Do Board Secretaries Influence Management Earnings Forecasts?

The literature shows that top managers influence management earnings forecasts. Baik et al. (2011) document the positive relationship between CEO ability and the likelihood, frequency and accuracy of management earnings forecasts. Stock markets respond more strongly to earnings forecasts issued by high-ability CEOs, which suggests that management earnings forecasts communicate information regarding CEOs’ ability to the market. Cassell et al. (2013) show that retiring CEOs are more likely to issue earnings forecasts in the final year of their tenure, and those final year forecasts are more likely to contain good news. This result is stronger when CEOs receive high equity holdings and when CEOs cut final year spending in R&D and capital expenditure, implying that retiring CEOs tend to manage final year earnings forecasts for their self-serving benefits. Kwak et al. (2012) find that Chief Counsels play an important role in forecast disclosures. Firms with a General Counsel in management are more likely to issue earnings forecasts, and their forecasts tend to be less optimistic and more accurate. They further show that the influence of General Counsels on forecast disclosures is more significant when the General Counsel also holds the role of company secretary or receives higher compensation.

The literature also documents the link between management styles and voluntary disclosure of earnings forecasts. Bamber et al. (2010) find that top managers exhibit idiosyncrasies to a significant degree in earnings forecast disclosures. Managers’ styles regarding earnings forecast disclosure are associated with their career path, age cohort, military experience, education and legal background. Similarly, Brochet et al. (2011) find that firms hiring new CEOs with previous forecasting experience are more likely to issue earnings forecasts. Among firms that have historically issued earnings forecasts, they document that a temporary break in forecast issuance follows CFO turnover, and that subsequent forecasts disclosed by a newly appointed CFO tend to be less precise, due to the CFO’s inexperience in the firm or industry.

The literature indicates that managers are likely to strategically manage earnings forecasts for their self-serving benefit. Cheng and Lo (2006) report that managers who plan to buy their firm’s stocks issue more bad news forecasts to decrease the purchase price, while managers who plan to sell their firm’s stocks do not change their forecasting strategy, due to the higher litigation risk associated with insider sales. CEO insider trading has a bigger influence on disclosure than insider trading in general. Likewise, Cheng et al. (2013) find a positive (negative) relationship between forecast news and forecast precision before managers sell (buy) their firm’s stocks, suggesting that managers issue good news forecasts with high (low) precision before insider sales (purchases). They also reveal that managers are less inclined to strategically manage forecast precision when large institutional investors exist or when their forecasting behaviour poses great litigation risk, and they are more likely to manage precision when that precision cannot be assessed by investors.

Previous research supports the notion that management earnings forecast outcomes also have implications for individual managers. Trueman (1986) reports that investors use management forecast quality to evaluate managers’ ability to adjust production plans according to foreseeable changes in the business environment. Lee et al. (2012) find that inaccurate management earnings forecasts result in the replacement of CEOs in firms with poor earnings performance. This implies that boards of directors in these firms use management earnings forecast accuracy to evaluate CEOs’ ability in uncertain business environments when making decisions on CEO replacement. The relation between management earnings forecast accuracy and CEO turnover is more pronounced among firms with less entrenched CEOs.

Research in corporate governance has long focused on the effects of top managers’ demographic characteristics on corporate financial performance (Nelson 2005; Kaplan et al. 2012). Bamber et al. (2010) show that top managers’ personal characteristics, including age, education and functional experience, influence management earnings forecasts. Since board secretaries are top management team members and are legally responsible for information disclosure by law, we argue that their demographic characteristics affect their competence in improving management forecasts. Specifically, we expect that board secretaries’ legal background, accounting background and international experience could influence their firms’ earnings forecast policies.

Board secretaries with legal expertise are more sensitive to litigation risks associated with information asymmetry, and they are more likely to have more of an intention to improve, rather than stating it outright. Furthermore, since board secretaries with legal expertise are more capable of advising other managers on issues related to legislative and regulatory compliance, they can in turn improve firms’ overall risk management and information disclosure quality. In addition, the issuance of earnings forecasts requires accounting knowledge. Board secretaries with accounting expertise can better understand the financial conditions of firms and more accurately predict future earnings.

International expertise has been identified as an important managerial characteristic. Chinese returnee managers have opportunities to learn about more advanced legal institutions and superior management practices during their overseas experience (Duan and Hou 2016; Cumming et al. 2016a, b). Giannetti et al. (2013) provide evidence that board directors with international experience transfer the advanced governance and management knowledge they acquire abroad to Chinese firms, which leads to the improved corporate governance and earnings performance of those local firms. In addition, managers’ international experience is found to improve corporate performance (Carpenter et al. 2001) and corporate social performance (Slater and Dixon-Fowler 2009). Therefore, we expect that Chinese board secretaries with international work or study experience are more likely to enhance the quality of management earnings forecasts, and we propose H1 as follows.

Politically connected managers in China have been found to undermine firm performance (Fan et al. 2007) because they lack business experience and tend to pursue social objectives rather than maximising shareholder interest. In addition, political connection brings privileges, such as job security. In other words, it increases managerial entrenchment and undermines managers’ accountability (Cao et al. 2011; You and Du 2012). We therefore expect that politically connected board secretaries are less capable of issuing high-quality forecasts, and we propose H2 as follows.

Finkelstein (1992) measures the structural power of managers by the number of official titles that they hold. He notes that top managers’ ability to influence corporate decisions is contingent on their power. We argue that board secretaries holding other senior executive titles are able to access more resources to provide high-quality management earnings forecasts. Board secretaries often take on an additional post of board director, CFO or another senior executive role (e.g. vice-president).⁵

In terms of the additional roles, first, board directorship could align the interests of board secretaries with shareholders and consequently enhance forecast quality. Organisation theory suggests that the consolidation of management and board roles promotes unity of command and leads to organisational effectiveness (Pfeffer and Salancik 1978; Donaldson and Davis 1991; Boyd 1990, 1995). Second, board secretaries who are also CFOs have greater financial insight and can use this information advantage to issue more accurate forecasts. Third, board secretaries who hold other senior executive positions, such as presidents and vice-presidents, tend to be extensively involved in daily management and operation. Such experience and power help to enhance management earnings forecasts. We hereby propose H3 as follows:

Producing high-quality forecasts can be costly, and board secretaries may choose to withhold information instead. Jensen and Meckling (1976) and Datta et al. (2005) point out that managerial equity holdings can alleviate agency problems and facilitate managers’ incentive alignment with investors (Jensen and Meckling 1976; Datta et al. 2005). Nagar et al. (2003) find that CEOs tend to provide more frequent management earnings forecasts when the interests of CEOs and shareholders are aligned. Thus, we argue that board secretaries’ equity holdings motivate them to reduce information asymmetry by issuing high-quality management earnings forecasts.

We hand-collect data on board secretary characteristics from their biographical information in the China Securities Market and Accounting Research (CSMAR) database for all listed firms on the Main Board of the Shenzhen Stock Exchange between 2001 and 2012.⁶ We obtain management earnings forecast data from the RESSET database. We use both quarterly and annual management earnings forecasts. Following Anilowski et al. (2007), we require that a quarterly management forecast is issued prior to the earnings announcement date of the forecast fiscal quarter and no more than 90 days prior to the end of that fiscal quarter. For annual forecasts, we require that the forecast is issued prior to the earnings announcement date of the forecast fiscal year and no more than 730 days prior to the end of that fiscal year.

Table 1A summarises the definition and data sources of all variables. After merging the management forecast data with data on board secretary characteristics, corporate governance and firm characteristics, the final sample contains 6840 firm-year observations to conduct management forecast occurrence and frequency analyses. The sample contains 5362 firm-years that have at least one management forecast issuance; these are used to analyse management forecast precision. When we examine management forecast accuracy and optimism, the sample size is further reduced to 4818 firm-years due to the availability of data on actual earnings and stock prices.

To examine the impact of board secretary characteristics on the properties of management earnings forecasts, we use the following regression model:where i indexes firm and t indexes year. Forecast is one of the management forecast properties: Occurrence, Frequency, Precision, Accuracy and Optimism. Occurrence is a dummy variable which is equal to one if there is at least one management forecast issuance in a given firm-year, and zero otherwise. Frequency is the total number of forecasts issued by a firm in a given year. Precision is coded as three if the forecast is a point forecast, two if it is a range forecast, one if it is an open-interval forecast and zero if it is a qualitative forecast. Accuracy is the absolute difference between the management forecast of net profits and the actual net profits, scaled by the market value of tradable shares one day prior to the forecast release date, and then multiplied by −1. A less negative value of Accuracy indicates higher management forecast accuracy. Optimism is coded as 1 if the management earnings forecast is higher than the actual net profits (optimistic bias), 0 if it is equal to the actual net profits, and −1 if it is less than the actual net profits (pessimistic bias). When firms issue multiple forecasts in a year, we take their average values to construct Precision, Accuracy and Optimism. We use point, range and open-interval forecasts to construct Accuracy and Optimism. For range forecasts, the midpoint of the range is used as the management forecast estimate (Kross et al. 2011). For open-interval forecasts, the value provided in the open-interval forecast is taken as the management forecast estimate (Yang 2012; Cassell et al. 2013).

We examine whether and how the properties of management earnings forecasts are affected by board secretary characteristics, including professional background, political connections, role duality and equity holdings. We construct three background variables for Expertise. Law is a dummy variable, equal to one if a board secretary holds a law licence, and zero otherwise. Accounting is a dummy variable, equal to one if a board secretary holds a professional certificate in accounting, and zero otherwise. ForeignExp is coded as 1 if a board secretary has work or study experience in foreign countries, 0.5 if she/he has work experience in Hong Kong, Macau or Taiwan, and 0 otherwise. Following Li et al. (2008), we use Chinese Communist Party membership to measure the political connection of board secretaries. PartyMeb is a binary indicator suggesting the existence of a board secretary’s political connection, which equals one if the secretary is a member of the Chinese Communist Party, and zero otherwise.

We also examine three forms of board secretary duality. Duality is proxied by Board_Duality, CFO_Duality and Mag_Duality. Board_Duality is a dummy variable, equal to one if a board secretary is a member of the firm’s board of directors, and zero otherwise. CFO_Duality is a dummy variable, equal to one if a board secretary serves as the firm’s CFO, and zero otherwise. Mag_Duality is a dummy variable, equal to one if a board secretary holds an additional non-accounting senior executive position in the firm, and zero otherwise. Finally, we examine the effect of the equity holdings of board secretaries on management forecast properties. The equity holdings of board secretaries are calculated as the change in value of the secretary’s stockholdings given a 1% increase in the firm’s stock prices (Bergstresser and Philippon 2006; Burns and Kedia 2006). EquityHold is defined as the natural logarithm of one plus the equity holdings.

In our regressions, we control for other board secretary demographic characteristics. Female is a dummy variable, equal to one if a board secretary is female, and zero otherwise. Age is the age of a board secretary. Ln(Tenure) is the natural logarithm of the number of days that a board secretary has held this position. If the board secretary was replaced during a firm-year, we use the characteristics of the board secretary holding the position at the year-end to construct these variables.

We also control for corporate governance characteristics, as prior research finds evidence suggesting that better corporate governance improves management forecast quality (Karamanou and Vafeas 2005; Ajinkya et al. 2005). We consider several corporate governance factors. BIndep is the proportion of independent directors on the board. BMeet is the number of board meetings. Duality is a binary variable which equals one if a CEO also serves as the chair of the board, and zero otherwise. InstHold is the proportion of shares held by institutional investors. GovHold is the proportion of shares held by the Chinese government. In addition, the model controls for a number of firm-level characteristics.⁷ Since Ajinkya et al. (2005) find a negative relation between ownership concentration and management forecast properties, we control for the concentration ratio (CR), which is measured as the proportion of stocks held by the firm’s ten largest blockholders. The price-to-book ratio (P/B) is included to account for a firm’s growth opportunities; Bamber and Cheon (1998) find that growth opportunities serving as an indicator of proprietary costs are related to firms’ forecasting choices. Ln(Assets) is calculated as the natural logarithm of a firm’s total assets and is included in our model because firm size has been found to affect forecast disclosures (Kasznik and Lev 1995; Baginski and Hassell 1997). We also control for Crisis, which is a dummy variable with a value equal to one if the forecast year is during the crisis period of 2007–2008, and zero otherwise. Moreover, the literature on voluntary information disclosure suggests that firms in different industries are exposed to different litigation costs, proprietary costs and information asymmetry and thus manage their forecasting policies using different strategies (Kasznik and Lev 1995; Bamber and Cheon 1998). Therefore, we include Industry dummies, defined as the first two digits of the firm’s Global Industry Classification Standard (GICS) code.

We estimate Eq. (1) using a Probit model if the dependent variable is Occurrence, a Poisson model if the dependent variable is Frequency, and an ordinary least squares (OLS) regression if the dependent variable is Precision, Accuracy or Optimism. For the regressions of Precision, Accuracy and Optimism, we add an additional control variable, Ln(Horizon), which is the natural logarithm of the number of days between the forecast release date and the actual earnings announcement date. Ln(Horizon) is included because the literature has found a negative relation between forecast horizon and forecast precision and accuracy (Pownall et al. 1993; Baginski and Hassell 1997; Xu 2010). If multiple forecasts are issued in a firm-year, we take the average horizon to generate this variable. On the right side of Eq. (1), the independent variables, except Ln(Horizon), are lagged by 1 year relative to the dependent variable to mitigate reverse causality. We winsorize all continuous variables at 1% and 99%.

Table 2 presents descriptive statistics on management earnings forecasts by classifying the management forecasts into annual and quarterly forecasts. The difference in mean and median tests shows that there is no obvious difference in forecast precision between the two types of forecasts. The annual forecasts are substantially higher (lower) than the quarterly forecasts in terms of forecast optimism and horizon (forecast accuracy), according to both the difference in mean and the difference in median.

Table 3 presents descriptive statistics for the variables used in our empirical analyses. The mean of Occurrence is 0.654, 0.720 and 0.787 for annual forecasts, quarterly forecasts and overall forecasts, respectively, suggesting that 65.4, 72 and 78.7% of firm-year observations have at least one forecast issuance for annual forecasts, quarterly forecasts and overall forecasts, respectively. The mean forecasting frequency is 0.850, 1.680 and 2.533 for annual forecasts, quarterly forecasts and overall forecasts, respectively, which indicates the prevalence of multiple forecast disclosures by Chinese firms. The 25th percentile of forecasting precision is 2.00, 2.00 and 1.75 for, respectively, annual forecasts, quarterly forecasts and overall forecasts, suggesting a high propensity of Chinese firms to issue quantitative forecasts. The median forecasting accuracy is −4.818 for overall forecasts, while the mean accuracy for overall forecasts surprisingly reaches −13.340. These results reveal that there is a high discrepancy in management forecast accuracy and that the forecasts issued by certain firms or in certain years could be extremely unreliable. The mean Optimism (−0.026 for overall forecasts) is a negative value, consistent with the idea that management forecasts, on average, are pessimistically biased in China.

We find that lots of board secretaries have a certificate in accounting or serve as a Party member in our sample even though overall it is rare for secretaries to have a law licence or have either foreign experience. In addition, more than half of board secretaries hold another senior role in Chinese listed firms; most hold another non-accounting management position. Over 75% of board secretaries do not have any equity holdings in their firms.

We examine the effects of board secretaries’ professional background, political connections, role duality and equity holdings on management forecast occurrence by using annual forecasts, quarterly forecasts and overall forecasts. The results are shown in Table 4. Panels A, B and C show results for testing of H1, H2, H3 and H4 by using annual forecast occurrence, quarterly forecast occurrence and overall forecast occurrence, respectively. Models 5, 10 and 15 report the results for testing all hypotheses together using annual forecasts, quarterly forecasts and overall forecasts, respectively. The coefficients of Law are significantly positive in models 6, 10, 11 and 15, indicating that board secretaries with legal expertise tend to issue quarterly earnings forecasts. The coefficients of Accounting are significantly positive in models 1, 5, 6, 10, 11 and 15, indicating that board secretaries with accounting expertise are more likely to issue both annual and quarterly earnings forecasts. The coefficient of ForeignExp is significantly positive in models 6 and 10, indicating that board secretaries with international experience are more likely to issue quarterly annual forecasts. The results generally support the prediction that board secretaries with more expertise are more likely to issue management forecasts, supporting our H1a, 1b and 1c. The coefficient on PartyMeb is significantly negative in models 2, 5, 7 and 10 at the 5% level, which provides evidence of the adverse effect of political connections on both annual and quarterly earnings forecast issuance and supports our H2. In models 3, 5, 8, 10, 13 and 15, we show that the coefficients on Board_Duality, CFO_Duality and Mag_Duality are positively significant. The results suggest that board secretaries sitting on the board are more likely to issue annual earnings forecasts and that board secretaries with a dual CFO or other senior manager title are more likely to issue both annual and quarterly earnings forecasts. Overall, board secretaries with a second senior position generally have more power and more inside information to issue earnings forecasts; this supports our H3a, 3b and 3c. Finally, the coefficient on EquityHold is significantly positive in models 4, 5, 9, 10, 14 and 15, at the 1% level, which indicates that board secretaries with stockholdings in their firms are more likely to voluntarily disclose both annual and quarterly earnings forecasts for the sake of outside investors. The evidence supports our H4.

Table 5 reports the test results for the effects of board secretaries’ professional background, political connections, dual roles and equity holdings on management forecast frequency. Panels A, B and C show results for testing of H1, H2, H3 and H4 by using annual forecast frequency, quarterly forecast frequency and overall forecast frequency, respectively. It shows that Law, Accounting and ForeignExp are significantly and positively associated with forecast frequency in models 1, 5, 6, 10, 11 and 15, which indicates that the professional expertise of board secretaries improves earnings forecast frequency, supporting H1a, 1b and 1c. The coefficient of PartyMeb is significantly negative in all models. The results suggest that political connections decrease board secretaries’ willingness to make frequent annual and quarterly forecast disclosures and support H2. Furthermore, the coefficient estimates on Board_Duality, CFO_Duality and Mag_Duality are significantly positively related to forecast frequency in all models, indicating that board secretary role duality leads to more frequent annual and quarterly forecast issuance. This evidence supports H3a, 3b and 3c. The coefficient estimates on EquityHold are significantly positive in all models, which support H4, about the positive incentive role of board secretaries’ stockholdings for both annual and quarterly forecast frequency.

Table 6 provides the test results for the effects of board secretaries’ professional background, political connections, role duality and equity holdings on management forecast precision. Panels A, B and C show results for testing of H1, H2, H3 and H4 by using annual forecast precision, quarterly forecast precision and overall forecasts precision, respectively. The coefficient estimates on Board_Duality, CFO_Duality and Mag_Duality are significantly positive on both annual and quarterly forecast precision in all models. This result suggests that board secretaries with role duality employ their expanded power and superior resources to generate more precise earnings forecasts, which supports H3a, 3b and 3c. In addition, the result shows that EquityHold significantly improves the precision of quarterly earnings forecasts and supports H4.

As for the accuracy of management forecasts, we examine its association with board secretary characteristics from two ways: the value (Accuracy) and the sign (Optimism). In Table 7, we present the results for the associations between board secretary characteristics and forecast Accuracy. Panels A, B and C show results for testing of H1, H2, H3 and H4 by using annual forecast accuracy, quarterly forecast accuracy and overall forecast accuracy, respectively. The coefficient estimate on Accounting is significantly positive in model 11. For the effect of duality on forecast accuracy, we find that the coefficients on CFO_Duality and Mag_Duality are significantly positive in models 3, 5, 13 and 15, which supports our H3b and 3c, that also holding an additional senior role in a firm contributes to board secretaries’ concentrated power and information advantage, with which they can produce more accurate estimates about future earnings. Furthermore, we report a significantly negative relation between Equityhold and Accuracy in all models, supporting H4.

We also examine whether and how board secretary characteristics affect the way in which management earnings forecasts are directionally biased, and we report our results in Table 8. Panels A, B and C show results for testing of H1, H2, H3 and H4 by using annual forecast optimism, quarterly forecast optimism and overall forecast optimism, respectively. The coefficient estimates on Board_Duality and Mag_Duality are significantly positive in models 9, 13 and 15. The results reveal that board secretaries who also serve as board members or non-accounting senior executives generally issue more optimistically biased quarterly earnings forecasts, which literally means that their forecasts tend to exceed actual earnings. The result suggests that board secretaries with higher hierarchical status within firms, as evidenced by their dual senior positions, overestimate firm profitability and issue more optimistic future earnings estimates. In addition, we document a significantly positive coefficient on EquityHold in all models, suggesting that board secretaries with stockholdings tend to issue optimistically biased forecasts. This could result from the self-serving tendency of board secretaries.⁸

Although the duties of board secretaries have been specified in the Guidance for the Articles of Listed Company since 1997,⁹ they were not covered in Company Law until 2005. In 2005, the revised Company Law regulated that a listed company must appoint a board secretary after 1 January 2006. We therefore examine whether the legislative change strengthens the impact of board secretary characteristics on management earnings forecasts. We construct Post, a dummy variable equal to one if the year is 2006 or onward, and zero otherwise and interact each board secretary characteristics variable with Post. We regress forecast occurrence, forecast frequency, forecast precision, forecast accuracy, optimism and bad news forecast, respectively, on these variables and present the results in the unpublished appendix. The untabulated results indicate that the promulgation of the revised company law of 2005 has little effect on the impact of board secretaries on management earnings forecasts, presumably because the duties of board secretaries had been already clearly defined by China Securities Regulatory Commission (CSRC) in Guidance for the Articles of Listed Company promulgated in 1997.

Managers issue more bad news forecasts and downward guidance than other forecasts because this could help decrease their legal costs and reputational costs (Skinner 1994; Heflin et al. 2016). Since board secretaries by law have important legal and regulatory duties, they could face more litigation risks arising from irresponsible forecast disclosure. Thus, the characteristics of board secretaries should have an impact on bad news issuance.¹⁰

Following Heflin et al. (2016), we classify a management forecast as a bad (good) news forecast if it is lower (higher) than the most recent consensus analyst forecast, where the consensus forecast is identified as the median analyst forecast issued within 180 days prior to the management forecast disclosure date. Analyst forecast data are obtained from the CSMAR database. However, only annual analyst earnings forecast data are available in the database. Therefore, our analyses on bad news versus good news management forecasts rely on annual analyst earnings forecasts only. We study three different properties of bad news management forecasts, following Heflin et al. (2016): (1) Pr(Bad news) is a dummy variable which is equal to one if the firm issues at least one bad news management earnings forecast in a given year, and zero otherwise; (2) #Bad news is the number of bad news management forecasts issued by a firm in a given year; and (3) %Bad news is the fraction of management earnings forecasts issued by a firm in a given year that convey bad news. In line with Kross et al. (2011) and Heflin et al. (2016), our sample for bad news versus good news management forecast analysis includes firm-years both with and without management forecast issued.¹¹

Table 9 presents the regression analysis results for the impact of board secretary characteristics on bad news and downward guidance. Panels A, B and C report Probit regression results on the occurrence of bad news management forecasts, Poisson regression results on the frequency of bad news management forecasts, and OLS regression results on the fraction of bad news management forecasts, respectively. The coefficient of Accounting is significantly positive in models 1 and 11, which suggests that board secretaries with accounting expertise are more likely to issue bad news and downward guidance and issue a higher fraction of bad news forecasts. The coefficient estimate on PartyMeb is negatively significant in all models. The results suggest that political connections decrease board secretaries’ willingness to issue bad news and downward guidance forecast in terms of occurrence, frequency and fraction. The coefficients of Board_Duality, CFO_Duality, Mag_Duality and EquityHold are significantly positive, offering further evidence that board secretaries with dual roles and equity holdings are associated with more frequent issuance of bad news and downward guidance management forecasts. Overall, board secretaries with professional expertise, dual roles and equity holdings are more likely to issue bad news and downward guidance. Their willingness to issue bad news and downward guidance decreases when they have political connections.

The literature reveals that management earnings forecasts are used by investors to evaluate managers’ abilities to adapt future production plans in response to foreseeable changes in the business environment (Trueman 1986; Baik et al. 2011; Yang 2012). Lee et al. (2012) further show that management earnings forecast errors increase the likelihood of poorly performing CEOs being replaced. These studies generally suggest that management earnings forecasts provide an important signal of managerial ability. Thus, in this paper we argue that firms use management earnings forecasts to evaluate board secretaries’ performance, because they have a duty to reduce information asymmetry between corporate insiders and outside investors by ensuring full and accurate forecast disclosure. Since investors may favourably evaluate firms with frequent forecast issuance and high forecast precision and accuracy, these firms that benefit from improved forecast disclosure are likely to reward their board secretaries with higher compensation. Therefore, we predict that board secretaries’ pay is positively related to the occurrence, frequency, precision and accuracy of management forecasts. To verify our prediction, we estimate regressions where the dependent variable is Ln(Pay), calculated as the natural logarithm of the sum of a board secretary’s salary and bonus. The independent variables are lagged by 1 year relative to the dependent variable.

Table 10 presents the regression results for management earnings forecasts and board secretary pay.¹² Panel A reports the results by using the OLS estimator. In models 1–6, we regress board secretary Pay on each of the forecast property variables, respectively, and then in model 7 we include all properties of management forecasts if there is at least one forecast issuance, i.e., the value of Occurrence is one. We report in model 1 that the coefficient on Occurrence is significantly positive at the 1% level, implying board secretaries that issue forecasts receive higher compensation than secretaries without forecast issuance. In models 2–6, which consider the situation where at least one forecast is disclosed by board secretaries in a given year, we find the coefficients on Frequency, Precision, Accuracy and %BadNews consistently significantly positive, consistent with the argument that pay is significantly higher for board secretaries that issue high-quality forecasts. The coefficient of Optimism is significantly negative, which suggests that board secretaries who issue less optimistically biased forecasts receive higher pay. The results also reveal the link between board secretary characteristics and pay. For example, board secretaries with foreign experience, dual roles and stock holdings receive more compensation, on average, while pay is significantly lower for secretaries with Party membership.