The Impact of Managers’ Background Characteristics on Disclosure of Earnings Forecasts
Name: Wang Huiyan Student number: 11450495
Thesis supervisor: Sikalidis, Alexandros Date: 24 June 2018
Word count: 12,554
MSc Accountancy & Control, variant Control Amsterdam Business School
Statement of Originality
This document is written by student Huiyan Wang who declares to take full responsibility for the contents of this document.
I declare that the text and the work presented in this document is original and that no sources other than those mentioned in the text and its references have been used in creating it.
The Faculty of Economics and Business is responsible solely for the supervision of completion of the work, not for the contents.
Abstract
This paper examines the impacts of managers’ background characteristics on earnings forecasts in a sample of 532 Chinese listed non-financial companies from 2013 to 2016. Followed, I provide hypothesis from managers’ gender, age, education level, and other career experiences, and use descriptive statistics, correlation analysis, and multivariate model to study their impacts on attributes of earnings forecasts (news and accuracy). According to the results, I find that there exist relationships between managers’ background characteristics and attributes of earnings forecasts. Specifically, managers’ gender has no impacts on attributes of earnings forecasts since most of managers are male; managers’ age is significantly negatively related to accuracy and news of earnings forecasts; managers’ education level is significantly positively related to accuracy of earnings forecasts; other working experience, especially financial background, is significantly positively related to accuracy of earnings forecasts.
Keywords: Managers’ background characteristics, earnings forecast, upper echelons
Contents
1. Introduction...6
2. Theory development... 9
2.1. Upper Echelons theory...9
2.2. Information Asymmetry...10
2.3. Signaling theory...11
2.4. Agency theory...12
3. Literature review...13
3.1. Literature review on earnings forecasts...13
3.1.1 Literature review on disclosure incentives of earnings forecasts...13
3.1.2 Literature review on disclosure strategies of earnings forecasts...16
3.1.3 Literature review on disclosure consequences of earnings forecasts.17 3.2. Literature review on managers’ background characteristics...18
3.2.1 Research object...19
3.2.2 Background characteristics... 20
4. Hypothesis development...21
4.1. Gender and earnings forecasts... 21
4.2. Age and earnings forecasts... 22
4.3. Education level and earnings forecasts...22
4.4. Other career experiences and earnings forecasts... 23
5. Research design... 25 5.1. Research model...25 5.2. Independent variables... 25 5.3. Dependent variables...26 5.4. Control variables...27 6. Results...29 6.1. Descriptive statistics... 29
6.1.1 Descriptive statistics of managers’ background characteristics...29
6.1.2 Descriptive statistics of attributes of earnings forecasts... 30
6.1.3 Descriptive statistics of control variables...31
6.2. Correlation result and analysis...32
6.3. Regression results and analysis...34
6.4. Robustness test...41 7. Conclusion... 42 7.1 Research conclusion... 42 7.2 Research limitation... 43 References...45 Appendices...50
1. Introduction
The “separation of ownership and control” in the modern ownership corporation is closely related to agency problem (Jensen & Meckling, 1976). Based on hypothesis of economic man, managers have incentives to make decisions that will not be aligned with shareholders’ interests. Information disclosure is a bridge connecting investors and companies, which facilitates investors to know about operating performance and financial situation of companies, and reduces information asymmetry. Therefore, companies can reduce costs of capital and increase company’ s value through information disclosure. As a type of information disclosure, earnings forecast is considered as an important tool in financial market and is researched widely. Hirst, Koonce, & Venkataraman (2008) confirm that earnings forecasts can establish or alter market earnings expectations, preempt litigation concerns, and influence their reputation for transparent and accurate reporting. Therefore, earnings forecasts of listed companies are essential to capital market and investors. However, articles researching on attributes of earnings forecasts are few. More researches focus on incentives of earnings forecasts disclosure and behavior, which decreases effectiveness of earnings forecasts as foundation of decision-making. As such, I will research and discuss earnings forecasts deeply based on information asymmetry and signaling theory, to reduce agency problem between management and shareholders and other stakeholders.
In addition, although some literature concentration on earnings forecasts of listed company management, heterogeneity of managers is missed. Most of prior literature ignore personal factors of managers. They believe that rational managers will make similar decisions facing similar situations, in accordance with neoclassical economic theory that people will not impact firms’ strategy by personal differences. However, this assumption is contradictory to people in reality. Hambrick & Mason (1984) first mention that top managers can partly reflect and predict organizational outcomes-strategic choices and performance. It indicates that people can have an impact on organization since their different personal experience and value will
influence their choices, especially under complicated circumstances. Previous research have proved that managers’ choices will be affected by age, education level and working experience. For example, older managers are more conservative in financial aspect (Bertrand & Schoar, 2002). Ambiguous information disclosure is more unacceptable to people who have prior accounting working experience (Holland, 1997).
That I emphasis on background characteristics of top management team, rather than on psychological factors, is based on the perspective of upper echelons by Hambrick (2007). It is hard or even unable to measure the personal cognitive bases, values and perceptions of top managers because managers are probably quite reluctant to participate in psychological batteries. However, research and application of management selection and development needs observable background data on managers. Therefore, even if demographic indicators can be influenced by external factors than psychological measures, they can still show important relationship with managers’ behavior and can stand test of time.
The development of capital market in China is slow and regulation construction is not mature. Securities Regulatory Commission promoted annual report early loss system in 1998, and earnings forecasting system was developed, including early loss, early warning, early increase in 2002. Until 2007 did Chinese Securities Regulatory Commission enact and implement information disclosure specification document “Administrative Measures for the Disclosure of Information of Listed Companies”, hoping to ensure fair, truthful, complete information disclosure of Chinese listed companies. However, there are differences in respond to disclosing earnings forecasts among listed companies, which drive our research questions: Is there any relationship between managers’ background characteristics and disclosure of earnings forecasts? Do managers have a preference when forecasting earnings? To answer these questions, I will research on earnings forecasts of Chinese listed companies to provide ideas for investors.
background characteristics, and discuss their effects on two dimensions of earnings forecasts systematically. Since there is little research on personal influence on earnings forecasts in China, this paper fills a gap of research on relationship between characteristics of entity of information disclosure and characteristics of information disclosure. It also expands research perspective of managers’ background characteristics and information disclosure theory.
Moreover, I hope to prove that heterogeneity of managers’ background characteristics will impact earnings forecasts of listed companies, which can interpret why there exist differences in information disclosure among listed companies in security market. This paper emphasizes the importance of factor of top management team in earnings forecasts of listed companies, which helps companies focus more on managers selection and their roles.
The article is organized as follows. Part 2 presents theory development. In the third section, I provide literature review regarding earnings forecasts disclosure and managers’ background characteristics, the primary concentration in my study. Part 4 shows hypothesis build on prior research and theory foundation. Part 5 presents my research design, including data collection process, sample, models, and variables definitions. Part 6 presents the results and part 7 concludes.
2. Theory development
2.1. Upper Echelons theoryNeoclassical economics had been the basis theory of research on economics for a long time. However, Simon pointed out the disadvantages of neoclassical economics and analyzed two main weak assumptions in 1940s: the assumption that future changes are in line with current situation; the assumption that the possible results of strategies and alternatives are known. They are not possible to happen in real life, which make neoclassical economics lose fundamental base. With the development of behavioral economics and psychology later on, people started to research the effects of top management’ s thinking ways on strategy choice based on bounded rationality (March & Simon, 1958). Upper echelons theory was developed first in 1984 by Hambrick & Mason and became one of the most popular topic in recent 30 years.
Upper echelons theory indicates that executives' experiences, values, and personalities (Finkelstein & Hambrick, 1990) greatly influence their interpretations of the situations they face and then affect their choices. Managers will make different choices based on their own cognitive base and values even if facing the similar operating environment. We can know that managers’ background characteristics can interpret and predict organization outcome to some extent, so organization reflects its top managers’ characteristics. Therefore, managerial background characteristics have important influences on strategy choices and performance levels of organizations. Upper echelons theory has important influences: 1. Most of research focus on top management teams because managers work together to have impacts in the organizations (Cyert & March, 1963). 2. Managers’ background characteristics are considered as indicators to represent their cognitive and value.
However, it is hard to measure personal values, personalities and cognitive even if they have great influence on decision-making. Based on old version, Hambrick (2007) update the upper echelons theory and introduce the demographic characteristics of executives, which can be used as proxy of managers’ cognitive mode since they have
measurable advantages. In this paper, I will apply demographic characteristics of managers, including gender, age, education level, and other career experiences, to examine their influences on earnings forecasts.
2.2. Information Asymmetry
Information asymmetry is developed by George, A. Michael, & Joseph (1970) due to existence of private information. It stems from transactions when one party has more information than the other in the market. Information asymmetry may lead to low efficiency of transactions or even market failure. In addition, information asymmetry may cause “adverse selection” (Buyers will choose lower price rather than higher quality because of lacking information, so that low-quality sellers are left in the market) and “moral hazard” (A person takes more risks because someone else can bears the cost of those risks).
Information asymmetry is widespread in security market, influencing effectiveness operation of security market. It may distort the resource allocation function of the securities market, especially when different investors have different professional background and different information. Information disclosure can avoid this problem, however, with lowering the asymmetry between investors and analysts and organization by weakening information advantages of managers, and improve organizational transparency. As for listed companies, however, information disclosure will generate disclosure cost, including direct costs and indirect costs. Direct costs are labor cost, financial cost, and time cost; while indirect costs are possible negative impacts from internal information disclosure. Therefore, government need to manage this situation by enacting disclosure regulation to force companies to disclose more information, increase transparency, and decrease information asymmetry.
Information asymmetry also exists in China. Investors always can acquire limited information, most of which is historical and cannot provide any effective help in decision-making. As such, disclosure of forecasting information by organization has important influence on investors’ decision-making, especially disclosure of reliable
earnings forecasts. In addition, Securities Regulatory Department pushes companies to disclose more reliable forecasts information. Chinese earnings forecasts regulation developed a lot and experienced different stages. Now it is more improved and mature.
2.3. Signaling theory
In the environment with information asymmetry, companies need to express information related to companies to public through some channels to influence investors’ decision-making. Meanwhile investors need to choose good investment based on information from companies. Signaling is generated in such situation when companies exchange information with external investors to decrease information asymmetry and maximize interests.
Signaling theory is developed by Spence (1973), that education as signaling device to convey high quality employees to employers. It focuses on disclosing true and valuable information to present organizational operating and financial state for investors to evaluate and make decisions in the case of information asymmetry. Since managers have more private information, investors require higher return of capital to avoid loss when trading at informational disadvantage relative to other traders (Bhattacharya & Spiegel, 1991). Signaling theory advocates that managers will lower future estimate uncertainty of organization through signaling to decrease investors’ estimated risks, hence required return of investment, namely costs of capital (Scott, 2006). In addition, information asymmetry leads to adverse-selection and Gresham's Law (bad money drives out good). Therefore, excellent performance companies will have strong incentives to convey good signaling through information disclosure. It explains why some organizations disclose information voluntarily without mandating information disclosure regulation, that is, organizations have incentive of voluntary disclosure. Some organizations consider voluntary disclosure as a signal of well-operating, which facilitates investors to identify them. Organizations that build good reputations on disclosure forecasts will improve financial ability because large
and reliable information can lower information uncertainty and investment risks, reducing return on investment of investors, hence costs of capital of organizations. Chinese Securities Regulatory Department demonstrates the “minimum threshold” for mandatory disclosure of listed companies. When it comes to earnings forecasts, reliable information can reduce costs of capital and increase value.
2.4. Agency theory
Jensen & Meckling (1976) first proposed agency theory that the separation of ownership and control leads to agency conflict and information asymmetry between owners and managers of organizations. Managers have incentives to make decisions that may is harmful to shareholders’ interests based on hypothesis of economic man. Since shareholders and managers form a principal-agent relationship through contracts, and shareholders do not participate in management and operation of companies directly, the manager actually controls the company’ s asset operations. Shareholders focus on value maximization, while managers focus on awards maximization, which lead to conflict when realizing their interests. In addition, since managers manage daily operating, investing, and financing activities, they have more information than shareholder do. Interests conflict and information asymmetry may induce managers to behave inconsistent with shareholder’ interests (e.g. adverse selection and moral hazard). Therefore, managers need incentive mechanism to be aligned with shareholders’ interest. Obviously, implementing such mechanism will generate agency costs, which lower managers’ compensation. Managers are willing to disclose information to reduce agency costs.
In addition, there are also agency costs between large shareholders and small shareholders, shareholders and creditors, which can make small shareholders and creditors suffer loss (e.g. debt overhang and under-investment problem). Therefore, small shareholders and creditors also will take some measures to protect themselves, such as signing strict debt contract, which will push managers or companies to disclose information to lower agency costs.
3. Literature review
3.1. Literature review on earnings forecasts
The paper review prior literature about earnings forecasts in three aspects: disclosure incentives of earnings forecasts, managers’ disclosure strategies of earnings forecasts, and disclosure consequences of earnings forecasts. Following is the framework of three aspects:
3.1.1 Literature review on disclosure incentives of earnings forecasts
Based on signaling theory, economic investors consider that companies with good news will disclose information proactively, leading to value increase, while companies that do not disclose information may hide bad news, leading to value decrease. Therefore, all the companies will proactively disclose relevant information (Grossman, 1981). But companies have not disclose information completely yet in reality, indicating some other factors impact information disclosure.
1. Internal factors
A. Financial position of companies. Financial position of companies including size, financial leverage, and profitability. Most literature of disclosure of earnings forecasts is based on voluntary disclosure of American management. Earlier researches tested whether there were differences between forecasting companies and non-forecasting companies (Ruland, Tung, & George, 1990). They find that forecasting companies are larger and less earnings volatility than non-forecasting, and that forecasting companies are more likely to issue stocks than non-forecasting companies.
Disclosure incentives Disclosure strategies Disclosure consequences Accuracy Precision Timeliness Nature of news Internal factors External factors Market reaction Cost of capital Earning management Analysts prediction
Many researches find size is related to voluntary disclosure with different conclusions. Agency theory intends that agency costs are higher with larger companies. Thus larger companies will have stronger incentives to disclose information, such as earnings forecasts, to reduce agency costs (Lang & Lundholm, 1993). In addition, larger companies have fewer fixed disclosure costs, resulting in high possibility and quality of disclosing earnings forecasts than smaller companies (Botosan, 1997). In the contrary, some researchers believe that agency problem is more serious in larger companies, making managers reduce frequency of information disclosure for their own interests.
Profitability is another important factor that impacts disclosure incentives. Companies with more stable operation and less profit volatility will be more likely to disclose earnings forecasts (Lang & Lundholm, 1993). Rogers (2005) find that companies are tend to disclose bad news with bad financial position, high competition and high litigation risk. Malone, Fries, and Jones (1993) show that financial leverage is positively related to information disclosure.
B. Governance structure of companies. Companies with higher level of governance are more likely to disclose earnings forecasts (Ajinkya, Bhojraj, & Sengupta, 2005). In addition, earnings forecasts is related to organizational performance. Miller (2002) find that companies are more likely to issue forecasts when improving performance, and less likely to issue forecasts when performance declines.
Ownership concentration. Although most researches advocate that ownership concentration is essential to information disclosure, findings are not consistent. Supporters hold that increasing shareholding ratio of large shareholders can reduce agency problems and improve the level of information disclosure (Kaplan & Minton, 1994). Shleifer & Vishny (1997) also think that excessive dispersion of ownership is not conducive to shareholders’ supervision of management. Ownership concentration can effectively suppress the self-interest of management and reduce agency costs. Contrary to the point above, Schadewitz & Blevins (1998) find that there is a negative relationship between the degree of concentration of institutional owners and the
quality of disclosure.
Board of directors. Board size affects effectiveness of company governance. On the one hand, expansion of board of directors can reduce CEO’ s control over the board of directors, which facilitates board to play the role of oversight management. Thus managers can be pushed to disclose more information. On the other hand, board size affects efficiency. Board expansion may lower efficiency of board (Jensen, 1993), so that board of directors cannot influence management.
In addition, improving independence of the board can increase efficiency of the board (Hermalin & Weisbach, 1991), and significantly affect disclosure of earnings forecasts of management. Fama & Jensen (1983) consider that the greater the proportion of independent directors, the higher efficiency of the board and higher possibility of voluntary information disclosure. Klein (2002) states that the greater the proportion of independent directors, the higher quality of financial information disclosure.
2. External factors
A. Media reports. Media plays two roles in capital market: information interpretation and corporate governance. On the one hand, media takes its advantage of information collection that collect, process, and disclose information of listed companies, to reduce information asymmetry and enhance investors’ discernment (Merton, 1987). On the other hand, although there are huge differences in institutional background and market development among different countries, media reports play important roles in corporate governance by improving corporate governance environment (Dyck & Zingales, 2002).
However, some researchers find that media reports can also have negative impact on information disclosure. Chen, Pantzalis & Park (2009) demonstrate that abnormal press coverage on listed companies will significantly exaggerate mispricing, that is, abnormal press coverage will generate strong emotional and overreaction among investors, which may result in the company’ s stock being mispriced.
B. Analysts tracking. Analysts acquire information through public and private ways, evaluate performance of tracking companies, and forecast the future to recommend to investors (Healy & Palepu, 2001). Analysts tracking can effectively suppress agency conflicts (Moyer, Chatfield & Sisneros, 1989), reduce opportunistic behavior of management, and improve organization value (Chung & Jo, 1996). Moreover, analysts tracking can reduce information asymmetry in capital market, reduce adverse selection of investors, and reduce cost of capital of listed companies (Bowen, Chen & Cheng, 2008). In addition, financing and capital structure can also be influenced by analysts. Since analysts can strength in collecting information, they have high possibility finding financial fraud (Dycket, 2008). Yu (2008) find that analysts tracking can impact decision-making on earnings management of organization directly.
3.1.2 Literature review on disclosure strategies of earnings forecasts
1. Accuracy. When management decide to disclose earnings forecasts, they will try to improve forecasting accuracy. Findings show that there are large differences in accuracy of earnings forecasts. Hassell & Jennings (1986) report that quarterly forecast is more accurate than annual forecast. Less experienced managers have less earnings forecasts (Chen, 2004). Over-confident managers are more likely to disclose optimistic error earnings forecasts (Hribar & Yang, 2006), and companies with better corporate governance are able to provide more accurate earnings forecasts (Karamanou & Vafeas, 2005).
2. Precision. There are two ways to disclose earnings forecasts: qualitative and quantitative ways. Quantitative way includes point forecasts and range forecasts with minimum and maximum number. Stephen, Edward, & John (1993) demonstrate that less than 20% companies applied point or range forecasts from 1983 to 1986. Recent research find that around 50% companies apply point or range forecasts (Stephen, John, & Michael, 2004). Most research focus on point forecasts and range forecasts is because it is better to present accuracy and error of earnings forecasts (Rogers & Stocken, 2005). Research find that management’ s over-confidence (Hribar & Yang,
2006), good corporate governance (Karamanou & Vafeas, 2005), and analysts tracking (Stephen & John, 1997) are positively related to precision of forecasts, respectively. There is also a finding that analysts tracking can promote companies to disclose more precise forecasts than media reporting because analysts tracking is more direct and professional (Bamber & Cheon, 1998). Stephen & John (1997) find that size, volatility of earnings, and length of forecasts period are negatively related to presicion of earnings forecasts. For instance, they find that bigger companies have higher agency cost, and management is tend to disclose less precise forecasts.
3. Timeliness. Compared to annual forecast, quarterly forecast is more timely. Waymire (1985) find that companies with higher earnings volatility are more likely to disclose late earnings forecasts. Pownall, Wasley & Waymire (1993) show that the shorter period between earnings forecast day and earnings announcement day, the more accuracy and the higher quality of earnings forecasts.
4. Nature of news. Managers can influence forecasting news in short run. Normally, researchers use the prevailing median analyst forecast as proxy of market forecast. Soffer, Thiagarajan, & Walther (2000) find that managers employ different strategies when disclosing different news. When disclosing bad news, managers will release all of their bad news so that there will be no negative surprise later, while they will release only some of their good news, leaving a positive surprise for the earnings announcement.
3.1.3 Literature review on disclosure consequences of earnings forecasts
1. Earnings forecasts and market reaction. Hutton & Stocken (2007) test the impact of earnings forecasts of companies with different reputation on investors’ response. They find that investors have stronger response to good news disclosed by companies with good reputation. In addition, investors have different response towards good news and bad news since bad news earnings forecasts are always informative, but good news forecasts are informative only when supplemented by verifiable forward-looking statements (Hutton, Miller, & Skinner, 2003).
2. Earnings forecasts and cost of capital (information asymmetry). Diamond & Verrecchia (1991) believe that voluntary disclosure can reduce information asymmetry. Decrease in information asymmetry can lead to decrease in cast of capital. Kim & Shi (2011) test the impact of earnings forecasts on cost of capital and find that bad news earnings forecasts significantly increase cost of capital, while good news earnings forecasts do not influence cost of capital.
3. Earnings forecasts and earning management. Kasznik (1999) demonstrate that managers who disclose annual earnings forecasts manage reported earnings toward their forecasts. Hayn (1995) find that if there is large difference between management earnings forecasts and actual EPS, management is likely to suffer loss from litigation risk. In order to avoid loss, managers have incentives to implement earnings management.
4. Earnings forecasts and analyst prediction. Management and analysts share predictive information with public. Research find that analysts will update their forecasts as a respond to management forecasts (Jennings, 1987). Cotter, Tuna, & Wysocki (2006) find that analysts quickly react to management earnings forecasts and are more likely to issue final meetable or beatable earnings targets when management provides public guidance. It suggests that management earnings forecasts play important roles in leading analysts toward achievable earnings targets.
3.2. Literature review on managers’ background characteristics
With the development of organizational behavior and psychology, especially application of upper echelons theory, many researchers try to find impact of managers’ background on organizational decision and outcomes. Research of managers includes individual manager and top management team, while research of background characteristics includes age, gender, education level, and tenure. Following is the framework of analysis:
3.2.1 Research object
1. Individual manager. Barker & Mueller (2002) study relationship between CEO personal characteristics and R&D expense and find that the younger the CEO with marketing background, the higher the R&D expense. They also find that CEO’ s education level is significantly related to R&D expense. Malmendier & Tate (2008) illustrate CEO’ s decision is related to his overconfidence, optimism, and risk preference. Vintil & Gherghina (2012) find that CEO’ s age is negatively related to price earnings ratio, CEO’ s tenure is positively related to return on assets and price earnings ratio.
2. Top management team. It is impossible that complex organization and responsibility is controlled by one person (Drucker, 1974). Hambrick & Mason (1984) suggest that relationship between top management team and organizational strategy should be tested. Therefore, research objects began to be transitioned from individual manager to top management team. Heterogeneity of managers’ background represents diversity of team cognitive, which means the management team can acquire information from various sources and have different point of view (Hambrick & Mason, 1984). Heterogeneity of management team is related to high level of creativity and innovation (Bantel & Jackson, 1989). This will promote the collision of different opinions among managers and help to form high-quality decisions (Sorenson, 1968). Individual manager Top management team Gender Age Education level Tenure Background characteristics
3.2.2 Background characteristics
Observable managerial characteristics are indicators of managers’ conception and cognitive (Hambrick & Mason, 1984). Based on demographic indicators research, there is a linkage between background characteristics and givens and behaviors. 1. Age. Managers in different ages have great differences in making decision. The reasons include different decision-making related capabilities and different attitudes towards risk. Taylor (1975) find that younger managers have stronger creativity and adaptability than older managers. As for information disclosure, younger managers are more adaptable to the changes in disclosure system and respond accordingly. They are better at integrating information and forecasting future performance. Moreover, the older the mangers, the more likely they will avoid risk decisions (Carlson & Karlsson, 1970).
2. Gender. Male managers and female managers have differences in ability of process information, attitudes towards to risks, confidence. Graham, Stendardi, Jr, Myers, & Graham (2002) state that female managers often deal with information synthetically, while male managers are selective in processing information. Martin, Nishikawa, & Williams (2009) prove that the market perceives female CEOs to be relatively risk averse, and that firms with high risk are more likely to appoint female CEOs so that risk might decrease. Peng & Wei (2007) find that male managers are more sensitive to cash flows because they are more confident than female managers.
3. Education level. High level of education is related to high ability to process information (Schroder, Driver, & Streufert, 1967). It is also related to acceptability for innovation (Becker, 1970).
4. Tenure. Katz (1982) find that when managers work together for a long time, they are tend to form standard communication and consistent ideas. But if the tenure is much longer, their communication level will decrease since they feel they have known each other very much. Finkelstein (1992) further find that longer tenure leads to more conservative management strategy.
4. Hypothesis development
Based on prior literature and research, I select four managers’ background characteristics, including gender, age, education level, and other working experience, to find their impacts on attributes of earnings forecasts (accuracy and news) of listed companies. The hypothesis are developed as following.
4.1. Gender and earnings forecasts.
Gender is an important indicator to distinguish different social groups. Different gender differs physically and psychologically and has huge impact on behavior of human. As for leadership of organization, gender difference influences directly managers’ leadership and management (Li, 2012). Research finds that female is less sensitive to probability change than male, and underestimates probability of yield. Female also tend to be negative facing uncertain outcome. These findings can be interpreted as female is less confident than male (Beckmann & Menkhoff, 2008). Compared to male, female has higher anxiety or worry towards uncertain environment (Fehr-duda, Degennaro, & Schubert, 2006). Peng and Wei (2007) find that male managers are more confident and take more risks compared to female managers. Especially when in competitive environment, female managers are more careful and prudent (Niederle & Vesterlund, 2007). Earnings forecasts estimate the near-future profitability of firms and impacts the decision-making of investors, leading to fluctuation of firm share price. At the same time, managers have pressure of making accurate forecasting in case of litigation risk and criticism of analysts. For more confidence of male managers, I suppose they are more likely to issue good news of earnings. As such, the following hypothesis are offered:
H1a: Male managers are more likely to disclose good news of earnings forecasts. H1b: Female managers are more likely to disclose higher accuracy earnings forecasts.
4.2. Age and earnings forecasts.
A few researchers have found that managerial youth appears to be associated with corporate growth, though there are not many studies on relationship between age and organizational situation. Age can show possible tendency of risks taken to some extent, affecting decision-making. Hambriek & Mason (1984) find that older managers are more conservative for three reasons. First because they tend to spend more time to find more information, and make decisions after analysis. While younger managers have greater ability to integrate information in making decision and have deep confidence (Taylor, 1975). Second because they have greater psychological commitment to the organizational status (Alutto & Hrebiniak, 1975; Stevens, Beyer, & Trice, 1978). Finally because they attach importance to financial security and career security in their ages, so they tend to avoid risk. Wiersema & Bantel (1992) indicate that managers’ recognition ability will be lower following the increased age, and the knowledge structure can be outdated. Therefore, aged managers do not have as strong confidence as young managers when making decision. So younger managers are more likely to take risk measures. Since disclosure of earnings forecasts will take some risks, older managers will less likely to forecast earnings and issue bad news to reduce risks of earnings forecasts. However, they probably have more experience to anticipate the earnings and will provide more accurate earnings forecasts. As such, the following hypotheses are offered:
H2a: The younger the managers are, the more likely they will disclose good news of earning forecasts.
H2b: The younger the managers are, the lower accuracy the disclosure is. 4.3. Education level and earnings forecasts.
Education shows to some extent someone’ s cognitive preferences, values. Education can also be related to innovation and some findings conclude that level of education is positively related to receptivity to innovation (Becker, 1970). Compared to managers with low level of education, managers with high education level are more likely to
acquire new idea (Barker & Mueller, 2002), and to learn new behavior and innovative method to solve complex problems (Bantel & Jackson, 1989). Managers with higher education level are more likely to be willing to accept new idea and have ability to adapt to new environment. They also have stronger ability to generate innovative method to solve problem, so they are easier to accept innovation (Barker & Mueller, 2002). Disclosure of earnings forecasts requires managers to be able to respond to consequence, such as fluctuation of share price and emotion of investors, thus managers with high education level will have better performance and tend to issue good news. In addition, accuracy of earnings forecasts requires high forecasting ability of managers (Tan, Libby, & Hunton, 2002), and requires strong ability of analysis of managers. As such, the following hypothesis are offered:
H3a: The higher the education level, the managers disclose more accurate earnings forecasts.
H3b: The higher the education level, the more likely the managers will disclose good news of earning forecasts.
4.4. Other career experiences and earnings forecasts
Managers’ knowledge, value and behavior will be influenced by the experiences they have had in other industry, organization or even department. Executives have limited perspective and cognition when they spend all career in one organization. Therefore, when the organization perform poorly, there will be outside succession to make some changes to the organization (Hambrick & Mason, 1984). Jensen & Zajac (2004) prove that individuals may choose functional areas that are suitable for their cognitive models and values, but also over time become socialized and inculcated with the area’ s dominant mode of thinking and acting. Especially the managers who have background experiences in finance are more likely to attach importance to growth through diversification and acquisitions. As a matter of fact, managers with financial experiences take more concentrate on details of data, and more sensitive to financial information, which shows that they are more likely to issue concrete earnings
forecasting. As such, following hypothesis is offered:
H4: Managers with financial experiences are more likely to disclose more accurate earnings forecasting.
5. Research design
My full sample of firms is identified using CSMAR database (to identify managers’ background characteristics and attributes of earnings forecasts). It is noticed that New Accounting Standards were implemented in 2007 in China, which impact financial decision for companies. To keep consistency of sample and accounting policy in the research period, and reduce the volatility after implementation, I collect earnings forecasts of 532 Chinese A-shared listed companies from 2013 to 2016 as research sample since there are large quantities of data of listed companies and managers cannot be found from 2008 to 2012 due to incomplete database, and data of 2017 has not been disclosed in database.
To test the impact of managers on disclosure of earnings forecasts, I track three top managers with highest compensation of each company to form a top management team, including board of directors, CEO, and CFO. And then I compute the average number of independent variable of each top management to represent managers’ characteristic background of each company. After taking out the incomplete data from companies and years, I find 1,811 observations from each variable.
5.1. Research model
To examine whether managers’ background characteristics play important roles in disclosure of earnings forecasts, I apply the OLS method based on research of Bamber & Cheon (1998). The empirical models to examine the relationships are as follows:
Acc=β0+ β Βackground characteristics+β1CONTROLS+ϵ Model 1 News=β0+β Βackground characteristics+β1CONTROLS+ϵ Model 2
5.2. Independent variables
Prior literature mainly focus on demographic characteristic when studying managers’ background characteristics, such as age (Schman & Scott, 1989), gender (Peng & Wei,
2007), education degree (Wiersema & Bantel, 1992), and career experience (Holland, 1997). These demographic indicators are much easier to study compared to psychological indicators. Generally speaking, psychological indicators of managers are hard to measure and acquire, and many managers do not accept psychological tests. In addition, mental activities and feelings are more subjective, which cannot be considered as reliable information. Therefore, I select gender, age, education level, and other career experience as managers’ background characteristics and select following proxies based on prior literature:
a. Gender ratio: Ratio of female managers to male managers in the top management team.
b. Average age: average age of top management team.
c. Average education level: divided into four levels: below bachelor=0, bachelor=1, master=2, doctor of philosophy=3.
d. Other career experience: whether managers have prior financial experience. 5.3. Dependent variables
I measure two attributes variables of earnings forecasts: news and accuracy according to Linda, John, & Isabel (2010). I measure news (News) as the average news conveyed by forecasts firm issued at the end of year, which could be positive, negative or uncertain. I measure forecast accuracy (Acc) as the absolute value of the difference between the management EPS forecast and actual EPS.
Exhibit 5.1 independent and dependent variables and definitions Independent
variables Definitions
gendrat Ratio of female managers to male managers in the top management group, from 0 to 1.
aveage Average age of managers in top management team below 30=0; between 30 and 40=1; between 40 and 50=2; between 50 and 60=3; and above 60=4.
aveedu Average education of top management team below high school=1, high school=2, bachelor=3, master=4, doctor of philosophy=5, and others=6
aveFinback coded as 1 if top management team has prior accounting or financial-related working experience; and 0 otherwise.
Dependent
variables Definitions
news average news conveyed by forecasts firm issued at the end of year. coded as 1 if forecasting Goodnews (increased EPS); coded as 0 if forecasting Badnews (decreased EPS or uncertain)
acc absolute value of the difference between the management forecast and actual EPS at the end of year.
5.4. Control variables
Apart from managers’ background characteristics, firm characteristics and governance can affect earnings forecasts. To control those impact of other possible factors, and reflect only the impact of managers’ background characteristics, I decide to control three groups of control variables according to prior literature:
1. Other managerial background characteristics variables. Apart from those variables I mention above, there are other factors of managerial background characteristics can have influence on financial decision-making. (a). whether managers have shares of company. (b). number of other companies managers work in at the same time. (c). whether managers have specific profession titles.
2. Organizational characteristics variables. (a). Industry characteristics: Stang & Belkao, Kahl & Cooke find that industry characteristics are significantly related to quality of information disclosure. If the company belongs to high-risk industries of biochemical pharmaceutical technology, computer, and electronics, takes the value of 1; otherwise 0.
3. Financial position variables. (a). B/M (book value to market value) can be used to control impact of corporate growth on information quality (Richard, Marilyn, & Karen, 2002). (b). Increase: if company’ current EPS is more than EPS of last year,
takes a value of 1; otherwise 0. (c). financial leverage (DEL): Scott (2006) find that managers can lower uncertainty of investors through signaling, and lower the required return of investment correspondingly. As such, managers have incentives to disclose information with high financial leverage. (d). P/E (price earnings ratio): Companies are willing to disclose information when they have high profitability because it can express good news to public and promote stock price. Delin (2004) indicate that profitability is positively related to earnings forecasts.
Exhibit 5.2 control variables and definitions Control
variables Definitions
SH Average shares managers have in top management team=1; and 0 otherwise.
Concur Number of companies where managers work at the same time..
Profession Coded as 1 if managers have profession titles; and 0 otherwise. Calculate average profession titles top management teams have.
Litig coded as 1 if the firm is a member of one of the following high-litigation-risk industries: biotechnology, computers, electronics, retailing, and R&D service, and suffers a 20 percent or greater decrease in earnings; and 0 otherwise.
P/E Price earnings ratio of companies in year.
B/M market value of firm’ s common equity divided by the book value of its common equity, at the end of year.
|ΔEPSit| absolute value of the change in firm i’ s earnings per share from year t-1 to t.
6. Results
6.1. Descriptive statistics
Based on my research sample, I develop descriptive statistics analysis of managers’ background characteristics, attributes of earnings forecasts, and control variables to reflect the situations of managers and information disclosure of Chinese listed companies.
6.1.1 Descriptive statistics of managers’ background characteristics Table 6.1 managers’ background characteristics
Notes: The table provides summary statistics for the independent variables for firm-year observations from fiscal years 2013-2016. All numbers are rounded up to third decimal place. Variable definitions are shown in the Appendix.
Table 6.1 provides descriptive statistics of managers’ background characteristics. We can see that: The mean and median of gender ratio of female to male manager in a top management team is very low, nearly 0. It shows that higher position of managers in listed companies are mainly male. But there are also some companies with female top management team, which is related to industry.
Top management teams’ ages mainly locate between 40 and 50, which is a normal phenomenon in career development. Most of people become managers when they reach the middle age because they have more experience and know companies better. But there are huge differences in ages of managers. The youngest average age of top management team locates between 30 and 40, while the oldest average age of Descriptive average characteristics of the sample
Variable Obs Mean P50 Std.Dev. Min P25 P75 Max
gendrat 1,811 0.105 0.000 0.173 0.000 0.000 0.333 1.000
aveage 1,811 2.606 3.000 0.541 1.000 2.000 3.000 4.000
aveedu 1,811 3.328 3.333 0.603 1.000 3.000 3.667 5.000
management team is above 60.
Top management teams’ education levels are more bachelor or even higher, indicating that Chinese top managers of listed companies have high education and knowledge. The minimum average education of top management team is below high school, but the amount of these companies is small.
There are few managers having financial prior working experience, only 4.7%. During data collection, I find that more managers have science and engineering working experience. Maybe it means the people who have professional technology are more likely to become core human resources and become managers. In fact, this result is consistent with research of Hambrick & Daven (1992) that managers with finance and accounting background cannot provide competitiveness in the long run, while managers with research and design background can.
6.1.2 Descriptive statistics of attributes of earnings forecasts Table 6.2 Attributes of earnings forecasts Descriptive accuracy and news of earnings forecasts
Variable Obs Mean P50 Std.Dev. Min P25 P75 Max
acc 1.811 0.078 0.040 0.149 0.000 0.020 0.090 2.712
news 1,811 0.890 1.000 0.301 0.000 1.000 1.000 1.000
Notes: The table provides summary statistics for the dependent variables for firm-year observations from fiscal years 2013-2016. All numbers are rounded up to third decimal place. Variable definitions are shown in the Appendix.
From table 1, we can see that:
Managers have more accurate earnings forecasts in general. It is surprised to have such accurate forecasts. It may because these forecasts are adjusted at the end of year. Companies can change the estimated numbers when tracking the situation where companies stay in and market volatility. In addition, the volatility of accuracy of earnings forecasts is low when seeing standard deviation of accuracy.
Managers are more likely to disclose good news when disclosing information because there are 89% of news are good. Their forecasting EPS is higher than actual EPS,
which is consistent with what we assumed before. When the earnings are positive, the companies will be over confident. When the earnings are negative, the companies will be over pessimistic. This leads to higher forecasting EPS than actual EPS.
6.1.3 Descriptive statistics of control variables
Listed companies take important roles in security market, and their performance characteristics are closely related to earnings forecasts. As such, I analyze descriptive statistics of main control variables to find what kind of companies can follow Chinese earnings forecasts regulation well and reflect their characteristics.
Table 6.3 Control variables Descriptive control variables of the sample
Variable Obs Mean Median Std. Min P25 P75 Max
Litig 1,811 0.202 0.000 0.401 0.000 0.000 0.000 1.000 DEL 1,811 0.347 0.317 0.212 0.000 0.194 0.483 3.985 P/E 1,731 79.416 52.180 103.784 -163.971 30.393 94.760 545.536 B/M 1,731 0.583 0.427 0.606 0.024 0.272 0.663 8.109 |ΔEPSit| 1,811 0.431 0.000 0.495 0.000 0.000 0.000 1.000 avecocur 1,811 1.647 0.667 2.202 0.000 0.000 2.333 17.333 SH 1,811 0.634 0.667 0.387 0.000 0.333 1.000 1.000 Profession 1,811 0.569 0.667 .365 0.000 0.333 1.000 1.000
Notes: The table provides summary statistics for the control variables for firm-year observations from fiscal years 2013-2016. All numbers are rounded up to third decimal place. Variable definitions are shown in the Appendix.
From table 1, we can see that:
Most listed companies that disclose earnings forecasts are not high-litigation-risk industries.
Average debt asset ratio (DEL) of listed companies is 34.7%, meaning the amount of listed companies with debt financing decreases. It is related to our tight monetary policy recent years, and it is hard for companies to acquire money from banks. It also
indicates that companies with lower risks are more likely to disclose earnings forecasts since high debt asset ratio means high risk, and high volatility of financial performance, which leads to high difficulty of forecasting.
The profitability of companies is good in general, but different companies performance very differently.
There are huge differences in number of companies that managers have managerial position. But most of managers do not work in other companies at the same time. 63% of managers hold shares in the companies, indicating that stock is becoming an important incentive method for managers in companies.
There are over half of managers having specific profession, indicating that profession title is popular in career.
6.2. Correlation result and analysis
Before regression analysis, I firstly apply Pearson correlation analysis to see correlation between managers’ background characteristics and control variables and attributes of earnings forecasts. Following table shows the correlation coefficient.
From table 6.4 we can see that there is no relationship between gender and accuracy and news of earnings forecasts. Manager’ s age is significantly positively related to accuracy of earnings forecasts, and financial background is significantly positively related to accuracy of earnings forecasts. When it comes to control variables, the coefficient of DEL and B/M and accuracy are positive and significant at the 1% level, while The coefficient of |ΔEPSit| and accuracy is negative (-0.108) and significant at the 1% level. In addition, the coefficient of shares and accuracy is also negative (-0.051) and significant at the 5% level.
We can also see that gender has no relationship with news of earnings forecasts. The coefficient of age and news is negative (-0.137) and significant at the 1% level. Average education level of top management team is positively related to news of earnings forecasts at the 5% level. The coefficient of |ΔEPSit| and news is negative (-0.108) and significant at the 1% level.
The correlation results show managers’ background characteristics may have impacts on attributes of earnings forecasts, but we need further test and analysis to have more accurate conclusion.
6.3. Regression results and analysis
Based on the variables and model of research design, I apply regression of managers’ background characteristics and attributes of earnings forecasts of listed companies to discuss the specific effects. To test the impact of each independent variable on attributes of earnings forecasts, I firstly put each independent variable into the model to test separately, and then put all independent variables into the model. As such, following tables are the results of regression.
Table 6.5 managers’ gender and earnings forecasts regression
Model 1 Model 2
acc news
gendrat -0.005 0.058
(-0.229) (1.363)
(-0.283) (0.620) avecocur 0.003 0.004 (1.595) (1.142) SH -0.014 -0.003 (-1.511) (-0.162) |ΔEPSit| -0.032*** -0.057*** (-4.318) (-3.855) DEL 0.040** 0.031 (2.098) (0.827) P/E -0.000 -0.000 (-1.180) (-1.164) B/M 0.006 -0.025* (0.813) (-1.826) Litig 0.012 0.013 (1.255) (0.735) _cons 0.084*** 0.913*** (6.366) (34.698) N 1725 1725 Adj.R-Square 0.016 0.037
Notes: This table examines gender and earnings forecasts attributes and reports the results of the relevant OLS regressions. The dependent variable in the first column is accuracy of earnings forecasts. The dependent variable in the second column is news of earnings forecasts In the second column. The sample consists of 1725 firm-year observations from fiscal years 2013–2016. All numbers are rounded up to third decimal place.
Variable definitions are shown in the Appendix. ∗∗∗ indicate significance at the 1% level. ∗∗ indicate significance at the 5% level. ∗ indicate significance at the 10% level.
From table 6.5, we can find that the first hypothesis H1a cannot be proved. Gender ratio cannot show the difference between male managers and female managers regarding to attributes of earnings forecasts. It mainly because the ratio of female managers to male managers is too low, meaning most of top managers in Chinese listed companies are male. Therefore, we cannot see whether gender can impact earnings forecasts now. In addition, H1b also cannot be proved due to the same reason. As such, Chinese companies should provide more opportunities for female to balance the gender ratio since female managers can also have advantages due to their own characteristics, as Niederle & Vesterlund (2007) mentioned that female managers are more careful than male managers.
Table 6.6 managers’ age and earnings forecasts regression Model 1 Model 2 acc news aveage 0.067*** -0.056*** (12.532) (-5.065) Profession -0.002 0.012 (-0.180) (0.606) avecocur 0.003 0.004 (1.599) (1.150) SH -0.015* -0.005 (-1.670) (-0.252) |ΔEPSit| -0.028*** -0.053*** (-3.946) (-3.618) DEL 0.038** 0.030 (2.099) (0.788) P/E -0.000 -0.000 (-0.631) (-0.880) B/M 0.009 -0.023* (1.304) (-1.725) Litig 0.010 0.013 (1.118) (0.719) _cons 0.253*** 1.063*** (13.807) (27.898) N 1725 1725 Adj.R-Square 0.099 0.021
Notes: This table examines age and earnings forecasts attributes and reports the results of the relevant OLS regressions. The dependent variable in the first column is accuracy of earnings forecasts. The dependent variable in the second column is news of earnings forecasts In the second column. The sample consists of 1725 firm-year observations from fiscal years 2013–2016. All numbers are rounded up to third decimal place.
Variable definitions are shown in the Appendix. ∗∗∗ indicate significance at the 1% level. ∗∗ indicate significance at the 5% level. ∗ indicate significance at the 10% level.
From table 6.6, we can see that H2a can be proved that managers’ ages are significantly negatively related to news at the 1% level (p<0.056). It means when making earnings forecasts, older managers are more conservative and more pessimistic. They are more likely to disclose bad news to avoid responsibility. The result is consistent with research of Bertrand & Schoar (2002) that older managers are more conservative in financial aspect.
accuracy of earnings forecasts at the 1% level (p<0.067). The older managers disclose more accurate earnings forecasts because they have more experience in reasonably estimating earnings of companies.
Table 6.7 managers’ average education and earnings forecasts regression
Model 1 Model 2 acc news aveedu 0.010* 0.019 (1.729) (1.640) Profession -0.003 0.011 (-0.276) (0.562) avecocur 0.003* 0.004 (1.684) (1.077) SH -0.015 -0.002 (-1.589) (-0.129) |ΔEPSit| -0.033*** -0.055*** (-4.405) (-3.745) DEL 0.041** 0.029 (2.156) (0.766) P/E -0.000 -0.000 (-1.236) (-1.075) B/M 0.006 -0.026* (0.839) (-1.905) Litig 0.013 0.012 (1.368) (0.669) _cons 0.116*** 0.858*** (5.041) (18.588) N 1725 1725 Adj.R-Square 0.034 0.037
Notes: This table examines education and earnings forecasts attributes and reports the results of the relevant OLS regressions. The dependent variable in the first column is accuracy of earnings forecasts. The dependent variable in the second column is news of earnings forecasts In the second column. The sample consists of 1725 firm-year observations from fiscal years 2013–2016. All numbers are rounded up to third decimal place.
Variable definitions are shown in the Appendix. ∗∗∗ indicate significance at the 1% level. ∗∗ indicate significance at the 5% level. ∗ indicate significance at the 10% level.
From table 6.7, we can easily understand that education of top management team is positively related to accuracy of earnings forecasts, which proves H3a. However, this relation is only significant at 10% level. It is because the higher education level of managers, the stronger ability of data process managers (Wiersema & Bantel, 1992),
and managers with higher education level are more likely to develop reasonable analysis and decision-making, inhibiting over-investment behavior of investors (Fuxiu, 2009). Managers with higher education level can have better knowledge of companies’ operating situation and financial performance, leading to more accurate estimating of earnings.
In addition, the result cannot support H3b that managers with higher education level are more likely to disclose good news of earnings forecasts. It does not show a significant relationship between education and news of earnings forecasts.
Table 6.8 managers’ other working experience and earnings forecasts regression
Model 1 Model 2 acc news aveFinback 0.086*** 0.025 (3.165) (0.462) Profession -0.003 0.011 (-0.279) (0.558) avecocur 0.002 0.004 (0.934) (1.052) SH -0.014 -0.004 (-1.496) (-0.209) |ΔEPSit| -0.033*** -0.057*** (-4.423) (-3.834) DEL 0.040** 0.031 (2.099) (0.819) P/E -0.000 -0.000 (-1.138) (-1.111) B/M 0.006 -0.026* (0.896) (-1.880) Litig 0.012 0.015 (1.346) (0.802) _cons 0.081*** 0.920*** (6.256) (35.559) N 1725 1725 Adj.R-Square 0.036 0.036
Notes: This table examines other working experience and earnings forecasts attributes and reports the results of the relevant OLS regressions. The dependent variable in the first column is accuracy of earnings forecasts. The dependent variable in the second column is news of earnings forecasts In the second column. The sample consists of 1725 firm-year observations from fiscal years 2013–2016. All numbers are rounded up to third decimal place. Variable definitions are shown in the Appendix.
∗∗ indicate significance at the 5% level. ∗ indicate significance at the 10% level.
From table 6.8 we can find that the result is accordance with H4 that financial working experience is significantly positively related to accuracy of earnings forecasts. Managers with financial background can disclose more accurate earnings forecasts because they cannot accept vague information disclosure (Holland, 1997). There is no relationship between financial working experience and news of earnings forecasts.
Table 6.9 managers’ background characteristics and earnings forecasts regression
Variables Model 1 Model 2
acc news gendrat -0.015 0.060 (0.020) (0.042) aveage 0.067*** -0.055*** (0.006) (0.011) aveedu 0.014*** 0.018 (0.006) (0.012) aveFinback 0.074*** 0.010 (0.026) (0.055) Profession -0.002 -0.013 (0.010) (0.020) avecocur -0.002 0.003 (0.002) (0.003) SH -0.017* 0.002 (0.009) (0.019) |ΔEPSit| -0.029*** -0.053*** (0.007) (0.015) DEL 0.039** 0.028 (0.016) (0.038) P/E -0.000 -0.000 (0.000) (0.000) B/M -0.009 -0.023* (0.007) (0.014) Litig 0.012 0.010 (0.009) (0.018) Constant -0.298*** 0.993*** (0.027) (0.056) Observations 1,725 1,725 R-squared 0.104 0.022
the results of the relevant OLS regressions. In the first column, the sample consists of 1725 firm-year observations from fiscal years 2013–2016, where the dependent variable is accuracy of earnings forecasts. In the second column, the sample consists of 1752 firm-year observations from fiscal years 2013–2016, where the dependent variable is news of earnings forecasts. The models estimated are as follows: News=β0+βBC+β1CONTROLS+ϵ, Acc=β0+ βΒC+β1CONTROLS+ϵ, where BC is each background characteristic. Robust standard errors clustered by firm are displayed in parentheses. All numbers are rounded up to third decimal place.
Variable definitions are shown in the Appendix. ***indicate significance at the 1% level. **indicate significance at the 5% level. *indicate significance at the 10% level.
We can see from table 6.9 that the results in the regression of all variables are in accordance with results in separate regression. The accuracy of earnings forecasts are influenced by average age, average education level, and average financial background of top management teams: The older managers disclose more accurate earnings forecasts; the managers with higher education level disclose more accurate earnings forecasts; the managers with financial background disclose more accurate earnings forecasts. The news of earnings forecasts are influenced by average age of top management team: the older managers are more likely to disclose bad news. Gender has little impact on attributes of earnings forecasts because most managers are male. When it comes to control variables, we can see the most obvious result is that |ΔEPSit| is significantly negatively related to accuracy and news of earnings forecasts whenever in regression. |ΔEPSit| means absolute value of the change in firm’ s earnings per share from year t-1 to t. So when there is small change between EPS in the two years, the companies will have more accurate earnings forecasts and good news of earnings forecasts. In addition, the amount of shares managers have is also negatively related to accuracy of earnings forecasts at 10% level. B/M is negatively related to news of earnings forecasts at 10% level. Financial leverage is significantly positively related to accuracy of earnings forecasts at 5% level, which indicates that companies with higher financial leverage need to disclose more accurate earnings forecasts to acquire trust of investors, and then reduce their uncertainty to lower cost of financing.
6.4. Robustness test
Test about accuracy of earnings forecasts.
In the prior tests, I use absolute value of the difference between the management forecast and actual EPS to represent accuracy. This time I use relative value, which means absolute value of the difference between the management forecast and actual EPS divided by actual EPS, to represent accuracy. The test result is consistent with the prior result.
Table 6.10 multiple regression
Variables Acc (new)
gendrat 0.684 (1.786) aveage 1.441** (0.574) aveedu 2.623*** (0.453) aveFinback 16.58*** (2.308) Profession 0.0254 (0.877) avecocur -0.141 (0.142) SH 0.357 (0.807) |ΔEPSit| -0.586 (0.623) DEL -0.0155 (1.603) P/E -3.83e-05 (0.000251) B/M -0.407 (0.577) Litig -1.226 (0.774) Constant -10.40*** (2.081) Observations 1,722 R-squared 0.387
7. Conclusion
7.1 Research conclusionDriven by conflicting views on whether managers have influence on corporate decisions (Weintraub 2002), and lack of research on manager factor in China, I examine the relation between managers’ background characteristics and earnings forecasts to infer manager-specific effects on the provision and formation of disclosure of Chinese listed companies from 2013 to 2016. Based on prior upper echelons theory (Hambrick & Mason, 1984), I estimate the effects of managers’ gender, and, education level, and other working experience on accuracy and news of earnings forecasts after controlling for firm- and industry-fixed effects.
The research results confirm that earnings forecasts of listed companies are not only influenced by firm- and industry-specific determinants, but also influences by managers’ background characteristics. Specifically:
1. Managers’ ages have significantly positive impact on accuracy, and have negative impact on news of earnings forecasts. In other words, the younger the manager, the more accurate earnings forecasts are disclosed. Older managers have experience and ability to disclose more accurate earnings forecasts, while they are also conservative and not positive at the same time so that they are more likely to disclose bad news. 2. Managers’ education level is significantly positively related to accuracy of earnings forecasts. In other words, the higher education level managers have, the more accurate earnings forecasts are disclosed. because they have more knowledge and ability to manage companies and deal with bad performance. However, there is no relationship between education level and news of earnings forecasts according to the results. 3. Managers’ other working experience, especially financial background, is significantly positively related to accuracy of earnings forecasts. It shows managers with financial working experience have better interpretation and forecasting ability than managers without financial background, so they can disclose more accurate earnings forecasts.
4. Managers’ gender has no impact on attributes of earnings forecasts due to the low ratio of female managers in top management team.
This paper stands from managers’ background characteristics point of view and studies their relationships with attributes of earnings forecasts. I mainly focus on impacts of gender, age, education level, and working experience, which expands the research area of information disclosure in China. I find that differences in managers’ background characteristics will have different impacts on earnings forecasts, which interprets in some degree reasons of information disclosure differences in Chinese security market.
7.2 Research limitation
There are some limitations in this paper, which can be a future research:
Research object. I only study information disclosure of listed companies based on earnings forecasts. In fact, earnings forecasts are important forward-looking information disclosure, but they cannot reflect all information disclosure of listed companies. There are management discussion and analysis in annual report that can be important forecasts of future performance of companies. In addition, I only consider ordinary forecasts without including updated forecasts. And this many influence the accuracy of earnings forecasts.
Data selecting. I only use data sample from 2013 to 2016. however, China started to have earnings forecasts-related regulation since 1998, and listed companies have experienced differences in disclosing. As such, it is better to research the whole time line of earnings forecasts. What’ s more, one manager may work in many companies at the same time in reality. As such, the best research method is to apply panel data to track managers’ disclosure style and earnings forecasts. In the research sample, managers should work as managers in at least two listed companies so that we can record whether their styles will appear in different companies. However, Chinese security market has a slow and short development compared to western security market, and regulation regarding information disclosure was issued in recent years,
leading to difficulty to acquire sample where managers work in at least two listed companies. In addition, we lack database that records detailed individual information of managers in listed companies. So I cannot use panel data, which is a limitation. Future research might explore the effects of managers changes on information disclosure of listed companies. Because forecasts in the period during managers changes may reflect managers’ self-interest motivation. In addition, future research may take various methods to analyze managers’ behavior and psychology, such as doing survey and interviews, to observe and understand managers’ decision-making process more directly.
