheless, the percentage of positive timers among the ESG funds in our study is substantially lower than the corresponding percentage reported by Jiang et al. (2021) for traditional US funds.

To confirm the robustness of our results, we use an alternative measure as a proxy for fund activeness. We deliberately avoid using returns-based models as a robustness check, to sidestep the low-turnover “penalty” inherent in their design. As argued in Section 2, traditional regressions frequently ascribe ESG managers' stable, high-conviction positions to a lack of timing skill. By utilizing the holdings-based measure and its interaction with the portfolio turnover ratio, we provide a test that is structurally aligned with the ESG investment horizon, ensuring that our findings are not distorted by the methodological drawbacks of returns-based frameworks. Following Yan and Zhang (2009), we define fund portfolio turnover ratio as follows:

where $Buysi,t and Sellsi,t$ are fund i's total purchases and sales of stocks in month t, and $TNA̅i,[t−1,t]$ is the fund i's average total net asset value over months t-1 and t. The cross-sectional distribution of the t-statistics for the timing coefficients (g_i) and the percentage of funds with t-statistics exceeding or falling below specific cutoffs are presented in  Appendix 1.

As shown, our results remain largely consistent when using portfolio turnover as an alternative measure of fund activeness. This reinforces our main finding that UK-domiciled ESG funds engage in timely trades, thereby displaying significant stock-selection timing, whereas French and German funds do not exhibit such skill.

We perform a bootstrapping approach to extract statistical inferences for stock-selection timing coefficients in the spirit of Jiang et al. (2021). The bootstrap analysis addresses problems associated with statistical inference in the cross-sectional distributions of stock-timing coefficients and their t-statistics when they are obtained from a finite sample under the normality assumption. This is because some funds may display significant t-statistics by random chance, even if they do not actually exhibit true timing skills (Kosowski et al., 2006). Another issue is that the independent and identically distributed assumption across timing measures can be violated due to correlation among fund activeness when funds follow similar investment strategies (Cao et al., 2013). For the precise methodology of the bootstrap estimation, refer to  Appendix 2. Table 4 presents the bootstrapped p-values alongside the cross-sectional Newey-West t-statistics of the stock-selection timing coefficients.

The table shows that the positive stock-selection timing coefficients generated by the top percentile funds are not purely attributable to luck. For instance, all sampled ESG funds exhibit high timing coefficients in the 90th, 95th, and 99th percentile of the distribution, equal to 3.04, 4.59, and 6.08, respectively. These results demonstrate significant stock-selection timing ability, since the equivalent p-values are close to zero. In other words, the t-statistics from the actual data are higher than the corresponding t-statistics based on pseudo datasets under the assumption of no timing ability, providing evidence of substantial stock-selection timing skills. On the other hand, the empirical findings also indicate that the bottom 10% of ESG funds with adverse stock-selection timing is attributed to a lack of skill and not merely to pure randomness. Similar patterns emerge when we examine the distributions of t-statistics for the UK market. Top-ranked funds display significant positive stock-selection timing abilities, while bottom-ranked funds demonstrate negative timing skills. In contrast, ESG funds domiciled in France and Germany lack such skills on average, as we find the corresponding timing coefficients at crucial cutoffs to be higher in absolute value for negative timers compared to positive ones.

We conduct the stock-selection test among conventional funds to identify any substantial disparities between ESG and non-ESG funds, given that traditional funds do not integrate ESG criteria into their investment decisions. Due to the vast number of conventional funds in circulation, we identify and select 309 funds that best approximate ESG fund attributes in terms of size, age, and investment focus, using the methodology provided by Hoberg et al. (2018). We represent the style characteristics of fund i in month t using an N-dimensional vector ^[9], denoted as V_i. Fund j is classified as a rival of fund i if the elements within its style characteristic vector V_j exhibit a high degree of proximity to the respective elements in the style characteristic vector V_i in terms of their absolute magnitudes. We define d_ij as the distance measurement between i and j. Assuming V_i [n] represents the nth element within the vector V_i, we can define the pairwise distance between funds i and j, denoted by d_ij, as follows:

Reduced distance scores between funds i and j are indicative of a greater probability of rivalry between these two entities. The balance diagnostics for the ESG and matched conventional funds are shown in  Appendix 3. As it is observed, the average fund size for ESG funds (€513.10 million) is statistically identical to that of the matched conventional funds (€508.45 million, t = 0.28, p = 0.780). Similarly, no statistically significant differences are detected for fund age (p = 0.734) or investment focus scores (p = 0.881). These results confirm that our Hoberg et al. (2018) matching methodology resulted in a perfectly balanced control sample of conventional funds, as the two samples are almost indistinguishable statistically in terms of their style characteristics. The distribution of t-statistics for the timing coefficients (g_i) for the rival conventional funds is shown in Table 5.

The empirical findings documented in Table 5 suggest that conventional funds outperform ESG funds in terms of stock-selection timing abilities. By applying the Fama and French (2015) six-factor model, around 23.95% of the funds exhibit significant timing ability (t-statistic of g_i ≥ 1.96). Conversely, about 11.00% of the funds demonstrate negative timing ability (t-statistic of g_i < −1.96), suggesting they are ineffective at timely picking winning stocks. Similar to ESG funds, the superior stock-selection timing skills of traditional funds are mainly concentrated among UK-domiciled entities. In particular, UK-based conventional funds have a much higher rate of positive and statistically significant timing coefficients (27.78%) compared to negative ones (8.59%). While UK conventional funds show strong timing abilities, French and German funds are more prone to mistiming stock selections, with 14.52% and 16.33% of each group showing negative timing, respectively. Nevertheless, traditional French and German funds still perform better than their ESG counterparts. Therefore, funds that do not prioritize ESG factors display stronger stock-selection timing abilities, lending support to Hypothesis 2 (H2).

The discrepancies reported could be attributed to the fact that conventional fund managers often have a broader investment universe, allowing them to react more quickly to market shifts. ESG funds may encounter limitations due to their focus on specific criteria (Parikh et al., 2023), which could constrain their ability to pivot rapidly. Furthermore, conventional fund managers often have more flexibility to change their investment styles or strategies based on market conditions. In contrast, ESG funds might face constraints due to their commitment to specific ESG principles (Dasilas, 2025), limiting their capacity to adapt to rapidly changing environments. In addition, some conventional fund managers emphasize short-term returns, making them more likely to adjust their portfolios based on near-term economic indicators or market sentiment (Hilliard et al., 2020). ESG funds, with a longer-term focus on sustainability, may be less reactive to short-term fluctuations.

Beyond these operational constraints, these differences may also be the result of underlying selection effects. First, manager self-selection may be a determining factor, as portfolio managers with specialized tactical trading skills may inherently prefer unconstrained conventional funds. Second, because ESG investors are generally less sensitive to short-term underperformance (Bollen, 2007), ESG managers face less pressure to aggressively time the market and experience lower redemption rates, which may alter manager incentives.

We apply the stock-selection timing test among ESG funds focusing on the period after the COVID-19 outbreak and subsequent geopolitical tensions in order to assess whether the stock-selection timing ability of ESG funds were compromised by these major events. In particular, the continuous five-year span from January 1, 2020, to December 31, 2024 is defined as this turbulent period. Although the 2020 pandemic outbreak and the 2022 Russia-Ukraine conflict are two different triggers within this period, macro-financial instability persisted because supply shocks caused by the pandemic directly led to geopolitical commodity shocks and historic monetary tightening. This environment posed unique challenges for fund managers, requiring them to make rapid and difficult investment decisions (Papathanasiou et al., 2025). Therefore, by evaluating stock-selection timing abilities separately for the post-pandemic period, we can gain valuable insights into how fund managers responded to a crisis of unprecedented magnitude and identify those who were particularly adept at navigating volatile markets. Table 6 presents the distribution of the timing coefficients (g_i) during this turbulent period.

The empirical evidence presented in Table 6 suggests that a significant portion of ESG funds exhibited a notable deficiency in their ability to accurately predict market movements and capitalize on price fluctuations through active trading strategies. In contrast, a smaller subset of these funds demonstrated a superior aptitude for identifying and exploiting favorable market conditions. A mere 8.39% of the ESG funds displayed t-statistics for their g_i coefficients that exceed the critical value of 1.96. Conversely, a substantial 24.16% exhibited t-statistics below −1.96, indicating a lack of stock-selection timing skills. A more in-depth examination reveals that the superior ability of UK-domiciled entities to time stock-selection opportunities diminished significantly after the outbreak of the pandemic. 22.05% of ESG funds based in the UK show negative and statistically significant timing coefficients, implying a tendency to enter the market too late. In contrast, only 7.69% of these funds have positive and statistically significant g_i estimates, suggesting an ability to time their stock purchases. Furthermore, we observe an even greater percentage of French funds (29.31%) mistiming their stock-selection decisions, while German funds closely followed suit with 26.67%. In conclusion, our findings suggest that COVID-19 and the Russian invasion of Ukraine exerted a severe negative impact on managers' capacity to accurately predict market movements and gain from price fluctuations, providing evidence in favor of Hypothesis 3 (H3).

We conduct a sensitivity analysis using a Heckman (1979) two-step selection model to address the possibility of survivorship bias during the volatile 2020–2024 period. This accounts for the possibility that funds surviving this highly volatile period may possess safer asset allocations or consistently better risk management skills than their liquidated counterparts, which could bias our timing coefficients upward. We estimate a survival probit based on the same fund characteristics used in our matching methodology (size, age, and investment focus) to generate an Inverse Mills Ratio (IMR). The IMR was then included as a control variable in our timing regression ACT_i,t = c_i + g_iSSO_t+1 + δIMR_i + ε_i,t+1. Results in  Appendix 4 show that across all regions, the estimates of the Inverse Mills Ratio are statistically insignificant (with p-values ranging from 0.356 to 0.588). This suggests that there is no substantial correlation between stock-selection timing skill and the unobserved factors that determine fund longevity, such as defensive positioning or idiosyncratic risk management strategies. Furthermore, the adjusted timing coefficients remain within a marginal range of the initial estimates (e.g. changing from −0.182 to −0.179 for the full sample), indicating that the cross-sectional distribution of t-statistics shown in Table 6 is not substantially affected by survivorship bias. The structural reality of failing funds, which usually lower active risk and compress toward benchmark weights before liquidation or merger, is consistent with this statistical outcome. Thus, it is highly unlikely that their exclusion will overstate the capabilities of the remaining sample or omit concealed timing skills. As a result, our baseline conclusions are not qualitatively distorted by the selection mechanism, which is isolated.

We also perform the stock-selection timing test on conventional funds to contrast the performance of ESG fund managers with that of their conventional peers during turbulent periods. Table 7 presents the distribution of the timing coefficients (g_i) for conventional funds.

Comparing the two samples, we observe that conventional funds underperformed their ESG counterparts in stock-selection timing. This result is likely attributed to the “flight to quality” effect reported in the literature (Fang and Parida, 2022), as managers pivoted toward more resilient, sustainable assets. The general underperformance of both fund types suggests that the heightened market volatility led to frequent and sudden price changes, making it challenging for managers to identify trading opportunities. However, the relatively poorer performance of conventional funds may stem from their greater exposure to sectors most susceptible to pandemic-related shocks, such as traditional energy or travel. In contrast, ESG-tilted portfolios may have benefited from lower idiosyncratic risk.

Furthermore, the economic shocks caused by the pandemic could have made it more difficult for fund managers to assess the fundamental value of companies and their future prospects. The additional stress and anxiety caused by these two “black swan” events may also have exacerbated behavioral biases among fund managers (Mohanty et al., 2024), leading to suboptimal decision-making and impaired timing abilities. Due to the lack of an ESG-based risk-filtering framework, conventional managers may have been more prone to these biases and the “noise” produced by rapid changes in monetary and fiscal policy (Samitas et al., 2022; Chudik et al., 2021). This could have led to more pronounced risk-averse behavior, such as a propensity to hold onto losing investments for too long or avoiding potentially profitable trades due to fear of incurring losses.

We proceed with our empirical findings by examining whether ESG funds with stock-selection timing skills perform better than ESG funds that do not exhibit such abilities. If stock-selection timing is a critical aspect of fund managerial skill, we anticipate that positive timers will outperform negative timers in the subsequent period. We estimate g_i using Eq. (2) and its associated t-statistic, relying on observations of fund activity and stock-selection opportunities over the previous 12 months. Specifically, we categorize funds into two groups—positive and negative timers—based on the significant t-statistic cutoffs (±1.96). Each month, we create portfolios of funds with positive and negative stock-selection abilities and then track the performance of these portfolios over the next 6, 12, 18, and 24 months, using both equal-weighted and value-weighted returns. Panel A of Table 8 shows the equally-weighted abnormal performance of positive and negative ESG timers, as well as the return spreads between them.

Our findings indicate that funds skilled in stock-selection timing consistently outperform those that are not over the subsequent six, twelve, eighteen, and twenty-four months. For example, with t-statistic cutoffs of ±1.96, positive timers among funds domiciled in France yielded risk-adjusted returns of 1.19%, 0.69%, 0.42%, and 0.39% over the subsequent 6-, 12-, 18-, and 24-month horizons, respectively. These returns are 2.72% (t = 2.27), 2.32% (t = 1.90), 1.82% (t = 1.52), and 1.86% (t = 1.55) higher than those of negative timers. For funds domiciled in Germany, positive timers also outperform negative timers, with abnormal returns being significantly higher by 4.47% (t = 3.71), 3.14% (t = 2.62), 2.14% (t = 1.78), and 0.89% (t = 1.19), respectively. An even larger return spread is documented for the funds based in the UK, with the performance difference being statistically significant regardless of the time horizon. These conclusions remain consistent when employing value-weighted portfolios (Panel B), further validating Hypothesis 4 (H4).

The abnormal performance documented in Table 8 is calculated excluding the impact of realistic trading frictions. To quantify transaction costs, we use a range of 50–100 bps to account for total execution costs, including bid-ask spreads and market impact. This threshold is frequently employed to account for liquidity restrictions in specialized equity segments. The magnitude of the return spread consistently exceeds these expanded implementation costs, suggesting that the outcomes represent robust net-of-cost gains rather than artifacts of frictionless rebalancing. For example, since portfolio sorting occurs at long horizons (6–24 months) and turnover is low, annualized costs would need to exceed 232 bps to entirely eliminate our lowest significant return spread (2.32% for France at the 12-month horizon). Nevertheless, the performance gap is most pronounced at the six-month horizon.

These findings can be ascribed to numerous contributing factors that restrict the persistence of stock-selection timing benefits. First, modern financial markets are highly efficient, making it increasingly difficult to consistently predict price movements. New information is often immediately incorporated into asset prices, limiting opportunities for long-term abnormal returns (Manser and Schmid, 2009; Florin et al., 2005). This holds true especially for ESG-related information, which frequently triggers sharp price changes as institutional interest grows, narrowing the window for profitable execution. Second, active trading, often associated with timing strategies, can incur significant transaction costs that can erode profits (Adams et al., 2022). Notably, the rapid alpha decay indicates that the stock-selection timing abilities depend more on transient, micro-level information than on structural market inefficiencies. Overall, the empirical findings highlight that stock-selection timing produces significant, albeit short-lived, economic value for ESG investors.

Table 9 reports the results pertaining to the turbulent period after the onset of COVID-19.

The empirical findings reveal the dissolution of the stock-selection timing premium, with the performance spread between positive and negative timers being statistically insignificant across all investment horizons. The systemic, non-diversifiable shocks during this period erased the micro-level information—the foundation of timing alpha—making short-term stock direction predictions highly challenging. Fund managers struggled to determine optimal entry and exit points, thereby neutralizing the disparity in performance between the two groups. Moreover, the continuous, albeit statistically non-significant, positive return spread implies that portfolios held by positive timers may have benefited from the documented “flight-to-quality” effect. Consequently, this resulted in passive risk resilience and downside protection throughout the crisis, rather than active timing alpha.

How much investors can profit from investing in funds with stock-selection timing skill is an interesting question. We build portfolios of positive and negative stock-selection timing funds and monitor their performance over time in order to provide an answer to this question. Funds are classified each month as positive or negative stock-selection timing funds based on Newey-West t-statistic cutoffs of ±1.96. Portfolio returns are computed in the subsequent month for positive and negative stock-selection timing funds, respectively. To clearly illustrate these dynamics, the cumulative returns of equally weighted positive and negative stock-selection timing fund portfolios are plotted in Figure 3. The starting value of each portfolio was €1 in January 2012. All UK-based fund returns were converted from British pounds (GBP) to euros (EUR) using daily spot exchange rates.

The plot shows that the portfolio of positive timing funds consistently outperforms the portfolio of negative timing funds. With an investment of €1 in January 2012, the positive timing fund portfolio reaches accumulated value of €2.06 at the end of December 2024, whereas the negative timing fund portfolio amounts to €1.81. Notably, following the onset of the pandemic, the performance gap between the positive and negative stock-selection timing portfolios narrowed significantly.

In this section, we examine the key factors shaping the ability of ESG managers to time stock-selection opportunities. To do so, we apply a panel regression to explore the association between stock-selection timing ability and several fund characteristics:

where $tˆ$_i,t represents the t-statistic of fund i's stock-selection timing coefficient (g_i) from the regression in Eq. (2) over the rolling [t, t+1] months and X_i,k,t-1 denotes the fund's k^th characteristic, including active share, fund size, fund age, expense ratio, fund return, return volatility, normalized fund flow ^[10], cash holdings, the number of stocks in the portfolio, and the ESG rating. All explanatory variables are lagged by one month, except for the ESG score, which is lagged by twelve months.

Following Jiang et al. (2021), we utilize lagged values for all independent variables to address potential endogeneity and simultaneity bias between stock-selection timing ability and fund characteristics. Specifically, using lagged active share ensures that fund attributes are pre-determined relative to the period in which timing skill is measured. This temporal separation reduces the likelihood that the observed relationship is driven by contemporaneous shocks. Additionally, we apply natural logarithmic transformations to fund size and fund age in accordance with Kacperczyk et al. (2008), which consistently minimizes scaling dependencies and colinear overlaps across fund structural parameters. Our panel specification also includes time fixed effects to further reduce multicollinearity. These effects absorb time-varying market dynamics, unobserved macro-level shocks, common market cycles, and sentiment shifts that may otherwise induce cross-sectional correlation among the independent variables. Furthermore, the use of robust standard errors clustered at the fund level accounts for potential heteroskedasticity and within-fund residual autocorrelation, ensuring that our statistical inferences are not distorted or artificially inflated by confounding data dependencies. The results for Eq. 7 are presented in Table 10.

As shown, active share is positively, though insignificantly, correlated with stock-selection timing skill across the full sample. This implies that funds with greater deviations from the benchmark weights tend to time stock-selection opportunities more effectively. To explore potential non-linearities and determine whether this relation is driven by high- or low-conviction funds, we follow Jiang et al. (2021) and include a dummy variable d, which equals 1 if the fund's active share is above the median and is 0 otherwise. The results indicate that the coefficient of the interaction term is positive and statistically significant (t = 2.01) for the entire sample. This suggests that the positive relation between active share and stock-selection timing is non-linear and driven by high active stock pickers. These findings align with Cremers et al. (2022) and Petajisto (2013) who argue that the most active stock pickers outperform closet indexers. In terms of economic magnitude, the full-sample coefficient of 0.561 indicates that high active share funds increase their stock-selection timing t-statistic by more than half a point relative to low active share funds. This mechanism is plausible given that ESG factors often involve rapidly evolving news and events, such as environmental disasters, social controversies, or regulatory changes. Consequently, a high active share allows ESG fund managers to quickly adjust their portfolios in response to these events, potentially capturing short-term price movements.

Furthermore, the positive and statistically significant coefficient for fund return— documented for both the UK funds and the entire sample—confirms that the stock-selection timing contributes significantly to performance, representing a key aspect of managerial ability. However, in contrast to Jordan and Riley (2015) who identify return volatility as a powerful predictor of manager skill, we find no statistically significant relation between stock-selection timing and return volatility.

Moreover, the effect of ESG ratings on stock-selection timing is negative and statistically significant across all countries. This suggests a potential disadvantage for high-rated ESG funds in achieving positive timing, thereby bolstering Hypothesis 5 (H5). Our empirical design includes key mechanisms to mitigate potential reverse causality, particularly the concern that successful funds may afford higher ESG integration. By utilizing a twelve-month lagged ESG score, we ensure that current timing outcomes cannot affect past ESG ratings. Additionally, whereas time fixed effects account for changes in market sentiment, our panel specification directly controls for lagged fund returns (Fund return_i,t-1), which absorbs the performance dynamics that might otherwise finance future ESG budgets. Furthermore, we control for fund size (TNA_i,t-1) to ensure that the negative ESG-timing relationship does not simply capture the trading constraints associated with larger asset bases. Our findings are consistent with the findings of Rompotis (2022) and Ghoul and Karoui (2017) who support the view that socially responsible investors pursue utility derived mainly from non-performance attributes, as high-ESG funds often display poorer performance. Our results are also in line with Bofinger et al. (2022) who suggest that higher fund sustainability is associated with fund overpricing, and, consequently, a lack of investment skill. From an economic perspective, the full-sample coefficient of −0.097 implies that a 10-point increase in a fund's ESG score reduces its timing t-statistic by 0.97, an adverse effect that remains significant across all regions, peaking at a 1.35 reduction for French funds. We recognize that measurement error may be introduced by historical ex-post modifications to LSEG ESG values. Such random measurement error usually causes an attenuation bias in econometrics, pushing the coefficients towards zero. As a result, our significant negative estimates provide a conservative lower bound, indicating that the observed negative ESG-timing association remains robust despite potential data rewriting.

The restricted investment universe may be the primary cause of the observed negative correlation. By excluding companies with lower ESG scores, fund managers have fewer “high-conviction” tactical opportunities to exploit. As a result, the requirement to uphold a high ESG profile may serve as a constraint, where the dedication to sustainability supersedes the exploitation of short-term market inefficiencies. This implies that the timing disadvantage is a strategic consequence of the ESG investment commitment, which prioritizes steady, ethical alignment over opportunistic trading, rather than a lack of innate ability.

Upon further scrutiny, it becomes apparent that there is a negative ^[11] but insignificant relation between stock-selection timing and fund size, implying that smaller funds may have an advantage in timing stock-selection opportunities compared to larger funds. These results align with the majority of the literature (Angelidis et al., 2021; Pástor and Stambaugh, 2012; Tang et al., 2012) highlighting that larger funds are often constrained by investment opportunities and higher transaction costs when buying or selling securities. The observed difference in statistical significance compared to the existing literature may lie in the nature of ESG funds, whose binding sustainability mandates alter traditional scale dynamics. Moreover, the findings indicate that younger funds exhibit better stock-selection timing skills, albeit without statistical significance. Younger funds may be more likely to embrace new investment strategies and technologies, providing a competitive edge in identifying and exploiting market opportunities (El Ammari et al., 2023; Ammann et al., 2010). The empirical results in Table 10 also show a negative correlation between stock-selection timing and cash holdings ^[12]. This suggests that managers holding less cash make timelier decisions compared to their peers, an inference in accordance with Khan et al. (2023) and Yan (2006). Managers overcome the cost of holding cash by making profitable investments (Dimitropoulos et al., 2020), while efficiently managing outflows and potential fire sales. Nevertheless, the estimate for cash holdings remains statistically insignificant.

The expense ratio is positively, though not statistically significantly, associated with stock-selection timing ability—not only for the entire sample, but also across all regions. These results are in agreement with Berkowitz and Kotowitz (2002) and Droms and Walker (1996) who argue that skilled fund managers charge higher fees. Similar inferences can be drawn regarding the number of stocks held in the portfolio, as the coefficients are positive in most cases, implying that a well-diversified portfolio contributes to more efficient stock-selection timing. A wider range of stocks can expose managers to more potential investment opportunities (Le Moigne and Savaria, 2006), making it easier to identify undervalued or overvalued securities and time their trades accordingly. Ultimately, the findings suggest that investors have a constrained capacity to identify funds exhibiting superior stock-selection timing abilities.

Another critical issue is whether ESG funds with superior stock-selection skills attract more capital in comparison to ESG funds lacking those skills. Thus, we build on the empirical findings of the previous section and attempt to further analyze the stock-selection timing-flow relation. For this purpose, we execute the following panel regression:

where Flow_i,t is the normalized fund flow at month t, defined as the monthly net fund flow divided by the fund's TNA at the beginning of the month. The terms d₁HighNeg_i,t-1, d₂MidNeg_i,t-1, and d₃LowNeg_i,_t-1 are indicator variables that take the value of 1 if the fund displays a high-negative (t ≤ −1.96), medium-negative (−1.96 < t < −1.28), or low negative (−1.28 ≤ t < 0) timing coefficient (g_i), and zero otherwise. Dummy variables d₄LowPos_i,t-1, d₅MidPos_i,t-1, and d₆HighPos_i,t-1 are defined analogously for the right tail of the timing estimates distribution; specifically, low-positive (0 ≤ t ≤ 1.28), medium-positive (1.28 < t < 1.96) and high-positive (t ≥ 1.96). X_i,k,t-1 includes all the remaining control variables mentioned in Table 10.

We employ a cell means model by estimating the regression without an intercept. Consequently, the resulting coefficients (d₁ through d₆) represent the estimated average fund flow for each respective category. This approach allows for the direct observation of the mean flow levels across different skill groups. We use a lagged structure in Eq. (8) to address potential reverse causality issues. Since all independent variables are measured in period t-1, they are predetermined relative to the dependent variable Flow_i,t. This arrangement ensures that we are evaluating how well past timing predicts future flows. We also control for time and fund fixed effects, which help mitigate potential omitted variable bias arising from unobserved, time-invariant fund characteristics. The results for Eq. (8) are presented in Table 11.

As shown, for funds based in the German market, a positive and statistically significant coefficient is found for high-positive timers (t = 2.01), indicating that investors recognize ESG funds characterized by strong stock-selection timing skills. In addition, the negative coefficient estimate for high-negative timers (t = −1.12) shows that investors avoid funds incapable of accurately timing stock-selection opportunities. Thus, we can classify investors in the ESG German market as sophisticated. Further evidence suggests that investors are also able to identify funds with superior stock-selection timing skills in the UK market. We find that all categories of positive timing funds (low, medium, high) have positive and statistically significant coefficients, while high-negative and medium-negative timers display negative and statistically significant estimates. By contrast, the French market deviates from this pattern, as empirical results show that the top stock-selection timers experience significant outflows (t = −2.45), whereas funds with adverse stock-selection abilities attract more capital (t = 2.22). These findings substantiate the assertions of Hypothesis 6 (H6).

These disparities in fund flows can be attributed to a complex interplay of factors. Given our previous finding that stock-selection timing ability is negatively correlated with ESG scores, the positive flows towards negative-timing French funds might be driven by investors' conscious pursuit of high-ESG-rated funds. This indicates a trade-off in which timing-based alpha generation is subordinated to non-financial utility. Since our empirical specification accounts for standard fund flow drivers, such as past returns, fund size, age, and fees within the X_i,k,t-1 vector, this ensures that this trend is not merely an artifact of omitted fund characteristics. Consequently, it is difficult to attribute the observed empirical flow patterns to structural variations or conventional performance chasing. This interpretation is supported by the fact that France has been at the forefront of ESG regulation and policy, implementing measures such the Pacte Law and Article 173–2 of the French Monetary and Financial Code, creating a regulatory framework more stringent than those of Germany and the UK. The French government has been particularly active in promoting ESG investing through policy and regulation ^[13]. This proactive regulatory approach generates structural demand; funds must maintain high ESG scores in order to reduce regulatory risk and attract capital pools constrained by ESG mandates. This emphasis on government intervention likely influences investor preferences (Nadler and Breuer, 2019). Furthermore, French funds may also be more effective in communicating their ESG credentials and positioning themselves as leaders in sustainable investing (Assaf et al., 2024). Thus, investors in the domestic market might possess a higher level of ESG awareness ^[14], leading them to prioritize ESG alignment over maximizing returns through timing strategies.

Additionally, France stands out among European nations with its robust support mechanisms for ESG investing. The “Plan d'Epargne en Actions” (PEA), a long-term equity savings plan, offers significant tax advantages, making it particularly attractive for investors in ESG funds. Furthermore, the French government has introduced various tax credits and incentives specifically for investments in green energy and sustainable projects. While not exclusively targeted at ESG funds, these initiatives indirectly bolster the appeal of high-ESG-focused investments by creating a more favorable environment for sustainable finance within the French market. These financial measures help compensating investors for the possible performance drag brought on by the limited selection available to high-ESG funds. On the other hand, comparable fiscal policies in Germany and the UK primarily focus on general investment promotion and research and development, with a less pronounced emphasis on specific ESG-related measures. The French institutional and retail markets structurally value the non-financial benefits of high ESG scores over the financial benefits of superior stock-selection timing, which is why this trade-off remains feasible.

Given the rising trend of sustainable investing and the excessive fluctuations observed in global financial markets recently, this study investigates whether ESG fund managers can time the stock-selection opportunities that emerge in the markets. To examine their stock-selection timing skills, we employ the recent and innovative measure of Jiang et al. (2021) within the European markets with the highest ESG concentration—the UK, France and Germany—for the period from January 1, 2012, to December 31, 2024. To separate skill from luck, we perform a bootstrap analysis on our data. We also examine whether ESG funds exhibit superior stock-selection timing skills compared to non-ESG funds. Considering the substantial volatility in asset prices induced by the COVID-19 pandemic and the recent geopolitical conflict, we assess whether these events had a discernible influence on managers' capacity to effectively time stock-selection opportunities. Moreover, we conduct a comparative analysis of the performance of positive and negative stock-selection timers across diverse subsequent periods, with the objective of quantifying the economic value accrued to investors. Lastly, we delve into the underlying factors influencing stock-selection timing ability, with a particular focus on its relationship with fund flows.

The empirical results show that 25.76% of UK funds successfully time their active trading, while only 9.60% fail to time stock-selection opportunities. Bootstrap analysis ensures that the aforementioned stock-selection timing abilities are driven by skill rather than pure luck. Notwithstanding this, the number of funds exhibiting positive timing ability is lower than that of their conventional counterparts. Conversely, empirical evidence suggests that French and German funds are unable to effectively time their active trading. The COVID-19 pandemic and the geopolitical conflict had a deleterious effect on the efficacy of fund managers' stock-selection timing strategies. While timing skills deteriorated during the crises, ESG funds marginally outperformed conventional funds in capturing stock-selection opportunities, reflecting greater resilience. We also demonstrate that positive timers consistently deliver significantly higher Fama-French-six-factor abnormal returns than negative timers at the short-term six-month horizon, with return spreads varying from 2.72% to 4.68% across all regions. Moreover, we find strong evidence that more active fund managers are better at timing the opportunities that arise in the market, whereas funds ranked as highly socially responsible exhibit negative stock-selection timing. Finally, we provide compelling evidence suggesting that investors have the ability to identify funds adept at capitalizing on stock-selection opportunities in the UK and German markets, yet fail to do so within the French market. Ultimately, our empirical evidence suggests a significant trade-off between tactical trading and ethical constraints in answering the core research question of “Skill or Sustainability?”. While a subset of managers in the UK navigate this successfully, the trade-off becomes particularly stark outside the UK, where stock-selection timing abilities are obviously compromised by sustainability mandates.

This study makes three original contributions to the literature. First, to the best of our knowledge, it is the first study to provide empirical evidence on the stock-selection timing skills of European ESG fund managers. Although the existing literature has thoroughly discussed the overall effectiveness of sustainable mandates, the precision and timing of individual trade executions remain largely underexplored. To address this gap, we investigate whether the specialized tactics employed by ESG managers translate into a functional, micro-level information advantage. This inquiry is critical as ESG assets under management continue to increase and investors demand greater transparency regarding the specific factors influencing alpha in sustainable portfolios.

Second, we offer a framework for assessing how ESG managers balance their long-term sustainability commitments with the demand for active, timely trading. This enables a more rigorous evaluation of whether tactical timing is constrained by ESG-related investment limitations or whether a specialized sustainability emphasis offers a unique informational advantage. Lastly, our study evaluates the impact of significant external shocks on fund portfolio dynamics. By examining the high-volatility period following the COVID-19 outbreak, we provide a comprehensive understanding of how ESG managers manage systemic crises in comparison to their conventional counterparts.

The practical implications derived from this study offer significant value for investors and market participants. The empirical evidence points to a strategic trade-off that can be mitigated by specific portfolio characteristics, rather than a fundamentally negative relationship between ESG commitment and stock-selection timing. Our findings on active share provide a clear path forward for investors who prioritize both sustainability and financial outperformance. Specifically, the “ESG-timing” penalty is primarily a risk for “closet indexers”, as evidenced by the positive interaction between active share and timing coefficients. The performance drag typically associated with a restricted investment universe can be offset by high-conviction managers who deviate significantly from benchmark weights in order to better utilize the micro-level information advantages inherent in evolving ESG data.

On the other hand, ESG funds demonstrate superior risk resilience during systemic crises, such as the COVID-19 pandemic or geopolitical turmoil. Asset managers can utilize these findings to implement a defensive hedge through ESG-tilted strategies. The lower idiosyncratic risk of sustainable assets offers a “passive” performance benefit that protects retail investors during market downturns, even when active timing alpha diminishes. The rapid decline of timing-based alpha, which mostly disappears within six months, provides a practical warning about “performance chasing”. Sophisticated investors should prioritize funds with recent timing success while remaining aware that these advantages are temporary. Fund flows show that, in contrast to investors in French funds, who show less sensitivity to tactical alpha, investors in UK-domiciled funds seem to reward short-term timing skill through capital allocations, reflecting these different behavioral dynamics. Investors in markets like France appear to be cognizant of this, as they rationally trade off short-term timing profits for long-term fiscal and ethical gains.

A structural friction between transparency and alpha generation is highlighted by the regional disparities observed between the UK and France/Germany. Strict disclosure requirements, such as the SFDR, effectively lessen information asymmetry and ensure that ESG considerations are fully “priced in”. Nevertheless, this greater efficiency erodes the potential for active managers to produce timing-based alpha. Since highly efficient markets may dampen the financial incentives for active sustainable management, policymakers must strike a balance between the necessity of market transparency and the resulting potential disincentives. The question of whether a more lenient, “light-touch” regulatory regime, like the UK's Sustainability Disclosure Requirements, is warranted is raised by this friction. A lighter strategy involves a fundamental trade-off; it undermines systemic market transparency and increases the risk of greenwashing, even while it maintains the informational asymmetries that enable active managers to produce tactical alpha and sustain financial incentives. Strict frameworks, on the other hand, maximize societal impact by ensuring authentic green capital allocation, but they do so at the expense of active management outperformance and local financial efficiency.

The “French Paradox” shows how private capital can be aligned with societal objectives through government involvement. Retail investors are effectively compensated for the potential loss of timing alpha through tax-advantaged structures such as the PEA and green investment credits. Other European authorities should consider comparable fiscal strategies to counteract the “constraint costs” of ESG rules to ensure the expansion of sustainable finance. The window for “informational arbitrage” will likely close as the UK implements its own Sustainability Disclosure Requirements. To prevent market efficiency from eroding active management expertise in the sustainable sector, policymakers should monitor this convergence.

Ultimately, the accessibility of sustainable investing determines the societal ramifications. Retail participation may falter if ESG portfolios are perceived as inherently disadvantaged in terms of financial timing. Our findings indicate that government support and lower downside risk reduce the societal “cost” of sustainability. Maintaining the long-term flow of capital toward the green transition requires that retail investors understand this trade-off, wherein greater systemic resilience and ethical utility are exchanged for lower tactical alpha.

While the study provides significant insights into active management strategies, it is not without limitations. The main limitation of this study lies in the geographical scope of the sample, which is restricted to ESG equity mutual funds domiciled in the UK, France, and Germany. The dataset is intrinsically constrained to relatively large-cap, developed-market stocks due to the benchmarking of our sample against the FTSE 100, CAC 40, and DAX 30. This domestic, large-cap focus restricts the external validity of our findings when compared to funds with more diverse investment scopes because many ESG funds operate under more expansive global or Europe-wide mandates. Thus, the generalizability of the findings is limited beyond the European context, particularly when contrasting them with different market settings such as the US or emerging markets. Yet, this specific focus was necessary to ensure a statistically adequate sample size and data homogeneity regarding European ESG adherence. Nevertheless, the cross-country comparisons in our study should be interpreted cautiously given the variations in sample sizes across the UK, French, and German fund subsamples. While the UK sample is significantly larger than the French and German cohorts, this distribution reflects the relative depth and maturity of the respective sustainable fund markets during the study period. Instead of using pooled regressions, our empirical design relies on independent, country-specific estimations to mitigate the impact of these sample disparities. Despite these imbalances, the French and German samples exceed the common thresholds required for robust statistical inference.

Our dataset's emphasis on active funds is another methodological constraint. Nonetheless, this decision aligns with our goal of assessing the long-term timing skills of committed ESG managers. By excluding liquidated or merged funds, we ensure that our sample consists of funds with a stable investment philosophy and a consistent track record, though we acknowledge that this introduces an explicit survivorship bias. As a result, our findings should be interpreted specifically as a performance benchmark for established, surviving ESG funds in the UK, France, and Germany. Additionally, users of ESG data should be aware of possible look-ahead issues as data providers may occasionally backfill or make ex-post modifications to prior ESG scores and ratings. Therefore, the findings should be applied prudently when considering ESG funds domiciled in smaller, emerging, or non-US markets where investor preferences and regulatory frameworks may differ substantially.

Furthermore, even though the use of monthly holdings data provides a higher frequency than typical quarterly disclosures, it remains subject to certain constraints. Very short-term timing decisions may remain unobserved, as monthly snapshots cannot capture intra-month trading activity or high-frequency adjustments to market exposure. Time-varying risk factors may also be partially reflected in the reported performance rather than pure timing ability, rendering the isolation of true managerial “skill” a major challenge in a multi-factor world. The absence of formal structural break tests around the COVID-19 period also remains a notable limitation of our analysis. Moreover, while the use of fixed effects and lagged variables helps attenuate omitted variable bias, these specifications do not fully resolve all endogeneity concerns. Consequently, the results should be interpreted as robust associations rather than strictly causal relationships.

Future research endeavors may attempt to extend this geographic scope to other regions, as high-quality data becomes available beyond the wider European context. How ESG timing dynamics unfold within small-cap stocks or funds with unrestricted, multi-regional mandates is another area that future research could potentially explore. Future studies could also enrich this framework by examining how individual manager characteristics, such as tenure, gender, or educational background, affect timing skills and ESG investment decisions.

To conduct the bootstrap procedure, we estimate the stock-timing test in Eq. (2) across all sampled ESG funds to obtain the parameter estimates ${cˆ$_i, $gˆ$_i} and the error terms for each fund {$εˆ$_i, t+1}:

We resample the entire vector of residuals across all funds for a given month synchronously with replacement in order to preserve the cross-sectional correlation structure across funds. This generates a time series of residuals that maintains the contemporaneous dependencies in the panel data. Next, by stipulating the null hypothesis of no stock-selection timing ability, we create a time series of bootstrapped activeness for each fund:

Subsequently, we run the stock-selection timing test using the pseudo-data and record the timing estimates and the corresponding t-statistics for each pseudo-fund:

Finally, we repeat this process for all funds 10,000 times to build the distributions of timing coefficients and their associated t-statistics for specific quartiles (25% and 75%), and extreme percentiles (1%, 5%, 10%, 90%, 95%, and 99%).

Under the null hypothesis of no stock-selection timing ability, we anticipate that the distributions of the bootstrapped statistics will match the distribution of the actual statistics. If the actual positive Newey-West t-statistic for a fund exceeds the equivalent bootstrapped value (i.e. the bootstrapped p-value is close to zero) the fund is assumed to possess positive stock-selection timing ability due to skill rather than pure luck. Conversely, if the actual negative Newey-West t-statistic for a fund is smaller than the corresponding bootstrapped value, we infer that the fund fails to accurately time stock-selection opportunities.