Long-term institutional shareholdings and stock price informativeness of analyst target prices Open Access

Sell-side analyst forecasts are important to investors in making investment decisions (Bonner, Hugon, & Walther, 2007). This is evident from investor responses to analyst earnings forecasts and stock recommendations (Womack, 1996; Green, 2006; Huang, Mian, & Sankaraguruswamy, 2009; Jung, Shane, & Yang, 2012). Although a large body of literature has investigated the informativeness of earnings forecasts and stock recommendations, the informative role of analyst target price revisions based on the investment horizon of institutional investors remains empirically unclear. In this study, therefore, we examine whether long-term institutional shareholdings affect the stock market response to target price revisions.

Sell-side analysts provide information regarding a firm’s short-term and long-term prospects. For instance, analysts typically report quarterly forecasts for firm-level earnings and stock recommendations. Target prices reflect analyst opinions on future stock price levels. These opinions are provided for a longer horizon, i.e. 99% of target price forecasts in the I/B/E/S database are for 12 months or longer horizons. Analysts derive their target prices from valuation models that rely on long-term fundamentals, such as multi-period models (Demirakos, Strong, & Walker, 2010). Unlike earnings forecasts and stock recommendations, target prices are relatively more verifiable, containing continuous and rich information about a firm’s intrinsic value that market participants can use while assessing the price multiples and long-term firm value (Brav & Lehavy, 2003; Asquith, Mikhail, & Au, 2005; Bradshaw, Brown, & Huang, 2013; Han, Kang, & Kim, 2022; Gu, Guo, & Zhang, 2022).

Target prices contain useful incremental information in addition to other analyst reports, such as earnings forecasts and stock recommendations (Brav & Lehavy, 2003; Asquith et al., 2005; Da & Schaumburg, 2011; Feldman, Livnat, & Zhang, 2012; Ho, Brownen-Trinh, & Xu, 2021; Han et al., 2022). Analyst perception of a stock’s intrinsic value is reflected through these target prices, thus containing incremental information for institutional investors (Gu et al., 2022). Despite institutional investors sharing important commonalities, they are heterogeneous in their investment objectives, styles, legal restrictions and competitive pressures (Yan & Zhang, 2009). These differences may lead to different institutional investment horizons. Therefore, the investment horizon of institutional investors could also affect the stock market response to target price revisions.

According to the first school of thought, non-long-term (or transient) investors may be better informed about their portfolio firms. For example, Ke and Petroni (2004) document that transient institutional investors are more skilled than long-term investors at predicting patterns, such as consecutive quarterly earnings increases. Moreover, changes in the ownership of such investors reflect private information about firms’ long-term earnings potential – an insight supported by Ke and Ramalingegowda (2005), Boehmer and Kelley (2009), Yan and Zhang (2009) and Cremers and Pareek (2009). Since these investors indulge in active trading, they contribute to more efficient price discovery in the capital markets (Bartov, Radhakrishnan, & Krinsky, 2000; Collins, Gong, & Hribar, 2003; Ke & Ramalingegowda, 2005; Yan & Zhang, 2009). From this perspective, we might not observe any relationship between long-term institutional shareholdings and stock price reactions to target price revisions. Therefore, we hypothesize that long-term institutional shareholdings are not associated with the stock market response to target price revisions.

According to the second school of thought, long-term institutional investors are more likely to monitor portfolio firms, emphasizing long-term fundamentals. They are generally better informed and place greater weight on long-term prospects. For instance, Bushee (2001) finds that short-term investors tend to overweight firms’ near-term earnings potential while underweighting their long-term prospects. Subsequent studies also support the latter argument (e.g. Bushee, 2004; Gaspar, Massa, & Matos, 2005; Chen, Harford, & Li, 2007; Elyasiani & Jia, 2008; Elyasiani, Jia, & Mao, 2010; Derrien, Kecskés, & Thesmar, 2013), showing that long-term investors place less emphasis on short-term performance and focus more on long-term value creation.

Given that long-term institutional investors – such as pension funds and mutual funds – often hold their investments until the firm’s long-term benefits are realized (Gaspar et al., 2005), we may expect firms with more long-term institutional shareholdings to exhibit stronger stock price reactions to target price revisions. Since analysts typically rely on multi-period valuation models to generate target prices, these revisions contain information that resonates more with the long-run orientation of long-term investors. Accordingly, in the presence of more long-term institutional shareholdings, target price revisions are likely to result in relatively larger stock price adjustments (Yan & Zhang, 2009). Therefore, we hypothesize that long-term institutional shareholdings are associated with a strong stock market response to target price revisions. On ex ante, whether long-term institutional shareholdings affect the stock market response to target price revisions remains unexplored.

We examine the relationship between long-term institutional shareholdings and the stock market response to target price revisions using a sample of 53,988 target prices from 2000 to 2019. By identifying long-term institutional investors based on their portfolio churn ratio, we find that firms with more long-term institutional shareholdings experience positive returns and react strongly to target price revisions. A one standard deviation increase in long-term institutional shareholdings is associated with around 26% increase in cumulative abnormal returns (CARs), suggesting that investors find value in target price revisions in the presence of more long-term institutional shareholders. Furthermore, our results are not influenced by certain investment styles, as we also control for different firm- and analyst-level characteristics, including the firm size, book-to-market ratio (B/M), leverage, analyst coverage, forecast dispersion and the momentum effect. Our findings remain robust to several alternative specifications, addressing self-selection, reverse causality, potential omitted variables and measurement error biases. We consider five additional empirical specifications to address these endogeneity concerns, including the active share, quasi-natural experiments, entropy balancing, alternative measures and the percent bias analysis.

We also find that the positive response of investors to target price revisions is restricted mainly to firms with more long-term “motivated” institutional shareholdings than long-term “non-motivated” institutional shareholdings (Fich, Harford, & Tran, 2015). Firms where long-term institutional investors are motivated and have higher monitoring incentives, i.e. firms in which the holding value of an institution is in the top 10% of its portfolio, experience positive returns on target price revisions. Investors of firms with more long-term motivated institutional shareholdings find value in target price revisions than those of firms with long-term non-motivated institutional investors. Further, investors find more value in target price revisions when issued by sophisticated analysts, i.e. with higher experience, higher earnings accuracy and a bigger brokerage size.

We also investigate the role of a firm’s information environment in explaining the relationship between long-term institutional shareholdings and the stock market response to target price revisions. We reason that information contained in target price revisions is more valuable for investors when there is a high level of information asymmetry. We consider firms’ idiosyncratic volatility and the generalized probability of informed trading (PIN) as our measures of the information environment (Duarte et al., 2008, 2020; Yang, Zhang, & Zhang, 2020). Consistent with our expectations, the strong price effect of target price revisions for firms with more long-term institutional shareholdings is pronounced mainly for firms with high idiosyncratic volatility and a higher PIN. In other words, the information contained in target price revisions is more valuable to investors when the information environment of portfolio firms is relatively weaker.

We also examine whether long-term institutional shareholdings affect the stock market response to stock recommendations and earnings forecast revisions. Unlike target price revisions, investors respond benignly to stock recommendation revisions in the presence of long-term institutional shareholdings. Also, investors remain indifferent to changes in earnings forecasts, suggesting that they find value only in target price revisions. These findings support our assertion that target price revisions contain incremental information for investors compared to stock recommendations and earnings forecast revisions for firms with more long-term institutional shareholdings.

Our study contributes to the extant literature on the informativeness of target price revisions (Brav & Lehavy, 2003; Asquith et al., 2005; Da & Schaumburg, 2011; Feldman et al., 2012; Ho et al., 2021; Han et al., 2022). To the best of our knowledge, this is the first study that examines the relationship between long-term institutional shareholdings and the stock market response to target price revisions. We show that firms with more long-term institutional shareholdings, particularly long-term motivated investors having higher monitoring incentives, experience positive returns on target price revisions. Furthermore, target price revisions are more informative to investors when firms’ information environment is relatively weaker.

We also contribute to the literature that examines how the investment horizon influences investor reliance on firm-level information, such as analyst target prices (e.g. Bartov et al., 2000; Bushee, 2001; Collins et al., 2003; Bushee, 2004; Gaspar et al., 2005; Ke & Ramalingegowda, 2005; Chen et al., 2007; Elyasiani & Jia, 2008; Yan & Zhang, 2009; Boehmer & Kelley, 2009; Cremers & Pareek, 2009; Elyasiani et al., 2010; Derrien et al., 2013). Our findings suggest that investors find value in target price revisions, especially for firms with more long-term institutional shareholdings. Analyst target prices contain information about a firm’s intrinsic value that investors can use.

We further extend the literature on earnings forecasts and stock recommendation revisions (e.g. Womack, 1996; Bonner et al., 2007; Bradshaw et al., 2013). Our findings suggest that target price revisions contain incremental information for investors compared to earnings forecasts and stock recommendation revisions in the presence of long-term institutional shareholdings. Firms with more long-term institutional shareholdings observe a benign response to stock recommendation revisions and remain indifferent to earnings forecast revisions. On the other hand, firms with more long-term institutional shareholdings experience a strong positive reaction to target price revisions, suggesting that investors find value in target price revisions compared to stock recommendation and earnings forecast revisions.

The rest of the article is organized as follows. Section 2 reports data and the sample overview, section 3 discusses our empirical findings, and lastly, section 4 concludes the paper.

We obtain target price forecasts from the I/B/E/S unadjusted US historical target price database. We use unadjusted forecasts to avoid the split rounding effect (Barber & Kang, 2002; Payne & Thomas, 2003). Specifically, we consider all target price forecasts issued between Jan 2000 and Dec 2019. Analyst earnings forecasts and stock recommendations are also obtained from the I/B/E/S for the same sample period. Following Diether, Malloy, and Scherbina (2002), data are obtained for one-quarter-ahead earnings forecasts from the unadjusted earnings forecasts file to minimize the rounding errors in the I/B/E/S. We adjust our unadjusted forecasts based on the same per-share basis, using the I/B/E/S split adjustment factors. A stock recommendation revision is defined as an analyst’s current rating minus the prior rating issued by the same analyst. Prior rating is assumed to be outstanding if it has not appeared in the I/B/E/S stopped recommendation file and is less than one year old based on the I/B/E/S review date (Ljungqvist, Malloy, & Marston, 2009). We also exclude anonymous analysts and stock recommendations with no prior outstanding recommendations.

Following Dechow and You (2020), we exclude financial firms (group 29 of Fama and French (1997) 30-industry classification) from the sample. We also exclude target price revisions issued around firm-news days to avoid analyst information getting confounded by the recent firm-specific announcements. Firm-news days are defined as three days centered around the Compustat earnings announcement date (or the earnings guidance date obtained from the I/B/E/S Earnings Guidance file) and days with multiple analysts issuing target prices. We also remove events where the lagged stock prices are less than $1 to mitigate the market microstructure-related concerns ^[1].

We obtain quarterly institutional portfolio holdings for all the common stocks traded on the New York Stock Exchange (NYSE), American Stock Exchange (AMEX) and NASDAQ from the fourth quarter of 1999 to the fourth quarter of 2019 from Thomson Financial. The Securities and Exchange Commission (SEC) requires investment managers (worth $100 million or more) to report all equity positions greater than 10,000 shares or $200,000 in their 13F quarterly filings. Institutional ownership (IO) for each stock is defined as the number of shares held by institutional investors divided by the total number of shares outstanding ^[2]. To measure the investment horizon of investors, we use the methodology employed in prior studies (e.g. Gaspar et al., 2005; Chen et al., 2007; Nguyen, Kecskés, & Mansi, 2020; Pathan, Haq, Faff, & Seymour, 2021). Following Gaspar et al. (2005), we calculate the investor-centric churn rate $CRi,t$ to measure the frequency of rotation and identify long-term investors. If we denote the set of firms held by investor i by Q, the churn rate of investor i at quarter t is defined as

Here, P_j,t and N_j,i,t represent the price and the number of shares, respectively, of firm j held by institutional investor i at quarter t. The numerator of Equation (1) calculates the sum of changes in the value of holdings for firm j, during quarter t, for all firms (Q) held by an investor i. The denominator of the equation shows the sum of the average values of shareholdings during quarter t, for firm j held by an investor i. It is calculated for all firms (Q) held by investor i. We sort investors into terciles each quarter based on their average churn rate over the past four quarters (Yan & Zhang, 2009; Pathan et al., 2021). Investors falling into the bottom tercile are categorized as long-term institutional investors. We then estimate long-term institutional shareholdings as the four-quarter average proportion of outstanding shares held by these long-term institutional investors. We also consider an alternative definition of long-term institutional shareholdings, i.e. dedicated institutional shareholdings, based on the investor classifications provided by Bushee (2001).

We obtain share price turnover and the number of shares outstanding from the Center for Research in Security Prices (CSRP) daily files for all the NYSE/AMEX/NASDAQ stocks. We obtain all firm-level fundamentals from the COMPUSTAT database. Earnings announcement data are from the I/B/E/S.

Our final sample contains 53,988 observations (target price revisions) from 2000 to 2019, excluding the COVID-19 pandemic period. Table 1 reports the summary statistics for our final sample, indicating the mean, standard deviation (SD), median (p50) and quartiles for all the relevant variables. To estimate the price impact of target price revisions, we estimate the CARs from the day before the target price announcement date to the day following the target price announcement date, which is the sum of the daily abnormal returns over the event window [−1,+1]. If the target price is issued on a non-trading day, day 0 is defined as the next trading date. We compute the daily abnormal return as the raw return of the common stock minus the daily return on a weighted-characteristics-matched Size, B/M and the Momentum portfolio. We use the DGTW portfolio approach introduced by Daniel, Grinblatt, Titman, and Wermers (1997) ^[3]. The CAR is then computed by adding the daily abnormal returns for the event window [−1,+1].

The average CAR is 0.3% for the sample, which is not noticeably different from the median (i.e. 0.3%). The magnitude and the direction of target price revisions are measured using TP∆/P, which is computed as the new target price minus the target price outstanding by the same analyst divided by the pre-announcement stock price (Brav & Lehavy, 2003). The mean (median) TP∆/P is recorded as −0.3% (2.6%), indicating that the average target price revision is negative. The average long-term institutional shareholding stands at around 23%. Table 2 depicts the correlation coefficients between the multitude of variables. The correlation coefficient between the target price revisions, i.e. TP∆/P and the CAR is positive and statistically significant. At the univariate level, this evidence is consistent with the notion that a positive target price revision is associated with a stronger stock price response.

To test the relationship between long-term institutional shareholdings and the stock market response to target price revisions, we adopt the following regression model:

Where CAR(−1,+1) is the sum of the daily abnormal returns over the event window [−1,+1] ^[4] and used as our main dependent variable to measure the stock market response to target price revisions; LT (%) denotes the four-quarter average proportion of outstanding shares held by long-term institutional investors; TP∆/P is the new target price minus the target price outstanding by the same analyst divided by the pre-announcement stock price (Brav & Lehavy, 2003); and Controls is for the firm- and analyst-level control variables, such as firm size (Size), B/M, leverage, IO, external financing (Ext Fin), brokerage size, analyst coverage (Coverage), analyst experience (Experience), forecast dispersion (Dispersion), the volatility index (∆VIX), earnings forecasts (EPS Frsct), recession indicator (Recession), momentum and the short-interest ratio (SIR). All the variable definitions are provided in the appendix (Table A1). We consider all the control variables from periods before the issuance of the target price and have skipped subscripts in Equation (2) for notational ease. We also control for the firm- and year-fixed effects to account for unobserved factors across the sample firms and years.

Our main coefficient of interest is $β3$ that captures the relationship between long-term institutional shareholdings and the stock market response to target price revisions. The interaction term in Equation (2) serves two main roles: (1) previous studies suggest that institutional investors may influence analysts and analysts may favor institutional investors (Gu, Li, & Yang, 2013; Bilinski, Cumming, Hass, Stathopoulos, & Walker, 2019; Chiu, Lourie, Nekrasov, & Teoh, 2021). Thus, analysts may revise their target prices in response to changes in long-term institutional shareholdings and other long-term institutional investors may simply herd (Grinblatt, Titman, & Wermers, 1995; Sias, 2004; Chen, Huang, & Jiang, 2017). In other words, it could be possible that both analysts and the market respond to changes in long-term institutional shareholdings; (2) also, one could argue that both long-term institutional investors and analysts react to the same new set of information instead of investors’ reaction to target price revisions.

The interaction term in Equation (2) accounts for this complex simultaneous relationship between changes in target prices and long-term institutional shareholdings. It also controls for the situation where target price revisions and long-term institutional shareholdings change in response to the same information. The combined effect of the coefficients $β2$ and $β3$ captures the informativeness of target price revisions, i.e. the CARs around the target price announcements, subject to the incremental contribution of long-term institutional shareholdings. Moreover, we also control for overall IO across all the regression models to account for potential interactions between long-term institutional shareholdings, analysts’ behavior and the stock market response to target price revisions.

Table 3 reports the baseline results of a multivariate model to estimate the effect of long-term institutional shareholdings on the stock market response to target price revisions. The coefficient of TP∆/P $,$ in column (1), is positive and statistically significant, indicating that firms experience positive returns (i.e. CARs) with the increase in target prices. Our coefficient of the interaction term, i.e. LT (%) $×$ TP∆/P, in column (2), is also positive and statistically significant, suggesting that firms with more long-term institutional shareholdings experience positive returns with the increase in target prices. In particular, the coefficient of TP∆/P is 0.0141, the coefficient of LT (%) $×$ TP∆/P is 0.0713 and the mean of long-term institutional shareholdings is 0.228; therefore, the stock market response to target price revisions at the mean level of long-term institutional shareholdings is around 0.0304 (0.0141 + 0.0713 $×$ 0.228). Holding all other variables constant, a one standard deviation increase in long-term institutional shareholdings is associated with around 26% (0.0713$×$ 0.114/0.0304) increase in the CARs, supporting that our results are also economically significant.

Putting together, our findings suggest that investors of firms with more long-term institutional shareholdings respond strongly to target price revisions. Earlier studies like Bushee (2004), Gaspar et al. (2005), Chen et al. (2007), Elyasiani and Jia (2008), Elyasiani et al. (2010) and Derrien et al. (2013) document that long-term investors tend to disregard short-term performance and emphasize more on the long-term earnings potential of portfolio firms. Our results are consistent with this argument that investors find value in target price revisions amidst higher long-term institutional shareholdings.

Analyst forecast dispersion (Dispersion) is strongly associated with the stock price response to target price revisions. It is an indication that when there is high uncertainty about a firm, new target prices tend to lead to a larger stock price reaction (Elyasiani et al., 2010). Firm size (Size) has a negative relationship with the stock price response to target price revisions, which is consistent with Bradshaw et al. (2013). We also account for the external financing activity of a firm (Dechow & You, 2020) by proxying for analysts’ investment banking incentives (Ext Fin). A higher degree of external financing would mean that there are larger investment banking-related activities. There is a negative relationship between external financing (Ext Fin) and the stock price response to target price revisions. The variable “institutional ownership” (IO) measures the overall institutional shareholdings, which also proxies for the brokerage incentives, i.e. higher IO would mean that overly optimistic target prices would lead to a larger loss of reputation. We find no evidence to suggest that IO has a significant association with the stock price response to target price revisions. We also find that the CARs are smaller with an increase in the short interest ratio (SIR).

Our results support that firms with more long-term institutional shareholdings experience positive returns on target price revisions. However, endogeneity issues due to self-selection, reverse causality, potential omitted variables and measurement errors could bias our results. Aside from firm-fixed effects across all the models (to account for unobserved time-invariant factors across sample firms), we employ five additional empirical specifications to collectively address these endogeneity issues, i.e. active share, quasi-natural experimental design, entropy balancing, alternative measures and the percent bias analysis.

One could argue that long-term institutional investors may choose to invest in a particular set of firms based on their firm- or analyst-level preferences. Therefore, our results could be affected by the self-selection bias. Following Pathan et al. (2021), we split our long-term institutional shareholdings into indexers and non-indexers groups. We expect indexers to be plausibly exogenous, as they can not choose firms they invest in while following an index and non-indexers to be endogenous (Derrien et al., 2013). In other words, while active trading exists across both categories, indexers do not have the freedom to choose their portfolio firms as they are required to follow a benchmark. Thus, although long-term indexers are exogenous to firm- or analyst-level preferences, investors could still find value in target price revisions for firms with more indexers (typically long-term), making our findings robust to self-selection bias (Gloßner, 2019; Aghion, Van Reenen, & Zingales, 2013).

We re-run our regression model in Equation (2), after categorizing long-term institutional shareholdings into two groups: long-term institutional shareholdings for indexers (LT Indexer (%)) and long-term institutional shareholdings for non-indexers (LT Non-Indexer (%)) based on the “active share” measure. In particular, we follow the three-step procedure (Cremers & Petajisto, 2009; Pathan et al., 2021). First, we estimate the active share of an investor i at quarter t as the sum of the absolute difference between the portfolio weight of the stocks (⁠$weightportfolio)$ and the benchmark weight of the stocks $(weightindex$⁠), across all the stocks in the portfolio, as shown in Equation (3). Following Ye (2012), we use the S&P 500 index as the benchmark for creating active shares.

Second, we categorize long-term investors into terciles, using their active share measure and denote long-term investors falling into the bottom tercile as long-term indexers and others as long-term non-indexers. Third, we estimate their shareholdings, i.e. the proportion of outstanding shares held by long-term indexers and long-term non-indexers. We append long-term indexer and long-term non-indexer shareholdings in Equation (2). Our findings in Table 4 show that the coefficients of both the interaction terms, i.e. LT Indexer (%) × TP∆/P and LT Non-Indexer (%) × TP∆/P, are positive and statistically significant, implying that the stock price response is positive to target price revisions regardless of the self-selection bias. This finding suggests that self-selection bias is unlikely to drive our results.

We also employ a quasi-natural experimental design, using two different exogenous variations to long-term institutional shareholdings to mitigate our concerns related to reverse causality, i.e. long-term institutional investors change their portfolio holdings in anticipation of the informativeness of target price revisions. First, following He and Huang (2017), we consider financial institution mergers as an exogenous variation of long-term institutional shareholdings. The authors argue that the fundamentals of portfolio firms are often unlikely to influence the merger decisions of financial institutions, such as banks, security brokers and asset management companies. In other words, the stock market response to target price revisions is less likely to influence the merger decision of financial institutions. After the merger, the acquirer institution takes over the existing blockholdings of the target institution, resulting in exogenous variations in the portfolio holdings of the merged entity. The merged entity maintains these holdings due to liquidity and other transaction costs.

The merger could lead to a cross-holding by the merged entity if a firm is blockheld by one of the merging institutions before the merger and one of its industry rivals is blockheld by the other party to the merger. He and Huang (2017) document that this act of cross-holding is associated with increased long-term institutional shareholdings, providing us with a quasi-natural experiment to tease out the variations in long-term institutional shareholdings and test the informativeness of target price revisions. We expect the merger of financial institutions to influence the informativeness of target price revisions only through its effect on the long-term institutional shareholdings after the merger.

Following He and Huang (2017), we use the Securities Data Company’s (SDC) Mergers and Acquisitions database along with 13F filings in the financial sector. Our sample is restricted to mergers that were completed within one year of the initial announcement and we focus on targets that cease to file 13F documents within one year of the deal’s completion. We adopt a difference-in-differences regression setting, using a seven-year window around the mergers, i.e. t-3 to t+3, including the year t as the year of the merger. In particular, we consider the following regression specification:

Where Treat is an indicator variable equal to 1 if the firm is a treatment firm and 0 for a control firm. A treatment firm is defined as blockheld by one of the merging institutions during the quarter immediately before the merger announcement date and at least one of its same industry rivals is blockheld by the other party to the merger during the same pre-merger quarter. A control firm is defined as blockheld by the same merging institution (blockholding the treatment firm) during the quarter immediately before the merger announcement date and one of its same industry rivals is not blockheld by the other party to the merger during the same pre-merger quarter. Post is an indicator variable equal to 1 for the post-event period and 0 otherwise. The latter identification strategy accounts for inherent differences in managerial styles and skills of merging institutions. We also control for merger-fixed effects to consider within-merger variations.

Our main coefficient of interest is $β12$ in Equation (4), i.e. the relationship between long-term institutional shareholdings and the stock market response to target price revisions after the merger of two financial institutions across the treatment and control firms. We expect long-term institutional shareholdings of treatment firms to increase post-merger relative to control firms because of the cross-holding structure. Our findings in Column (1) of Table A2 (in appendix) support the latter assertion, highlighting an increased sensitivity of long-term institutional shareholdings to target price revisions after the merger for treatment firms compared to control firms. Further, Column (1) of Table 5 reports our results for the quasi-natural experiment based on the merger of two financial institutions. The coefficient of $Treat×Post×(TP∆P)×(LT(%))$⁠, is positive and statistically significant, supporting that investors of treatment firms respond more positively to target price revisions than control firms after the merger. This finding suggests that after an exogenous variation in the sensitivity of long-term institutional shareholdings to target price revisions, treatment firms experience a more positive response to the increase in target prices than control firms after the merger.

Second, following Gloßner (2019) and Aghion et al. (2013), we exploit the changes in S&P 500 index membership as part of an exogenous variation to IO, especially long-term institutional shareholdings. The latter studies argue that firms that are part of the S&P 500 index have higher long-term institutional shareholdings, as non-index funds often use the S&P 500 index as their benchmark, while many index funds track the performance of the bellwether S&P 500 index. The changes in the S&P 500 index are exogenous as a firm becomes a member of the index only if it crosses a certain market capitalization-based threshold, reviewed on a timely basis with some discretion of the index committee members (Gloßner, 2019). Thus, the stock market response to target price revisions is unlikely to influence a firm’s inclusion in the S&P 500 index. Therefore, we expect the S&P 500 index membership to affect the informativeness of target price revisions only through its effect on long-term institutional shareholdings. Specifically, we employ the following two-way fixed-effect linear difference-in-differences regression setting to explore variations in long-term institutional shareholdings and the stock market response to target price revisions:

Where S&P 500 is an indicator variable equal to 1 if a firm is included in the S&P 500 index (treatment firm) and 0 otherwise (control firm). We also append firm- and year-fixed effects and three additional control variables, such as average stock returns (over the past 12 months), stock liquidity and the change in the book value of assets, accounting for some of the non-randomness in the S&P 500 index membership. Our main coefficient of interest is $β7$ in Equation (5), i.e. the relationship between long-term institutional shareholdings and the stock market response to target price revisions for treatment firms after the inclusion in the S&P 500 index relative to control firms. We expect increased long-term institutional shareholdings for firms becoming part of the index. Consistent with Gloßner (2019), our findings in Column (2) of Table A2 (in appendix) support increased long-term institutional shareholdings and a positive sensitivity of long-term institutional shareholdings to target price revisions for treatment firms than control firms after the inclusion in the S&P 500 index.

Column (2) of Table 5 reports our results for the quasi-natural experiment based on the inclusion of firms in the S&P 500 index. The coefficient of $S&P500×(TP∆P)×(LT(%))$⁠, is positive and statistically significant, supporting that investors of treatment firms with increased long-term institutional shareholdings respond more positively to target price revisions than control firms after the inclusion in the S&P 500 index.

Overall, our findings remain consistent after using two quasi-natural experiments involving exogenous variations in institutional shareholdings and sensitivity to target price revisions ^[5]. Firms with more long-term institutional shareholdings experience positive returns on the increase in target prices after these exogenous shocks, suggesting that investors find value in the information contained in target price revisions, especially for firms with more long-term institutional shareholdings ^[6].

One could argue that long-term institutional investors fundamentally differ from other investors in portfolio selection. To further combat these concerns, we execute an entropy balancing approach by having a matched sample of long-term versus other investors based on firm- and analyst-level characteristics. We remove the fundamental differences through an entropy reweighting scheme (Hainmueller, 2012). For this analysis, we specify the determinants of the likelihood of long-term institutional shareholdings that satisfy the balancing conditions without losing information and maintaining efficiency for the rest of the analyses (Fei, 2022; Chen, Ma, Teng, & Wu, 2022). In particular, we refer to the weights based on the first moment of the covariates and match the means of the covariates for the long-term institutional shareholdings to the means of the covariates for other investors in the reweighted data.

After having a sample of matched observations, we re-examine the role of long-term institutional shareholdings in influencing the stock market response to target price revisions. Table 6 provides the results for the entropy balancing. Panel A of Table 6 reports the summary statistics for the original and the reweighted samples. After reweighting the sample, the standardized differences between the treatment and control firm- and analyst-level characteristics are close to 0, suggesting that the entropy balancing has corrected imbalances in the first moment of the covariates. Panel B of Table 6 reports the results for the matched (reweighted) sample. The coefficient of the interaction term, i.e. LT (%) × TP∆/P, remains positive and statistically significant, implying that investors of firms with more long-term institutional shareholdings find value in target price revisions.

The measurement error could also lead to endogeneity issues. We use an alternative measure of long-term institutional shareholdings to mitigate this concern. We classify institutional investors into “dedicated” and “transient” categories based on the investor classifications proposed by Bushee (2001). Dedicated institutional investors concentrate on a few stocks and are long-term investors rather than transient investors who are likely to be short-term. Since dedicated institutional investors are more likely to be long-term and monitor their portfolio firms, we expect investors of firms with more long-term dedicated shareholdings to find value in target price revisions.

Like long-term institutional shareholdings, we determine dedicated institutional shareholdings (the proportion of outstanding shares held by dedicated institutional investors) and include them in the regression model. We include dedicated institutional shareholdings (Dedicated (%)) and its interaction term, i.e. Dedicated (%) × TP∆/P, to examine the relationship between dedicated institutional shareholdings and the stock market response to target price revisions. We also append transient institutional shareholdings (Transient (%)) and its interaction term, i.e. Transient (%) × TP∆/P, to simultaneously investigate its impact on the stock market response to target price revisions. Table 7 provides the results for our alternative measure of long-term institutional shareholdings. The coefficient of Dedicated (%) × TP∆/P is positive and statistically significant, indicating that our findings remain robust to the inclusion of an alternative measure of long-term institutional shareholdings. Investors of firms with more dedicated institutional shareholdings respond positively to target price revisions.

We also examine the relationship between long-term institutional shareholdings and the stock market response to target price revisions for an alternative dependent variable. Table 8 presents the regression results for an alternative dependent variable. We use “Influence” as our alternative dependent variable (Loh & Stulz, 2011). A target price change is considered influential if the CAR is in the same direction as the target price change and is 1.96 times larger than expected based on the past three-months’ idiosyncratic volatility of the stock. Loh and Stulz (2011) argue that measuring the average stock price impact using the CARs does not show whether the analyst report meaningfully affects stock prices. The variable “Influence” takes the value of 1 only if the firm’s stock price is visibly impacted and 0 otherwise. Therefore, unlike an average effect, we measure whether target price revisions lead to a visible change in stock prices based on long-term institutional shareholdings.

We use a probit regression model, which predicts the probability of an influential stock price effect in response to target price revisions amidst long-term institutional shareholdings. We use the absolute target price change (|TP∆/P|) because the “Influence” variable takes the value of 1 if the change in target price is influential according to the definition of Loh and Stulz (2011). The coefficient of the interaction term, i.e. LT (%) × |TP∆/P|, is positive and statistically significant, indicating that a change in target price is associated with a higher degree of influence for firms with more long-term institutional shareholdings. Overall, our findings remain consistent even after using an alternative dependent variable.

We also perform the percent bias analysis to combat potential omitted variable bias concerns (Frank, Maroulis, Duong, & Kelcey, 2013; Cinelli & Hazlett, 2020). The latter approach estimates the percentage observations of LT (%) × TP∆/P that we need to replace with observations for which the effect of LT (%) × TP∆/P on the CARs is 0. Our findings report that we may need to replace around 54% of LT (%) × TP∆/P observations to invalidate its effect on the CARs. Thus, omitted variables are unlikely to drive our results, as they must be severe (54%) to overturn our main results.

Thus far, we have found evidence that firms with more long-term institutional shareholdings experience positive returns on target price revisions. The underlying argument assumes that all long-term institutional investors have equal monitoring incentives. However, institutions invest across several firms, having different portfolio weights. Thus, it could be possible that the positive returns on target price revisions are mainly pronounced for firms where institutions have higher monitoring incentives. In other words, we expect institutions to allocate more monitoring efforts to relatively important firms based on the holding value of the firm in their portfolio. Following Fich et al. (2015), we categorize our long-term institutional shareholdings into “motivated” and “non-motivated” groups, considering the holding values of the firms in their portfolios.

If the holding value of the firm is in the top 10% of the portfolio, we categorize such long-term institutional investors as “motivated” and others as “non-motivated”. We then find the proportion of outstanding shares held by motivated and non-motivated long-term institutional investors. Table 9 provides the results for long-term motivated (LT Motivated (%)) and long-term non-motivated institutional shareholdings (LT Non-Motivated (%)). We append LT Motivated (%) and its interaction term, i.e. LT Motivated (%) × TP∆/P, in the regression model to account for the role of monitoring incentives in influencing the CARs. Similarly, we also include LT Non-Motivated (%) and its interaction term, i.e. LT Non-Motivated (%) × TP∆/P, in the regression model. Our findings report that the positive response of investors to target price revisions is mainly restricted to firms with more long-term motivated institutional shareholdings. Firms where long-term institutional investors have higher monitoring incentives, i.e. firms in which the holding value of an institution is in the top 10% of their portfolio, experience significant positive returns on target price revisions than firms with long-term non-motivated institutional investors.

One could also argue that analyst sophistication, i.e. analyst experience, brokerage size and past accuracy, could influence investors’ responses to target price revisions. We expect target price revisions to be more informed when issued by sophisticated analysts. Thus, investors of firms with more long-term institutional shareholdings could respond more strongly to target price revisions when issued by sophisticated analysts. Therefore, we also examine the moderating role of analyst sophistication and how it affects the relationship between long-term institutional shareholdings and the stock market response to target price revisions. We measure analyst sophistication using three different proxies, i.e. the experience of the issuing analyst (Mikhail, Walther, & Willis, 1997), brokerage size (Clement & Tse, 2005; Huang, Zang, & Zheng, 2014) and past earnings accuracy (Hong, Kubik, & Solomon, 2000; Bonner, Walther, & Young, 2003).

Table 10 presents our results for the effect of long-term institutional shareholdings on the informativeness of target price revisions after considering the role of analyst sophistication. We introduce a triple interaction term to understand the moderating contribution of analyst sophistication to our baseline results. High Experience is an indicator variable equal to 1 if the analyst experience is in the top tercile and 0 otherwise. Large Broker is an indicator variable equal to 1 if the brokerage size is in the top tercile and 0 otherwise. High Past Accuracy is an indicator variable equal to 1 if the relative earnings accuracy is in the top tercile and 0 otherwise (Hong et al., 2000). The coefficient of the interaction term, i.e. LT (%) × TP∆/P, is positive and statistically significant across all the analyst sophistication proxies. Investors of firms with more long-term institutional shareholdings respond positively to target price revisions. However, they react more strongly to target price revisions when issued by sophisticated analysts, as evidenced by statistically significant and positive coefficients of our triple interaction terms.

Overall, our results suggest that firms, where long-term institutional investors have higher monitoring incentives and where target price revisions are issued by more sophisticated analysts, experience significant positive returns on target price revisions.

We also examine the role of firms’ information environment and how it affects the relationship between long-term institutional shareholdings and the stock market response to target price revisions. We reason that information contained in target price revisions is more valuable for investors when there is a high level of information asymmetry. We consider firms’ idiosyncratic volatility and the generalized probability of informed trading (PIN) as our measures of the information environment (Duarte et al., 2008, 2020; Yang et al., 2020). Table 11 presents our subsample results for the effect of the firm-level information environment.

Panel A of Table 11 examines the role of idiosyncratic volatility. The past 90-day idiosyncratic volatility is calculated three days before the target price release. The “High” idiosyncratic volatility group is defined based on the top tercile and the remaining observations are classified as the “Low” idiosyncratic volatility group. The coefficient of the interaction term, i.e. LT (%) × TP∆/P, is positive and statistically significant for the high idiosyncratic volatility group. Firms with high idiosyncratic volatility reflect more firm-specific private information incorporated into stock prices (Grossman & Stiglitz, 1980; Roll, 1988; Yang et al., 2020). Thus, our results suggest that a weaker information environment prompts investors to rely on target price revisions, especially for firms with more long-term institutional shareholdings.

Panel B of Table 11 presents the results for the subsamples created based on the PIN measure. The PIN captures the probability of an informed trade, highlighting the incorporation of more firm-specific private information into stock prices (Duarte et al., 2008, 2020). The “High” PIN group is defined based on the top tercile and the remaining observations are classified as the “Low” PIN group. We find that the coefficient of the interaction term, i.e. LT (%) × TP∆/P, is positive and statistically significant for the high PIN group. The results show that firms with more long-term institutional shareholdings experience positive returns on target price revisions amidst a weak information environment. Across both panels, we also find that the difference between the coefficients across the low and high groups is statistically significant. Investors of firms with more long-term institutional shareholdings respond positively to target price revisions when the information environment is weak.

Target prices are not the only type of information released by the analysts. They also issue quarterly earnings forecasts and stock recommendations. Generally, earnings forecasts and stock recommendations are issued for periods less than a year, unlike target prices, which are long-term in nature. Panel A of Table 12 shows the results for long-term institutional shareholdings and the stock market response to stock recommendation revisions. Our recommendation revision sample consists of 58,913 recommendation changes between 2000 and 2019. The “Rec Chg” variable is calculated as the current rating on a specific firm minus the previous rating of the firm by the same analyst. We use changes in the recommendation ratings rather than the ratings themselves because Womack (1996) finds that changes in analyst recommendations are more informative.

We observe that the coefficient of the interaction term, i.e. LT (%) × Rec Chg, is negative and statistically significant. These results are opposite to the results that we observe in Table 3 and suggest that firms with more long-term institutional shareholdings experience positive returns when analysts issue target price revisions. Panel B of Table 12 examines long-term institutional shareholdings and the stock market response to earnings forecast revisions. Our earnings forecast revision sample consists of 408,343 observations. We have a larger sample of earnings forecasts because earnings forecasts are issued quarterly. The variable “∆EPS/P” is computed as the new earnings forecast minus the previous earnings forecast by the same analyst divided by the pre-announcement stock price. The coefficient of the interaction term, i.e. LT (%) × ∆EPS/P, is statistically insignificant, implying that investors of firms with more long-term institutional shareholdings remain indifferent to earnings forecast revisions.

Overall, target price revisions contain incremental information for investors, given that target prices are more verifiable relative to earnings forecasts and stock recommendations.

This study examines whether long-term institutional shareholdings affect the stock market response to target price revisions. Although a large body of literature has investigated the informativeness of earnings forecasts and stock recommendations, the informativeness of target price revisions based on long-term institutional shareholdings remains an empirical question. Using institutional investor portfolio holdings, we find that investors of firms with more long-term institutional shareholdings respond positively to target price revisions. A one standard deviation increase in long-term institutional shareholdings is associated with around 26% increase in the CARs. Our findings remain robust to endogeneity issues and are pronounced for firms with (1) more long-term motivated institutional shareholdings, (2) high idiosyncratic volatility and (3) a high PIN. In other words, the informative role of target price revisions is noticeable for firms with a high level of information asymmetry and institutions with higher monitoring incentives. We also find that investors find more value in target price revisions when issued by sophisticated analysts.

Our results have three important implications: first, target price revisions provide incremental information to investors, highlighting portfolio value implications compared to other analyst reports such as earnings forecasts and stock recommendation revisions. Second, our findings are specifically relevant for motivated long-term institutional investors, i.e. those with significant portfolio exposure, as target price revisions are more informative in the presence of such long-term investors. Lastly, our study also spotlights the role of the firm’s information environment while accounting for the relationship between long-term institutional shareholdings and the informativeness of target price revisions. Investors find value in target price revisions when the information environment is weak.

As part of future research, it would be worth examining whether investors and analysts react to the same information using more granular transaction-level data. We also leave an empirical assessment of long-term institutional shareholdings, analysts’ top picks and the informativeness of target price revisions for future research. Finally, in the future, one could undertake an experimental study to investigate how long-term institutional shareholdings influence the informativeness of target price revisions.

The supplementary material for this article can be found online.