Multiple Blockholdings and Auditor Behavior

Multiple Blockholdings and Auditor Behavior K.K. Raman 1, Chunlai Ye 2, Lin-Hui Yu 3 ABSTRACT This paper examines whether multiple blockholdings are associated with auditor behavior. Using a sample of S&P 1500 firms, we examine whether multiple blockholdings of institutional investors play a role in audit fees and audit report lags. First, we find that while auditors spend more time preparing audit reports for firms monitored by multiple blockholders, auditors do not charge higher fees. Second, we find that fee pressure exists in firms with more experienced multiple blockholders as manifested by having industry knowledge and accumulated monitoring experiences. Finally, we find that while auditors work longer hours for firms with more experienced multiple blockholders, they charge lower fees when firms report higher returns on assets and have higher leverage, suggesting that firms monitored by more experienced institutional investors know how to bargain more effectively based on financial positions. Overall, this paper provides findings that are relevant to the current capital market, in which institutional investors and multiple blockholders have become more influential. 1 The University of Texas at San Antonio, San Antonio, USA 2 Texas A&M University-Corpus Christi, Corpus Christi, USA 3 National Taiwan University, Taipei City, Taiwan 1

I. Introduction An agency problem arises when there is a separation of ownership and control in a company. Prior research has identified several mechanisms to improve corporate governance and to reduce agency problems. For example, Shleifer and Vishny (1986) show that a large stake in ownership held by outside shareholders is a potential solution, as large investors are more incentivized to monitor. Jensen and Meckling (1976) and Watts and Zimmerman (1983) indicate that an audit by an independent auditor is another mechanism for mitigating agency conflicts and should reduce managers opportunistic behavior. Because both large shareholders and auditors play a pivotal role in monitoring management and in reducing agency costs, prior research has shown interest in investigating the interaction between these two parties. For example, Cassell et al. (2018) find that audit fees are higher for firms with more institutional investors. Similarly, Raghunandan and Rama (2007) suggest that the demand for a high quality audit is higher for firms with higher levels of outside blockholder ownership and document that audit committee meeting frequency is higher for firms with higher levels of blockholdings. While prior research enhances our understanding of how institutional investors affect audit behavior, it largely ignores the fact that institutional investors serve as common owners in multiple firms. Whether cross-holding of institutional investors affects auditor behavior warrants further analysis, as institutional investors and multiple blockholders have become more influential in the current capital market. Institutional investors own 70%-80% of equity of U.S. publicly traded firms and institutional cross-holding has rapidly increased in recent years (Ben David et al. 2016; Azar et al. 2018). The increase in institutional holdings and crossownership increases the concern about whether multiple blockholdings affect the pricing power of companies (Solomon 2016; Azar et al. 2018). Motivated by this void in the literature and the 2

importance of this concern, we investigate whether multiple blockholdings of institutional investors affects auditor behavior audit fees and audit report lags and study whether two characteristics of multiple blockholders industry knowledge and investment horizon explain these associations. In this study, we examine S&P 1500 firms from 2003 to 2017 and adopt the approach proposed by Kang et al. (2018) to obtain the predicted and residual number of blockholdings by a firm s largest institutional investor. This approach is advantageous in that it controls for an institutional investor s fund size. Thus, a large residual from the prediction model reflects the extra number of blockholdings an institutional investor holds and not merely the fact that this institutional investor has more resources to manage its portfolio so that it has more blockholdings. Next, we identify a firm as monitored closely by a multiple blockholder if the residual blockholdings of the firm s largest institutional investor is higher than the median of the sample. We first examine whether multiple blockholdings are associated with audit pricing and propose two competing arguments. On one hand, if multiple blockholders acquire more privileged access to management in their portfolio firms, they may assess managers true effort and monitor managers behavior more effectively. Thus, they may rely less on monitoring through financial statement numbers but directly communicate and influence corporate behavior (Hölmstrom 1979; Ke et al. 1999). This argument predicts that the demand for a high quality audit is lower for firms whose investors cross-hold substantially in other companies. On the other hand, it is also possible that because of time constraints and limited attention, multiple blockholders are less likely to develop close ties with management and need to monitor through accounting numbers, increasing the demand for a high quality audit (Kempf et al. 2017; Liu et al. 2017). Furthermore, institutional investors with blockholdings may better understand the benefits 3

of a high quality audit and thus be more willing to pay higher fees for auditors assurance service (Kane and Velury 2004; Raghunandan and Rama 2007). The latter argument leads to a positive association between multiple blockholdings and audit fees. Next, we examine whether multiple blockholdings are associated with auditor effort. The analysis of the association between multiple blockholdings and auditor behavior is not complete if we ignore auditor effort. If both audit fees and auditor effort increase for firms whose institutional investors own multiple blockholdings, it indicates that auditors are compensated for the extra effort requested by multiple blockholders. This evidence is consistent with the view that multiple blockholders value high-quality audits and are willing to pay for the extra effort. On the other hand, if audit fees remain unchanged or decrease and audit report lags increase, it indicates that auditors face fee pressure and are unable to raise audit fees to compensate for extra work. This finding may raise the concern about whether multiple blockholders have a negative impact on auditors long-term profitability. Empirical results show that while multiple blockholdings are positively associated with audit report lags, auditors do not charge higher audit fees. In other words, while multiple blockholders may be associated with higher litigation risk, increasing the workload of auditors, auditors do not raise audit fees based on the additional working hours. This evidence is consistent with the views that common ownership may have a negative influence on pricing in an industry (Solomon 2016; Azar et al. 2018) and that common ownership increases audit fee pressure. To find out whether characteristics of institutional investors explain the associations above, we examine whether industry knowledge and investment horizons help multiple blockholders obtain price concessions. We find that audit report lags are longer for firms 4

monitored by multiple blockholders, regardless of industry knowledge. We also find marginal evidence to suggest that auditors are able to raise fees only for firms whose largest institutional investors do not have industry knowledge. Regarding investment horizons, we find that audit report lags are longer only for firms whose largest institutional investor is a long-term investor; however, we do not find evidence to suggest that auditors are able to charge higher fees accordingly. Taken together, our findings suggest that auditors react differently to different characteristics of multiple blockholders. Specifically, when auditors exert greater effort for firms whose largest investor has a diversified portfolio, they are compensated for additional working hours. When auditors work additional hours in the presence of multiple blockholders with industry knowledge and long investment horizons, auditors make fee concessions and do not charge higher fees. We further examine whether a firm s financial position plays a role in the association between multiple blockholdings and auditor behavior, focusing on firms leverage and profitability. Consistent with prior literature, we first document that audit fees are positively associated with leverage and negatively associated with profitability. However, auditors react differently toward leverage and profitability in the presence of multiple blockholders. We find that firms with more experienced multiple blockholders pay lower audit fees when they report higher leverage and higher return on assets. However, these associations do not hold for firms with less experienced multiple blockholders. Our results are consistent with the view that more experienced multiple blockholders perceive high leverage to be associated with lower agency costs, and thus are less willing to pay for high quality audits. Our results also indicate that while the presence of more experienced multiple blockholders increases auditors litigation risk, auditors perceive the risks to be lower when companies are more profitable. 5

Finally, we perform sensitivity tests. Our results are robust to alternative measures of multiple blockholdings. First, we re-identify a firm as monitored closely by a multiple blockholder if a firm s Top 5 institutional investors have residual blockholdings higher than the median of the sample. We repeat the analyses and focus on the Top 10 institutional investors. Our inferences remain unchanged. Furthermore, to address the concern that our results are driven by firms with different reporting schedules, we restrict our analysis to firms having the same reporting deadline of 75 days after the fiscal year ends. While our sample size decreases substantially, the results remain constant. The remainder of this article is organized as follows: Section 2 reviews the literature and develops the hypotheses; Section 3 describes the sample and the empirical measures; Section 4 presents the empirical results; Section 5 provides additional analyses; Section 6 presents the sensitivity tests; and Section 7 concludes the paper. II. Literature Review and Hypothesis Development 2.1 Auditors, institutional investors, and blockholdings Jensen and Meckling (1976) suggest that agency problems arise when there is a separation of ownership and control and when managers act at the expense of shareholders. Shareholders can reduce agency costs through monitoring activities. However, it is costly to gather and analyze information and to take actions to influence management. Therefore, small investors are less interested in monitoring managers but free-ride on large investors monitoring efforts. As the ownership structure becomes more diffused, the free-rider problem becomes more severe (Grossman and Hart 1980; Shleifer and Vishny 1986; Ang et al. 2007). Different from small shareholders with diffused ownership, institutions with large holdings are more 6

willing to devote resources and effort in monitoring activities as they have more at stake and enjoy higher net benefits from active monitoring (Chen et al. 2007). Consistent with this view, prior literature documents that large shareholders actively monitor management and subsequently enhance firms performance, as manifested by higher stock prices, higher profitability, and better innovative performance (Chen et al. 2007; Aghion et al. 2013). Prior literature also finds that large shareholders influence executive compensation, increasing payfor-performance sensitivity and decreasing the level of compensation (Hartzell and Starks 2003), and that they are more likely to fire CEOs for poor performances (Kang et al. 2018). While a large body of literature documents how large shareholders monitor management and influence corporate behavior, the role or auditing in institutional investors monitoring activities has received much less attention. In fact, auditing represents a monitoring mechanism and auditors adjust the scope of the audit based on the agency conflicts between management and shareholders. Agency costs are associated with audit risk and auditor business risk, which in turn should affect auditor behavior. Given the importance of auditing, some prior research examines whether blockholdings are positively or negatively associated with audit fees; however, the empirical evidence remains mixed and warrants further analyses (Hay et al. 2006). On one hand, institutional investors with blockholdings are sophisticated investors and are more likely to understand that credible financial statements reduce agency costs and provide governance benefits, and thus to value auditor assurance services (Kane and Velury 2004; Raghunandan and Rama 2007). In addition, if auditors perceive that institutional investors with blockholdings are associated with investor activism and increase audit engagement risk, they may charge higher audit fees (Cassell et al. 2018). These views support the arugment that blockholdings are positively associated with audit fees. On the other hand, as blockholders have privileged access 7

to managers; they can directly communicate with management and intervene in corporate decisions and thus rely less on monitoring through accounting numbers (Hölmstrom 1979; Ke et al. 1999). In addition, as agency conflict is lower in the presence of investors with blockholdings, the audit engagement risk should decrease (Francis and Wilson 1988; Ang et al. 2007). These latter two arguments predict a negative association between blockholdings and audit fees. Taken together, whether the presence of institutional investors with blockholding is positively or negatively associated with audit fees is an empirical question. 2.2 Multiple blockholdings and audit fees While prior literature enhances our understanding of how institutional blockholders affect audit fees, it often ignores the fact that institutions may have large blockholdings in multiple firms at the same time. This fact is important as multiple blockholdings affect institutional investors incentives and effort to monitor management, which in turn should affect auditor behavior. To address this void in the literature, we examine whether multiple blockholdings are associated with audit fees and audit report lag. Regarding audit fees, we discuss two competing arguments. First, multiple blockholdings enable institutions to develop privileged access to the management teams and gain insider information from their portfolio companies. Because institutional investors improve their monitoring skills through the monitoring experience and information obtained from multiple blockholdings, they may be more effective in direct monitoring and rely less on monitoring through audited financial statements (Kang et al. 2018). This argument predicts a negative association between multiple blockholdings and audit fees. On the other hand, multiple blockholdings may distract institutional investors and make it less likely that they will develop close ties with management. Prior literature provides evidence that 8

institutional investors face time constraints and have a limited attention span. For example, Kempf et al. (2017) find that when institutional investors monitoring attention is diverted, managers react to the looser monitoring and engage in value-destroying acquisitions. Similarly, Liu et al. (2017) document that distracted institutional investors are less effective in disciplining directors and are associated with poor board oversight. Thus, to the extent that institutional investors are distracted and less likely to directly monitor managers and intervene in corporate decisions, they may rely more on monitoring through accounting numbers. Stronger reliance on financial statements increases the value of auditing services, leading to a positive association between multiple blockholdings and audit fees. Taken together, whether multiple blockholdings are positively or negatively associated with audit fees is an empirical question and warrants further investigation. We also examine whether the characteristics of multiple blockholders play a role in audit pricing. We focus on institutional investors industry knowledge and investment horizons, and expect that the association between blockholdings and audit fees varies across the characteristics of blockholders. Prior literature documents that institutional owners gain industry knowledge and improve monitoring effectiveness through multiple blockholdings in other portfolio firms operating in the same industry. Multiple investments in the same industry reduces information asymmetries between institutions and their portfolio firms, because multiple blockholders learn more about the private probability of an average company s success in the industry and are more able to identify managers true effort (Eisenhardt 1989; Cressy et al. 2007). Furthermore, firms in the same industry share commonalities, enabling multiple blockholding institutional investors to accumulate industry-specific information relevant to monitoring firms. In other words, multiple blockholding institutional investors can get access to and compare financial 9

policies and performance information across firms in the same industry, increasing the effectiveness of their monitoring. Thus, to the extent that multiple blockholding institutional investors develop and improve their industry-specific monitoring skills, they are more likely to directly monitor managers and rely less on monitoring through financial statements, decreasing their willingness to pay auditors. Regarding investment horizons, we follow prior literature and predict that multiple blockholders may monitor investee firms more effectively when they hold shares for a longer horizon. A longer investment horizon provides not only the incentives to monitor but also opportunities to improve monitoring skills (Gaspar et al. 2005; Chen et al. 2007; Ramalingegowda and Yu 2012; McCahery et al. 2016). In addition, institutions are more likely to develop close ties and exert greater influence on managers when they hold shares for a longer time. Thus, compared to multiple blockholders with short investment horizons, those with long investment horizons should prefer monitoring managers directly to monitoring them through accounting information. As investors monitoring abilities and reliance on monitoring through financial statements vary across investment horizons, we expect that the willingness of institutional investors to pay for the auditing service varies with investment horizons. Overall, we expect that multiple blockholdings play a role in audit pricing and that the characteristics of multiple blockholders explain the variation in audit fees. On the basis of the discussion above, we propose the following set of hypotheses: H1. There is an association between multiple blockholdings and audit fees. H1a: Characteristics of institutional investors are associated with the association between multiple blockholdings and audit fees. 10

2.3 Multiple blockholdings and auditor effort Next, we examine whether multiple blockholdings is associated with auditor effort. This analysis is important as it is possible that audit fees and auditor effort do not change in the same direction (i.e., audit fees and auditors effort exhibit an increase or a decrease independently when a firm s largest institutional investor is a multiple blockholder). Specifically, it is possible that auditors make pricing concessions for firms with multiple blockholders because multiple blockholders have enhanced capabilities and incentives to monitor audit pricing and demand lower audit fees. However, this price concession may not necessarily suggest that auditors exert low effort. In fact, auditors may still exert enough or even higher effort when firms are monitored by multiple blockholders, as engagement risk may increase in the presence of sophisticated institutional investors (Cheng et al. 2010; Cox and Thomas 2006; Cassell et al. 2018). To proxy for auditor effort, we follow prior literature (Knechel and Payne 2001; Ettredge et al. 2006; Knechel et al. 2009; Blankley et al. 2013) and examine whether the audit report lag differs for firms whose largest institutional investor has multiple blockholdings. Similar to H1, we also predict that audit report lags vary with characteristics of multiple blockholders. Based on the discussion above, we predict the following: H2. There is an association between multiple blockholdings and audit report lags. H2a: Characteristics of institutional investors are associated with the association between multiple blockholdings and audit report lags. III. Sample and Empirical Measures 3.1 Sample construction 11

We construct our sample using the following data sources: (1) institutional holding information from Thomson Reuters (13F); (2) financial statement data from COMPUSTAT; (3) audit, going concern opinion, and internal control weakness information from Audit Analytics; and (4) price information from the Center for Research in Security Prices (CRSP). We begin with all firms included in the S&P 1500 index and firms for which we have the necessary data to derive the control variables. We exclude firms in financial institutions (SIC 6000-6999) because these firms face a different reporting environment. Our final sample consists of 14,022 firm-year observations, covering 1,740 firms during the 2003 2017 period. Table 1 presents the annual and industry distributions of our sample. 3.2 Measure of multiple blockholdings Consistent with Kang et al. (2018), we measure the monitoring effectiveness of multiple blockholdings using the residual approach. First, we estimate the following regression using institution-quarter observations: LNRAW_BLOCKHOLDINGjq = β0 + β1inst_total_mvholdingjq + εjq (1) Subscripts j and q indicate institutional investors and quarters, respectively. The dependent variable, LNRAW_BLOCKHODING, is the natural logarithm of the number of firms for which an institutional investor simultaneously owns more than 5% of the firm s shares and is set to zero if the institutional investor, j, does not have any blockholdings in firms in quarter q. The independent variable, INST_TOTAL_MVHOLDING, is measured as the average market value of equity managed by institutional investor j during the previous four quarters, scaled by the Consumer Price Index in 2000. 12

Next, we calculate the residual of multiple blockholdings at the investee firm-year level. For each S&P 1500 constituent, we identify firm i s largest (Top 1), five largest (Top 5), and ten largest (Top 10) institutional investors based on the holding information at the end of year t 1 and use the residuals from Equation (1) to construct TOP1_INTOWN, TOP5_INTOWN, and TOP10_INTOWN. Specifically, TOP1_INTOWN is measured as the residual blockholding number owned by firm i s largest institutional investor. TOP5_INTOWN (TOP10_INTOWN) is measured as the weighted average of the residual blockholding number owned by firm i s Top 5 (Top 10) institutional investors. The weights are assigned based on the ownership proportion, measured as institutional investor q s ownership in firm i divided by the total ownership owned by Top 5 (Top 10) institutional owners in firm i. 3.3 The empirical model for H1 and H2 To examine whether multiple blockholdings are associated with audit fees and audit report lags, we estimate the following regression: DEPit = α0 + β1multi_holdit + α1weightit + α2avg_retit + α3avg_churnit +α4intownit + α5lnatit + α6assetgrowthit + α7bmit + α8ageit +α9goingconit + α10decit + α11invrecit + α12big4it + α13specialistit + α14lossit + α15stdcfoit + α16bus_segit+ α17foreign_segit + α18mait + α19leverageit + α20roait + industry and year fixed effects + ε (2) The dependent variable is alternatively defined as LNAF (H1) and AULAG (H2). LNAF is the natural logarithm of total audit fees and AULAG is the natural logarithm of days between fiscal year-end date and the audit report date. MULTI_HOLD is an indicator variable that equals one if the residual blockholding number owned by firm i s largest institutional investor 13

(TOP1_INTOWN) is higher than the median of the sample and zero otherwise. A positive (negative) coefficient on MULTI_HOLD indicates that audit fees are higher (lower) and audit report lag is longer (shorter) for firms whose largest institutional investor is a multiple blockholder. Next, we include variables that control for characteristics of institutional investors. WEIGHT is the value of investment in firm i relative to the total holding value managed by firm i s largest institutional investor. INTOWN is the percentage of ownership owned by institutional investors. AVG_RET is the average buy-and-hold portfolio return for firm i s largest institutional investor over four quarters in year t. AVG_CHURN is the average churn rates of firm i s largest institutional investor over four quarters in year t. The churn rate is measured as aggregate purchases plus aggregate sales minus the absolute value of net flows scaled by lagged equity asset holding value (Gaspar et al. 2005). We also include control variables for previously documented firm characteristics likely to affect auditor behavior (Simunic 1980; Francis and Simon 1987; Geiger and Rama 2003; Whisenant et al. 2003; Hay et al. 2006; Ghosh and Pawlewicz 2009; Ettredge et al. 2014). LNAT is the natural logarithm of total assets. ASSET_GROWTH is the asset growth rate and is measured as the difference between assets in year t and year t 1, scaled by lagged total assets. BM, the book-to-market ratio, is measured as the book value of common equity divided by the market value of common equity. AGE is the natural logarithm of the firm s age in years. GOINGCON is an indicator variable that equals one if a firm has received going concern modifications in year t and zero otherwise. DEC is an indicator variable that equals one if a firm has a December year-end and zero otherwise. INVREC is the amount of inventory and receivables scaled by total assets. BIG4 is an indicator variable that equals one if a firm is audited by a Big 4 auditor and zero otherwise. SPECIALIST is an indicator variable that equals 14

one if a firm is audited by a specialist auditor and zero otherwise. We define a specialist auditor as the one who collected the highest total fees in an SIC 2-digit industry. LOSS is an indicator variable that equals one if a firm experiences a loss in year t and zero otherwise. STDCFO is the standard deviation of cash flow from operations scaled by total assets in the previous 5 years. BUS_SEG is the number of business segments. FOREIGN_SEG is the number of foreign segments. MA is an indicator variable that equals one if a firm has merger and acquisition events and zero otherwise. LEVERAGE is measured as total debt scaled by total assets. ROA is net income scaled by total assets. Finally, we include industry and year fixed effects to control for systematic variation in audit fees across industries and years, and cluster standard errors by firm. All continuous variables are winsorized at the 1% and 99 % levels to mitigate the effect of outliers. 3.4 The empirical models for H1a and H2a To examine whether characteristics of multiple blockholders are associated with auditor behavior, we estimate the following regression models: DEP = α0 + β1same+ β1diff + Controls + ε (3) DEP = α0 + β1long+ β1short + Controls + ε (4) In Equations (3) and (4), we replace MULTI_HOLD with proxies for the following characteristics of firm i s largest institutional investor: SAME, DIFF, LONG, and SHORT. We first set these four indicator variables to zero if a firm s largest institutional investor is not a multiple blockholder (that is, MULTI_HOLD equals zero). Next, we check whether a firm s multiple blockholder (that is, MULTI_HOLD equals one) has the characteristic we are interested in. Specifically, SAME equals one if firm i s largest institutional investor has at least one additional blockholding in another firm in the same SIC 2-digit industry during the previous 15

three years and zero otherwise. DIFF equals one if firm i s largest institutional investor has no blockholding in other firms in the same SIC 2-digit industry during the previous three years and zero otherwise. LONG equals one if firm i s largest institutional investor has continuously served as another portfolio firm s blockholder for at least a year during the previous three years and zero otherwise. SHORT equals one if firm i s largest institutional investor has not continuously served as another portfolio firm s blockholder for at least a year in the previous three years and zero otherwise. A significant coefficient on SAME, DIFF, LONG, and SHORT supports H1a and H2a, in which we predict that characteristics of institutional investors explain the variation in audit fees and audit report lag. For example, a positive (negative) coefficient on SAME indicates that audit fees are higher (lower) and the audit report lag is longer (shorter) for firms whose largest institutional investor is a multiple blockholder and has additional blockholdings in other firms in the same industry. IV. Empirical results 4.1 Descriptive statistics Table 2 presents the descriptive statistics. Columns (2) and (3) show the characteristics for firms whose largest institutional investor owns a substantial number of blockholdings in the market (that is, MULTI_HOLD =1) and for firms whose largest institutional investor does not greatly cross-hold (that is, MULTI_HOLD =0), respectively. Column (4) reports the difference between these groups. We find that audit fees are lower and audit report lags are longer in firms with multiple blockholders. Interestingly, while Hay et al. (2006) summarize the large body of literature and find that the dominant determining factor in audit pricing is client firm size, we do 16

not find that these two groups are different in size. We further find that firms that are monitored by multiple blockholders have lower leverage, lower profitability, higher book-to-market ratios, and fewer segments. In addition, they are younger and are less likely to be audited by reputational auditors. These univariate tests lend some support to the view that multiple blockholders invest in firms where their monitoring helps the manager maximize shareholder value (Brav et al. 2008). [Insert Table 2 Here] 4.2 Regression results for H1 and H2 Table 3, Panel A presents the regression results from examining the association between multiple blockholdings and auditor behavior. Column (1) reports the results for the firm characteristic control variables only when the dependent variable is audit fees (LNAF). Consistent with prior literature, we find that audit fees are higher for firms owned by institutional investors (INTOWN = 0.147, t-statistic = 2.34), with larger assets (LNAT = 0.551, t- statistic = 59.22), and that are older (AGE = 0.058, t-statistic = 3.05), have a December fiscal year end (DEC = 0.093, t-statistic = 3.73), have a higher percentage of receivables over total assets (INVREC = 0.925, t-statistic = 8.88), appoint reputational auditors (BIG4 = 0.148, t- statistic = 3.46 ; SPECIALIST = 0.059, t-statistic = 3.18), report a loss (LOSS = 0.087, t-statistic= 3.68 ), have more business segments (BUS_SEG = 0.012, t-statistic = 3.27), and have more foreign segments (FOREIGN_SEG = 0.020, t-statistic = 6.11). Audit fees are lower for firms with a higher growth rate in assets (ASSETGROWTH = 0.138, t-statistic = 5.43) and with higher profitability (ROA = 0.770, t-statistic = 5.35). 17

Column (2) reports the regression results when we include MULTI_HOLD and variables controlling for characteristics of institutional investors. The empirical results reveal that the coefficient on MULTI_HOLD is not significant and that the coefficient on WEIGHT is significantly positive (0.360, t-statistic = 1.98). This evidence indicates that audit fees are not different between firms whose largest institutional investor cross-holds substantially in other firms and firms whose largest institutional investor has a limited number of blockholdings; however, audit fees increase when the firm accounts for a larger fraction of the largest institutional investor s portfolio. This finding is consistent with the view that the willingness to pay auditors varies with the relative importance of stocks in institutional investors portfolio, and thus firms impose different fee pressures on the auditor (Fich et al. 2015). Next, we examine whether multiple blockholdings are associated with auditor effort and present the results in Table 3, Panel B. Similar to the analysis above, we first run a regression controlling for the firm characteristic variables only and use audit report lag as a dependent variable (AULAG). Column (1) reveals that audit report lags are longer for firms that have a higher growth rate in assets (ASSETGROWTH = 0.030, t-statistic = 3.16), have less growth potential (BM = 0.081, t-statistic = 8.55), receive a going-concern opinion (GOINGCON = 0.195, t-statistic = 4.72), have a December fiscal year end (DEC = 0.031, t-statistic = 3.93), report a loss (LOSS = 0.017, t-statistic= 2.01 ), have M&A transactions (MA = 0.011, t-statistic = 2.10), and have higher leverage (LEVERAGE = 0.089, t-statistic = 4.03). Audit report lags are shorter for firms with more institutional investors (INTOWN = 0.074, t-statistic = 3.58), with larger assets (LNAT = 0.047, t-statistic = 15.46), and with higher profitability (ROA = 0.176, t-statistic = 3.88). 18

Column (2) shows the results of the regression of both the test variable, MULTI_HOLD, and control variables on audit report lags. The empirical results reveal that the coefficient on MULTI_HOLD is significantly positive (0.016, t-statistic = 3.28), indicating that the audit report lag is longer for a firm whose largest institutional investor owns a substantially larger number of blockholdings in the market. Similar to the results in Panel A, we also find that the coefficient on WEIGHT is significantly positive (0.102, t-statistic = 1.92). Taken together, the results in Panels A and B indicate that auditors exert greater effort and spend more time preparing audit reports for firms with multiple blockholders; however, auditors do not charge higher fees for additional working hours. Our evidence is in line with the concern that common ownership may influence pricing in an industry (Azar et al. 2018) and that multiple blockholdings may hurt auditors profitability. [Insert Table 3 Here] 4.3 Regression results for H1a and H2a Table 4 reports the results for H1a and H2a, which propose that the association between multiple blockholdings and auditor behavior varies with two characteristics of institutional investors. We first examine whether financial investors industry knowledge plays a role in audit fees and estimate Equation (3), in which we decompose MULTI_HOLD into same- and differentindustry multiple blockholding indicators (that is, SAME and DIFF). Panel A, Columns (1) and (2) report the results when the dependent variable is defined as LNAF and AULAG, respectively. We find that, while DIFF is marginally significant and positive (0.040, t-statistic = 1.68), SAME is not significant (0.015, t-statistic = 2.94) in the audit fee model. We further find that both SAME and DIFF (SAME = 0.015, t-statistic = 2.94; DIFF = 0.020, t-statistic = 2.59) are 19

significantly positive in the audit report lag model. These results indicate that auditors exert greater effort both for firms whose largest institutional investors cross-hold in the same industry and for firms whose largest institutional investors do not have additional blockholdings in the same industry; however, auditors are compensated for additional working hours only for firms whose multiple blockholders do not cross-hold in the same industry. These results show that the fee pressure of institutions multiple blockholdings comes primarily from the information advantage that multiple blockholders acquire through investing in other firms in the same industry. Next, we focus on investment horizons and estimate Equation (4), in which we decompose MULTI_HOLD into long- and short-term multiple blockholding indicators (that is, LONG and SHORT). We find that both LONG and SHORT are insignificant (LONG= 0.006, t- statistic = 0.41; SHORT = 0.040, t-statistic = 0.71) when the dependent variable is LNAF. However, LONG is significantly positive (0.016, t-statistic = 3.27) when the dependent variable is AULAG. These results indicate that while auditors do not adjust audit fees according to investment horizons, they work longer hours for their reports for firms whose largest institutional investor is a long-term investor. These results show that the fee pressure of multiple blockholdings comes primarily from the information advantage that multiple blockholders acquire through accumulated investment experience. Collectively, the results in Table 4 support H1a and H2a. We find that the association between multiple blockholdings and auditor behavior varies with characteristics of multiple blockholders. Our evidence is consistent with the notion that the information advantage acquired through investing in other firms in the same industry and through investing for a long horizon 20

enables multiple blockholders to place fee pressure on auditors, and that auditors spend more time preparing reports in the presence of more experienced multiple blockholders. [Insert Table 4 Here] V. Additional Analyses Exploring the Association among Financial Positions, Multiple Blockholdings, and Auditor Behavior To enhance the understanding of the association between multiple blockholdings and auditor behavior, we extend prior literature and examine whether a firm s financial position plays a role in the associations we document in Section 4. Regarding financial position, we focus on firm leverage and profitability and expect that the association between multiple blockholdings and audit fees and auditor effort varies across firms financial position. Jensen (1986) indicates that because firms issuing bonds need to make periodic payments, bonds reduce managers control over free cash flow and their incentives to engaging in nonoptimal projects. In support of Jensen (1986), empirical studies suggest that the use of debt mitigates agency costs (Harris and Raviv 1991; Bathala et al. 1994) and that auditors react and adjust audit fees based on firms leverage ratio (Gul and Tsui 2001; Gul and Tsui 1997). Following these studies, we predict that because a higher debt ratio is associated with lower agency costs, multiple blockholders should be less concerned about managers opportunistic behavior and be less likely to demand high-quality audits. In addition, auditors face lower inherent risks and may be more willing to reduce audit fees. Next, we extend two studies: (1) Brav et al. (2008), who suggest that institutional investors serve a monitoring role to reduce agency costs and to improve ex-post operating performances and (2) Cassell et al. (2018), who document that auditors engagement risk is 21

higher in the presence of institutional owners. We examine whether better performance, which is associated with greater monitoring effort from institutional owners and lower litigation risk, reduces audit fees and auditor effort. To examine whether the association between multiple blockholdings and audit fees varies with a firm s financial position, we add interaction terms MULTI_HOLD LEVERAGE and MULTI_HOLD ROA into Equation (2); SAME LEVERAGE, SAME ROA, DIFF LEVERAGE and DIFF ROA into Equation (3); and LONG LEVERAGE, LONG ROA, SHORT LEVERAGE, and SHORT ROA into Equation (4). Table 5, Panel A presents the empirical results from estimating Equation (2). Column (1) reports the results for the audit fee model. Consistent with prior literature, we find that audit fees are higher for firms with higher leverage and lower profitability. We find that, different from the insignificant coefficient on MULTI_HOLD in the audit report lag model as reported in Table 3, in Column (2), MULTI_HOLD is significantly positive. Furthermore, we find that the interaction terms MULTI_HOLD LEVERAGE and MULTI_HOLD ROA, are both significantly negative. These results indicate that higher profitability and higher leverage mitigate the positive association between multiple blockholdings and audit fees. These results are consistent with the notion that multiple blockholders perceive debts as mitigating agency costs (Harris and Raviv 1991; Bathala et al. 1994) and thus firms with multiple blockholders are less willing to pay higher fees in the presence of reduced agency costs. The results are also consistent with the view that while auditors perceive the litigation risk to be high for firms with multiple blockholders, the risk decreases with profitability. Column (2) reports the results for the audit report lag model. We find that the interaction term MULTI_HOLD ROA is significantly negative, confirming that 22

the fee reduction associated with higher profitability in firms with multiple blockholders is driven by reduced working hours. Next, we investigate whether the association between the characteristics of institutional investors and auditor behavior varies with a firm s financial position. Table 5, Panel B presents the empirical results from estimating Equation (3). Regarding industry knowledge, we find that SAME LEVERAGE and SAME ROA are both significantly negative when the dependent variable is LNAF, and that SAME ROA is significantly negative when the dependent variable is AULAG. However, the results are weaker when we focus on institutional investors without such knowledge. Specifically, we find that only DIFF LEVERAGE is significantly negative in the audit fee model and that DIFF LEVERAGE and DIFF ROA are both insignificant. Panel C presents the empirical results from estimating Equation (4), in which we focus on investment horizons. Similar to the findings for industry knowledge in Panel B, we find that LONG LEVERAGE and LONG ROA are both significantly negative in the audit fee model, and that LONG ROA is significantly negative in the audit report lag model. Taken together, these results are consistent with the findings in Table 4, which suggest that the fee pressure of multiple blockholdings comes primarily from industry expertise and accumulated monitoring experience. These results indicate that auditors perceive profitable firms monitored by multiple blockholders with industry expertise and long monitoring experience to have lower inherent risks, and thus auditors exert less audit effort and charge lower audit fees. The results also indicate that more experienced multiple blockholders perceive higher leverage to be associated with less severe agency problems, and thus their willingness to pay for auditing decreases, pushing their investee firms to negotiate with auditors more aggressively. However, 23

we do not find evidence to suggest that auditors exert lower effort in response to reduced audit fees, as auditors are still responsible to creditors. [Insert Table 5 Here] VI. Robustness Checks 6.1 Alternative measure of multiple blockholdings In this section, we re-assess the association between multiple blockholdings and auditor behavior using two alternative measures of multiple blockholdings. We redefine the largest institutional investor in Equations (2), (3), and (4) using TOP5_INTOWN and TOP10_INTOWN ( discussed in Section 3.2). Table 6, Panel A presents the results from estimating our regression models based on a firm's Top 5 institutional investors and Panel B based on the Top 10 institutional investors. Similar to the results in Table 3, we find that MULTI_HOLD is insignificant when the dependent variable is LNAF and significantly positive when the dependent variable is AULAG. We also repeat the analyses in Table 3 using the original value of TOP1_INTOWN, the residual of multiple blockholdings at the investee firm-year level. Untabulated results reveal that TOP1_INTOWN is insignificant in the audit fee model and is significantly positive in the audit report lag model. In other words, the results based on an indicator variable (Table 3) are similar to those based on original residuals. Overall, the results in Table 6 confirm our main inferences that auditors exert greater effort for firms with multiple blockholders; however, we do not find evidence to suggest that auditors charge higher audit fees for additional working hours. [Insert Table 6 Here] 24

6.2 Alternative sample firms with same reporting schedules Our main sample consists of firms with different filing periods, which raises the concern that different filing schedules may bias the results. To address this concern, we restrict our sample to firms with the same filing schedules. First, we delete non-accelerated filer firms, as they are exempted from SOX Section 404 (b) compliance and are required to file annual reports within 90 days after the fiscal year ends. Next, we delete accelerated filer firms that do not have information on the opinion of internal control weakness. We further delete accelerated filer firms with market capitalization greater than $700 million if their fiscal year ends after December 15, 2006, because these firms have a deadline of 60 days. After these procedures, our subsample consists of firms having the same filing period of 75 days. Table 7 presents the results using the subsample. We again find that MULTI_HOLD is insignificant in the audit fee model and is significantly positive in the audit report lag model. Overall, the sensitivity tests using the subsample suggest that our main findings are unlikely to be driven by a subset of firms with different reporting schedules. [Insert Table 7 Here] VII. Conclusion This paper examines the association between multiple blockholdings and auditor behavior. In particular, we examine whether auditors charge different fees and exert different levels of effort for firms monitored by institutional investors with substantial cross-holdings in the market. Using a sample of S&P 1500 firms from 2003 to 2017 and the residual approach developed by Kang et al. (2018), we find that while auditors spend more time preparing audit 25

reports for firms monitored by multiple blockholders, auditors do not charge higher fees. This finding is consistent with the notion that cross-holding enables institutional investors to acquire more governance-relevant information and monitoring experience, which gives them an information advantage with which to bargain more aggressively with auditors. Next, we examine whether the characteristics of multiple blockholders explain the association between multiple blockholdings and auditor behavior. We find that firms with moreexperienced multiple blockholders place great fee pressure on auditors. Specifically, auditors do not charge higher audit fees for longer working hours for clients monitored by multiple blockholders (1) who also serve as blockholders in other firms in the same industry and (2) who hold shares for a long horizon. In other words, industry expertise and the accumulated monitoring experience of multiple blockholders create fee pressure for auditors in the fee negotiation process. We further find that firms with more experienced multiple blockholders pay lower audit fees when they report higher returns on assets and higher leverage. However, we do not find such evidence for firms with less-experienced multiple blockholders. These results provide additional evidence to suggest that firms monitored by institutional investors with more experience know how to bargain more effectively based on financial position. 26

References Aghion, P., J. Van Reenen, and L. Zingales. 2013. Innovation and institutional ownership. American Economic Review 103 (1): 277 304. Ang, J. S., R. A. Cole, and J. W. Lin. 2007. Agency costs and ownership structure. The Journal of Finance 55 (1): 81 106. Azar, J., M. C. Schmalz, and I. Tecu. 2018. Anticompetitive effects of common ownership. The Journal of Finance 73 (4): 1513 1565. Bathala, C. T., K. P. Moon, and R. P. Rao. 1994. Managerial ownership, debt policy, and the impact of institutional holdings: An agency perspective. Financial Management 23 (3): 38 50. Ben David, I., F. A. Franzoni, R. Moussawi, and J. Sedunov. 2016. The granular nature of large institutional investors. Working paper. Available at: https://www.nber.org/papers/w22247. Blankley, A. I., D. N. Hurtt, and J. E. MacGregor. 2013. The relationship between audit report lags and future restatements. AUDITING: A Journal of Practice & Theory 33 (2): 27 57. Brav, A., W. E. I. Jiang, F. Partnoy, and R. Thomas. 2008. Hedge fund activism, corporate governance, and firm performance. The Journal of Finance 63 (4): 1729 1775. Cassell, C. A., M. S. Drake, and T. A. Dyer. 2018. Auditor litigation risk and the number of institutional investors. AUDITING: A Journal of Practice & Theory 37 (3): 71 90. Chen, X., J. Harford, and K. Li. 2007. Monitoring: Which institutions matter? Journal of Financial Economics 86 (2): 279 305. Cheng, A., H. He Huang, Y. Li, and G. Lobo. 2010. Institutional monitoring through shareholder litigation. Journal of Financial Economics 95 (3): 356 383. Cox, J. D., and R. S. Thomas. 2006. Does the plaintiff matter-an empirical analysis of lead 27

plaintiffs in securities class actions. Columbia Law Review 106: 1587 1640. Cressy, R., F. Munari, and A. Malipiero. 2007. Playing to their strengths? Evidence that specialization in the private equity industry confers competitive advantage. Journal of Corporate Finance 13 (4): 647 669. Eisenhardt, K. M. 1989. Agency theory: An assessment and review. The Academy of Management Review 14 (1): 57 74. Ettredge, M., E. E. Fuerherm, and C. Li. 2014. Fee pressure and audit quality. Accounting, Organizations and Society 39 (4): 247 263. Ettredge, M. L., C. Li, and L. Sun. 2006. The impact of SOX Section 404 Internal Control Quality Assessment on audit delay in the SOX era. AUDITING: A Journal of Practice & Theory 25 (2): 1 23. Fich, E. M., J. Harford, and A. L. Tran. 2015. Motivated monitors: The importance of institutional investors portfolio weights. Journal of Financial Economics 118 (1): 21 48. Francis, J. R., and D. T. Simon. 1987. A test of audit pricing in the small-client segment of the US audit market. The Accounting Review 62 (1): 145 157. Francis, J. R., and E. R. Wilson. 1988. Auditor changes: A joint test of theories relating to agency costs and auditor differentiation. The Accounting Review 63 (4): 663 682. Gaspar, J.-M., M. Massa, and P. Matos. 2005. Shareholder investment horizons and the market for corporate control. Journal of Financial Economics 76 (1): 135 165. Geiger, M. A., and D. V Rama. 2003. Audit fees, nonaudit fees, and auditor reporting on stressed companies. AUDITING: A Journal of Practice & Theory 22 (2): 53 69. Ghosh, A., and R. Pawlewicz. 2009. The impact of regulation on auditor fees: Evidence from the Sarbanes Oxley Act. AUDITING: A Journal of Practice & Theory 28 (2): 171 197. 28