CC BY-NC-ND
0
DOI: https://doi.Org/10.17323/j.jcfr.2073-0438.18.1.2024.5-19 JEL classification: G11, G20
fee]
The Impact of ESG Ratings on Exchange-Traded Fund Flows
Yury Dranev ki
PhD, Leading Research Fellow, National Research University Higher School of Economics, Moscow, Russia, ydranev@hse.ru, ORCID
Mikhail Miriakov
PhD, leading expert, National Research University Higher School of Economics, Moscow, Russia, mmiryakov@hse.ru, ORCID
Elena Ochirova
PhD, Research Fellow, National Research University Higher School of Economics, Moscow, Russia, eochirova@hse.ru, ORCID
Gennadii Baranovskii
Leading expert, ESG Consulting Agency, Moscow, Russia, G.baranovsky83@gmail.com, ORCID
The aim of our paper is to examine the impact of environmental, social, and governance (ESG) ratings on investment decisions in the pre-pandemic US bond and equity exchange-traded fund (ETF) markets. We measure the attractiveness of investments in the ETF as net fund flows and estimate whether the attractiveness varies with the ESG score. For empirical estimations, we employ the regression analysis methodology; specifically, we use linear mixed-effect model to analyze time-series dataset and ordinary least squares to analyze the cross-section data. On the one hand, we found that, on average, ETFs which comply with ESG criteria attracted additional net assets per month as compared to conventional ETFs. Thus, the results of our study indicate that investors demonstrate collective preference towards ESG investments and pay attention to the information on whether the ETF complies with the ESG criteria. On the other hand, we found mixed evidence that higher ESG score always leads to larger investments: differences in scores could not explain the variation in net fund flows. Overall, our study shows that ETF market investments are not directed by the risk-return profile only, and investors also have non-pecuniary motives for their decisions. The results have several practical implications. First, our findings offer business entities useful insight into the fact that incorporation of ESG policy can increase the attractiveness of their business for potential investors. Second, it shows that the market participants would benefit from increasing transparency and unification of rating methodology.
Keywords: exchange-traded funds, sustainable finance innovation, ESG score, ESG compliance
For citation: Dranev Y., Miriakov M., Ochirova E., Baranovskii G. (2024) The Impact of ESG Ratings on Exchange-Traded Fund Flows. Journal of Corporate Finance Research. 18(1): 5-19. https://doi.org/10.17323/j. jcfr.2073-0438.18.1.2024.5-19
The journal is an open access journal which means that everybody can read, download, copy, distribute, print, search, or link to the full texts of these articles in accordance with CC Licence type: Attribution 4.0 International (CC BY 4.0 http://creativecommons.org/licenses/by/4.0/).
Abstract
Introduction
Since firms significantly increase expenses for environmental, social, and governance (ESG hereafter) activities, the financial market's assessment of the shift toward sus-tainability and social responsibility gains importance. The reallocation of fund flows to ESG assets has major implications for investment decisions [1], and several studies suggested the introduction of investor's personal tastes into the asset pricing model [2] - particularly, the inclusion of preferences for sustainable investment [3]. However, while some investors may have strong inclinations towards highly rated ESG assets because of non-pecuniary motives, others may consider the information on risk-return profiles as a framework for their decisions [4]. The overall reaction of market participants to ESG-related information remains a debatable issue and requires additional theoretical and empirical examination [e.g. 3; 5].
In this study, we attempt to assess the ESG preferences of investors and the impact of the ESG rating on the attractiveness of exchange-traded funds (ETF hereafter). ETFs are investment entities that track an index or a basket of assets [6]. For the past decade, the ETF industry has become a primary competitor for actively managed funds [7]. Since the shift of conventional wisdom in favor of passive investment strategies, the total net assets of ETFs have been growing rapidly [8]. The rise of the ETF market has been studied by numerous researchers, but relatively little attention has been heeded to the relationship between ESG policies and investments in the ETF market. Recently, several financial scholars examined the impact of the ESG rating on the financial performance and riskiness of ETF investments [1; 9-11]. The primary focus of our study is on fund flows as an indicator of ETF attractiveness for investors [e.g. 5]. We use two measures to capture the ESG-related information. First, the fact for an ETF of being compliant with ESG criteria is obtained from the MSCI ESG Score and the Morningstar's list of socially conscious funds. Second, the difference in MSCI ESG Score of ETFs measures the ability of underlying assets to manage risks and opportunities arising from ESG factors. These metrics are used to assess (1) whether ESG ETFs attract more investments as compared with the conventional ETFs, and (2) whether a higher level of ESG score is associated with the higher level of investments.
The main contribution of this paper is twofold. Firstly, the results indicate that ETFs that comply with ESG criteria attracted more investments in US bond and equity ETF markets from 2018 to 2020. Thus, our study provides evidence of nonfinancial incentives of investors in ETF: overall, the financial market rewards ESG ETFs with additional investment flows. Secondly, we could not find evidence that market participants consider the differences in the ESG score. The ESG score of ETFs does not explain the variation in the fund flows. Such investment behavior is consistent with previous findings that investors tend to react to basic sus-tainability metrics [e.g. 5] and often ignore complicated information in their decision-making process [e.g. 12].
The rest of the paper proceeds as follows. Second section offers a review of academic literature concerning ESG information in financial decision-making. In this section, we state the main hypotheses concerning the non-pecuniary motives of ETF investors and the role of the ESG score in decision-making. Sections three and four describe the methodology and data. Section five outlines the empirical results of the econometric analysis. Finally, Section six concludes with the discussion of results and its theoretical and practical implications as well as the limitations of our study and avenues for further research.
Development of hypotheses
Do investors in ETFs have non-pecuniary motives?
While ESG-compliant assets attract more funding, the important question concerns the reasons behind this tendency: whether it is a reflection of the attractiveness of related segments of the financial market, or a shift from conventional instruments to ESG-motivated investments. As the share of sustainable investments increases [13], a growing number of studies have examined the factors influencing the attractiveness of such financial instruments [14]. Several studies analyzed the market performance of ESG-compliant financial instruments. However, the evidence is mixed [e.g. 14]. Some empirical studies discovered that the ESG investing may reduce risk and provide superior returns. The attractiveness of investments in ESG assets was confirmed by T. Kanamura, A. Borgers et al., and T. Barko et al. [1; 15-16]. A. Amel-Zadeh and G. Serafeim showed that for investors the key motivation to use ESG information is its relevance to investment performance [17]. Other studies found evidence of low returns on socially responsible investments [18-23]: these authors suggest that ESG-motivated investors underperform in the market due to the non-pecuniary utility, which means sacrificing returns in order to invest responsibly.
To reconcile these contradictory empirical results, several studies explicitly incorporated non-financial incentives into modern portfolio theory. A prominent example of such theoretical research is E. Fama and K. French, who studied how the personal preferences of investors may affect asset prices in a real-world economy [2]. In a recent study L. Pedersen et al. developed an asset pricing model by including the ESG attitude of investors and proposed an ESG-adjusted asset pricing model [3]. Their model predicts that the proportion of different types of investors affect both the returns and resource allocation in the financial market.
Recent literature treats the attitude of investors toward ESG as an important factor that affects market resource allocation [e.g., 5]. In our study, we assume that ETF market investors are aware of ESG policy and pay attention to the general ESG-related information. The fact that an ETF complies with ESG criteria is important information in making investment decisions. Hence the first hypothesis states:
H1a: The compliance of a bond ETF with ESG criteria positively affects ETF flows.
H1b: The compliance of an equity ETF with ESG criteria positively affects ETF flows.
Do investors pay attention to the ESG Score?
Despite the progress that companies have made in disclosing their ESG performance during the last decade, the assessment of ESG factors usually entails high costs [24-26]. Therefore, rating agencies play an important mediatory role between firms and investors, provide information influencing investors' decisions and may thus direct fund flows in the financial market [27-28].
Several studies emphasized various challenges that ranking agencies had to deal with [11; 29]. First, investors often do not behave as rational agents, and look for simpler signals while making a decision [e.g., 12]. For ESG performance, the literature suggests that investors tend to respond to the highly ranked assets and ignore the others [e.g., 5; 30]. Some researchers warn that naive use of primary information on ESG ranking may be misleading [31], since non-expert investors face difficulties in linking numerous sustainability concepts in a coherent way [32]. Second, the uncertainty of ESG-related information constitutes an additional obstacle in decision-making. There are no uniform standards in ESG information disclosure, and rating agencies provide various ESG scores using opaque methods; the variability of approaches to the ESG ratings of firms may lead to biased investors' decisions in cases of information abundance [33]. The lack of unified methodology for assigning company-specific ratings increases the gap between the ESG scores of different ESG rating providers [31; 34].
Thus, to test whether a high ESG score increases the attractiveness of ETFs for investors, we developed the second hypothesis as follows:
H2a: A ESG score positively affects flows to bond ETFs. H2b: A ESG score positively affects flows to equity ETFs.
Methodology
Modeling the ESG compliance effect
We tested hypotheses H1a and H1b using linear mixed-effect model [e.g., 35]. In order to estimate the impact of ESG compliance on fund flows, we use the following model specification for ETF i and month t: FlowTNA ,t = Po + PESG Compliance^ + P2 ERt + + P3Return t + P4LogHoldingi t + P5PriceNAVi t + + P6 Log Agei t + P7Spread Pricei t + + P8Log Turnover t ■ (!)
Table 1 provides the definition of variables. The dependent variable is the one-year fund flow to net total assets ratio (Flow TNAit), which is a proxy for the attractiveness
of the ETF. Since one of the major advantages of passive investments is low managerial fees, we control the model for expense ratio (ERi, t) and assume that even a small increase is associated with a fall of fund flows [36]. High returns (Returnit) for the previous period, as one of the major motives to invest, positively affects the attractiveness of an ETF [1]. The number of underlying securities (Log Holdingit) is assumed to have a positive effect, since investors may have concerns about small numbers of holdings [37]. The ratio of the fund's market price to its book value (Price NAVi,t) may represent the inflows to ETFs. The assets of the newly launched ETF are expected to grow faster in percentage terms, indicating the larger inflows. Thus, the age of the fund (Log Agei,t) is expected to have a negative impact on asset-weighted fund flows [36]. The turnover (Log Turnoverit) controls for fund liquidity, which should have a positive effect [36]. Likewise, the bid-ask spread (Spread Pricei,t) shows the fund's liquidity.
Table 1. List of variables (ESG compliance effect modelling)
Variable Description
Dependent variable
Flow TNA The ratio of monthly fund flow divided by total net assets (TNA), %
Independent variable
ESG compliance Dummy variable, 1 - the fund complies with the ESG criteria, 0 -otherwise
Control variables
ER Expense ratio set by the fund, %
Return Aggregated monthly return lagged for one month, %
Log Holdings Natural logarithm of the number of securities owned by the fund
Price NAV Price of the ETF to the fund's Net Asset Value, %
Log Age Natural logarithm of the age of the fund, months
Spread Price Ratio of the ETF's price spread to its price, %
Log Turnover Natural logarithm of turnover divided by the total amount traded
We structure the panel data set of ESG-compliant ETFs and conventional ETFs using data provided by MSCI for March 2020 (available at ETF Database - ETFdb.com). Only ETFs included in both MSCI data and Morningstar's list were considered to be ESG compliant. To construct a
comparison subsample of conventional ETFs, we followed es for ESG ETFs. In the second step, we conducted further
the procedure described below. First, we identified the list matching based on asset-adjusted fund flows, exploited
of issuers of ESG-compliant ETFs. Therefore, all conven- age, expense ratio, and the number of holdings, following
tional ETFs were combined in the pool of potential match- [38-39]:
(Flow TNAj - Flow TNAj )2 (Agei - Age} )2 (ER - ERj )2 (Holdingsi - Holdings] )2 Matchi, j = 3 + 3 + 3 + 3 , (2)
CT
Age UER u Holdings
where, a is the cross-sectional deviation.
Following L. Renneboog et al., we restricted potential matches among conventional ETFs to be no more than 2 years older or younger than the ESG-compliant ETF [39]. This prevents an estimation bias of life-cycle effects and macroeconomic time-series effects. To construct panel A, for each ESG compliant ETF, we added one conventional ETF using the matching measure. Similarly, we constructed panel B by matching one ESG-compliant ETF to two conventional ETFs. Since several ESG-compliant ETF providers had less than two conventional ETFs, some matches have different issuers. The final subsamples of ESG compliant bond ETFs and ESG compliant equity ETFs covers 15 and 42 funds respectively. The lists of conventional and ESG compliant funds are provided in Appendix A.
Modeling the ESG score effect
In order to test the effect of ESG score, we estimate the following regression model using the ordinary least squares (OLS hereafter) method: FlowTNAi =p0 + frESG Scoret + P2 ER + + P3 Return + LogVolumei + SDi + + P6 Log Agei + j37Volatilityi. (3)
In the case of heteroscedasticity, we applied OLS with Hu-ber-White robust standard errors (the results of heteroscedasticity testing are in the Appendix). Table 2 provides the definition of variables of the regression equation. As in the case of the time-series model, the dependent variable is the one-year fund flow to net total assets ratio (Flow TNAi■). We considered five proxies of ESG measures for different model specifications. In Model 1, ETFs' ESG scores are provided by MSCI for March 2020 (available at ETF Database - ETFdb.com). The MSCI Inc. dominates the market of ESG ranking data providers, covering about 40% of the entire market [40]. In Model 2, the ESG score peer percentile (ESG Peer ) normalizes the ESG score to other ETFs in the same peer group. In model 3, the ESG score global percentile (ESG Globali) normalizes the ESG score to all funds in the MSCI ESG Fund Metrics coverage. In Model 4, SRI exclusion criteria (ESG Exclusion ) allows us to identify the level of funds' exposure to companies involving at least one SRI exclusion factor (e.g., alcohol, gambling, weapons, etc.). In Model 5, sustainable impact solutions (Sustainable Impacti) is the portfolio weighted average of each company's percentage of revenue generated by Sustainable Impact Solutions goods and services. In the cross-section model, we additionally control for a standard deviation of return ( SDi ) which is a measure of invest-
ment riskiness that is expected to have a negative impact on fund flows [9]. The average traded volume of a fund (Log Volumei) demonstrates the overall activity [41]. It is expected to have a positive effect. Finally, we expect the positive relationship between adjusted fund flows and fund volatility (Volatilityi) for the last 200 days, compared to its peer group in ETFdb.com [39].
Table 2. List of variables (ESG score effect modelling)
Variable Description
Dependent variable
Flow TNA The ratio of one-year fund flow divided by total net assets (TNA), %
Independent variable
ESG Score MSCI ESG score, 1 to 10
ESG Score Peer Percentile Measure of how the ESG score of ETF ranks relative to other funds in the
same peer group, %
ESG Score Global Percentile Measure of how the ESG score of ETF ranks relative to all funds in MSCI ESG Fund Metrics coverage, %
SRI Exclusion ETF's exposure to companies flagged for at least one SRI exclusion factors (e.g., alcohol, gambling, weapons), %
Sustainable impact Portfolio weighted average of each company's percent of revenue generated by Sustainable Impact Solutions goods and services, %
Control variables
ER Expense ratio set by the fund, %
Return Aggregated annual return for the previous year, %
Log Volume Logarithm of a fund's average traded volume, $
SD Standard deviation of a fund's returns, %
Log Age Logarithm of Age of fund, months
Volatility Volatility of the fund for last 200 days, compared to its peer group in ETFdb. com, %
Data
For the purposes of empirical testing, we collected 2 data samples for each model. The first sample covers the period from March 2018 to March 2020. A significant part of ESG-compli-ant ETFs were founded in 2015 and later, thus, it is impossible to collect earlier data appropriate for empirical study in the case
Table 3. Descriptive statistics of bond ETFs based on panel data
Panel A: Bond ETFs 1-1
Variables Mean St.Dev (1) (2) (3) (4) (5) (6) (7) (8) (9)
(1) Flow TNA 0.017 0.128 1
(2) ESG compliance 0.500 0.500 0.180*** 1
(3) ER 0.003 0.002 -0.014 -0.038 1
(4) Return 0.369 1.269 0.049 0.063* -0.047 1
(5) Log Holdings 5.347 1.837 0.057 0.230*** -0.230*** 0.005 1
(6) Price NAV 1.000 0.003 0.120*** 0.200*** 0.080** 0.078** -0.049 1
(7) Log Age 3.541 0.903 -0.130*** -0.130*** 0.021 0.070* -0.180*** -0.170*** 1
(8) Spread Price 0.304 8.140 -0.005 -0.037 0.014 -0.012 -0.085** -0.013 0.020 1
(9) Log Turnover 15.597 2.428 0.150*** 0.280*** 0.080** 0.073** 0.270*** 0.059 0.090** -0.036 1
No of obs: 750. * indicates significance at 10%, ** indicates significance at 5%, *** indicates significance at 1%.
Panel B: Bond ETFs 1-2
Variables Mean St.Dev (1) (2) (3) (4) (5) (6) (7) (8) (9)
(1) Flow TNA 0.019 0.110 1
(2) ESG compliance 0.333 0.472 0.140*** 1
(3) ER 0.002 0.003 -0.013 0.110*** 1
(4) Return 0.337 1.243 0.083*** 0.064** -0.043 1
(5) Log Holdings 5.339 1.612 0.062** 0.190*** -0.140*** -0.0003 1
(6) Price NAV 1.000 0.003 0.140*** 0.110*** 0.009 0.078*** -0.045 1
(7) Log Age 3.708 0.852 -0.077*** -0.230*** -0.084*** 0.075** -0.097*** -0.086*** 1
(8) Spread Price 0.203 6.646 -0.005 -0.021 0.017 -0.009 -0.079*** -0.013 0.012 1
(9) Log Turnover 16.594 2.645 0.100*** -0.085*** -0.180*** 0.041 0.180*** 0.033 0.260*** -0.039 1
No of obs: 1125. * indicates significance at 10%, ** indicates significance at 5%, *** indicates significance at 1%.
of ESG ETFs [42]. According to Statista, the value of Global ESG ETF assets started growing rapidly in 2017-2018 [43]. Besides, the sample is limited to the beginning of 2020, due to the Covid-19 pandemic's harsh impact on the economy and financial markets [11].
We use balanced panel data with financial information from the Bloomberg database. We employ the fund flow to the net total assets ratio as a dependent variable. Return of the funds, age and expense ratio are also included as independent variables. Additionally, we control for the number of securities owned by
the ETFs, the ratios of the ETF's price to net assets, the ETF's price spread to its price, and the turnover ratio of the funds. Table 3 presents the descriptive statistics for bond ETFs based on panel data. For the majority of variables, both panels have similar results.
Table 4 shows the descriptive statistics for equity ETFs. In comparison, bond ETFs demonstrated a higher average return than equity ETFs. The spread price difference was also higher for bond ETFs. Table 4. Descriptive statistics of equity ETFs based on panel data
Panel A: Equity ETFs 1-1
Variables Mean St.Dev (1) (2) (3) (4) (5) (6) (7) (8) (9)
(1) Flow TNA 0.003 0.164 1
(2) ESG compliance 0.500 0.500 0.074*** 1
(3) ER 0.004 0.002 -0.150*** 0.085*** 1
(4) Return -0.308 5.908 0.058*** 0.021 0.007 1
(5) Log Holdings 4.460 1.410 0.160*** 0.020 -0.480*** -0.008 1
(6) Price NAV 1.000 0.003 0.067*** 0.053** -0.097*** 0.160*** 0.064*** 1
(7) Log Age 4.259 0.701 -0.081*** -0.240*** 0.420*** -0.001 -0.330*** -0.140*** 1
(8) Spread Price 0.054 1.044 0.045** 0.031 -0.006 0.027 -0.012 0.025 -0.030 1
(9) Log Turnover 16.766 1.861 0.030 -0.310*** -0.210*** -0.110*** 0.140*** -0.055** 0.350*** -0.066*** 1
No of obs: 2100. * indicates significance at 10%, ** indicates significance at 5%, *** indicates significance at 1%.
Panel B: Equity ETFs 1-2
Variables Mean St.Dev (1) (2) (3) (4) (5) (6) (7) (8) (9)
(1) Flow TNA 0.002 0.149 1
(2) ESG compliance 0.333 0.471 0.062*** 1
(3) ER 0.005 0.002 -0.150*** 0.052*** 1
(4) Return -0.333 6.059 0.086*** 0.017 0.006 1
(5) Log Holdings 4.566 1.363 0.130*** -0.041** -0.480*** -0.013 1
(6) Price NAV 1.000 0.003 0.075*** 0.067*** -0.095*** 0.160*** 0.044** 1
(7) Log Age 4.400 0.681 -0.081*** -0.320*** 0.380*** -0.010 -0.220*** -0.120*** 1
(8) Spread Price 0.049 1.098 0.039** 0.024 0.019 0.016 -0.010 0.019 -0.017 1
(9) Log Turnover 17.283 1.905 0.015 -0.410*** -0.210*** -0.110*** 0.160*** -0.053*** 0.420*** -0.051*** 1
No of obs: 3150. * indicates significance at 10%, ** indicates significance at 5%, *** indicates significance at 1%.
The second data sample is obtained from ETFdb.com on the US ETF market. The sample does not cover inverse and leveraged ETFs because of the differences in investment strategies. The overall sample consists of 206 bonds and 1,095 equity ESG ETFs. Table 5 presents the descriptive statistics for cross-sectional data. The average ESG score for bond ETFs is 4.914, while for equity ETFs this score is 5.185. ESG Score Peer Percentile and ESG Score Global Percentile variables do not differentiate substantially between bond and equity funds.
Table 5. Descriptive statistics of ESG ETFs based on cross-sectional data
ESG Bond ETFs
Variables Mean St.Dev (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12)
(1) Flow TNA 0.186 0.340 1
(2) ER 0.002 0.002 -0.180** 1
(3) Return 0.044 0.087 0.052 -0.570*** 1
(4) Log Volume 11.862 2.409 -0.031 -0.330*** 0.180*** 1
(5) SD 0.016 0.017 -0.160** -0.030 0.530*** 0.180*** 1
(6) Log Age 4.201 0.701 -0.440*** -0.140** 0.220*** 0.620*** 0.330*** 1
(7) Volatility 0.161 0.090 -0.052 0.290*** 0.015 -0.002 0.490*** 0.056 1
(8) ESG_Score 4.914 1.265 -0.026 -0.550*** 0.600*** 0.180*** 0.024 0.260*** -0.450*** 1
(9) ESG-Peer 0.557 0.302 0.0002 -0.230*** 0.340*** 0.075 0.130* 0.200*** -0.170** 0.640*** 1
(10) ESG-Global 0.419 0.252 -0.030 -0.520*** 0.580*** 0.190*** 0.055 0.270*** -0.440*** 0.980*** 0.680*** 1
(11) ESG-Exclusion 0.049 0.040 0.150** -0.100 0.062 -0.190*** -0.069 -0.240*** 0.210*** -0.200*** -0.370*** -0.280*** 1
(12) Sustainable-Impact 0.022 0.020 0.092 0.072 -0.066 -0.180** -0.083 -0.260*** 0.250*** -0.260*** -0.320*** -0.320*** 0.640*** 1
No of obs: 206. * indicates significance at 10%, ** indicates significance at 5%, *** indicates significance at 1%.
ESG Equity ETFs
Variables Mean St.Dev (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) I
(1) Flow TNA -0.025 0.550 1
(2) ER 0.004 0.002 -0.160*** 1
(3) Return -0.145 0.136 0.180*** -0.110*** 1
(4) Volume 10.959 2.440 0.088*** -0.260*** 0.038 1
(5) SD 0.043 0.036 -0.046 -0.310*** 0.023 0.260*** 1
(6) Log Age 4.348 0.805 -0.260*** -0.035 -0.095*** 0.540*** 0.460*** 1
(7) Volatility 0.485 0.099 -0.019 -0.011 -0.460*** 0.220*** 0.210*** 0.210*** 1
(8) ESG_Score 5.185 1.408 0.094*** -0.190*** 0.290*** 0.075** -0.064** 0.001 -0.230*** 1
(9) ESG-Peer 0.428 0.287 0.066** -0.170*** 0.220*** 0.079*** 0.032 0.013 -0.150*** 0.670*** 1
(10) ESG-Global 0.466 0.269 0.100*** -0.180*** 0.300*** 0.068** -0.087*** -0.007 -0.250*** 0.980*** 0.670*** 1
(11) ESG-Exclu-sion 0.076 0.096 0.019 -0.082*** 0.088*** 0.031 0.026 0.046 -0.094*** 0.330*** 0.160*** 0.330*** 1
(12) Sustainable-Impact 0.062 0.067 0.064** 0.059** 0.270*** -0.066** -0.054* -0.001 -0.180*** 0.200*** 0.170*** 0.200*** -0.074** 1
No of obs: 1095. * indicates significance at 10%, ** indicates significance at 5%, *** indicates significance at 1%.
Empirical results
ESG compliance and fund flows.
The time-series model addresses the hypothesis that the ESG compliance criteria affect the flows of the ETF positively and significantly. Tables 6 and 7 show the results of econometric analysis. To check whether the results are robust, we estimated two panels (A and B) with pooled OLS models.
Table 6 shows that the bond ESG ETFs attracted more investments than conventional ETFs: the dummy variable for ESG is statistically significant. Thus, H1a (compliance of a bond ETF with ESG criteria significantly and positively affects ETF flows) cannot be rejected at a 1% level of significance. This result is consistent: both panels confirmed a positive and significant relationship between ESG compliance and fund flows. Moreover, the robustness test also confirms the positive effect of ESG compliance on fund flows.
Table 6. ESG compliance and fund flows of bond ETFs: econometric analysis results
Panel A: Bond ETF 1-1 Panel B: Bond ETF 1-2
Dependent Variable Fund flow to TNA
Independent Variables Pooled OLS Mixed model Pooled OLS Mixed model
Intercept -2.347* -2.331 -3.983*** -3.028***
ESG Compliance 0.031*** 0.029** 0.028*** 0.029***
ER -2.149 -2.206 -0.453 0.167
Return 0.003 0.008** 0.006** 0.009***
Log Holdings -0.002 -0.002 0.001 0.0005
Price NAV 2.320 2.274 3.936*** 2.960***
Log Age -0.017*** -0.013* -0.010** -0.008
Spread Price 0.00008 0.00006 0.0001 0.00003
Log Turnover 0.007*** 0.008*** 0.005*** 0.007***
ETF effects No Yes No Yes
Time effects No Yes No Yes
No of obs. 750 750 1 125 1 125
R2 0.063 0.129 0.058 0.133
F-test 6.232*** 8.535***
Note: The table shows the results of panel regression models created to identify the impact of ESG compliance on US bond ETFs. The dependent variable is the fund flows to total net assets ratio. R2 for mixed linear models are conditional.
* Indicates significance at 10%.
** Indicates significance at 5%.
*** Indicates significance at 1%.
Table 7 reports the results of the H1b hypothesis' tests. According to the regression analysis, equity ESG ETFs, on average, attracted more investments than conventional ETFs. Both A and B panels confirmed a positive and significant relationship between ESG compliance and equity ETF flows.
Additional analysis using pooled OLS methodology indicates that the results are robust. As in the case of the bond ETF market, tests confirm that H1b (compliance of an equity ETF with ESG criteria significantly and positively affects ETF flows) cannot be rejected at a 1% level of significance.
Table 7. ESG compliance and fund flows of equity ETFs: econometric analysis results
Panel A: Equity ETF 1-1 Panel B: Equity ETF 1-2
Dependent Variable Flow to TNA
Independent Variables Pooled OLS Mixed model Pooled OLS Mixed model
Intercept -2.023* -2.005* -2.033** -1.926**
ESG Compliance 0.029*** 0.031*** 0.023*** 0.025***
ER -7.831*** -7.966*** -6.76*** -6.904***
Return 0.001** -0.002* 0.002*** 0.0004
Log Holdings 0.013*** 0.013*** 0.009*** 0.009***
Price NAV 1.933* 1.904* 1.988** 1.881**
Log Age 0.005 0.007 -0.002 -0.00005
Spread Price 0.007** 0.007** 0.005** 0.005**
Log Turnover 0.002 0.002 0.002 0.002
ETF effects No Yes No Yes
Time effects No Yes No Yes
No of obs. 2100 2100 3150 3150
R2 0.048 0.084 0.042 0.078
F-test 13.11*** 17.40***
Note: The table shows the results of panel regression models created to identify the impact of ESG compliance on US equity funds. The dependent variable is the fund flows to total net assets ratio. R2 for mixed linear models are conditional.
* Indicates significance at 10%.
** Indicates significance at 5%.
*** Indicates significance at 1%.
The overall evidence strongly confirms the positive link between the ETFs flows and the compliance with ESG criteria.
ESG score and fund flows
Tables 8 and 9 present the results for bond and equity ETF markets, respectively. We used five proxies of ESG perfor-
mance to estimate the impact on fund flows. The overall MSCI ESG score has no significant impact on fund flows on equity and bond markets. Moreover, two additional measures of ESG performance - ESG score peer percentile and ESG exclusion criteria - also have no influence on ETF flows.
Table 8. ESG score and bond ETFs' flows: econometric analysis results
Dependent Variable Flow_Assets
Independent Variables Model 1 Model 2 Model 3 Model 4 Model 5
Intercept 1.042*** 0.955*** 1.000*** 0.974*** 1.022***
ER -22.108 -20.988 -21.588 -20.330 -21.575
Return 0.600 0.409 0.541 0.489 0.515
Log Volume 0.048*** 0.050*** 0.048*** 0.049*** 0.0048***
SD -2.334 -2.217 -2.219 -2.045 -2.380
Dependent Variable Flow_Assets
Independent Variables Model 1 Model 2 Model 3 Model 4 Model 5
Log Age -0.319*** -0.332*** -0.322*** -0.322*** -0.328***
Volatility 0.200 0.316 0.230 0.221 0.317
ESG_Score -0.011
ESG-Peer 0.092
ESG-Global -0.025
ESG-Exclusion 0.185
Sustainable-Impact -0.678
No of obs. 206 206 206 206 206
R2 0.325 0.330 0.324 0.325 0.326
Robust st.error No No No No Yes
F-test 13.60*** 13.90*** 13.57*** 13.59*** 15.95***
Ramsey RESET 0.078 0.035 0.092 0.100 0.074
p-value 0.780 0.853 0.761 0.752 0.785
Note: This table reports the regression analysis of the ESG score on the fund flow of US bond ETFs. The dependent variable is the ratio of one-year fund flow divided by total net assets.
* Indicates significance at 10%.
** Indicates significance at 5%.
*** Indicates significance at 1%.
On the equity ETF market, sustainable impact solutions and ESG-Global Percentile have a significant and positive effect on fund flows. We additionally tested our regression models for specification errors, and the Ramsey test indicated the absence of omitted variables. Moreover, robust standard errors are used when the assumption of homoscedasticity is violated. The results for heteroscedasticity are provided in Appendix B.
Table 9. ESG score and equity ETFs' flows: econometric analysis results
Dependent Variable Flow_Assets
Independent Variables Model 1 Model 2 Model 3 Model 4 Model 5
Intercept 0.426*** 0.508*** 0.450*** 0.515*** 0.480***
ER -16.190* -17.686** -15.695** -17.880** -18.917**
Return 0.589*** 0.615*** 0.577*** 0.622*** 0.553***
Log Volume 0.060*** 0.059*** 0.059*** 0.059*** 0.061***
SD 0.652 0.559 0.698* 0.559 0.619
Log Age -0.289*** -0.287*** -0.290*** -0.288*** -0.292***
Volatility 0.425** 0.404** 0.439** 0.405** 0.413**
ESG_Score 0.016
ESG-Peer 0.028
Dependent Variable Flow_Assets
Independent Variables Model 1 Model 2 Model 3 Model 4 Model 5
ESG-Global 0.108*
ESG-Exclusion 0.092
Sustainable-Impact 0.533**
No of obs. 1095 1095 1095 1095 1095
R2 0.17 0.1688 0.171 0.1689 0.1724
Robust st.error Yes Yes Yes Yes Yes
F-test 34.81*** 35.03*** 35.04*** 35.06*** 36.17***
Ramsey RESET 2.096 1.939 1.867 2.227 2.184
p-value 0.148 0.164 0.171 0.136 0.140
Note: This table reports the regression analysis of ESG score on fund flow of US equity ETFs. The dependent variable is the ratio of one-year fund flow divided by total net assets.
* Indicates significance at 10%.
** Indicates significance at 5%.
*** Indicates significance at 1%.
Thus, empirical models provide mixed results. The majority of ESG performance measures do not explain the variation in the ETF flows. The sustainable impact index and ESG-Global Percentile positively affect only equity ETF flows. Overall, empirical results do not confirm hypotheses H2a and H2b, which postulate the positive effects of ESG scores on the flows of bond and equity ETFs.
Conclusion and Discussion
The financial market plays a crucial intermediary role in the saving-investment process, and the determination of factors directing investors' resources is highly relevant for both academic discussion and practical implication. In this study, we focus on ESG preferences of ETF market investors and assess the impact of ESG ranking on the attractiveness of exchange-traded funds. We found that, on average, ETFs that comply with ESG criteria attracted additional net assets per month as compared to conventional ETFs. Thus, our results may indicate that investors pay attention to ESG-related information and have strong preferences toward ESG investing. We also found mixed evidence that ESG ranking measures affect the allocation of resources in the financial market. Our analysis suggests that a higher ESG score is not a prerequisite of the larger investments: differences in scores could not explain the variation in fund flows. Taken together, our findings confirm that ETF market fund flows are not limited by the risk-return profile, and that investors have non-pecuniary motives for their decisions. At the same time, the decision-making process largely ignores ESG scores and follows a simpler behavioral pattern, which is consistent with the previous findings [5; 30].
Since investors have ESG preferences, social and environmental responsibility is one of the factors that should steer companies in allocating their limited resources. Thus, it is of high importance for a firm's management to incorporate ESG policy and increase the attractiveness of their business for potential investors. Ignoring ESG factors may have a negative impact on a firm's performance. Our evidence also emphasizes the need for additional control of ESG information flows. Generally, investors have limited capacities in processing ESG-related information and are looking for a simple signal as to whether the ETF is compliant with ESG criteria or not. However, even though the ESG objective is becoming one of the key factors for asset allocation, the average investor makes decisions in the absence of a unique and transparent methodology behind ESG measurement. The ESG score value may be biased because firms still make misleading ESG disclosures [e.g. 44]. Moreover, most non-institutional investors may not be familiar with the internal procedures behind the ESG rating approach [45]. Thus, market participants would benefit from increasing transparency and unification of rating methodology [46].
Our research has several limitations. First of all, we did not distinguish between professional investors (e.g., institutional investors) and less sophisticated, household investors. Since we focused on the ETF market dominated by household investors, our results may mostly describe the behavior of non-professional investors in ESG assets. The way experts incorporate ESG compliance in their decision-making process may differ significantly, since institutional investors have the capacity to develop their own ESG-related goals and to avoid externally assigned scores.
Secondly, we restricted our sample to the beginning of the Covid-19 pandemic, because of its harsh effect on financial markets and the global economy. Our research revealed the pre-pandemic patterns of decision-making, while the pandemic could have caused dramatic changes in the preferences and behavior of household investors. These limitations suggest avenues for further research.
Acknowledgement
The article was prepared within the framework of the HSE University Basic Research Program.
References
1. Kanamura T. Risk mitigation and return resilience for high yield bond ETFs with ESG components. Finance Research Letters. 2021;41:101866. https://doi. org/10.1016/j.frl.2020.101866
2. Fama E.F., French K.R. Disagreement, tastes, and asset prices. Journal of financial economics. 2007;83(3):667-689. https://doi.org/10.1016/]. jfineco.2006.01.003
3. Pedersen L.H., Fitzgibbons S., Pomorski L. Responsible investing: The ESG-efficient frontier. Journal of Financial Economics. 2021;142(2):572-597. https://doi.org/10.1016/j.jfineco.2020.11.001
4. Broadstock D.C., Chan K., Cheng L.T., et al. The role of ESG performance during times of financial crisis: Evidence from COVID-19 in China. Finance research letters. 2021;38:101716. https://doi.org/10.1016/'. frl.2020.101716
5. Hartzmark S.M., Sussman A.B. Do investors value sustainability? A natural experiment examining ranking and fund flows. The Journal of Finance. 2019;74(6):2789-2837. https://doi.org/10.1111/ jofi.12841
6. Pagano M., Sánchez Serrano A., Zechner J. Can EFTs Contribute to Systemic Risk? ESRB: Advisory Scientific Committee Reports. 2019;9. https://doi. org/10.2139/ssrn.3723329
7. Ben-David I., Franzoni F., Moussawi R. Exchange-traded funds. Annual Review of Financial Economics. 2017;9:169-189. https://doi.org/10.1146/annurev-financial-110716-032538
8. Cremers K.M., Fulkerson J.A., Riley T.B. Challenging the conventional wisdom on active management: A review of the past 20 years of academic literature on actively managed mutual funds. Financial Analysts Journal. 2019;75(4):8-35. https://doi.org/10.1080/001 5198X.2019.1628555
9. Kanuri S. Risk and return characteristics of environmental, social, and governance (ESG) equity ETFs. The Journal of Index Investing Fall. 2020;11(2):66-75. https://doi.org/10.3905/ jii.2020.1.092
10. Pavlova I., de Boyrie M.E. ESG ETFs and the COVID-19 stock market crash of 2020: Did clean funds fare better? Finance Research Letters. 2022;44:102051. https://doi.org/10.1016/j. frl.2021.102051
11. Folger-Laronde Z., Pashang S., Feor L., et al. ESG ratings and financial performance of exchange-traded funds during the COVID-19 pandemic. Journal of Sustainable Finance & Investment. 2022;12(2):490-496. https://doi.org/10.1080/20430795.2020.1782814
12. Ben-David I., Li J., Rossi A., Song Y. What do mutual fund investors really care about?. The Review of Financial Studies. 2022;35(4):1723-1774. https://doi. org/10.1093/rfs/hhab081
13. Global Sustainable Investment Alliance. Global Sustainable Investment Review 2020. URL: http:// www.gsi-alliance.org/trends-report-2020/ (Accessed on 07.06.2023)
14. Gillan S.L., Koch A., Starks L.T. Firms and social responsibility: A review of ESG and CSR research in corporate finance. Journal of Corporate Finance. 2021;66:101889. https://doi.org/10.1016/j-. jcorpfin.2021.101889
15. Borgers A., Derwall J., Koedijk K., et al. Do social factors influence investment behavior and performance? Evidence from mutual fund holdings. Journal of Banking & Finance. 2015;60:112-126. https://doi.org/10.1016/-. jbankfin.2015.07.001
16. Barko T., Cremers M., Renneboog L. Shareholder engagement on environmental, social, and governance performance. Journal of Business Ethics. 2021;180:777-812. https://doi.org/10.1007/s10551-021-04850-z
17. Amel-Zadeh A, Serafeim G. Why and How Investors Use ESG Information: Evidence from a Global Survey. Financial Analysts Journal. 2018;74(3):87-103. https://doi.org/10.2469/faj.v74.n3.2
18. Hong H., Kacperczyk M. The price of sin: The effects of social norms on markets. Journal of financial economics. 2009;93(1):15-36. https://doi. org/10.1016/j.jfineco.2008.09.001
19. Auer B.R., Schuhmacher F. Do socially (ir) responsible investments pay? New evidence from international ESG data. The Quarterly Review of Economics and Finance. 2016;59:51-62. https://doi. org/10.1016/j.qref.2015.07.002
20. Martin P.R., Moser D.V. Managers' green investment disclosures and investors' reaction. Journal of Accounting and Economics. 2016;61(1):239-254. https://doi.org/10.1016/j-.jacceco.2015.08.004
21. Karpf A., Mandel A. Does it pay to be green? Socially Responsible Investment eJournal. 2017. https://doi. org/10.2139/ssrn.2923484
22. Baker M., Bergstresser D., Serafeim G., et al. Financing the response to climate change: The pricing and ownership of US green bonds. National Bureau of Economic Research; 2018. https://doi.org/10.3386/w25194
23. Rompotis G.G. The ESG ETFs in the UK. Journal of Asset Management. 2022;23(2):114-129. https://doi. org/10.1057/s41260-021-00251-z
24. Hockerts K., Moir L. Communicating corporate responsibility to investors: The changing role of the investor relations function. Journal of Business Ethics. 2004;52:85-98. https://doi.org/10.1023/ B:BUSI.0000033109.35980.16
25. Adams C.A., Frost G.R. Integrating sustainability reporting into management practices. Accounting forum. 2008;32(4):288-302. https://doi.org/10.1016/]. accfor.2008.05.002
26. Searcy C., Buslovich R. Corporate perspectives on the development and use of sustainability reports. Journal of business ethics. 2014;121:149-169. https:// doi.org/10.1007/s10551-013-1701-7
27. Slager R., Chapple W. Carrot and stick? The role of financial market intermediaries in corporate social performance. Business & Society. 2016;55(3):398-426. https://doi.org/10.1177/0007650315575291
28. Widyawati L. A systematic literature review of socially responsible investment and environmental social governance metrics. Business Strategy and the Environment. 2020;29(2):619-637. https://doi. org/10.1002/bse.2393
29. Chatterji A.K., Levine D.I., Toffel M.W. How well do social ratings actually measure corporate social responsibility? Journal of Economics & Management Strategy. 2009;18(1):125-169. https://doi.org/10.1111/ j.1530-9134.2009.00210.x
30. Hartzmark S.M. The worst, the best, ignoring all the rest: The rank effect and trading behavior. The Review of Financial Studies. 2015;28(4):1024-1059. https:// doi.org/10.1093/rfs/hhu079
31. Gyönyörovä L., Stachoñ M., Stasek D. ESG ratings: relevant information or misleading clue? Evidence from the S&P Global 1200. Journal of sustainable finance & investment. 2023;13(2):1075-1109. https:// doi.org/10.1080/20430795.2021.1922062
32. Rook D.P. How can we know if investors are coherently linking sustainability concepts?. Journal of Sustainable Finance & Investment. 2012;2(3-4):198-221. https://doi.org/10.1080/20430795.2012.742634
33. Cornell B. ESG preferences, risk and return. European Financial Management. 2021;27(1):12-19. https://doi. org/10.1111/eufm.12295
34. Avramov D., Cheng S., Lioui A., et al. Sustainable investing with ESG rating uncertainty. Journal of Financial Economics. 2022;145(2B):642-664. https:// doi.org/10.1016/j.jfineco.2021.09.009
35. Krasnikov A., Jayachandran S. The relative impact of marketing, research-and-development, and operations capabilities on firm performance. Journal of marketing. 2008;72(4):1-11. https://doi. org/10.1509/jmkg.72.4.001
36. Clifford C.P., Fulkerson J.A., Jordan B.D. What drives ETF flows? Financial Review. 2014;49(3):619-642. https://doi.org/10.1111/fire.12049
37. Oztekin A. Information fusion-based meta-classification predictive modeling for ETF performance. Information Systems Frontiers. 2016;20:223-238. https://doi.org/10.1007/s10796-016-9704-4
38. Bollen N.P. Mutual fund attributes and investor behavior. Journal of financial and quantitative analysis. 2007;42(3):683-708. https://doi.org/10.1017/ S0022109000004142
39. Renneboog L., Ter Horst J., Zhang C. Is ethical money financially smart? Nonfinancial attributes and money flows of socially responsible investment funds. Journal of Financial Intermediation. 2011;20(4):562-588. https://doi.org/10.1016/j.jfi.2010.12.003
40. Simpson C., Rathi A. How to get an ESG rating upgrade. Bloomberg; December 21, 2021. URL: https://www.blo omb erg.com/news/ articles/2021-12-21/how-to-get-an-esg-rating-upgrade (Accessed 06.07.2023).
41. Lee C.C., Chen M.P., Lee C.C. Investor attention, ETF returns, and country-specific factors. Research in International Business and Finance. 2021;56:101386. https://doi.org/10.1016/j-.ribaf.2021.101386
42. The rise of the sustainable fund market and its role in financing sustainable development. Geneva; 2021. URL: https://unctad.org/system/files/official-document/diae2021d1_en.pdf (Accessed 06.07.2023).
43. Statista Research Department. Global ESG ETF assets 2006-2022. 2022. URL: https://www.statista.com/ statistics/1297487/assets-of-esg-etfs-worldwide/ (Accessed 06.07.2023).
44. Yu E.P., Van Luu B., Chen C.H. Greenwashing in environmental, social and governance disclosures. Research in International Business and Finance. 2020;52:101192. https://doi.org/10.1016/-. ribaf.2020.101192
45. Friede G. Why don't we see more action? A metasynthesis of the investor impediments to integrate environmental, social, and governance factors. Business Strategy and the Environment. 2019;28(6):1260-1282. https://doi.org/10.1002/bse.2346
46. In S.Y., Rook D., Monk A. Integrating alternative data (also known as ESG data) in investment decision making. Global Economic Review. 2019;48(3):237-260. https://doi.org/10.1080/1226508X.2019.1643059
Appendix A: Issuers of ETFs
Table A1. Issuers of bond ETFs (panels A and B)
Issuer ESG ETFs (Panels A and B) Non-ESG ETFs (Panel A) Non-ESG ETFs (Panel B)
Inspire Investing 1 0 0
IShares 2 3 8
Sage Advisory 1 0 0
J.P. Morgan 3 0 0
Nuveen 1 1 1
Hartford Funds 2 0 0
Vaneck 1 2 2
Invesco 3 3 13
DWS 1 6 6
Total 15 15 30
Table A2. Issuers of equity ETFs (panels A and B)
Issuer ESG ETFs (Panels A and B) Non-ESG ETFs (Panel A) Non-ESG ETFs (Panel B)
Columbia Threadneedle Investments 4 1 1
Ishares 7 9 21
State Street SPDR 5 7 15
FlexShares 1 4 4
Inspire Investing 2 0 0
Global X 2 3 6
Nuveen 5 0 0
ETF Managers Group 1 1 1
VanEck 2 0 0
First Trust 4 5 12
Invesco 7 10 22
Strategy Shares 1 1 1
Tortoise Capital 1 1 1
Total 42 42 84
Appendix B: Results of Breusch-Pagan tests for Heteroscedasticity
Table B1. Breusch-Pagan tests ESG ETFs based on cross-sectional data
Bond ETFs
Model 1 Model 2 Model 3 Model 4 Model 5
BP 10.75 11.053 10.761 10.894 13.067*
p-value (0.1499) (0.1363) (0.1494) (0.1433) (0.0705)
Equity ETFs
Model 1 Model 2 Model 3 Model 4 Model 5
BP 42.417*** 39.774*** 41.378*** 39.936*** 39.305***
p-value (0.000) (0.000) (0.000) (0.000) (0.000)
* Indicates significance at 10%. ** Indicates significance at 5%. *** Indicates significance at 1%.
We reject the null hypothesis and conclude that all regression models for Equity ETFs and Model 5 for Bond ETFs violate the homoscedasticity assumption. Therefore, for these models we apply robust standard error to obtain unbiased standard errors of OLS coefficients under heteroscedasticity.
Contribution of the authors: the authors contributed equally to this article. The authors declare no conflicts of interests.
The article was submitted 06.01.2024; approved after reviewing 08.02.2024; accepted for publication 29.02.2024.
