Introduction

Climate change is an imminent and critical global issue with long-term implications for the sustainable development of all countries. Investors’ demands for climate and other sustainable information have soared in recent years, showing that an increasing number of investors use these criteria in order to screen their potential investments. ESG criteria refer to environmental standards (adaptation of environmental-friendly practices), social ones (promotion of ethical and social conscious themes) and finally governance standards (usage of accurate and transparent accounting methods to ensure accountability) which a company must adhere to. ESG investing is a rapidly rising trend in finance (Koutsokostas and Papathanasiou 2017; Wong et al. 2021; De Jong and Rocco 2022), surpassing $40 trillion of assets under management globally in 2022, according to Global Sustainable Investment AssociationFootnote 1. Europe is one of the leading regions in ESG concentration (Papathanasiou et al 2022; Samitas et al. 2022), with asset under management poised to reach €9 trillion by 2025Footnote 2. In particular, ESG funds will constitute over 50% of total European mutual fund assets, up from 37% in end-2021, according to PwC Luxembourg research.

A common practice of fund managers is to artificially inflate the performance of the investment portfolio, a practice known as portfolio pumping or “painting the tape,” which is considered by market regulators to be illegal. This is typically done by purchasing additional stocks the fund already holds, at an inflated price, during the last days of a quarter, in order to augment the value of the existing positions and report better fund performance. The improved fund return reported will attract more flows, increase assets under management and, eventually, compensate fund managers (Ouyang and Cao 2020).

The existing literature shows that fund managers do involve in such trading activities and that price spikes are higher at year-ends rather than quarter-ends (Carhart et al. 2002; Bernhardt and Davies 2005; Gallagher et al. 2009; Agarwal et al. 2011; Ben-David et al. 2013; Lee et al. 2014; Li and Wu 2019; Shackleton et al. 2020). As concerns the characteristics of funds that distort the performance of their managed portfolios, the results provided by the literature are not clear, as Ben-David et al. (2013), Lee et al. (2014) and Li and Wu (2019) suggest that marking up is induced by top performing funds, whereas Gallagher et al. (2009) and Shackleton et al. (2020) show that these practices are pervasive among the worst-performing funds. However, no similar investigation has been conducted for ESG mutual funds. Thus, we intend to fill this gap in the literature, by examining the portfolio pumping activities of ESG fund managers for the first time.

In this paper, we explore quarter-end spikes in returns of funds that conform to ESG criteria in the leading European countries in sustainable assets, i.e., the UK, France and Germany, for the period 1/1/2010–12/31/2022. We choose the aforementioned markets as they are among: a) the largest markets in sustainable investmentsFootnote 3; b) the leading markets on ESG concentrationFootnote 4 (higher proportion of ESG assets compared to non-ESGFootnote 5) and c) the first to impose mandatory climate-related disclosures for listed companiesFootnote 6. We use Bernhardt and Davies (2005) measure as a proxy of portfolio pumping and detect its presence by conducting fixed effects panel regression, congruent with Li and Wu (2019), for each region and for each quarter. We also examine whether the trend for price inflation is more pronounced among ESG funds in relation to their conventional peers. The rationale behind this approach is to document whether significant differences exist, since conventional funds do not take ESG criteria into consideration when making investment decisions. Furthermore, as COVID-19 had a significant impact on asset price volatilities, we investigate if the manipulating trading strategies were affected during the coronavirus pandemic. For the robustness of our results, we execute a two-stage regression in the spirit of Carhart et al. (2002) to further decipher whether performance reverses are stronger around quarter-ends or year-ends. Finally, we perform panel regression to shed light on the characteristics of funds that are more prone to portfolio pumping activities, in order to inform investors about funds that purposefully inflate securities’ prices.

Thus, our study aims to provide sufficient evidence to answer the following research questions:

  1. RQ1:

    Do ESG equity fund managers pump and dump the stocks included in their portfolios? Is portfolio pumping more prevalent among ESG funds or non-ESG funds?

  2. RQ2:

    Was portfolio pumping during the infectious disease strengthened or weakened?

  3. RQ3:

    Is price manipulation stronger at quarter-ends or year-ends?

  4. RQ4:

    What types of funds engage in gaming behavior?

The contribution of our study lies in the following aspects. To the best of our knowledge, this study is the first to examine portfolio pumping in the ESG mutual fund industry. As “G” in ESG refers to governance factors that include business ethics and fair dealings with stakeholders, it is of great interest to examine whether ESG fund managers adhere to fair and accurate disclosure of quarter-end portfolio values. Our initiative is motivated by the possibility that these types of ethical investing may engage in improper strategies that provide investors with misleading indications of their relative performance. Documenting evidence of portfolio pumping is imperative for policy makers and regulators to shape appropriate policies in order to reduce its magnitude. In addition, reporting unusual end-of-quarter NAV price jumps is crucial for investors as it assists them in having an accurate impression of their portfolio performance. Second, by investigating the determinants of portfolio pumping, we are providing ESG investors with explicit schemes in order to avoid funds that deliberately manipulate their portfolios.

The remainder of this paper is organized as follows. In Section "Literature review", a literature review on portfolio pumping is included. A description of the data and techniques used in this study is presented in Section "Data and methodology". The empirical findings and conclusions of our research are reported in Sections "Empirical results" and “Conclusions”, respectively.

Literature review

The body of the literature regarding portfolio pumping is not extensive. The majority of the studies indicate that portfolio pumping is existent and more prominent around year-ends rather than quarter-ends. Carhart et al. (2002) were the first to provide proof that fund managers inflate portfolio prices at quarter’s end by making last minute purchases of stocks they already hold. They document that price manipulation is significantly higher at the end of the year, ranging from 0.5% for large-cap funds to over 2% for small-cap funds annually, while no effect at month-ends that do not coincide with quarter-ends is found. Hillion and Suominen (2004) also report evidence of price manipulation in order for a better performance to be documented. Fund managers’ incentives to manipulate asset prices at the end of the quarter are so strong that their transactions have a significant impact even on the aggregate market returns (Bernhardt and Davies 2005). Even after adjusting for risk in the time series and cross-section, Agarwal et al. (2011) find the returns during December to be significantly higher than returns during the rest of the year for funds with high incentives and more opportunities to inflate returns. Funds strategically steer their returns upward in order to charge larger fees, by underreporting returns earlier in the year. However, this excessive trading may erode long-term fund performance, as suggested by Bhattacharyya and Nanda (2013).

Another strand of the literature focuses on whether winner or loser funds exhibit stronger portfolio pumping patterns. Ben-David et al. (2013) highlight stock price manipulation by some hedge funds on crucial reporting dates. The final day of the quarter in their study witnessed abnormal returns of 0.30% for stocks in the top quartile of hedge fund holdings, followed the next day by a reversal of 0.25%. The final few minutes of trading are when a sizable portion of the return is achieved. They provide additional evidence in support of manipulation by the analysis of intraday volume and order imbalance. For funds with greater incentives to outperform their peers, these trends are more pronounced. Lee et al. (2014) also explore the existence of portfolio pumping in the Korean equity fund market and outline the elements that support this practice. Their empirical findings show that fund managers manipulate fund performance by pumping up their portfolios at year-ends, when their accomplishments are evaluated, and this manipulation is stronger for small funds and those with higher past returns. In addition, portfolio pumping is more frequent in the products of small or foreign asset management firms, and fund managers are tempted to manipulate prices of holdings characterized by low liquidity. In addition, Li and Wu (2019) examine portfolio pumping from a performance-based perspective by utilizing information from 60 Chinese mutual fund companies on compensation policies. They show that pumping is more widespread among funds that rank at the top performance distribution points and at crucial cutoffs that companies set to evaluate the bonus amounts for the managers. Their analysis indicates that the flow-performance relationship described in earlier studies is not the driving force behind portfolio pumping, but rather performance ranking.

On the other hand, Gallagher et al. (2009) and Shackleton et al. (2020) show that portfolio pumping is more profound among funds that underperform relatively to their peers. Gallagher et al. (2009) find significantly higher abnormal returns on the last day of quarter-ends, especially at financial year-ends. They suggest that gaming trades occur mostly among smaller and less liquid stocks which are easier to manipulate. They report that poor-performing managers are most likely to engage in gaming behavior in order to retain their jobs. Similarly, Shackleton et al. (2020) analyze price inflation practices by fund managers in China and highlight that equity funds artificially enhance portfolio performance at quarter-ends, particularly at year-ends. They further point out that the worst-performing managers experience more severe NAV inflation and stocks in which fund managers have greater shares show a more distinct pattern of price inflation around quarter- and year-ends than other stocks. Finally, Duong and Meschke (2020) investigate how regulatory scrutiny of portfolio pumping affected the way US mutual funds trade. They prove that the imposition of fines and the subsequent reputational harm has resulted in a decline in last minute price spikes around quarter-ends. These declines are largest in magnitude for small-cap and better-performing funds.

Data and methodology

Data

The dataset of our empirical study is obtained from LSEG’s Lipper Database, including actively managed equity mutual funds that comply with the ESG criteria, since passive funds replicate a market index, investing in the constituents of the underlying index in the same proportion as they are present in the index. We select funds operating in the UK, France and Germany, because not only do these markets have the highest assets in sustainable funds in the continent, but also they were one of the first to require of listed corporations to disclose information relating to climate change. Our sample consists of funds that invest domestically and do not hold positions in foreign stocks to a great extent. Moreover, we take into account funds that have FTSE 100, CAC 40 and DAX 30 as a primary investment benchmark when they are domiciled in the UK, France and Germany, respectively. The rationale behind this approach is to secure the reliability of our results, as, when a fund tracks multiple benchmarks, it is difficult to detect if there is a spike in fund returns in relation to the benchmark’s returnsFootnote 7. We remove funds that have less than 2 years of operation throughout the sample period, as it usually requires months for a manager to develop his/her investment plan. We also exclude funds that hold less than 10 stocks in their portfolios, to better capture the portfolio pumping behavior. Therefore, our final sample comprises 271 ESG equity mutual funds, 198 of which are concentrated in the UK, 50 in France and 23 in Germany. Timespan covers the period from January 2010 to December 2022. Thus, it is of great interest to investigate a time period after the dissemination of the study of Carhart et al. (2002) in order to confirm if this phenomenon has been eliminated, as denoted by Duong and Meschke (2020). Appendix 1 provides a full description of the sampled data (asset names with the corresponding identification codes).

In Table 1, the time series averages of cross-sectional mean and median of fund characteristics, including total net assets, fund age, expense ratio, cash holdings, normalized fund flow, number of shares contained in the portfolio and ESG score, are presented. For the estimation of cash and stocks included in the portfolios, 3-month portfolio holdings were used throughout the sample period.

Table 1 Descriptive statistics of fund characteristics/ESG funds
Full size table

As shown, funds operating in the UK include more stocks in their portfolios and attract more flows. French funds incorporate sustainable assets to a greater extent and have a higher expense ratio. Funds domiciled in Germany are larger in magnitude, older in age and hold more cash in their portfolios. Finally, in most cases, the means of the funds’ characteristics are greater than the equivalent medians, indicating a high possibility of the distributions to be right skewed.

Methodology

We investigate portfolio pumping, focusing on ESG funds’ NAV anomalies during quarter-ends. We estimate fund abnormal return as follows:

$$ ARNAV_{i,t} = RNAV_{i,t} - RMkt_{t} , $$
(1)

where ARNAVi,t is the abnormal return of fund i on day t, RNAVi,t is the raw return of fund i on day t and RMktt is the value-weighted marketFootnote 8 return on day t. RNAVi,t is computed through the following formula:

$$ RNAVi,t = \frac{{NAV_{i,t} - NAV_{i,t - 1} }}{{NAV_{i,t - 1} }}, $$
(2)

where NAVi,t and NAVi,t−1 are the net asset value per unit share of fund i on day t and t−1, respectively, adjusted by share split and dividend.

If fund managers engage in portfolio pumping, we should expect them to exhibit an abnormally high return at the end of the quarter due to placing a large volume of orders for a stock at an inflated bid price. Contrariwise, in the 1st days of the next quarter, the fund’s NAV is anticipated to revert, as the manipulation is commonly short-term, and the fund managers will dump the stocks by selling their positions. We select a time span of 6 days around the turn of the quarter, 3 days before the end of the quarter and 3 days after the end of the quarter, and evaluate the pumping and reversal of the fund’s abnormal return as a proxy for portfolio pumping.

We measure portfolio pumping as the difference between the quarter-end pumping and the following reversal divided by two, in accordance with Bernhardt and Davies (2005):

$$ Blip_{i,T} = \frac{{Pump_{i,T} - Reversal_{i,T + 1} }}{2}, $$
(3)

where Blipi,T symbolizes the proxy of portfolio pumping for fund i at the end of quarter T. Pumpi,T is the fund i’s average abnormal return in the last 3 days of quarter T, and Reversali,T+1 is the fund i’s average abnormal return in the first 3 days of quarter T+1, which are calculated, respectively, through the following equations:

$$ Pump_{i,T} = \frac{{ARNAV_{i,T}^{3rdToLastDay} + ARNAV_{i,T}^{2ndToLastDay} + ARNAV_{i,T}^{LastDay} }}{3} $$
(4)
$$ Reversal_{i,T + 1} = \frac{{ARNAV_{i,T + 1}^{1stDay} + ARNAV_{i,T + 1}^{2ndDay} + ARNAV_{i,T + 1}^{3rdDay} }}{3}, $$
(5)

where \(ARNAV_{i,T}^{3rdToLastDay}\), \(ARNAV_{i,T}^{2ndToLastDay} \) and \(ARNAV_{i,T}^{LastDay}\) are the fund i’s abnormal return on the 3rd-to-last day, 2nd-to-last day and last day of quarter T, respectively. In the same manner, \(ARNAV_{i,T + 1}^{1stDay}\), \(ARNAV_{i,T + 1}^{2ndDay} \) and \( ARNAV_{i,T + 1}^{3rdDay}\) denote the fund i’s abnormal return on the 1st, 2nd and 3rd day of quarter T+1, respectively.

To examine whether fund managers inflate the price of shares included in their portfolios at the end of the quarter, we execute the following panel regression, as in Li and Wu (2019):

$$ \begin{aligned} ARNAV_{i,t} & = \alpha_{0} + \alpha_{1} Day_{t}^{3rdToLast} + \alpha_{2} Day_{t}^{2ndToLast} + \alpha_{3} Day_{t}^{Last} + \alpha_{4} Day_{t}^{1st} + \alpha_{5} Day_{t}^{2nd} \\ & \quad + \alpha_{6} Day_{t}^{3rd} + \alpha_{7} LogFundTNA_{i,t} + \alpha_{8} LogFundAge_{i,t} + f.e. + \varepsilon_{i,t} , \\ \end{aligned} $$
(6)

where ARNAVi,t is defined as above, and \(Day_{t}^{3rdToLast}\), \(Day_{t}^{2ndToLast}\) and \(Day_{t}^{Last}\) are dummy variables that are equal to 1 if t is the 3rd-to-last, 2nd-to-last and last day of a quarter, respectively, and 0 otherwise. The dummy variables \(Day_{t}^{1st}\), \(Day_{t}^{2nd}\) and \(Day_{t}^{3rd}\) are defined analogously, indicating the first 3 days of the next quarter. The control variables encompass fund size and fund age, which the existing literature proves to have a significant effect on fund performance (Chen et al 2004; Pollet and Wilson 2008; Cremers and Petajisto 2009; Massa and Patgiri 2009; Huang et al. 2011). LogFundTNAi,t is the logarithm of the fund i’s monthly total net assets. LogFundAgei,t is defined as the logarithm of the total months that span from fund i’s launch date until the end of the sample period. We also control for time and fund fixed effects.

We further investigate the existence of portfolio pumping by switching abnormal return data with Blip data around the end of the quarter. We replace the ARNAV data on the last day of quarter T with Blip data, and we delete the ARNAV observations for the remaining days around the turn of the quarter (the 3rd- and 2nd-to-last days of the quarter and the 1st, 2nd and 3rd days of the following quarter). Subsequently, we execute the following regression for each country and for each quarter:

$$ ARNAV_{i,t} = \alpha_{0} + \alpha_{1} LastDay_{t} + \alpha_{2} LogFundTNA_{i,t} + \alpha_{3} LogFundAge_{i,t} + f.e. + \varepsilon_{i,t} , $$
(7)

where ARNAVi,t, LogFundTNAi,t and LogFundAgei,t are defined as above, and LastDayt is an indicator variable that takes the value of 1 if t is the last day of the quarter and 0 otherwise.

While proceeding with our empirical results, we also explore whether the positive return at the end of the quarter is associated with the negative return at the beginning of the next quarter, by implementing a two-stage regression in the spirit of Carhart et al. (2002). In the first stage, we regress Pump values on Reversal values for every month of our sample period:

$$ Pump_{i,t} = \alpha_{0,t} + \alpha_{1,t} Reversal_{i,t + 1} + \varepsilon_{i,t} , $$
(8)

In the second-stage regression, we test whether the slope coefficients α1 obtained from Eq. (8) are significantly lower when t is the last day of a quarter or a year, as follows:

$$ \alpha_{1,t} = \beta_{0} + \beta_{1} QuarterEnd_{t} + \beta_{2} YearEnd_{t} + \varepsilon_{t} , $$
(9)

where QuarterEndt is a dummy variable that is equal to 1 if t is the last day of March, June or September and 0 otherwise. Analogously, YearEndt takes the value of 1 if t is the last day of December and 0 otherwise.

Finally, we apply the following panel regression to investigate the relation between portfolio pumping and several fund characteristics:

$$ Blip_{i,T} = c + \delta_{k} X_{i,k,T - 1} + \varepsilon_{i,t} , $$
(10)

where Blipi,T is defined as above (Eq. 3), and Xi,k,T−1 denotes the fund i’s characteristics, which include fund return, fund TNA, age, expense ratio, cash holdings, normalized fund flow, number of shares contained in the portfolio and ESG score. All explanatory variables are lagged by 3 monthsFootnote 9.

Empirical results

ESG portfolio pumping

In Table 2, the summary statistics on the main variables of our study (RNAV, RMkt, ARNAV, Pump, Reversal and Blip) are reported.

Table 2 Summary statistics of portfolio pumping measures/ESG funds
Full size table

As shown, ESG funds based in France and the UKFootnote 10 outperform the market’s return. The average of funds’ NAV per share (RNAV) is 2.728 basis points (bp) and 2.378 bp for French and British funds, higher than the value-weighted market return (RMkt), which stands at 2.357 bp and 1.517 bp, respectively. On the other hand, ESG funds in Germany underperform, on average, the market. The above findings translate into a positive average abnormal return (ARNAV) for funds operating in France (ARNAV = 0.243 bp) and the UK (ARNAV = 0.947 bp), and a negative one for German funds (ARNAV = −0.646 bp). Furthermore, we observe a significant pumping-reversal pattern of portfolio pumping for the UK funds. The average Pump value at quarter-end is 3.040 bp, while the average Reversal at the beginning of the quarter is −9.215 bp, which results in a Blip value of 6.127 bp on average. On the contrary, we do not receive indications of portfolio pumping for French funds, as the Blip value is extremely low (0.084 bp), while the Reversal at the beginning of the quarter is positive (1.148 bp). Finally, weak evidence of the occurrence of portfolio pumping is reported for German funds at quarter-ends; however, it is possible that fund managers may inflate the stocks in their portfolios only in year-ends.

Table 3 presents the results of the panel regression described in Eq. 6 which shows the level of abnormal returns around the quarter-end.

Table 3 Regression results for abnormal returns around quarter-ends/ESG funds
Full size table

In the case of the UK ESG mutual funds, we observe significantly positive abnormal returns in the final 3 days of the quarter, followed by negative abnormal returns in the first 3 days of the next quarter, corroborating evidence of strong portfolio pumping. The coefficients \(Day_{t}^{3rdToLast}\), \(Day_{t}^{2ndToLast}\) and \(Day_{t}^{Last}\) are positive and statistically significant at 5% level, showing that the UK fund managers inflate portfolio prices by approximately 9 bp in the last 3 days of the quarter, with \(Day_{t}^{3rdToLast}\) being the day when the stronger price inflation occurs (\(Day_{t}^{3rdToLast}\) = 3.6144). On the other hand, the estimations on \(Day_{t}^{1st}\), \(Day_{t}^{2nd}\) and \(Day_{t}^{3rd}\) are negative and statistically significant at 5% level, suggesting that fund abnormal returns reverse nearly 28 bp in the 3 following days after the turn of the quarter. \(Day_{t}^{3rd}\) is the day that experiences the highest reversal (\(Day_{t}^{3rd}\)= −13.0219). On the contrary, no evidence that managers pump and dump the stocks included in their portfolios are documented for French and German ESG funds, as all the coefficients are statistically insignificant and do not follow a similar pattern. Furthermore, we find that size has a significant positive impact on fund abnormal returns in the case of French funds, whereas age negatively affects the performance of the UK funds.

ESG funds vs conventional funds

We examine the propensity for portfolio pumping among non-ESG funds to assess whether investment strategies differ for ESG funds. Given the large number of conventional funds in circulation during the sample period, we include in the sample through the screening process 300 funds that meet the characteristics of ESG funds to the extent feasible, in particular in terms of total assets under management and the number of securities encompassed in their portfolios. Descriptive statistics of the characteristics of conventional funds and the main variables of the sample are presented in Tables 4 and 5, respectively.

Table 4 Descriptive statistics of fund characteristics/conventional funds
Full size table
Table 5 Summary statistics of portfolio pumping measures/conventional funds
Full size table

Conventional funds based in France and the UKFootnote 11 also perform better than the market index, while the performance of German funds lags the return of DAX 30. This results in an average positive abnormal return of 0.329 bp and 1.038 bp for French and British funds, respectively, and a negative abnormal return for German funds (ARNAV = −0.697 bp). German funds do not pump equity prices (Pump = −0.450 bp), while French funds’ returns do not revert in the 1st days of the next quarter (Reversal = 1.208 bp). In contrast, both price inflation (Pump = 6.415 bp) and price deflation (Reversal = −13.583 bp) are of larger magnitude for the UK conventional funds compared to ESG funds.

The outcomes of the panel regression showing the level of abnormal returns around quarter-ends for the conventional mutual funds in the sample are displayed in Table 6.

Table 6 Regression results for abnormal returns around quarter-ends/conventional funds
Full size table

Strong portfolio pumping is supported by the highly positive abnormal returns we observe in the last 3 days of the quarter for the UK conventional funds, followed by negative abnormal returns in the first 3 days of the following quarter. The results show that portfolio returns strengthen by around 19 bp at the end of the quarter, double the amount observed in the case of ESG UK funds and, at the beginning of the following quarter, fund prices fall even further, by about 39 bp. Thus, funds that do not adhere to ESG criteria participate in more severe portfolio pumping. Documented differences may be due to the fact that ESG funds probably follow different standards for stock screening. Different standards impose different constraints on stock selection, which, in turn, impose different constraints on pumping stock prices. Moreover, as in the case of German and French ESG funds, no evidence is reported that non-ESG fund managers are involved in pumping activities, since all the coefficients are not statistically significant and exhibit a random motivation.

ESG portfolio pumping during COVID-19

As the majority of stock markets across the globe crashed after the emergence of COVID-19 pandemic, we are intrigued to probe whether manipulating trading strategies of ESG fund managers have changed after the spread of the disease. The summary data for the portfolio pumping measures during COVID-19 are provided in Table 7.

Table 7 Summary statistics of portfolio pumping measures during COVID-19/ESG funds
Full size table

As observed, the performance of ESG funds during the COVID-19 is slightly inferior compared to the markets’ in any case, since we find the raw return equal to 0.609 bp, 2.102 bp and 0.733 bp for the UK, French and German funds, while the value-weighted market return reaches 0.739 bp, 2.286 bp and 2.038 bp, respectively. Thus, ESG funds perform worse, on average, than the market, delivering a negative abnormal return in the UK (ARNAV = −0.222 bp), in France (ARNAV = −0.445 bp) and in Germany (ARNAV = −1.452 bp). Moreover, high Pump values are documented in each group of funds, especially in the case of the UK. The average Pump value at quarter-end is 14.373 bp, while the average Reversal at the beginning of the quarter is −13.939 bp, which results in a Blip value of 14.156 bp on average. Contrariwise, returns remain positive in the 1st days of the next quarter for German funds (Reversal = 2.254 bp), or they are marginally negative for French funds (Reversal = −0.228 bp). Therefore, results provide evidence in support of extreme price manipulation for the UK ESG funds, while no such proof is reported for the remaining fund categories.

Table 8 presents the panel regression results for the ESG mutual funds in the sample, indicating the level of abnormal returns at quarter-ends during COVID-19.

Table 8 Regression results for abnormal returns around quarter-ends during COVID-19/ESG funds
Full size table

The empirical findings connote that the UK managers have steered fund returns upwards after the surge of COVID-19 to a greater extent compared to the overall sample period, as fund prices are inflated by around 43 bp at the end of the quarter. At the beginning of the following quarter, portfolio returns decline by about the same amount (42 bp), substantiating evidence of excessive portfolio pumping. We also find the coefficients of \(Day_{t}^{3rdToLast}\), \(Day_{t}^{2ndToLast}\) and \(Day_{t}^{Last}\) positive and statistically significant at 5% level in the case of French and German funds; however, none of the coefficients of \(Day_{t}^{1st}\), \(Day_{t}^{2nd}\) and \(Day_{t}^{3rd}\) are negative and statistically significant at 5% level. Thus, we infer that pumping activities are non-significant in French and German ESG market.

ESG portfolio pumping at quarter-ends and year-ends

We proceed with our empirical findings concerning ESG pumping by providing in Table 9; the results of the regression described in Eq. 7, after substituting ARNAV data with Blip data in the last day of the quarter and deleting the ARNAV data for the remaining days around the turn of the quarter.

Table 9 Regression results for the occurrence of ESG portfolio pumping for each country
Full size table

As observed, the coefficient on LastDayi,t is statistically significant (t = 4.52) for the UK ESG funds at 5% level, verifying that managers distort the performance of their portfolios. On the other hand, for funds operating in the German and French market, a statistically insignificant coefficient is reported (t = 0.22 and t = 0.11, respectively). The empirical results in Table 9 support the findings from the previous section that the UK ESG funds managers engage in portfolio pumping activities, whereas German and French funds do not exhibit any irregular returns during quarter-ends.

Table 10 reports the results of aforementioned regression (Eq. 7) for each quarter of the sample period.

Table 10 Regression results for the occurrence of ESG portfolio pumping for each quarter
Full size table

Results provide further evidence that ESG fund managers in the UK inflate quarter-end NAV prices, as the coefficient on LastDayi,t is statistically significant at 5% level for each quarter of the sample period (with the exception of the third quarter) and markedly higher at year-ends. These findings are consistent with the majority of the literature (Carhart et al. 2002; Gallagher et al. 2009; Agarwal et al. 2011; Ben-David et al. 2013; Lee et al. 2014; Li and Wu 2019; Shackleton et al. 2020) highlighting that fund managers manipulate stock prices at quarter-ends, particularly at year-ends. As fund managers’ evaluation is usually based on their yearly performance, it is reasonable to assume that they have a greater incentive to conduct portfolio pumping at year-ends. Contrariwise, our findings confirm the absence of gaming behavior in the German and French ESG market, as all the estimations on LastDayi,t are not statistically significant. These results are in line with Duong and Meschke (2020), who suggest that portfolio pumping was reduced due to increased regulatory scrutiny after the study of Carhart et al. (2002). Therefore, our empirical findings denote cross-country differences in the level of portfolio pumping. These diverging results could lie in regulatory differences among these countries or perhaps the lack of efficient supervision in the UK market. Another reason for the occurrence of portfolio pumping among the UK funds could be the different incentive schemes managers face that motivate them to perform such activities.

Robustness considerations

As a robustness test, we carry out a two-stage regression as in Carhart et al. (2002), in order to determine whether performance reverses are more prevalent around quarter-ends or year-ends. In Table 11, the results of the second-stage regressions described in Eq. 9 are illustrated.

Table 11 Pumping-reversal pattern of ESG portfolio pumping
Full size table

The coefficient on YearEndt is statistically significant at 5% level in the case of ESG funds domiciled in the UK, confirming that the positive return at the end of the year is associated with the negative return at the beginning of the next year. On the other hand, the coefficient on QuarterEndt is also negative but statistically insignificant for the UK funds. These findings confirm previous evidence that the UK portfolio pumping is larger in magnitude at year-ends rather than quarter-ends.

What types of ESG funds exhibit stronger pumping practices?

Table 12 shows the results of the panel regression (Eq. 10) in order to detect the type of funds that have a stronger tendency to inflate the performance of their portfolios at quarter-ends.

Table 12 Key drivers of ESG portfolio pumping
Full size table

Results show that portfolio pumping is more prominent among the worst-performing funds, as the coefficient on RNAVi,T−1 is negative and statistically significant at 5% level (t = −8.28). These results contradict Carhart et al. (2002), Ben-David et al. (2013), Lee et al. (2014) and Li and Wu (2019) who argue that portfolio pumping is mainly driven by managers pursuing bonuses linked with performance-based compensation contracts. On the contrary, our empirical findings are in accordance with Gallagher et al. (2009) and Shackleton et al. (2020) who suggest that underperforming funds tend to inflate portfolio prices in relation to their peers. ESG poor-performing managers display greater evidence of portfolio pumping. This gaming behavior could be attributed to the fact that loser funds do not want to lag behind other funds or managers wanting to avoid potential penalties imposed by fund companies due to bad performance. Furthermore, we find that ESG rating is negatively correlated with portfolio pumping, as the coefficient on ESG Scorei,T−1 is significantly negative (t = −2.72) . This means that low-rated ESG funds show stronger evidence of marking up activity. Given the fact that the abidance to the ESG criteria implies the implementation of accurate accounting methods and accountability to shareholders, this outcome is expected to some extent, as stronger commitment to these criteria could lead to the elimination of unethical practices. Surprisingly, funds that charge investors with lower expenses appear to conduct more pumping, as we find the estimation on Expense Ratioi,T−1 negative and statistically significant (t = −3.24). These results are in contrast with the results of Agarwal et al. (2011) who argue that funds strategically manipulate their returns in order to impose larger fees. The low-cost strategy can be an effective marketing tool used by fund companies to attract new clients, especially for new entrant funds. However, this is not the case in our analysis, as we find a negligible positive coefficient for Fund Flowi,T−1. Therefore, funds that engage in portfolio pumping do not ultimately succeed in increasing fund inflows. The cost of conducting portfolio pumping outweighs the marginal benefits the funds receive.

The empirical results in Table 12 also show that pumping is primarily the focus of younger rather than older funds. Perhaps, it is a tactic used by funds when entering an overcrowded market. We further find that pumping is stronger in less-diversified portfolios; this finding is unsurprising, as the smaller the number of stocks included in the portfolio, the easier their manipulation becomes. On the other hand, pumping is positively associated with cash holdings and fund size. This indicates that managers who hold more cash in their portfolios can more easily purchase additional stocks, they already hold at an inflated price. In addition, our findings suggest that larger funds tend to inflate asset prices to a greater extent than smaller funds, probably because they have more incentives to do so, due to the higher value of assets being managed. However, the estimated coefficients for fund age, number of shares, cash holdings and fund size are all statistically insignificant at 5% level.

Conclusions

In this paper, we investigate quarter-end spikes in returns of funds that apply ESG factors in the major European markets of the UK, France and Germany, for the period 1/1/2010–12/31/2022. We utilize Bernhardt and Davies (2005) proxy for portfolio pumping, and we analyze its occurrence by carrying out fixed effects panel regression (Li and Wu 2019) for each region and for each quarter. Results show that the UK fund NAV prices experience extremely abnormal increases at quarter-ends, especially at year-ends. However, compared to their traditional counterparts, the UK ESG managers tend to inflate fund returns less. Contrariwise, no evidence of portfolio pumping is found for ESG funds in Germany and France. The COVID-19 pandemic significantly increased fund managers' inclination to artificially boost the performance of the investment portfolios. Our findings are robust when applying the Carhart et al. (2002) two-stage regression model in order to ascertain the magnitude of turn-of-quarter/year effect. We also find that loser funds engage in stronger portfolio pumping and that price manipulation is negatively related to expense ratio and ESG rating.

Our study adds to the body of literature in two ways. First, to the best of the authors’ knowledge, this study is the first to explore portfolio pumping across ESG equity mutual funds. Examining if ESG fund managers provide accurate disclosure of portfolio values in financial statements is an extremely relevant issue, as “G” accounts for governance considerations which include business ethics and fair interactions with stakeholders. The motivation behind our initiative lies in the fact that these types of ethical investments might employ unethical tactics that misguide investors about their relative portfolio performance. It is essential for policymakers and regulators to be aware of the occurrence of portfolio pumping in order to develop policy frameworks that would lessen its impact. Investors must also disclose any exceptional end-of-quarter NAV price increases since it helps them to accurately assess the performance of their portfolios. Second, by examining the factors that drive portfolio pumping, we are giving ESG investors clear guidelines for avoiding funds that purposefully manipulate their portfolios.

The conclusions that are drawn from this research are the following. First, ESG funds based in Germany and France maintain compliance with regulations regarding portfolio pumping, as no evidence of spikes in fund portfolio values at quarter-ends is found. Thus, investors can derive utility from socially responsible funds domiciled in these markets through non-performance attributes, as these funds develop ethical business practices, including financial transparency and business integrity. Second, as the UK ESG funds engage in frequent and strong portfolio pumping, it is evident that the Financial Conduct Authority (FCA) should strengthen its supervision to ensure lucidity in the domestic financial market. Third, depending on the timing of their entry into and withdrawal from the funds, certain investors may benefit at the expense of others if the reported NAVs of some ESG funds differ from their true NAVs. Future research should deal with the investigation of the cross-country differences found in our study regarding the level of portfolio pumping between the UK and the remaining markets.