Next Article in Journal
On Fintech and Financial Inclusion: Evidence from Qatar
Previous Article in Journal
Accounting Manipulation and Value Creation: An Empirical Study of EVA and Accounting Quality in NYSE and NASDAQ Companies
Previous Article in Special Issue
Navigating the Green Frontier: Dynamic Risk and Return Transmission Between Clean Energy ETFs and ESG Indexes in Emerging Markets
 
 
Font Type:
Arial Georgia Verdana
Font Size:
Aa Aa Aa
Line Spacing:
Column Width:
Background:
Article

Forecast Bias in Analysts’ Initial Coverage: The Influence of Firm ESG Disclosures

by
Mohammadali Fallah
1,
Sulei Han
2 and
Le Zhao
1,*
1
Department of Finance, Real Estate and Business Law, Craig School of Business, California State University, Fresno, CA 93720, USA
2
Sykes College of Business, The University of Tampa, Tampa, FL 33606, USA
*
Author to whom correspondence should be addressed.
J. Risk Financial Manag. 2025, 18(10), 585; https://doi.org/10.3390/jrfm18100585
Submission received: 15 August 2025 / Revised: 8 October 2025 / Accepted: 14 October 2025 / Published: 15 October 2025

Abstract

This study examines analyst forecast bias during the initial coverage of a firm using a sample of 631,660 firm-quarter analyst forecasts from 2007 to 2022. Estimating OLS regressions with firm, analyst, and time fixed effects, we find that analysts’ initial EPS forecasts are closer to the consensus than ongoing coverage forecasts. Our results suggest that ESG considerations influence analysts’ initial assessments of the firm. Higher ESG disclosure scores attenuate this tendency to issue consensus-aligned forecasts, particularly for analysts with a more favorable assessment of the firms than the consensus. We observe this effect when EPS forecast dispersion is high, indicating that ESG disclosures influence analysts’ initial assessments when uncertainty and disagreement among analysts are high. Our findings are robust to restricting the sample to small brokerage firms where analyst coverage assignments are more likely to be exogenous. We also find that analysts issue more optimistic price target estimates for firms with higher ESG disclosure scores.

1. Introduction

Analysts play a crucial role in capital markets by reducing information asymmetry between firm management and market participants, which can improve market efficiency and facilitate optimal capital allocation. However, several studies have documented that analyst forecasts are subject to cognitive limitations and behavioral biases.1 Moreover, career incentives and institutional pressures can compromise analysts’ objectivity and affect their recommendations.
When analysts first start covering a firm, they often lack established relationships with management and have limited firm-specific expertise, potentially making them more susceptible to behavioral biases. Analysts initiating coverage must balance multiple competing objectives such as establishing credibility with investors, building relationships with corporate management, and preserving professional reputation. Understanding analysts’ behavior at coverage initiation is particularly important, given that first impressions formed during this period can lead to lasting biases in analysts’ subsequent forecasting decisions (Hirshleifer et al., 2021).
Beyond cognitive limitations and behavioral biases, career-related incentives may lead analysts to strategically issue biased forecasts. Ke and Yu (2006) document that biased forecasts serve as a strategic tool for analysts to please corporate management in order to gain access to private information, which in turn increases the accuracy of analysts’ future forecasts and affects their job security. Malmendier and Shanthikumar (2014) document that analysts employ a strategic “two-tongues” behavior to balance conflicting incentives. Specifically, they demonstrate that analysts issue overly optimistic recommendations to attract retail investors and gain favor with corporate management while providing conservative earnings forecasts to maintain credibility with institutional investors. The incentive to avoid adverse career outcomes may also discourage analysts from deviating from consensus forecasts (Hong et al., 2000; Clement & Tse, 2005).
The evolving importance of Environmental, Social, and Governance (ESG) considerations in investment decisions adds a new factor that may affect analysts’ decision making. If ESG factors materially impact firm value, analysts should consider them in their forecasts. Derrien et al. (2022), Roger (2024), and Bancel et al. (2025) provide evidence that analysts increasingly incorporate ESG scores and related news into their valuation models. Despite this growing evidence, little is known about how ESG information affects analysts’ initial assessments of the firm when they lack firm-specific expertise and established relationships with management. This distinction matters since initiating analysts face informational disadvantage relative to continuing analysts and may rely on ESG information differently in their initial assessments. Our study addresses this gap by investigating whether ESG disclosures influence analysts’ assessments at coverage initiation.
We first examine whether analysts exhibit systematic differences in their forecasts when initiating coverage compared to ongoing coverage. We find that analysts issue more conservative EPS forecasts during initial coverage, with these forecasts deviating less from the consensus than the average ongoing coverage forecast in our sample. In contrast, analysts issue price target estimates that deviate more from consensus forecasts when initiating coverage. This pattern of conservative earnings forecasts along with optimistic price targets is consistent with the strategic “two-tongues” behavior documented by Malmendier and Shanthikumar (2014), who show that analysts’ misalignment of incentives makes them tailor their communications to different audiences. While optimistic recommendations and price targets can stimulate individual investors’ trading activities, institutional investors pay closer attention to the accuracy of earnings forecasts and discount overly optimistic recommendations. Corporate management reinforces this pattern by favoring positive stock recommendations but guiding analysts toward conservative earnings forecasts that are easier to meet.
We next investigate whether ESG disclosures influence analysts’ assessments during the initial coverage period. We combine the I/B/E/S unadjusted detailed history dataset with Bloomberg ESG disclosure scores for the period of 2007 to 2022. Our results indicate that, when analysts initiate coverage of firms with higher ESG scores, they systematically issue more optimistic earnings forecasts. Specifically, we find that, while analysts’ earnings forecasts exhibit lower absolute bias during initial coverage, this conservative tendency reverses for firms with high ESG scores. These results are primarily driven by analysts whose forecasts are more optimistic than consensus (the positive EPS bias subsample), indicating that the effect of ESG disclosures is more significant when analysts have a relatively more favorable assessment of the firm compared to the consensus view. Furthermore, we find that this effect is more pronounced when EPS forecast dispersion is higher, suggesting that ESG disclosures have a greater influence on shaping analysts’ initial assessments when there is higher uncertainty or disagreement among analysts about the firm’s earnings prospects.
It is possible that analysts’ decisions to initiate coverage of a firm might be correlated with unobserved factors that affect their forecast bias. To address this potential endogeneity issue, we restrict our sample to small brokerage firms and re-estimate our regressions. Analysts in small brokerage firms have limited autonomy in their coverage decisions, thereby mitigating concerns about selection bias (Jannati et al., 2025). Our findings remain robust when limiting the sample to smaller brokerage firms. Additionally, we employ propensity score matching (PSM) to account for potential unobservable confounding variables that could influence the relationship between coverage initiation and forecast bias. We match treatment observations (forecasts for newly covered firms) with control observations (forecasts for firms with ongoing coverage) based on propensity scores derived from a logistic regression model that incorporates all control variables. Our results remain robust across different matching specifications.
We also examine the effect of ESG disclosures on target price estimates, as we expect target prices to incorporate longer-term considerations where ESG factors should have greater influence. Our results show that, while analysts’ target price estimates have higher absolute bias during initial coverage, this optimism becomes even more pronounced for firms with higher ESG scores. The impact of ESG disclosure appears to be more significant for analysts who are optimistic relative to consensus. Collectively, these results highlight the important role of ESG disclosure in shaping analyst behavior during initial coverage. Stronger ESG disclosure appears to reduce uncertainty for analysts initiating coverage, particularly those with favorable views of the firm.
Our study makes several contributions to prior literature. First, we add to the literature on analyst forecast herding by providing evidence that this behavior occurs during the initial coverage, in which analysts seek to establish credibility with the firm’s investors but face an informational disadvantage relative to continuing analysts. While prior studies have identified analysts’ tendency to herd toward consensus forecasts as a mechanism to avoid reputational damage and adverse career outcomes (Hong et al., 2000; Clement & Tse, 2005; Huang et al., 2017), our results indicate that the herding toward consensus forecasts is amplified at coverage initiation as we compare initiating coverage with continuing coverage in our estimations.
Second, we find that the “two-tongues” behavior identified by Malmendier and Shanthikumar (2014) is more salient during initial coverage. Initiating analysts issue significantly bolder price targets (further from consensus) while providing more conservative earnings forecasts (closer to consensus), consistent with analysts employing biased forecasts strategically to balance multiple objectives, including access to management, attracting small investors, and preserving credibility with sophisticated market participants, as documented in prior research (Ke & Yu, 2006; Malmendier & Shanthikumar, 2014).
Most importantly, we extend the growing literature on how analysts integrate ESG factors into their assessments by demonstrating that ESG disclosures significantly influence analyst forecast bias during coverage initiation. Our findings suggest that ESG disclosures reduce information asymmetry for initiating analysts, leading to more deviation from consensus forecasts, particularly when there is greater disagreement among analysts. Given that forecasts deviating from consensus are, on average, more informative (Clement & Tse, 2005), our results indicate that more comprehensive ESG disclosures enhance forecast informativeness. This interpretation aligns with Boulton (2024), who finds cross-country evidence of ESG disclosure mandates reducing information asymmetry, and Wang and Xing (2025), who document that ESG information disclosure improves stock price informativeness in the Chinese stock market.
The remainder of the paper proceeds as follows. Section 2 develops our hypotheses by integrating insights from behavioral finance and ESG literature. Section 3 describes our empirical design, including data sources, variable construction, regression specifications, and summary statistics. Section 4 presents our main results and robustness tests addressing endogeneity concerns and heterogeneity across different levels of forecast dispersion. Section 5 provides additional evidence from price target estimates. Section 6 contains our conclusions.

2. Hypothesis Development

This study examines the behavior of financial analysts when initiating coverage on firms, focusing on their earnings per share (EPS) forecasts. Analysts’ earnings forecasts are known to significantly influence investor decision making, yet they are subject to biases driven by both cognitive limitations and institutional incentives.
For example, studies have shown that analysts frequently rely on cognitive shortcuts such as anchoring, herding, and representativeness heuristics, especially in situations involving market uncertainty or decision fatigue (Jegadeesh & Kim, 2010; Hirshleifer et al., 2019; F. Li et al., 2021).
Beyond the above well-documented behavioral mechanisms, three additional behavioral forces are particularly relevant in financial analyst forecasts. Reputational bias can lead analysts to avoid issuing contrarian or pessimistic forecasts out of concern for career consequences or strained client relationships, resulting in systematically optimistic recommendations and a reluctance to downgrade (Lim, 2001; Hong et al., 2000). First-impression anchoring further contributes to inertia: initial assessments at coverage initiation (often favorable) become a reference point that disproportionately influences subsequent forecasts and recommendation revisions, yielding gradual adjustments even when fundamentals shift (McNichols & O’Brien, 1997; Demiroglu & Ryngaert, 2010). Finally, cognitive asymmetry in information processing leads analysts to respond unevenly to good versus bad news. Analysts tend to overreact to positive signals while underweighting negative information, amplifying the optimistic skew observed in consensus forecasts (Easterwood & Nutt, 1999; Hilary & Menzly, 2006). Taken together, these behavioral biases provide a foundation in behavioral finance for why analyst outputs exhibit optimistic levels, delayed incorporation of adverse signals, and herding around prevailing views.
In addition, analysts may also issue biased forecasts due to institutional incentives to gain favor with management or advance their professional reputation (Ke & Yu, 2006; Hong & Kubik, 2003). Company management typically has access to more detailed information about the firm’s value than external analysts. This difference in information leads analysts to manage their interactions with management to obtain additional non-public data. As a result, analysts may sometimes provide optimistic forecasts to maintain good relations and ensure future information sharing.
When initiating coverage, analysts have an informational disadvantage compared to analysts already covering the same firm. To manage the reputational risks associated with their initial forecasts of a firm, they may prefer to issue estimates that closely align with the existing consensus. This conformity behavior arises from career-related incentives. For example, Hong et al. (2000) demonstrate that inexperienced analysts are more likely to conform to consensus forecasts to avoid reputational penalties. Clement and Tse (2005) suggest that analysts face a trade-off: while herding toward consensus reduces the risk of adverse career outcomes, bold forecasts that deviate from consensus are both more informative and more accurate. Similarly, Huang et al. (2017) show that, when analysts are more concerned about their reputations, they may prioritize conformity over informativeness, issuing more consensus-aligned forecasts. Therefore, we conjecture that, at the initiation of coverage, when credibility is being established with investors following a specific firm, analysts conform with the consensus to avoid reputational harm.
Additionally, analysts are also susceptible to cognitive pressures. Cai et al. (2024) show that analysts are susceptible to confirmation bias, giving disproportionate weight to prevailing market sentiment, especially when peer forecast dispersion is low. Furthermore, Malmendier and Shanthikumar (2014) offer evidence of analysts’ “two-tongues” strategy, where analysts may issue optimistic public recommendations to attract retail investors while maintaining more cautious, consensus-aligned EPS forecasts to preserve credibility with institutional investors.
Furthermore, accounting conservatism may influence analyst forecasting behaviors. Prior research shows that conservative accounting practices shape the way analysts interpret and forecast earnings. For instance, Sohn (2012) finds that financial analysts may incorporate conservatism into their forecasts, experiencing greater difficulty predicting earnings accurately for firms with less conservative accounting practices. Similarly, Helbok and Walker (2004) find that analysts tend to anchor their initial forecasts on permanent earnings and revise them later to incorporate temporary effects arising from conservative reporting.
Collectively, these arguments lead to our first hypothesis regarding analyst forecasts when initiating coverage of a firm, namely that the magnitude of deviation in analyst EPS forecasts from the consensus tends to be smaller, as analysts seek to reduce the risk of standing out under uncertainty.
H1: 
Analysts’ EPS forecasts tend to be more moderate and closer to consensus when they initiate coverage of a firm.
Building on this, in the second hypothesis, we investigate whether a firm’s environmental, social, and governance (ESG) disclosures affect analysts’ forecast behavior when they initiate coverage of a firm. A growing body of literature demonstrates that ESG performance affects firm valuation and investor perception. ESG commitment can enhance trust, reduce litigation risk, and improve capital allocation. For instance, Dhaliwal et al. (2012) find that ESG commitment enhances investor trust and improves capital allocation by signaling responsible and sustainable business practices. Serafeim and Yoon (2023) show that high ESG ratings signal lower ESG-related risk, which influences investor expectations and stabilizes stock prices. In addition, H. Zhang et al. (2024) suggest that better ESG performance is significantly associated with reduced firm litigation risk.
ESG performance is also linked to lower cost of capital, improved valuation accuracy, and reduced market short-termism. For instance, Eichholtz et al. (2019) show that firms with higher environmental performance typically benefit from lower costs of debt, especially noticeable in environmentally certified real estate and property companies’ corporate debt. Wong et al. (2021) find that the ESG-certified firms in emerging markets experience significant financial benefits with reduced cost of capital and an increased Tobin’s Q. Bofinger et al. (2022) further confirm that improvements in ESG profiles significantly influence firm valuation by reducing misvaluation. Specifically, better ESG performance is associated with correcting undervaluation and, notably, expanding existing overvaluation. Additionally, del Río et al. (2023) find that firms with persistent enforcement of sustainability reporting practices experience less market myopia, a prevalent behavioral bias in which investors overly emphasize short-term earnings at the expense of long-term performance. Such practice underscores the role of transparency in promoting efficient market pricing and encouraging longer-term strategic planning.
Analysts appear to respond to ESG-related information in various ways. ESG disclosures have been shown to improve forecast accuracy by reducing informational asymmetry and operational uncertainty (Luo & Wu, 2022; Acheampong & Elshandidy, 2025). Park et al. (2025) and Bancel et al. (2025) suggest that analysts incorporate ESG risks into valuation by adjusting discount rates, further supporting the relevance of ESG information in analysts’ models.
However, comprehensive ESG disclosures might also introduce biases during initial coverage. Hirshleifer et al. (2021) discuss first-impression bias, where initial observations overly influence later judgments. As such, strong ESG disclosures could create anchoring effects or foster unwarranted optimism, potentially offsetting analysts’ initial conservatism. If ESG disclosures shape perceived risk profiles, they may encourage analysts to issue more aggressive forecasts even during the typically cautious initiation phase. Thus, we hypothesize:
H2: 
Higher ESG disclosures mitigate analysts’ conservative EPS forecast bias during coverage initiation, especially in high-dispersion environments.

3. Empirical Design

3.1. Data Sources and Sample Description

We compile data on U.S. firms from four main databases: Institutional Brokers’ Estimate System (I/B/E/S), Bloomberg, Compustat, and the Center for Research in Security Prices (CRSP), covering the sample period from 2007 to 2022. Our sample period begins in 2007, when Bloomberg started systematically publishing ESG disclosure data for a broader set of firms.
We focus primarily on analysts’ EPS forecasts and incorporate price target forecasts as supplementary evidence. Both types of forecasts are obtained from the I/B/E/S Detail History file. For the EPS forecast sample, we follow Hirshleifer et al. (2021) and focus on one-quarter ahead earnings forecasts issued within 90 days prior to the actual earnings announcement for the corresponding firm and fiscal quarter. If an analyst issues multiple forecasts for the same firm and quarter, we retain only the first forecast within the 90-day window. For the price target forecast sample, we include only forecasts with a 12-month horizon and retain the first forecast issued by each analyst for a firm within a given calendar quarter.
Firms’ ESG scores are sourced from Bloomberg. Bloomberg’s ESG Disclosure Scores range from 0 to 100 and reflect the extent to which a firm discloses ESG-related information in a given year. A higher score indicates greater disclosure, which we interpret as a signal of a firm’s commitment to transparency and a factor potentially linked to specific corporate outcomes.
The ESG Disclosure Score is calculated from over 120 data points collected by Bloomberg from sources including annual reports, sustainability reports, and company websites. These data points are categorized into three pillars: Environmental, which captures the company’s environmental issues such as pollution, energy and water use, and waste management; Social, which covers employee treatment, diversity, health and safety, and community relations; and Governance, which includes board composition, executive compensation, shareholder rights, and business ethics. The final score is computed as a weighted generalized mean of the three pillars, with weights assigned based on Bloomberg Intelligence’s evaluation of financial materiality within industry peer groups to enhance the relevance of the measure for each company.2 The Bloomberg ESG score has served as a critical tool used by firms to manage and comply with evolving sustainability regulations.3 It has also been widely used in many academic studies (Demartis & Rogo, 2024; Buchanan et al., 2018; Christensen et al., 2022; Y. Li et al., 2018).
To construct control variables related to firm characteristics, we collect data from CRSP, restricting the sample to common stocks with a share code of 10 or 11, as well as from the Compustat quarterly files. After excluding observations with missing values for the required variables, the EPS forecast sample consists of 631,660 analyst-firm-quarter observations.

3.2. Variable Definitions and Methodology

Following Hirshleifer et al. (2021), we construct our main dependent variable, EPS Bias, which captures the relative optimism of an analyst’s forecast compared to the consensus forecast for the same firm and forecast period. We also create an absolute bias variable, Abs EPS Bias, which captures the absolute magnitude of the EPS forecast bias. Specifically, the two variables are defined as follows:
E P S   B i a s i , j , t = E P S i , j , t C o n s e n s u s   E P S j , t S t d .   D e v .   ( E P S j , t ) ,
A b s   E P S   B i a s i , j , t = E P S   B i a s i , j , t ,
where E P S i , j , t is the first earnings forecast issued by analyst i for firm j in fiscal quarter t , C o n s e n s u s   E P S j , t is the mean forecast calculated based on the first forecasts issued by all analysts covering firm j for the same fiscal quarter, and S t d .   D e v .   ( E P S j , t ) is the standard deviation of those forecasts.4
Our main independent variables include New Firm, ESG Score, and their interaction term. New Firm is a dummy variable that equals 1 if an analyst issues an EPS forecast or price target for a firm for the first time, provided the analyst has been listed in the I/B/E/S database for more than two calendar quarters. Observations where the analyst has been listed for two or fewer calendar quarters are excluded from the analysis. ESG Score is the Bloomberg ESG Disclosure Score from the calendar year preceding the fiscal quarter associated with an analyst’s EPS forecast.
We test Hypotheses H1 and H2 by estimating the Ordinary Least Squares (OLS) regression model (1) on the EPS forecast sample. We follow prior studies (Hirshleifer et al., 2019, 2021; Gao et al., 2025; Tang et al., 2025) in selecting additional control variables.5
E P S   B i a s i , j , t = β 0 + β 1 N e w   F i r m i , j , t + C o n t r o l s + F i r m   F E + A n a l y s t   F E + Y e a r   Q u a r t e r   F E + ε i , t .
In Equation (1), E P S   B i a s i , j , t represents analyst i ’s relative optimism in their EPS forecast for firm j in fiscal quarter t , compared to the consensus forecast. N e w   F i r m i , j , t is equal to 1 if the corresponding EPS forecast is the first ever issued by analyst i for firm j , and 0 otherwise.
The control variables include two sets. The forecast-related variables ln 1 + A n a l y s t   C o v , E P S   D i s p , ln 1 + F o r e c a s t   A g e , and l n 1 + A n a l y s t   E x p are measured based on the forecasts for fiscal quarter t . The remaining control variables, l n ( 1 + N o .   o f   F i r m s ) , T u r n o v e r , R e t u r n   V o l , l n ( 1 + M k t   C a p ) , R O A , l n ( 1 + B / M ) , and S a l e s   G r o w t h are lagged by one fiscal quarter. The variables expressed in logarithmic form represent the natural logarithm of one plus the corresponding raw values.
Analyst Cov is the number of analysts issuing EPS forecasts for a firm for a given fiscal quarter, while EPS Disp is the standard deviation of analysts’ forecasts for that firm, scaled by the mean forecast. Forecast Age is the number of days between the announcement date of an analyst’s EPS forecast for a firm for a given fiscal quarter and the corresponding earnings forecast date. Analyst Exp is the number of years from the first time the analyst appears in the I/B/E/S EPS forecast dataset to the announcement date of the EPS forecast under consideration. No. of Firms represents the number of firms an analyst has covered up to the previous calendar quarter. Turnover is the average turnover ratio over the prior 12 months. Return Vol is the standard deviation of a stock’s monthly returns over the prior 12 months. Mkt Cap, ROA, and B/M represent the firm’s market capitalization, return on assets, and book-to-market ratio, respectively, measured as of the previous fiscal quarter. Sales Growth represents the growth rate in a firm’s sales over the previous fiscal quarter. The detailed definitions of all variables are provided in Appendix A, Table A1.
In addition to using EPS Bias as our dependent variable, we also employ its absolute value, Abs EPS Bias, to examine how analysts’ biases during their initial coverage differ from those of their peers. Our hypothesis H1 posits that analysts’ EPS forecasts are closer to the consensus when they initiate coverage of a firm. Specifically, if initiating analysts exhibit a positive (negative) bias, the magnitude of their bias should be smaller than that of other analysts who are also optimistic (pessimistic). In line with this hypothesis, we expect β 1 to be negative (positive) in the subsample of positively (negatively) biased forecasts when EPS Bias is the dependent variable. Additionally, we expect the coefficient β 1 to be negative when Abs EPS Bias is used as the dependent variable.
To test Hypothesis H2 regarding the impact of ESG disclosure scores on analysts’ forecast bias during initial coverage, we estimate the following regression model:
E P S   B i a s i , j , t = β 0 + β 1 N e w   F i r m i , j , t + β 2 E S G   S c o r e j , t 1 + β 3 N e w   F i r m i , j , t × E S G   S c o r e j , t 1 + C o n t r o l s + F i r m   F E + A n a l y s t   F E + Y e a r   Q u a r t e r   F E + ε i , t .  
In Equation (2), all variables are defined consistently with those in Equation (1). E S G   S c o r e j , t 1 is the Bloomberg ESG score for firm j in the calendar year preceding the fiscal quarter t . To assess how ESG disclosure scores influence the magnitude of forecast bias, we also estimate these models using the absolute bias measure, Abs EPS Bias, as the dependent variable.
Our main variable of interest is the interaction term N e w   F i r m i , j , t × E S G   S c o r e j , t 1 , which measures whether ESG transparency mitigates forecast bias. By combining the initiation indicator with ESG disclosure scores, we isolate how ESG transparency alters analysts’ incentives to herd toward consensus when they lack private information channels with management at coverage initiation. To our knowledge, this is the first study to examine whether ESG transparency influences forecast bias precisely when analysts establish their first impressions of a firm. Our hypothesis posits that higher ESG scores mitigate analysts’ conservative bias when initiating coverage of a firm. If this hypothesis holds, we expect the coefficient β 3 in Equation (2) to have the opposite sign of β 1 .

3.3. Summary Statistics

Table 1 presents the summary statistics for all variables in the EPS forecast sample. As constructed, EPS Bias has a mean value of 0. Its 75th percentile value is 0.63, and its standard deviation is approximately 0.91, which aligns with the findings reported in Hirshleifer et al. (2021). The mean value of Abs EPS Bias is 0.74, with a standard deviation of 0.55, which reflects the average magnitude of analysts’ forecast bias.
Additionally, the mean value of New Firm is 0.048, indicating that 4.8% of the 631,660 forecasts with a non-missing value correspond to analysts covering a firm for the first time. The mean ESG Score is 28.24, with a standard deviation of 16.27, reflecting considerable variation across firms and years. On average, each firm is covered by 13 analysts with an average of 12 years of experience, and these analysts have previously covered 39 different firms before issuing the EPS forecast in question. The average dispersion of forecasts pertaining to the same firm and quarter is 0.28, and forecasts are generally issued approximately 67 days before the actual earnings announcement, consistent with prior evidence (e.g., Diether et al., 2002; Hirshleifer et al., 2021). The summary statistics on firm characteristics, including share turnover, return volatility, market capitalization, ROA, and sales growth, also reveal substantial variation across the sample.

4. Results

4.1. Analyst Initial Forecast Bias

To examine whether the analysts exhibit any forecast bias in their initial coverage of a firm and to test our first hypothesis, we estimate the OLS regression model presented in Equation (1). Table 2 presents the estimation results, with EPS Bias as the dependent variable in Columns (1) through (6) and Abs EPS Bias in Columns (7) and (8). Our primary variable of interest is New Firm, an indicator variable equal to 1 if an analyst issues a forecast for a firm for the first time, and 0 otherwise.
The positive and statistically significant coefficients on New Firm in Columns (1) and (2) appear to suggest that initiating analysts exhibit greater forecast bias. However, partitioning the sample based on the direction of forecast deviation reveals a different pattern. Columns (3) and (4) restrict the analysis to observations with positive bias (optimistic forecasts), while columns (5) and (6) focus on negative bias (pessimistic forecasts).
In the positive-bias subsample (EPS Bias > 0), the coefficients on New Firm are negative and significant, indicating that, among optimistic forecasts, initial coverages are relatively more conservative and closer to consensus. Specifically, since forecast bias is defined as the difference between an analyst’s forecast and the consensus, scaled by the standard deviation of all forecasts, the coefficient of 0.014 on New Firm in Column (3) indicates that, on average, forecasts from optimistic initiating analysts are 0.014 standard deviations lower and therefore closer to the consensus forecast compared to other optimistic forecasts.
In the negative-bias subsample (EPS Bias < 0), where the dependent variable takes only negative values, the significantly positive coefficients on New Firm suggest that initial coverage forecasts are less pessimistic and closer to consensus relative to other forecasts. Specifically, the coefficient of 0.022 on New Firm in Column (5) implies that forecasts from pessimistic initiating analysts are 0.022 standard deviations higher and therefore closer to the consensus. The opposite signs on New Firm between the positive- and negative-bias subsamples indicate that analysts initiating coverage are not systematically more optimistic or pessimistic. Rather, they tend to herd toward the consensus forecast.
The results for absolute EPS forecast bias in Columns (7) and (8) are consistent with the above interpretation. The estimated coefficient on New Firm is negative and statistically significant at the 1% level, indicating that analysts initiating coverage issue EPS forecasts with smaller forecast bias magnitudes than forecasts from continuing analysts in our sample. These initial EPS forecasts tend to be closer to the consensus, which supports our Hypothesis H1. This result suggests that initiating analysts deviate less from consensus to avoid reputational harm at the time they are building credibility with investors following the firm but have an informational disadvantage compared with their incumbent peers, consistent with Hong et al. (2000) and Huang et al. (2017). In addition, by issuing earnings forecasts that are not overly pessimistic but are also easier for the firm to meet, analysts position themselves to gain access to company management, which aligns with findings of Ke and Yu (2006), Hong and Kubik (2003) and Malmendier and Shanthikumar (2014).

4.2. ESG Score and Analyst Initial Forecast Bias

To evaluate the second hypothesis regarding the moderating effect of ESG disclosure scores on analysts’ initial forecast bias, Figure 1 presents the absolute forecast bias of initiating versus non-initiating analysts across ESG score terciles. The results show that (1) the absolute magnitude of forecast bias for initiating analysts is smaller than that for non-initiating analysts, consistent with Hypothesis H1, and (2) the overall magnitude of forecast bias among initiating analysts increases with ESG scores. More importantly, the difference between initiating and non-initiating forecasts narrows as ESG scores rise, decreasing from 0.025 in the low-ESG tercile to 0.004 in the high-ESG tercile. These results suggest that, although initiating analysts generally issue forecasts closer to the consensus, this tendency weakens as ESG scores increase, consistent with Hypothesis H2.
We next estimate the regression model specified in Equation (2) to incorporate additional control variables. Specifically, we include New Firm, ESG Score, and an interaction term New Firm × ESG Score to examine whether ESG scores influence analysts’ conservatism during the initial coverage observed in Table 2. In these regressions, ESG scores are rescaled by 100 so that the coefficients capture meaningful effects on forecast bias.
Table 3 reports the results. The coefficients on the interaction term New Firm × ESG Score are positive and marginally significant for the entire sample. And it remains consistently positive and significant when we include firm fixed effects (Columns (2), (4), (6), and (8)), indicating that our findings are robust to alternative specifications controlling for unobserved firm heterogeneity. In Column (2), the coefficient of 0.074 on the interaction term implies that a one-standard-deviation increase in ESG Score is associated with a 0.013-standard-deviation additional increase in forecast bias for initiating analysts relative to non-initiating analysts. The effects are more pronounced in the positive-bias subsample (EPS Bias > 0), where the interaction coefficients are positive and highly significant at the 1% level. In Column (4), the coefficient of 0.105 suggests that, among analysts with optimistic forecasts, a one-standard-deviation increase in ESG Score corresponds to a 0.019-standard-deviation additional increase in forecast bias for initiating analysts relative to their non-initiating counterparts.6 These findings suggest that high ESG scores substantially offset analysts’ inclination toward conservative forecasting during the initiation of coverage. Higher-quality ESG disclosures are associated with lower information asymmetry and uncertainty (Boulton, 2024), thereby boosting analysts’ confidence in issuing bolder forecasts, despite the absence of established information channels.
However, the coefficients of the interaction term for the negative-bias subsample (Columns (5) and (6)) are negative but insignificant. This result implies that ESG disclosures do not meaningfully affect analysts’ behavior when they hold pessimistic views of the firm compared to the consensus view.
In Columns (7) and (8), where absolute EPS forecast bias is the dependent variable, the results are similar to those for the positive-bias subsample. The significant positive coefficients on the interaction terms indicate that higher ESG scores amplify the magnitude of deviation from consensus during initial coverage, driven primarily by the reduction in herding and conservatism among optimistic analysts. Taken together, these results support Hypothesis H2, demonstrating that ESG disclosures mitigate analysts’ conservative forecasting behavior when initiating coverage. Rather than uniformly affecting all analysts, ESG information appears to provide confidence specifically to those with positive fundamental views, enabling them to deviate from the cautious approach associated with coverage initiation.7
Additionally, Table 3 shows that ESG disclosure scores do not have a significant effect on analyst EPS forecast bias when analysts are not initiating coverage of a firm, suggesting that ESG scores exert limited influence on their forecasts in such cases. We conjecture that, for analysts with established coverage relationships, ESG information appears to be already incorporated into their forecasting models or considered less relevant than their existing information channels.

4.3. Robustness Tests

The findings in Section 4.1 and Section 4.2 are consistent with our hypotheses. To further evaluate the robustness of these results, we conduct four additional analyses: (1) restricting the sample to forecasts from analysts in small brokerage firms (Section 4.3.1), applying propensity score matching (Section 4.3.2), (3) performing a placebo test using future forecast bias as the dependent variable (Section 4.3.3), and (4) comparing forecasts for firms with high versus low earnings forecast dispersion (Section 4.3.4). The results from these robustness checks lend further support to our hypotheses.

4.3.1. Addressing the Endogeneity Concerns

One potential endogeneity concern is that analysts’ decisions to initiate coverage of new firms may be influenced by unobserved factors that are also related to their forecast biases. To address this issue, we re-estimate the regressions using only forecasts issued by analysts at small brokerage firms. Following Jannati et al. (2025), we argue that analysts at small brokerage firms have less discretion in choosing which companies to cover and are more likely to be assigned new firms by their employers, thereby reducing the concern of self-selection bias. A brokerage firm is classified as small if it has fewer analysts than the sample median in each quarter. Table 4 presents the results for this subsample, which closely align with the main findings reported in Table 3.
The results again show that the coefficients on New Firm are significantly negative in the positive-bias subsample (Columns (3) and (4)) and significantly positive in the negative-bias subsample (Columns (5) and (6)). Consistently, the results for absolute EPS forecast bias in Columns (7) and (8) also show significantly negative coefficients on New Firm, indicating a smaller magnitude of forecast bias when analysts initiate coverage. In particular, Column (8) shows that the coefficient of –0.036 on New Firm implies that forecasts from initiating analysts at small brokerage firms are 0.036 standard deviations closer to the consensus.8
Meanwhile, the coefficients on the interaction term New Firm × ESG Score exhibit the opposite pattern, suggesting that higher ESG scores encourage analysts to issue bolder forecasts during coverage initiation. In particular, the coefficient of 0.071 in Column (8) implies that a one-standard-deviation increase in ESG Score is associated with a 0.013-standard-deviation increase in initial forecast bias for initiating analysts compared with non-initiating analysts. Overall, these findings provide further support for Hypotheses H1 and H2, suggesting that, while analysts’ EPS forecasts tend to herd towards consensus during the initial coverage period, this tendency is weakened when they cover firms with high ESG scores.

4.3.2. Propensity Score Matching (PSM) Analysis

We also perform a propensity score matching (PSM) analysis to control for unobservable confounding factors. This analysis helps address potential endogeneity concerns by ensuring that differences in forecast bias are not driven by underlying firm or analyst characteristics that influence coverage decisions. Each quarter, treatment forecasts (New Firm = 1) are matched with comparable control forecasts (New Firm = 0) using propensity scores estimated from a logistic regression that includes all control variables as covariates. We then estimate the regression model specified in Equation (2) using the matched samples and present the results in Table 5.
Across matching specifications from 1-to-1 to 1-to-3 nearest neighbors, the findings remain consistent with the main results reported in Table 3. Panels A and B present the estimates from 1-to-1 and 1-to-3 nearest-neighbor matching, respectively. Similar to our previous results, the coefficients on New Firm and its interaction with ESG Score differ between the positive-bias and negative-bias subsamples.
In Column (2), the negative coefficient on New Firm and the positive coefficient on New Firm × ESG Score indicate that optimistic initiating analysts tend to issue forecasts that deviate less from the consensus (Hypothesis H1), but this tendency is attenuated when the covered firms have higher ESG scores (Hypothesis H2). In Column (3), the positive coefficient on New Firm and the negative coefficient on New Firm × ESG Score suggest that pessimistic initiating analysts issue forecasts that are less negative and closer to the consensus. However, higher ESG scores do not appear to significantly influence initial assessments of analysts with more pessimistic opinions relative to the consensus. Finally, when the two subsamples are combined, the results in Column (4) show that forecasts from initiating analysts are 0.055 standard deviations closer to the consensus, but this effect is attenuated when ESG scores are higher.
These results provide further support for Hypotheses H1 and H2, suggesting that, while initiating analysts tend to be conservative in their initial forecasts, they become more willing to issue forecasts that deviate further from the consensus when covering firms with higher ESG scores, particularly when they hold optimistic views.

4.3.3. Placebo Test

To further ensure that our results are not spurious, we conduct a placebo outcome test using the lead value of earnings forecast bias from the same analyst for the same firm as the dependent variable. Specifically, we replace the dependent variable in Equation (2) with E P S   B i a s i , j , t + 1 and A b s   E P S   B i a s i , j , t + 1 and replicate the analysis in Table 3. The corresponding results are presented in Table 6.
Across all columns (1)–(8), the coefficients on our main variables, New Firm and New Firm × ESG Score, are statistically insignificant. This absence of significance in the placebo tests confirms that the results in Table 3 are not driven by spurious associations but instead reflect initiating analysts’ tendency to herd toward consensus, which becomes less pronounced as ESG scores increase.

4.3.4. Heterogeneity Analysis

Our findings thus far indicate that higher ESG disclosure scores enhance initiating analysts’ confidence in issuing forecasts that deviate more from the consensus. We posit that these disclosures provide valuable information to analysts who lack established relationships with firm management, which can reduce their informational disadvantage relative to continuing analysts during the initiation of coverage. Building on this logic, the impact of ESG disclosures should be more pronounced for firms with greater earnings uncertainty or information asymmetry. Such firms tend to exhibit higher dispersion in analysts’ earnings forecasts, and prior research has widely used earnings forecast dispersion as a proxy for market disagreement (e.g., Diether et al., 2002) and information uncertainty (e.g., Barron & Stuerke, 1998; X. F. Zhang, 2006). Therefore, the information embedded in ESG scores is likely to be more valuable in firms with higher earnings forecast dispersion (Barron & Stuerke, 1998; Diether et al., 2002; X. F. Zhang, 2006). Accordingly, in this subsection, we examine whether the impact of ESG disclosure scores on analyst initial forecast bias varies with earnings forecast dispersion. Each quarter, we sort firms into tercile portfolios—High, Medium, and Low—based on their earnings forecast dispersion and estimate the regression model specified in Equation (2) for each group. Table 7 presents the results for the high- and low-dispersion subsamples separately.
We find that the effect of ESG disclosure scores on analysts’ initial EPS forecast bias is indeed more pronounced among high-dispersion firms. As reported in Columns (1) and (2), in the full sample, the coefficient on the interaction term New Firm × ESG Score is positive and statistically significant for high-dispersion firms but insignificant for low-dispersion firms. This suggests that higher ESG scores have a significant positive impact on analysts’ initial forecast bias only when the firms are associated with high information asymmetry and uncertainty.
Meanwhile, in the positive-bias subsample (Columns (3) and (4)), the corresponding coefficient is positive and statistically significant for both low- and high-dispersion firms but is larger in magnitude and more statistically significant for high-dispersion firms. In contrast, in the negative-bias subsample (Columns (5) and (6)), the coefficient is insignificant for both high- and low-dispersion firms. This pattern aligns with the overall findings reported in Table 3 and suggests that ESG disclosure scores have a limited impact on EPS forecast bias when analysts hold negative views about a firm’s earnings.
Consistently, the amplifying effect of ESG disclosure scores on initiating analysts’ absolute EPS forecast bias appears only among high-dispersion firms, as shown in Columns (7) and (8), and is primarily driven by the positive-bias subsample. On average, forecasts from initiating analysts are 0.050 standard deviations closer to the consensus for high-dispersion firms, compared with only 0.030 standard deviations for low-dispersion firms. Moreover, ESG scores have no significant impact on analysts’ initial forecasts for low-dispersion firms, but they attenuate the herding tendency for high-dispersion firms, as evidenced by a statistically significant interaction term at the 1% level. Overall, these results further support the view that higher ESG disclosure scores increase analysts’ confidence in issuing more optimistic forecasts.

5. Additional Analyses: Analysts’ Price Target Forecasts

While our main analyses focus on EPS forecasts, analysts’ price target (PTG) forecasts also offer valuable insights into analyst forecasting behavior during coverage initiation. PTG forecasts are inherently forward-looking valuations that incorporate longer-horizon inputs such as multiples, PEG ratios, and P/E components reflecting risk and growth prospects (Da et al., 2016; Bradshaw, 2002). Analysts also employ broader modeling approaches, including cash-flow forecasts, to improve PTG accuracy and integrate non-earnings information into valuations (Hashim & Strong, 2018). Previous studies suggest that analysts are incentivized to issue more aggressive and distinctive price targets when initiating coverage to build reputational capital, demonstrate expertise, and stimulate investor engagement (Irvine, 2003; Demiroglu & Ryngaert, 2010; Ertimur et al., 2011; Malmendier & Shanthikumar, 2014).
Building on this literature, we posit that ESG disclosures may further amplify analysts’ optimistic forecast bias in PTG forecasts during initial coverage. Comprehensive ESG disclosures signal a firm’s strong commitment to ESG-related initiatives, enhancing its perceived attractiveness and lowering its perceived investment risk. Consequently, analysts may respond by issuing more positive and distinct price targets that align with the favorable market narrative and take advantage of increased investor interest.
To test this hypothesis, we construct two bias variables, PTG Bias and Abs PTG Bias, defined as follows:
P T G   B i a s i , j , t = P T G i , j , t C o n s e n s u s   P T G j , t S t d .   D e v .   ( P T G j , t ) ,
A b s   P T G   B i a s i , j , t = | P T G   B i a s i , j , t | ,
where P T G i , j , t is the first price target forecast issued by analyst i for firm j in the calendar quarter t , C o n s e n s u s   P T G j , t is the mean forecast calculated based on the first forecasts issued by all analysts covering firm j in the same calendar quarter, and S t d .   D e v .   ( P T G j , t ) is the standard deviation of those forecasts. The variable, Abs PTG Bias, captures the absolute magnitude of the price target forecast bias.
We then apply the following regression models, where PTG Bias serves as the dependent variable9:
P T G   B i a s i , j , t = β 0 + β 1 N e w   F i r m i , j , t + C o n t r o l s + F i r m   F E + A n a l y s t   F E + Y e a r   Q u a r t e r   F E + ε i , t .
P T G   B i a s i , j , t = β 0 + β 1 N e w   F i r m i , j , t + β 2 E S G   S c o r e j , t 1 + β 3 N e w   F i r m i , j , t × E S G   S c o r e j , t 1 + C o n t r o l s + F i r m   F E + A n a l y s t   F E + Y e a r   Q u a r t e r   F E + ε i , t .
In Equations (3) and (4), P T G   B i a s i , j , t represents analyst i ’s relative optimism in their price target forecast announced for firm j in calendar quarter t, compared to the consensus forecast. All other variables are defined in the same manner as in our main analysis. ESG Score is the Bloomberg ESG Disclosure Score from the calendar year preceding the calendar quarter associated with the analyst’s price target forecast. Forecast Age is excluded, as it is not applicable to the price target forecasts, which all pertain to a 12-month horizon.
Table 8 presents the regression results. Columns (1) through (6) report results using PTG Bias as the dependent variable. Specifically, Columns (1) and (2) use the full sample, Columns (3) and (4) focus on the positive bias subsample (PTG Bias > 0), and Columns (5) and (6) examine the negative bias subsample (PTG Bias < 0). Columns (7) and (8) report results using the absolute value of PTG bias (Abs PTG Bias) as the dependent variable.
The positive and significant coefficients of New Firm in Columns (1), (3), and (4), combined with the negative and significant coefficients in Columns (5) and (6), suggest that analysts’ initial PTG forecasts tend to be bolder and deviate more from the consensus forecast compared with ongoing coverage forecasts in our sample. This is reinforced by the positive coefficients on New Firm in Columns (7) and (8), indicating that the magnitude of forecast bias, regardless of direction, is greater at the time of initiation.
Importantly, the coefficients on the interaction term New Firm × ESG Score are positive and statistically significant in Columns (2) and (4), suggesting that higher ESG disclosure intensifies analysts’ optimistic bias when they initiate coverage. This effect is most evident in the full sample and the positive bias subsample, indicating that ESG-related signals may enhance perceived firm quality or reduce perceived risk, thereby encouraging analysts to issue more optimistic price targets. In contrast, the coefficient on the interaction term is not statistically significant in the negative bias subsample (Column (6)) or when the absolute bias is used as the dependent variable (Column (8)), indicating that ESG disclosure primarily affects the optimistic direction of forecast bias rather than its overall magnitude.
Collectively, these results highlight the important role of ESG disclosure in shaping analyst behavior during initial coverage. Analysts appear to interpret stronger ESG signals as indicative of firm strength or reduced downside risk, which in turn encourages them to provide more aggressive price targets when initiating coverage.
In summary, the additional analyses on price target forecasts provide complementary evidence to our main findings on EPS forecasts and underscore the importance of ESG disclosure in shaping analyst forecast behavior during coverage initiation. We find that analysts tend to issue EPS forecasts that are closer to the consensus and price target forecasts that deviate more from the consensus when initiating coverage of a firm. These findings are consistent with Malmendier and Shanthikumar (2014), who document that analysts strategically issue favorable stock recommendations to attract retail investors’ attention but conservative earnings forecasts to establish or maintain favorable relationships with management. This behavior is particularly important during initial coverage periods when analysts need to secure access to company information. Furthermore, our findings reveal that higher ESG disclosure scores play a significant role in influencing these forecasts: they mitigate conservative bias in EPS forecasts and amplify optimistic bias in price target forecasts.

6. Conclusions

We investigate analysts’ behavior when initiating coverage on firms and how ESG disclosures influence their forecasting decisions during this period. Using a comprehensive sample of analyst forecasts from 2007 to 2022, we find that analysts’ EPS forecasts are closer to consensus during the initial coverage period compared to an average ongoing coverage forecast in our sample. Issuing conservative EPS forecasts that are more achievable targets is consistent with analysts’ incentives to curry favor with corporate management and gain access to information.
Our study provides evidence that more comprehensive ESG disclosures mitigate analysts’ herding tendencies during coverage initiation. This result is driven by analysts who are more optimistic relative to consensus, suggesting that ESG disclosures provide meaningful information that reduces uncertainty during coverage initiation and leads to more optimistic forecasts. We also document that ESG disclosures play a more pronounced role in shaping analysts’ initial assessments of the firm when the EPS forecast dispersion is high, which suggests that these disclosures narrow the information gap between initiating and continuing analysts in the absence of private information channels with the covered firm.
Our findings are robust to restricting the sample to small brokerage firms, where analysts have limited autonomy in their firm coverage choices, mitigating concerns about selection bias. In addition to including analyst, firm, and time fixed effects in our regressions, we conduct a propensity score matching analysis to address a potential endogeneity concern that our results are driven by differences in firm or analyst characteristics between initiated and ongoing coverages. Re-estimating the baseline model on the matched samples yields consistent results across multiple specifications.
Our analysis of price target forecasts provides complementary evidence on the impact of ESG disclosures on analysts’ decision making during the initiation of coverage. In contrast to the conservative EPS forecasts, we document that price target estimates display the greater deviation from consensus. This bias becomes more amplified for firms with higher ESG disclosure scores, specifically among analysts who hold more favorable views about the firm compared to the consensus.
Taken together, conservative EPS forecasts coupled with optimistic price targets are consistent with the incentive structures documented by Ke and Yu (2006) and Malmendier and Shanthikumar (2014). Most importantly, our results underscore the value of comprehensive ESG disclosure in shaping analyst perceptions during the critical initial coverage period, when first impressions are formed and can have lasting effects on analysts’ future decisions (Hirshleifer et al., 2021). Our results indicate that ESG disclosures reduce the tendency of initiating analysts to herd toward consensus forecasts, thereby increasing the informativeness of their forecasts for capital market participants, as forecasts that deviate more from consensus tend to be more accurate and informative.
The findings of this study have important policy implications and contribute to ongoing regulatory discussions surrounding ESG disclosure in the United States. The SEC’s climate disclosure rules, first approved in 2024 but subsequently scaled back and now facing abandonment of legal defense in 202510, highlight ongoing debate over company requirements for reporting ESG-related risks. Our research provides timely evidence on this matter. We show that voluntary ESG disclosures already play an important role in the information environment, particularly by reducing information asymmetry for analysts during the critical coverage initiation period. Accordingly, regulatory actions to weaken, pause, or withdraw support for disclosure mandates may unintentionally remove a valuable tool that analysts rely on to assess firm fundamentals, potentially resulting in less informative forecasts and greater market uncertainty. Overall, our results suggest that the benefits of comprehensive ESG disclosure in fostering more informed and efficient capital markets should remain a central consideration for regulators.
Beyond regulatory debates, our findings also have practical implications for financial professionals. Since ESG disclosures help analysts reduce uncertainty and offset herding tendencies during coverage initiation, brokerage firms could enhance their research quality by incorporating ESG disclosure analysis into training programs for junior analysts. Such steps would help shape more independent and informative forecasts and strengthen the firm’s credibility with clients.
A limitation of our study is that we rely on aggregate ESG disclosure measures. Grewal et al. (2021) suggest that markets primarily respond to financially material ESG disclosures (as identified by SASB), which improve stock price informativeness, while non-material disclosures do not. Moreover, ESG ratings and disclosures are subject to significant disagreement across providers, which can weaken the predictive power of ESG information and mute market reactions to ESG news (Serafeim & Yoon, 2023). Therefore, future research could build on our findings by disentangling material from non-material disclosures and by considering how rating disagreement influences the extent to which ESG information affects analyst forecast behavior.
Beyond this limitation, several extensions could deepen our understanding of how ESG considerations affect analysts’ assessment of the firm. Future research could examine tone-based ESG metrics to capture the sentiment ESG disclosures. It is also important to explore how varying ESG regulatory frameworks and disclosure requirements across international markets shape analyst forecasting behavior.
Finally, while our research focuses on analyst behavior, future work could explore investor-level dynamics by examining trading reactions of different investor types to analyst forecasts conditional on firms’ ESG disclosures. Such studies would shed light on how investor interpretation shapes the market impact of analyst reports.

Author Contributions

Conceptualization, M.F., S.H. and L.Z.; methodology, M.F., S.H. and L.Z.; software, S.H.; validation, M.F. and L.Z.; formal analysis, M.F., S.H. and L.Z.; data curation, M.F., S.H. and L.Z.; writing—original draft preparation, M.F., S.H. and L.Z.; writing—review and editing, M.F., S.H. and L.Z.; project administration: L.Z. All authors have read and agreed to the published version of the manuscript.

Funding

This research received no external funding.

Institutional Review Board Statement

Not applicable.

Informed Consent Statement

Not applicable.

Data Availability Statement

The data presented in this study are available upon request from the corresponding author. The data are not publicly available due to subscription requirements and are subject to non-disclosure agreements.

Conflicts of Interest

The authors declare no conflict of interest.

Appendix A

Table A1. Variable Definitions and Data Sources.
Table A1. Variable Definitions and Data Sources.
Variable Definition Data Sources Reference
Dependent Variables
EPS BiasThe difference between an analyst’s EPS forecast for a firm for a given fiscal quarter and the mean EPS forecast for that firm, scaled by the standard deviation of all analysts’ EPS forecasts for that fiscal quarter. Forecasts issued more than 90 days before the earnings announcement are excluded. If an analyst issues multiple forecasts for a firm for the same fiscal quarter, only the first forecast issued within the 90-day window is retained.I/B/E/SHirshleifer et al. (2021)
Abs EPS BiasThe absolute difference between an analyst’s EPS forecast for a firm for a given fiscal quarter and the mean EPS forecast for that firm, scaled by the standard deviation of all analysts’ EPS forecasts for that fiscal quarter. Forecasts issued more than 90 days before the earnings announcement are excluded. If an analyst issues multiple forecasts for a firm for the same fiscal quarter, only the first forecast issued within the 90-day window is retained.I/B/E/SHirshleifer et al. (2021)
PTG BiasThe difference between an analyst’s price target forecast for a firm announced in a given calendar quarter and the mean price target forecast for that firm, scaled by the standard deviation of all analysts’ price target forecasts for the firm announced in the same quarter. If an analyst issues multiple forecasts for a firm in the same calendar quarter, only the first forecast is retained.I/B/E/SHirshleifer et al. (2021)
Abs PTG BiasThe absolute difference between an analyst’s price target forecast for a firm announced in a given calendar quarter and the mean price target forecast for that firm, scaled by the standard deviation of all analysts’ price target forecasts for the firm announced in the same quarter. If an analyst issues multiple forecasts for a firm in the same calendar quarter, only the first forecast is retained.I/B/E/SHirshleifer et al. (2021)
Independent Variables
New FirmA dummy variable that equals 1 if an analyst issues an EPS forecast or price target for a firm for the first time, provided the analyst has been listed in the I/B/E/S database for more than two calendar quarters. If the analyst has been listed in the I/B/E/S database for two or fewer calendar quarters, the value is coded as missing.I/B/E/S
ESG ScoreThe firm’s weighted generalized mean of environmental, social, and governance rating for year t on a scale of 0 to 100. The score for a given year t is defined as the ESG Disclosure Score from the calendar year preceding the fiscal quarter to which an analyst’s EPS forecast pertains, or the calendar quarter to which an analyst’s price target pertains.BloombergWong et al. (2021)
Analyst CovThe number of analysts issuing EPS forecasts for a firm for a given fiscal quarter (for the EPS sample), or the number of analysts issuing price target forecasts for a firm in a given calendar quarter (for the price target sample). EPS forecasts issued more than 90 days before the earnings announcement are excluded.I/B/E/SDiether et al. (2002); Hirshleifer et al. (2019); Gao et al. (2025)
EPS DispThe standard deviation of all analysts’ EPS forecasts for a firm for a given fiscal quarter, scaled by the mean forecast. Forecasts issued more than 90 days before the earnings announcement are excluded. If an analyst issues multiple forecasts for a firm for the same fiscal quarter, only the first forecast issued within the 90-day window is retained.I/B/E/SDiether et al. (2002); Tang et al. (2025)
PTG DispThe standard deviation of all analysts’ price target forecasts for a firm announced in a given calendar quarter, scaled by the mean forecast. If an analyst issues multiple forecasts for a firm in the same calendar quarter, only the first forecast is retained.I/B/E/SDiether et al. (2002); Tang et al. (2025)
Forecast AgeThe number of days between the announcement date of an analyst’s EPS forecast for a firm for a given fiscal quarter and the earnings announcement date for the corresponding quarter.I/B/E/SHirshleifer et al. (2019); Hirshleifer et al. (2021)
Analyst ExpThe number of years from the first time the analyst appears in the I/B/E/S EPS forecast database (for the EPS sample) or the price target database (for the price target sample) to the announcement date of the EPS or price target forecast under consideration.I/B/E/SHirshleifer et al. (2021)
No. of FirmsThe number of firms an analyst has covered up to the calendar quarter immediately preceding the announcement date of the EPS or price target forecast under consideration.I/B/E/SHirshleifer et al. (2019); Hirshleifer et al. (2021)
TurnoverThe average turnover ratio over the 12-month period ending one month before the start of the calendar quarter in which an analyst’s EPS forecast or price target forecast is announced. The turnover ratio for a stock in a given month is defined as the monthly trading volume divided by the number of shares outstanding.CRSPHirshleifer et al. (2019); Tang et al. (2025)
Return VolThe standard deviation of a stock’s monthly returns over the 12-month period ending one month before the start of the calendar quarter in which an analyst’s EPS forecast or price target forecast is announced.CRSPHirshleifer et al. (2019)
Mkt CapMarket capitalization of a firm measured in millions of dollars, calculated as the stock price (prccq) multiplied by the number of shares outstanding (cshoq), measured at the end of the fiscal quarter preceding the fiscal quarter to which the analyst’s EPS forecast pertains (for the EPS sample), or at least four months prior to the announcement of the price target forecast (for the price target sample), to ensure the information was available to analysts at the time of forecasting. If the data is unavailable in Compustat, market capitalization is calculated using CRSP data. All values are adjusted to 2022 U.S. dollars.CRSP; CompustatHirshleifer et al. (2019)
ROAThe Return on Assets (ROA) ratio, calculated as quarterly income before extraordinary items (ibq) divided by total assets (atq), measured at the end of the fiscal quarter preceding the fiscal quarter to which the analyst’s EPS forecast pertains (for the EPS sample), or at least four months prior to the announcement of the price target forecast (for the price target sample), to ensure the information was available to analysts at the time of forecasting.CompustatHirshleifer et al. (2019); Gao et al. (2025); Tang et al. (2025)
B/MThe Book-to-Market (BM) ratio of a firm, calculated as the book value of equity divided by market capitalization, measured at the end of the fiscal quarter preceding the fiscal quarter to which the analyst’s EPS forecast pertains (for the EPS sample), or at least four months prior to the announcement of the price target forecast (for the price target sample), to ensure the information was available to analysts at the time of forecasting.CompustatDiether et al. (2002)
Sales GrowthThe growth rate in a firm’s sales (salesq) over a fiscal quarter, measured at the end of the fiscal quarter preceding the fiscal quarter to which the analyst’s EPS forecast pertains (for the EPS sample), or at least four months prior to the announcement of the price target forecast (for the price target sample), to ensure the information was available to analysts at the time of forecasting.CompustatHirshleifer et al. (2019); Gao et al. (2025)

Notes

1
For instance, see Jegadeesh and Kim (2010); Hirshleifer et al. (2019) and F. Li et al. (2021) among others.
2
For a detailed methodology of the Bloomberg ESG Disclosure Score, including its components, data collection process, and weighting scheme, see the official documentation available on the Bloomberg Professional Services website at https://www.bloomberg.com/professional/products/data/enterprise-catalog/esg/#overview (accessed on 13 October 2025).
3
4
In our main tests, analyst forecast bias is measured relative to the mean forecast. As an untabulated robustness check, we also use the median forecast as the consensus, and the results remain highly consistent. These results are available upon request.
5
Most variables, with the exception of the ESG scores and the dummy variable New Firm, are winsorized at the 1st and 99th percentiles.
6
According to Table 1, the standard deviations of EPS Bias and the rescaled ESG Score (divided by 100) are 0.908 and 0.163, respectively. This implies that a one-standard-deviation increase in ESG Score is associated with a 0.019-standard-deviation increase in EPS Bias ( 0.105 × 0.163 / 0.908 = 0.019 ).
7
To alleviate concerns about potential multicollinearity between the ESG score and the interaction term, we calculated their correlation, which is 0.067, suggesting that multicollinearity is not a significant issue. In addition, we computed variance inflation factors (VIFs) for the regressions in Table 2 and Table 3. All VIFs are below 5, further confirming that multicollinearity is not a concern in our results.
8
EPS Bias is defined as the difference between an analyst’s EPS forecast and the mean forecast, scaled by the cross-sectional standard deviation of all forecasts for the same firm-quarter.
9
We also conduct robustness checks addressing endogeneity and heterogeneity concerns for the PTG forecast sample, and our results remain consistent across these tests.
10
See “SEC Adopts Rules to Enhance and Standardize Climate-Related Disclosures for Investors”, 6 March 2024, https://www.sec.gov/newsroom/press-releases/2024-31 (accessed on 13 October 2025).
And “SEC votes to end defense of climate disclosure rules”, 27 March 2025. https://www.sec.gov/newsroom/press-releases/2025-58 (accessed on 13 October 2025).

References

  1. Acheampong, A., & Elshandidy, T. (2025). Does sustainability disclosure improve analysts’ forecast accuracy? Evidence from European banks. Financial Innovation, 11(1), 25. [Google Scholar] [CrossRef]
  2. Bancel, F., Glavas, D., & Karolyi, G. A. (2025). Do ESG factors influence firm valuations? Evidence from the field. Financial Review, 60(4), 1191–1223. [Google Scholar] [CrossRef]
  3. Barron, O. E., & Stuerke, P. S. (1998). Dispersion in analysts’ earnings forecasts as a measure of uncertainty. Journal of Accounting, Auditing & Finance, 13(3), 245–270. [Google Scholar]
  4. Bofinger, Y., Heyden, K. J., & Rock, B. (2022). Corporate social responsibility and market efficiency: Evidence from ESG and misvaluation measures. Journal of Banking & Finance, 134, 106322. [Google Scholar] [CrossRef]
  5. Boulton, T. J. (2024). Mandatory ESG disclosure, information asymmetry, and litigation risk: Evidence from initial public offerings. European Financial Management, 30(5), 2790–2839. [Google Scholar] [CrossRef]
  6. Bradshaw, M. T. (2002). The use of target prices to justify sell-side analysts’ stock recommendations. Accounting Horizons, 16(1), 27–41. [Google Scholar] [CrossRef]
  7. Buchanan, B., Cao, C. X., & Chen, C. (2018). Corporate social responsibility, firm value, and influential institutional ownership. Journal of Corporate Finance, 52, 73–95. [Google Scholar] [CrossRef]
  8. Cai, H., Yao, T., & Zhang, X. (2024). Confirmation bias in analysts’ response to consensus forecasts. Journal of Behavioral Finance, 25(3), 334–355. [Google Scholar] [CrossRef]
  9. Christensen, D. M., Serafeim, G., & Sikochi, A. (2022). Why is corporate virtue in the eye of the beholder? The case of ESG ratings. The Accounting Review, 97(1), 147–175. [Google Scholar] [CrossRef]
  10. Clement, M. B., & Tse, S. Y. (2005). Financial analyst characteristics and herding behavior in forecasting. The Journal of Finance, 60(1), 307–341. [Google Scholar] [CrossRef]
  11. Da, Z., Hong, K. P., & Lee, S. (2016). What drives target price forecasts and their investment value? Journal of Business Finance & Accounting, 43(3–4), 487–510. [Google Scholar] [CrossRef]
  12. del Río, C., López-Arceiz, F. J., & Muga, L. (2023). Do sustainability disclosure mechanisms reduce market myopia? Evidence from European sustainability companies. International Review of Financial Analysis, 87, 102600. [Google Scholar] [CrossRef]
  13. Demartis, S., & Rogo, B. (2024). The relationship between ESG scores and Value-at-Risk: A vine copula–GARCH based approach. Journal of Risk and Financial Management, 17(11), 517. [Google Scholar] [CrossRef]
  14. Demiroglu, C., & Ryngaert, M. (2010). The first analyst coverage of neglected stocks. Financial Management, 39(2), 555–584. [Google Scholar] [CrossRef]
  15. Derrien, F., Krüger, P., Landier, A., & Yao, T. (2022). ESG news, future cash flows, and firm value. Swiss finance institute research paper No. 21-84, HEC Paris research paper No FIN-2021-1441. Available online: https://ssrn.com/abstract=3903274 (accessed on 13 October 2025). [CrossRef]
  16. Dhaliwal, D. S., Radhakrishnan, S., Tsang, A., & Yang, Y. G. (2012). Nonfinancial disclosure and analyst forecast accuracy: International evidence on corporate social responsibility disclosure. The Accounting Review, 87(3), 723–759. [Google Scholar] [CrossRef]
  17. Diether, K. B., Malloy, C. J., & Scherbina, A. (2002). Differences of opinion and the cross section of stock returns. The Journal of Finance, 57(5), 2113–2141. [Google Scholar] [CrossRef]
  18. Easterwood, J. C., & Nutt, S. R. (1999). Inefficiency in analysts’ earnings forecasts: Systematic misreaction or systematic optimism? The Journal of Finance, 54(5), 1777–1797. [Google Scholar] [CrossRef]
  19. Eichholtz, P., Holtermans, R., Kok, N., & Yönder, E. (2019). Environmental performance and the cost of debt: Evidence from commercial mortgages and REIT bonds. Journal of Banking & Finance, 102, 19–32. [Google Scholar] [CrossRef]
  20. Ertimur, Y., Muslu, V., & Zhang, F. (2011). Why are recommendations optimistic? Evidence from analysts’ coverage initiations. Review of Accounting Studies, 16(4), 679–718. [Google Scholar] [CrossRef]
  21. Gao, X., Luo, J., & Zhao, Q. (2025). Impact of equity pledges on analysts earnings forecasts. Finance Research Letters, 71, 106380. [Google Scholar] [CrossRef]
  22. Grewal, J., Hauptmann, C., & Serafeim, G. (2021). Material sustainability information and stock price informativeness. Journal of Business Ethics, 171(3), 513–544. [Google Scholar] [CrossRef]
  23. Hashim, N. A., & Strong, N. C. (2018). Do analysts’ cash flow forecasts improve their target price accuracy? Contemporary Accounting Research, 35(4), 1816–1842. [Google Scholar] [CrossRef]
  24. Helbok, G., & Walker, M. (2004). On the nature and rationality of analysts’ forecasts under earnings conservatism. The British Accounting Review, 36(1), 45–77. [Google Scholar] [CrossRef]
  25. Hilary, G., & Menzly, L. (2006). Does past success lead analysts to become overconfident? Management Science, 52(4), 489–500. [Google Scholar] [CrossRef]
  26. Hirshleifer, D., Levi, Y., Lourie, B., & Teoh, S. H. (2019). Decision fatigue and heuristic analyst forecasts. Journal of Financial Economics, 133(1), 83–98. [Google Scholar] [CrossRef]
  27. Hirshleifer, D., Lourie, B., Ruchti, T. G., & Truong, P. (2021). First impression bias: Evidence from analyst forecasts. Review of Finance, 25(2), 325–364. [Google Scholar] [CrossRef]
  28. Hong, H., & Kubik, J. D. (2003). Analyzing the analysts: Career concerns and biased earnings forecasts. The Journal of Finance, 58(1), 313–351. [Google Scholar] [CrossRef]
  29. Hong, H., Kubik, J. D., & Solomon, A. (2000). Security analysts’ career concerns and herding of earnings forecasts. The RAND Journal of Economics, 31(1), 121–144. [Google Scholar] [CrossRef]
  30. Huang, R., Krishnan, M., Shon, J., & Zhou, P. (2017). Who herds? Who doesn’t? Estimates of analysts’ herding propensity in forecasting earnings. Contemporary Accounting Research, 34(1), 374–399. [Google Scholar] [CrossRef]
  31. Irvine, P. J. (2003). The incremental impact of analyst initiation of coverage. Journal of Corporate Finance, 9(4), 431–451. [Google Scholar] [CrossRef]
  32. Jannati, S., Kumar, A., Niessen-Ruenzi, A., & Wolfers, J. (2025). In-group bias in financial markets. Management Science. [Google Scholar] [CrossRef]
  33. Jegadeesh, N., & Kim, W. (2010). Do analysts herd? An analysis of recommendations and market reactions. The Review of Financial Studies, 23(2), 901–937. [Google Scholar] [CrossRef]
  34. Ke, B., & Yu, Y. (2006). The effect of issuing biased earnings forecasts on analysts’ access to management and survival. Journal of Accounting Research, 44(5), 965–999. [Google Scholar] [CrossRef]
  35. Li, F., Lin, C., & Lin, T. (2021). Salient anchor and analyst recommendation downgrade. Journal of Corporate Finance, 69, 102033. [Google Scholar] [CrossRef]
  36. Li, Y., Gong, M., Zhang, X., & Koh, L. (2018). The impact of environmental, social, and governance disclosure on firm value: The role of CEO power. The British Accounting Review, 50(1), 60–75. [Google Scholar] [CrossRef]
  37. Lim, T. (2001). Rationality and analysts’ forecast bias. The Journal of Finance, 56(1), 369–385. [Google Scholar] [CrossRef]
  38. Luo, K., & Wu, S. (2022). Corporate sustainability and analysts’ earnings forecast accuracy: Evidence from environmental, social and governance ratings. Corporate Social Responsibility and Environmental Management, 29(5), 1465–1481. [Google Scholar] [CrossRef]
  39. Malmendier, U., & Shanthikumar, D. (2014). Do security analysts speak in two tongues? The Review of Financial Studies, 27(5), 1287–1322. [Google Scholar] [CrossRef]
  40. McNichols, M., & O’Brien, P. C. (1997). Self-selection and analyst coverage. Journal of Accounting Research, 35, 167–199. [Google Scholar] [CrossRef]
  41. Park, M., Yoon, A., & Zach, T. (2025). Sell-side analysts’ assessment of ESG risk. Journal of Accounting and Economics, 79(2–3), 101759. [Google Scholar] [CrossRef]
  42. Roger, T. (2024). Do financial analysts care about ESG? Finance Research Letters, 63, 105289. [Google Scholar] [CrossRef]
  43. Serafeim, G., & Yoon, A. (2023). Stock price reactions to ESG news: The role of ESG ratings and disagreement. Review of Accounting Studies, 28(3), 1500–1530. [Google Scholar] [CrossRef]
  44. Sohn, B. C. (2012). Analyst forecast, accounting conservatism and the related valuation implications. Accounting & Finance, 52, 311–341. [Google Scholar]
  45. Tang, S., Qin, C., & Qi, F. (2025). Circuit courts and capital market transparency: Evidence from analysts’ earnings forecasts. Finance Research Letters, 77, 107150. [Google Scholar] [CrossRef]
  46. Wang, Z., & Xing, T. (2025). ESG information disclosure, stock price informativeness and corporate digital transformation. Applied Economics, 57(6), 600–616. [Google Scholar] [CrossRef]
  47. Wong, W. C., Batten, J. A., Ahmad, A. H., Mohamed-Arshad, S. B., Nordin, S., & Adzis, A. A. (2021). Does ESG certification add firm value? Finance Research Letters, 39, 101593. [Google Scholar] [CrossRef]
  48. Zhang, H., Zhang, H., Tian, L., Yuan, S., & Tu, Y. (2024). ESG performance and litigation risk. Finance Research Letters, 63, 105311. [Google Scholar] [CrossRef]
  49. Zhang, X. F. (2006). Information uncertainty and stock returns. The Journal of Finance, 61(1), 105–137. [Google Scholar] [CrossRef]
Figure 1. Analysts’ Absolute Forecast Bias Across ESG Score Terciles. *** indicates statistical significance at the 1% level.
Figure 1. Analysts’ Absolute Forecast Bias Across ESG Score Terciles. *** indicates statistical significance at the 1% level.
Jrfm 18 00585 g001
Table 1. Summary Statistics. This table presents the summary statistics for all variables in the EPS forecast sample. The detailed definitions of all variables are provided in Appendix A, Table A1. Most variables, except for Bloomberg ESG scores and the dummy variable New Firm, are winsorized at the 1st and 99th percentiles based on annual breakpoints. Firms’ market capitalizations are expressed in millions of dollars and adjusted to 2022 dollars.
Table 1. Summary Statistics. This table presents the summary statistics for all variables in the EPS forecast sample. The detailed definitions of all variables are provided in Appendix A, Table A1. Most variables, except for Bloomberg ESG scores and the dummy variable New Firm, are winsorized at the 1st and 99th percentiles based on annual breakpoints. Firms’ market capitalizations are expressed in millions of dollars and adjusted to 2022 dollars.
VariableObs.MeanStd. Dev.P25MedianP75
EPS Bias631,6600.0000.908−0.6340.0000.627
Abs EPS Bias631,6600.7360.5520.3040.6311.051
New Firm631,6600.0480.2140.0000.0000.000
ESG Score631,66028.24516.27214.87623.14139.669
Analyst Cov631,66012.5727.3387.00011.00017.000
EPS Disp631,6600.2820.8450.0350.0730.185
Forecast Age631,66066.96527.15050.00081.00089.000
Analyst Exp631,66012.3688.2926.00011.00018.000
No. of Firms631,66039.22123.26822.00035.00052.000
Turnover631,6600.2520.2010.1320.1940.300
Return Vol631,6600.1020.0590.0630.0870.124
Mkt Cap631,66020,290.93448,835.2551587.1594702.83516,003.228
ROA631,6600.0090.0330.0020.0100.022
BM631,6600.5180.5990.2160.3930.669
Sales Growth631,6600.0410.226−0.0380.0220.090
Table 2. Initial Coverage and Analyst EPS Forecast Bias. This table presents the results from OLS regressions of analysts’ EPS forecast biases on New Firm and a set of control variables, using data from 2007 to 2022. The dependent variable in Columns (1) through (6) is EPS Bias. Columns (3) and (4) restrict the sample to forecasts with positive bias, while Columns (5) and (6) focus on those with negative bias. Columns (7) and (8) use absolute EPS bias (Abs EPS Bias) as the dependent variable. The detailed definitions of all variables are documented in Appendix A, Table A1. All t-statistics reported in the parentheses are calculated based on robust standard errors clustered at the analyst level. *, **, and *** indicate significance at 10%, 5% and 1% level, respectively.
Table 2. Initial Coverage and Analyst EPS Forecast Bias. This table presents the results from OLS regressions of analysts’ EPS forecast biases on New Firm and a set of control variables, using data from 2007 to 2022. The dependent variable in Columns (1) through (6) is EPS Bias. Columns (3) and (4) restrict the sample to forecasts with positive bias, while Columns (5) and (6) focus on those with negative bias. Columns (7) and (8) use absolute EPS bias (Abs EPS Bias) as the dependent variable. The detailed definitions of all variables are documented in Appendix A, Table A1. All t-statistics reported in the parentheses are calculated based on robust standard errors clustered at the analyst level. *, **, and *** indicate significance at 10%, 5% and 1% level, respectively.
VariableEPS BiasEPS Bias > 0EPS Bias < 0Abs EPS Bias
(1)(2)(3)(4)(5)(6)(7)(8)
New Firm0.013 **0.012 **−0.014 ***−0.014 ***0.022 ***0.020 ***−0.019 ***−0.018 ***
(2.10)(2.05)(−2.74)(−2.70)(4.67)(4.10)(−5.17)(−4.90)
ln(1 + Analyst Cov)−0.012 ***−0.004−0.026 ***−0.018 ***0.025 ***0.025 ***−0.011 ***−0.010 ***
(−3.03)(−0.95)(−8.21)(−4.92)(8.44)(7.32)(−4.74)(−3.58)
EPS Disp0.0010.001−0.008 ***−0.008 ***0.006 ***0.005 ***−0.006 ***−0.006 ***
(0.50)(0.39)(−6.24)(−5.73)(4.32)(3.58)(−6.24)(−5.89)
ln(1 + Forecast Age)0.045 ***0.045 ***0.032 ***0.033 ***0.004 **0.0020.014 ***0.015 ***
(18.27)(18.03)(15.68)(15.91)(2.25)(1.25)(8.38)(8.86)
ln(1 + Analyst Exp)0.0090.009−0.004−0.0040.0100.009−0.006−0.004
(0.69)(0.68)(−0.44)(−0.42)(1.13)(0.94)(−0.86)(−0.53)
ln(1 + No. of Firms)−0.032 **−0.035 ***0.0060.006−0.023 **−0.023 **0.014 *0.013 *
(−2.47)(−2.63)(0.58)(0.60)(−2.49)(−2.52)(1.94)(1.78)
Turnover−0.020 *−0.004−0.018 **−0.0130.004−0.008−0.0080.001
(−1.89)(−0.29)(−2.14)(−1.07)(0.59)(−0.73)(−1.38)(0.10)
Return Vol−0.044−0.123 ***−0.054 *−0.046−0.032−0.036−0.0060.002
(−1.10)(−2.78)(−1.89)(−1.41)(−1.18)(−1.14)(−0.31)(0.09)
ln(1 + Mkt Cap)−0.010 ***−0.006−0.008 ***−0.0050.0010.001−0.005 ***−0.002
(−5.96)(−1.41)(−6.48)(−1.52)(1.00)(0.31)(−4.64)(−0.78)
ROA−0.075−0.0150.0270.039−0.016−0.0170.0410.041
(−1.33)(−0.24)(0.62)(0.81)(−0.43)(−0.38)(1.30)(1.21)
VariableEPS BiasEPS Bias > 0EPS Bias < 0Abs EPS Bias
(1)(2)(3)(4)(5)(6)(7)(8)
ln(1 + B/M)−0.009−0.003−0.018 ***−0.0130.0050.015 **−0.011 ***−0.013 **
(−1.17)(−0.28)(−3.10)(−1.57)(0.91)(1.97)(−2.62)(−2.15)
Sales Growth0.0060.004−0.000−0.0020.010 **0.005−0.004−0.002
(0.98)(0.62)(−0.00)(−0.34)(2.27)(1.23)(−1.24)(−0.77)
Intercept0.037−0.0070.751 ***0.697 ***−0.766 ***−0.753 ***0.719 ***0.684 ***
(0.89)(−0.12)(23.06)(16.20)(−25.33)(−19.46)(28.71)(21.13)
 
Obs.631,307631,304311,458311,411314,962314,914631,307631,304
Firm FENoYesNoYesNoYesNoYes
Analyst FEYesYesYesYesYesYesYesYes
Year-Quarter FEYesYesYesYesYesYesYesYes
Adjusted R20.0360.0340.0480.0520.0420.0460.0350.034
Table 3. Initial Coverage, ESG Score, and Analyst EPS Forecast Bias. This table presents the results from OLS regressions of analysts’ EPS forecast biases on New Firm, ESG Score, their interaction term, and a set of control variables, using data from 2007 to 2022. The dependent variable in Columns (1) through (6) is EPS Bias. Columns (3) and (4) restrict the sample to forecasts with positive bias, while Columns (5) and (6) focus on those with negative bias. Columns (7) and (8) use absolute EPS bias (Abs EPS Bias) as the dependent variable. The detailed definitions of all variables are documented in Appendix A, Table A1. ESG Scores are scaled by 100 for ease of coefficient interpretation. All t-statistics reported in the parentheses are calculated based on robust standard errors clustered at the analyst level. * and *** indicate significance at 10% and 1% level, respectively.
Table 3. Initial Coverage, ESG Score, and Analyst EPS Forecast Bias. This table presents the results from OLS regressions of analysts’ EPS forecast biases on New Firm, ESG Score, their interaction term, and a set of control variables, using data from 2007 to 2022. The dependent variable in Columns (1) through (6) is EPS Bias. Columns (3) and (4) restrict the sample to forecasts with positive bias, while Columns (5) and (6) focus on those with negative bias. Columns (7) and (8) use absolute EPS bias (Abs EPS Bias) as the dependent variable. The detailed definitions of all variables are documented in Appendix A, Table A1. ESG Scores are scaled by 100 for ease of coefficient interpretation. All t-statistics reported in the parentheses are calculated based on robust standard errors clustered at the analyst level. * and *** indicate significance at 10% and 1% level, respectively.
VariableEPS BiasEPS Bias > 0EPS Bias < 0Abs EPS Bias
(1)(2)(3)(4)(5)(6)(7)(8)
New Firm−0.004−0.006−0.040 ***−0.039 ***0.030 ***0.029 ***−0.035 ***−0.036 ***
(−0.40)(−0.55)(−4.61)(−4.57)(3.45)(3.35)(−5.55)(−5.53)
ESG Score−0.004−0.013−0.021 *−0.022−0.007−0.007−0.010−0.009
(−0.32)(−0.55)(−1.88)(−1.16)(−0.63)(−0.41)(−1.26)(−0.66)
New Firm × ESG Score0.069 *0.074 *0.105 ***0.105 ***−0.030−0.0370.066 ***0.071 ***
(1.81)(1.93)(3.21)(3.21)(−0.96)(−1.19)(2.75)(2.92)
 
Obs.631,307631,304311,458311,411314,962314,914631,307631,304
ControlsYesYesYesYesYesYesYesYes
Firm FENoYesNoYesNoYesNoYes
Analyst FEYesYesYesYesYesYesYesYes
Year-Quarter FEYesYesYesYesYesYesYesYes
Adjusted R20.0360.0340.0480.0520.0420.0460.0350.034
Table 4. Initial Coverage, ESG Score, and Analyst EPS Forecast Bias: Small Brokerage Firms. This table reports the results of OLS regressions based on EPS forecasts issued by analysts at small brokerage firms. A brokerage firm is classified as small if its number of analysts falls below the quarterly median. Analysts’ EPS forecast biases are regressed on New Firm, ESG Score, their interaction term, and a set of control variables, using data from 2007 to 2022. The dependent variable in Columns (1) through (6) is EPS Bias. Columns (3) and (4) restrict the sample to forecasts with positive bias, while Columns (5) and (6) focus on those with negative bias. Columns (7) and (8) use absolute EPS bias (Abs EPS Bias) as the dependent variable. The detailed definitions of all variables are documented in Appendix A Table A1. ESG Scores are scaled by 100 for ease of coefficient interpretation. All t-statistics reported in the parentheses are calculated based on robust standard errors clustered at the analyst level. *, **, and *** indicate significance at 10%, 5% and 1% level, respectively.
Table 4. Initial Coverage, ESG Score, and Analyst EPS Forecast Bias: Small Brokerage Firms. This table reports the results of OLS regressions based on EPS forecasts issued by analysts at small brokerage firms. A brokerage firm is classified as small if its number of analysts falls below the quarterly median. Analysts’ EPS forecast biases are regressed on New Firm, ESG Score, their interaction term, and a set of control variables, using data from 2007 to 2022. The dependent variable in Columns (1) through (6) is EPS Bias. Columns (3) and (4) restrict the sample to forecasts with positive bias, while Columns (5) and (6) focus on those with negative bias. Columns (7) and (8) use absolute EPS bias (Abs EPS Bias) as the dependent variable. The detailed definitions of all variables are documented in Appendix A Table A1. ESG Scores are scaled by 100 for ease of coefficient interpretation. All t-statistics reported in the parentheses are calculated based on robust standard errors clustered at the analyst level. *, **, and *** indicate significance at 10%, 5% and 1% level, respectively.
VariableEPS BiasEPS Bias > 0EPS Bias < 0Abs EPS Bias
(1)(2)(3)(4)(5)(6)(7)(8)
New Firm0.0050.004−0.042 ***−0.043 ***0.047 ***0.047 ***−0.044 ***−0.045 ***
(0.36)(0.29)(−3.56)(−3.53)(3.96)(3.92)(−4.97)(−4.95)
ESG Score0.015−0.008−0.028 *−0.0190.028 *0.019−0.033 ***−0.021
(0.72)(−0.23)(−1.69)(−0.69)(1.92)(0.74)(−2.75)(−1.06)
New Firm × ESG Score0.0170.0190.080 *0.085 *−0.085 *−0.098 **0.079 **0.086 **
(0.31)(0.34)(1.71)(1.81)(−1.86)(−2.13)(2.20)(2.38)
 
Obs.296,270296,233146,758146,652146,786146,702296,270296,233
ControlsYesYesYesYesYesYesYesYes
Firm FENoYesNoYesNoYesNoYes
Analyst FEYesYesYesYesYesYesYesYes
Year-Quarter FEYesYesYesYesYesYesYesYes
Adjusted R20.0430.0450.0540.0590.0460.0510.0390.041
Table 5. Initial Coverage, ESG Score, and Analyst EPS Forecast Bias: PSM Analysis. This table presents the results of the propensity score matching (PSM) analysis. For each quarter, treatment forecasts (New Firm = 1) are matched to comparable control forecasts (New Firm = 0) based on their estimated propensity scores. Propensity scores are estimated using a logistic regression that includes all control variables as matching covariates. Using the matched samples, analysts’ EPS forecast biases are regressed on New Firm, ESG Score, their interaction term, and all control variables. Panel A reports results from 1-to-1 nearest neighbor matching, and Panel B reports results from 1-to-3 nearest neighbor matching. The dependent variable in Columns (1) through (3) is EPS Bias. Column (2) restricts the sample to forecasts with positive bias, while Column (3) focuses on those with negative bias. Column (4) uses absolute EPS bias (Abs EPS Bias) as the dependent variable. The detailed definitions of all variables are documented in Appendix A Table A1. ESG Scores are scaled by 100 for ease of coefficient interpretation. All t-statistics reported in the parentheses are calculated based on robust standard errors clustered at the analyst level. *, **, and *** indicate significance at 10%, 5% and 1% level, respectively.
Table 5. Initial Coverage, ESG Score, and Analyst EPS Forecast Bias: PSM Analysis. This table presents the results of the propensity score matching (PSM) analysis. For each quarter, treatment forecasts (New Firm = 1) are matched to comparable control forecasts (New Firm = 0) based on their estimated propensity scores. Propensity scores are estimated using a logistic regression that includes all control variables as matching covariates. Using the matched samples, analysts’ EPS forecast biases are regressed on New Firm, ESG Score, their interaction term, and all control variables. Panel A reports results from 1-to-1 nearest neighbor matching, and Panel B reports results from 1-to-3 nearest neighbor matching. The dependent variable in Columns (1) through (3) is EPS Bias. Column (2) restricts the sample to forecasts with positive bias, while Column (3) focuses on those with negative bias. Column (4) uses absolute EPS bias (Abs EPS Bias) as the dependent variable. The detailed definitions of all variables are documented in Appendix A Table A1. ESG Scores are scaled by 100 for ease of coefficient interpretation. All t-statistics reported in the parentheses are calculated based on robust standard errors clustered at the analyst level. *, **, and *** indicate significance at 10%, 5% and 1% level, respectively.
Panel A: 1-to-1 Nearest Neighbor Matching
VariableEPS BiasEPS Bias > 0EPS Bias < 0Abs EPS Bias
(1)(2)(3)(4)
New Firm0.009−0.049 ***0.023 *−0.055 ***
(0.55)(−3.50)(1.66)(−5.48)
ESG Score−0.038−0.137 **−0.104 *0.024
(−0.51)(−2.10)(−1.67)(0.54)
New Firm × ESG Score0.0220.103 **−0.0020.102 ***
(0.39)(2.11)(−0.04)(2.90)
 
Obs.57,66828,55428,90257,668
ControlsYesYesYesYes
Firm FEYesYesYesYes
Analyst FEYesYesYesYes
Year-Quarter FEYesYesYesYes
Adjusted R20.0370.0490.0320.0384
Panel B: 1-to-3 Nearest Neighbor Matching
VariableEPS BiasEPS Bias > 0EPS Bias < 0Abs EPS Bias
(1)(2)(3)(4)
New Firm0.011−0.039 ***0.038 ***−0.045 ***
(0.88)(−3.55)(3.37)(−5.68)
ESG Score−0.009−0.054−0.006−0.017
(−0.17)(−1.14)(−0.12)(−0.50)
New Firm × ESG Score0.0000.090 **−0.0490.097 ***
(0.00)(2.30)(−1.24)(3.39)
 
Obs.106,04851,34151,966106,048
ControlsYesYesYesYes
Firm FEYesYesYesYes
Analyst FEYesYesYesYes
Year-Quarter FEYesYesYesYes
Adjusted R20.0520.0880.0740.0544
Table 6. Placebo Tests Using Lead Forecast Bias. This table presents the results from OLS regressions of analysts’ EPS forecast biases in period t + 1 on New Firm, ESG Score, their interaction term, and a set of control variables. The dependent variable in Columns (1) through (6) is EPS Bias. Columns (3) and (4) restrict the sample to forecasts with positive bias, while Columns (5) and (6) focus on those with negative bias. Columns (7) and (8) use absolute EPS bias (Abs EPS Bias) as the dependent variable. The detailed definitions of all variables are documented in Appendix A, Table A1. ESG Scores are scaled by 100 for ease of coefficient interpretation. All t-statistics reported in the parentheses are calculated based on robust standard errors clustered at the analyst level. * indicates significance at 10% level.
Table 6. Placebo Tests Using Lead Forecast Bias. This table presents the results from OLS regressions of analysts’ EPS forecast biases in period t + 1 on New Firm, ESG Score, their interaction term, and a set of control variables. The dependent variable in Columns (1) through (6) is EPS Bias. Columns (3) and (4) restrict the sample to forecasts with positive bias, while Columns (5) and (6) focus on those with negative bias. Columns (7) and (8) use absolute EPS bias (Abs EPS Bias) as the dependent variable. The detailed definitions of all variables are documented in Appendix A, Table A1. ESG Scores are scaled by 100 for ease of coefficient interpretation. All t-statistics reported in the parentheses are calculated based on robust standard errors clustered at the analyst level. * indicates significance at 10% level.
VariableEPS BiasEPS Bias > 0EPS Bias < 0Abs EPS Bias
(1)(2)(3)(4)(5)(6)(7)(8)
New Firm0.0090.006−0.004−0.0030.0120.011−0.009−0.008
(0.82)(0.53)(−0.44)(−0.29)(1.31)(1.22)(−1.31)(−1.21)
ESG Score0.0080.008−0.021 *−0.024−0.0020.009−0.013−0.019
(0.52)(0.33)(−1.72)(−1.21)(−0.14)(0.49)(−1.49)(−1.29)
New Firm × ESG Score−0.040−0.031−0.010−0.011−0.024−0.0280.0080.010
(−1.03)(−0.80)(−0.28)(−0.30)(−0.72)(−0.82)(0.35)(0.40)
 
Obs.564,081564,063278,052277,988281,990281,942564,081564,063
ControlsYesYesYesYesYesYesYesYes
Firm FENoYesNoYesNoYesNoYes
Analyst FEYesYesYesYesYesYesYesYes
Year-Quarter FEYesYesYesYesYesYesYesYes
Adjusted R20.0360.0340.0480.0520.0420.0460.0350.034
Table 7. Initial Coverage, ESG Score, and Analyst EPS Forecast Bias: High vs. Low Earnings Forecast Dispersion. This table presents the results from OLS regressions conducted separately for low-, medium-, and high-dispersion subsamples. For each subsample, analysts’ EPS forecast biases are regressed on New Firm, ESG Score, their interaction term, and a set of control variables, using data from 2007 to 2022. Each quarter, all firms in each sample are sorted into tercile portfolios, Low, Medium, and High, based on their earnings forecast dispersion. To conserve space, we present results only for the small and large firm subsamples. The dependent variable in Columns (1) through (6) is EPS Bias. Columns (3) and (4) restrict the sample to forecasts with positive bias, while Columns (5) and (6) focus on those with negative bias. Models (7) and (8) use absolute EPS bias (Abs EPS Bias) as the dependent variable. The detailed definitions of all variables are documented in Appendix A, Table A1. ESG Scores are scaled by 100 for ease of coefficient interpretation. All t-statistics reported in the parentheses are calculated based on robust standard errors clustered at the analyst level. The statistics reported in parentheses in these columns are the corresponding p-values. *, **, and *** indicate significance at 10%, 5% and 1% level, respectively.
Table 7. Initial Coverage, ESG Score, and Analyst EPS Forecast Bias: High vs. Low Earnings Forecast Dispersion. This table presents the results from OLS regressions conducted separately for low-, medium-, and high-dispersion subsamples. For each subsample, analysts’ EPS forecast biases are regressed on New Firm, ESG Score, their interaction term, and a set of control variables, using data from 2007 to 2022. Each quarter, all firms in each sample are sorted into tercile portfolios, Low, Medium, and High, based on their earnings forecast dispersion. To conserve space, we present results only for the small and large firm subsamples. The dependent variable in Columns (1) through (6) is EPS Bias. Columns (3) and (4) restrict the sample to forecasts with positive bias, while Columns (5) and (6) focus on those with negative bias. Models (7) and (8) use absolute EPS bias (Abs EPS Bias) as the dependent variable. The detailed definitions of all variables are documented in Appendix A, Table A1. ESG Scores are scaled by 100 for ease of coefficient interpretation. All t-statistics reported in the parentheses are calculated based on robust standard errors clustered at the analyst level. The statistics reported in parentheses in these columns are the corresponding p-values. *, **, and *** indicate significance at 10%, 5% and 1% level, respectively.
VariableEPS BiasEPS Bias > 0EPS Bias < 0Abs EPS Bias
Low Forecast DispersionHigh
Forecast Dispersion
Low
Forecast Dispersion
High
Forecast Dispersion
Low
Forecast Dispersion
High
Forecast Dispersion
Low
Forecast Dispersion
High
Forecast Dispersion
(1)(2)(3)(4)(5)(6)(7)(8)
New Firm−0.0020.000−0.050 ***−0.047 ***0.0100.048 ***−0.030 **−0.050 ***
(−0.12)(0.02)(−3.05)(−3.33)(0.62)(3.28)(−2.50)(−4.80)
ESG Score−0.008−0.0300.010−0.061 *0.002−0.0100.006−0.029
(−0.21)(−0.67)(0.34)(−1.69)(0.07)(−0.29)(0.27)(−1.09)
New Firm × ESG Score0.0340.141 **0.123 **0.194 ***−0.005−0.0400.0610.125 ***
(0.53)(1.97)(2.15)(3.19)(−0.10)(−0.71)(1.51)(2.80)
Obs.209,922209,994101,724103,933104,900104,229209,922209,994
ControlsYesYesYesYesYesYesYesYes
Firm FEYesYesYesYesYesYesYesYes
Analyst FEYesYesYesYesYesYesYesYes
Year-Quarter FEYesYesYesYesYesYesYesYes
Adjusted R20.0400.0380.0550.0650.0430.0630.0330.037
Table 8. Initial Coverage, ESG Score, and Analyst Price Target Forecast Bias. This table presents the results from OLS regressions of analysts’ price target forecast biases on New Firm, ESG Score, their interaction term, and a set of control variables, using data from 2007 to 2022. The dependent variable in Columns (1) through (6) is PTG Bias. Columns (3) and (4) restrict the sample to forecasts with positive bias, while Columns (5) and (6) focus on those with negative bias. Columns (7) and (8) use absolute EPS bias (Abs PTG Bias) as the dependent variable. The detailed definitions of all variables are documented in Appendix A, Table A1. ESG Scores are scaled by 100 for ease of coefficient interpretation. All t-statistics reported in the parentheses are calculated based on robust standard errors clustered at the analyst level. * and *** indicate significance at 10% and 1% level, respectively.
Table 8. Initial Coverage, ESG Score, and Analyst Price Target Forecast Bias. This table presents the results from OLS regressions of analysts’ price target forecast biases on New Firm, ESG Score, their interaction term, and a set of control variables, using data from 2007 to 2022. The dependent variable in Columns (1) through (6) is PTG Bias. Columns (3) and (4) restrict the sample to forecasts with positive bias, while Columns (5) and (6) focus on those with negative bias. Columns (7) and (8) use absolute EPS bias (Abs PTG Bias) as the dependent variable. The detailed definitions of all variables are documented in Appendix A, Table A1. ESG Scores are scaled by 100 for ease of coefficient interpretation. All t-statistics reported in the parentheses are calculated based on robust standard errors clustered at the analyst level. * and *** indicate significance at 10% and 1% level, respectively.
VariablePTG BiasPTG Bias > 0PTG Bias < 0Abs PTG Bias
(1)(2)(3)(4)(5)(6)(7)(8)
New Firm0.047 ***0.0180.064 ***0.050 ***−0.024 ***−0.026 ***0.041 ***0.035 ***
(7.25)(1.59)(13.37)(5.86)(−5.23)(−3.28)(12.07)(5.82)
ESG Score −0.004 0.008 0.015 0.002
(−0.12) (0.43) (0.71) (0.10)
New Firm × ESG Score 0.117 *** 0.057 * 0.007 0.023
(3.15) (1.85) (0.28) (1.11)
Obs.556,597556,597276,077276,077274,406274,406556,597556,597
ControlsYesYesYesYesYesYesYesYes
Firm FEYesYesYesYesYesYesYesYes
Analyst FEYesYesYesYesYesYesYesYes
Year-Quarter FEYesYesYesYesYesYesYesYes
Adjusted R20.1080.1080.0880.0880.0970.0970.0630.063
Disclaimer/Publisher’s Note: The statements, opinions and data contained in all publications are solely those of the individual author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to people or property resulting from any ideas, methods, instructions or products referred to in the content.

Share and Cite

MDPI and ACS Style

Fallah, M.; Han, S.; Zhao, L. Forecast Bias in Analysts’ Initial Coverage: The Influence of Firm ESG Disclosures. J. Risk Financial Manag. 2025, 18, 585. https://doi.org/10.3390/jrfm18100585

AMA Style

Fallah M, Han S, Zhao L. Forecast Bias in Analysts’ Initial Coverage: The Influence of Firm ESG Disclosures. Journal of Risk and Financial Management. 2025; 18(10):585. https://doi.org/10.3390/jrfm18100585

Chicago/Turabian Style

Fallah, Mohammadali, Sulei Han, and Le Zhao. 2025. "Forecast Bias in Analysts’ Initial Coverage: The Influence of Firm ESG Disclosures" Journal of Risk and Financial Management 18, no. 10: 585. https://doi.org/10.3390/jrfm18100585

APA Style

Fallah, M., Han, S., & Zhao, L. (2025). Forecast Bias in Analysts’ Initial Coverage: The Influence of Firm ESG Disclosures. Journal of Risk and Financial Management, 18(10), 585. https://doi.org/10.3390/jrfm18100585

Article Metrics

Back to TopTop